Show More
@@ -1,165 +1,164 b'' | |||
|
1 | 1 | # perf.py - performance test routines |
|
2 | 2 | '''helper extension to measure performance''' |
|
3 | 3 | |
|
4 | 4 | from mercurial import cmdutil, match, commands |
|
5 | 5 | import time, os, sys |
|
6 | 6 | |
|
7 | 7 | def timer(func, title=None): |
|
8 | 8 | results = [] |
|
9 | 9 | begin = time.time() |
|
10 | 10 | count = 0 |
|
11 | 11 | while 1: |
|
12 | 12 | ostart = os.times() |
|
13 | 13 | cstart = time.time() |
|
14 | 14 | r = func() |
|
15 | 15 | cstop = time.time() |
|
16 | 16 | ostop = os.times() |
|
17 | 17 | count += 1 |
|
18 | 18 | a, b = ostart, ostop |
|
19 | 19 | results.append((cstop - cstart, b[0] - a[0], b[1]-a[1])) |
|
20 | 20 | if cstop - begin > 3 and count >= 100: |
|
21 | 21 | break |
|
22 | 22 | if cstop - begin > 10 and count >= 3: |
|
23 | 23 | break |
|
24 | 24 | if title: |
|
25 | 25 | sys.stderr.write("! %s\n" % title) |
|
26 | 26 | if r: |
|
27 | 27 | sys.stderr.write("! result: %s\n" % r) |
|
28 | 28 | m = min(results) |
|
29 | 29 | sys.stderr.write("! wall %f comb %f user %f sys %f (best of %d)\n" |
|
30 | 30 | % (m[0], m[1] + m[2], m[1], m[2], count)) |
|
31 | 31 | |
|
32 | 32 | def perfwalk(ui, repo, *pats): |
|
33 | 33 | try: |
|
34 | 34 | m = cmdutil.match(repo, pats, {}) |
|
35 | 35 | timer(lambda: len(list(repo.dirstate.walk(m, [], True, False)))) |
|
36 | 36 | except: |
|
37 | 37 | try: |
|
38 | 38 | m = cmdutil.match(repo, pats, {}) |
|
39 | 39 | timer(lambda: len([b for a, b, c in repo.dirstate.statwalk([], m)])) |
|
40 | 40 | except: |
|
41 | 41 | timer(lambda: len(list(cmdutil.walk(repo, pats, {})))) |
|
42 | 42 | |
|
43 | 43 | def perfstatus(ui, repo, *pats): |
|
44 | 44 | #m = match.always(repo.root, repo.getcwd()) |
|
45 | 45 | #timer(lambda: sum(map(len, repo.dirstate.status(m, [], False, False, False)))) |
|
46 | 46 | timer(lambda: sum(map(len, repo.status()))) |
|
47 | 47 | |
|
48 | 48 | def perfheads(ui, repo): |
|
49 | 49 | timer(lambda: len(repo.changelog.heads())) |
|
50 | 50 | |
|
51 | 51 | def perftags(ui, repo): |
|
52 | 52 | import mercurial.changelog, mercurial.manifest |
|
53 | 53 | def t(): |
|
54 | 54 | repo.changelog = mercurial.changelog.changelog(repo.sopener) |
|
55 | 55 | repo.manifest = mercurial.manifest.manifest(repo.sopener) |
|
56 | 56 | repo._tags = None |
|
57 | 57 | return len(repo.tags()) |
|
58 | 58 | timer(t) |
|
59 | 59 | |
|
60 | 60 | def perfdirstate(ui, repo): |
|
61 | 61 | "a" in repo.dirstate |
|
62 | 62 | def d(): |
|
63 | 63 | repo.dirstate.invalidate() |
|
64 | 64 | "a" in repo.dirstate |
|
65 | 65 | timer(d) |
|
66 | 66 | |
|
67 | 67 | def perfdirstatedirs(ui, repo): |
|
68 | 68 | "a" in repo.dirstate |
|
69 | 69 | def d(): |
|
70 | 70 | "a" in repo.dirstate._dirs |
|
71 | 71 | del repo.dirstate._dirs |
|
72 | 72 | timer(d) |
|
73 | 73 | |
|
74 | 74 | def perfmanifest(ui, repo): |
|
75 | 75 | def d(): |
|
76 | 76 | t = repo.manifest.tip() |
|
77 | 77 | m = repo.manifest.read(t) |
|
78 | 78 | repo.manifest.mapcache = None |
|
79 | 79 | repo.manifest._cache = None |
|
80 | 80 | timer(d) |
|
81 | 81 | |
|
82 | 82 | def perfindex(ui, repo): |
|
83 | 83 | import mercurial.changelog |
|
84 | 84 | def d(): |
|
85 | 85 | t = repo.changelog.tip() |
|
86 | repo.changelog = mercurial.changelog.changelog(repo.sopener) | |
|
87 | repo.changelog._loadindexmap() | |
|
86 | repo.invalidate() | |
|
88 | 87 | timer(d) |
|
89 | 88 | |
|
90 | 89 | def perfstartup(ui, repo): |
|
91 | 90 | cmd = sys.argv[0] |
|
92 | 91 | def d(): |
|
93 | 92 | os.system("HGRCPATH= %s version -q > /dev/null" % cmd) |
|
94 | 93 | timer(d) |
|
95 | 94 | |
|
96 | 95 | def perfparents(ui, repo): |
|
97 | 96 | nl = [repo.changelog.node(i) for i in xrange(1000)] |
|
98 | 97 | def d(): |
|
99 | 98 | for n in nl: |
|
100 | 99 | repo.changelog.parents(n) |
|
101 | 100 | timer(d) |
|
102 | 101 | |
|
103 | 102 | def perflookup(ui, repo, rev): |
|
104 | 103 | timer(lambda: len(repo.lookup(rev))) |
|
105 | 104 | |
|
106 | 105 | def perflog(ui, repo, **opts): |
|
107 | 106 | ui.pushbuffer() |
|
108 | 107 | timer(lambda: commands.log(ui, repo, rev=[], date='', user='', |
|
109 | 108 | copies=opts.get('rename'))) |
|
110 | 109 | ui.popbuffer() |
|
111 | 110 | |
|
112 | 111 | def perftemplating(ui, repo): |
|
113 | 112 | ui.pushbuffer() |
|
114 | 113 | timer(lambda: commands.log(ui, repo, rev=[], date='', user='', |
|
115 | 114 | template='{date|shortdate} [{rev}:{node|short}]' |
|
116 | 115 | ' {author|person}: {desc|firstline}\n')) |
|
117 | 116 | ui.popbuffer() |
|
118 | 117 | |
|
119 | 118 | def perfdiffwd(ui, repo): |
|
120 | 119 | """Profile diff of working directory changes""" |
|
121 | 120 | options = { |
|
122 | 121 | 'w': 'ignore_all_space', |
|
123 | 122 | 'b': 'ignore_space_change', |
|
124 | 123 | 'B': 'ignore_blank_lines', |
|
125 | 124 | } |
|
126 | 125 | |
|
127 | 126 | for diffopt in ('', 'w', 'b', 'B', 'wB'): |
|
128 | 127 | opts = dict((options[c], '1') for c in diffopt) |
|
129 | 128 | def d(): |
|
130 | 129 | ui.pushbuffer() |
|
131 | 130 | commands.diff(ui, repo, **opts) |
|
132 | 131 | ui.popbuffer() |
|
133 | 132 | title = 'diffopts: %s' % (diffopt and ('-' + diffopt) or 'none') |
|
134 | 133 | timer(d, title) |
|
135 | 134 | |
|
136 | 135 | def perfrevlog(ui, repo, file_, **opts): |
|
137 | 136 | from mercurial import revlog |
|
138 | 137 | dist = opts['dist'] |
|
139 | 138 | def d(): |
|
140 | 139 | r = revlog.revlog(lambda fn: open(fn, 'rb'), file_) |
|
141 | 140 | for x in xrange(0, len(r), dist): |
|
142 | 141 | r.revision(r.node(x)) |
|
143 | 142 | |
|
144 | 143 | timer(d) |
|
145 | 144 | |
|
146 | 145 | cmdtable = { |
|
147 | 146 | 'perflookup': (perflookup, []), |
|
148 | 147 | 'perfparents': (perfparents, []), |
|
149 | 148 | 'perfstartup': (perfstartup, []), |
|
150 | 149 | 'perfstatus': (perfstatus, []), |
|
151 | 150 | 'perfwalk': (perfwalk, []), |
|
152 | 151 | 'perfmanifest': (perfmanifest, []), |
|
153 | 152 | 'perfindex': (perfindex, []), |
|
154 | 153 | 'perfheads': (perfheads, []), |
|
155 | 154 | 'perftags': (perftags, []), |
|
156 | 155 | 'perfdirstate': (perfdirstate, []), |
|
157 | 156 | 'perfdirstatedirs': (perfdirstate, []), |
|
158 | 157 | 'perflog': (perflog, |
|
159 | 158 | [('', 'rename', False, 'ask log to follow renames')]), |
|
160 | 159 | 'perftemplating': (perftemplating, []), |
|
161 | 160 | 'perfdiffwd': (perfdiffwd, []), |
|
162 | 161 | 'perfrevlog': (perfrevlog, |
|
163 | 162 | [('d', 'dist', 100, 'distance between the revisions')], |
|
164 | 163 | "[INDEXFILE]"), |
|
165 | 164 | } |
@@ -1,1938 +1,1935 b'' | |||
|
1 | 1 | # localrepo.py - read/write repository class 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 node import bin, hex, nullid, nullrev, short |
|
9 | 9 | from i18n import _ |
|
10 | 10 | import repo, changegroup, subrepo, discovery, pushkey |
|
11 | 11 | import changelog, dirstate, filelog, manifest, context |
|
12 | 12 | import lock, transaction, store, encoding |
|
13 | 13 | import util, extensions, hook, error |
|
14 | 14 | import match as matchmod |
|
15 | 15 | import merge as mergemod |
|
16 | 16 | import tags as tagsmod |
|
17 | 17 | import url as urlmod |
|
18 | 18 | from lock import release |
|
19 | 19 | import weakref, errno, os, time, inspect |
|
20 | 20 | propertycache = util.propertycache |
|
21 | 21 | |
|
22 | 22 | class localrepository(repo.repository): |
|
23 | 23 | capabilities = set(('lookup', 'changegroupsubset', 'branchmap', 'pushkey')) |
|
24 | 24 | supportedformats = set(('revlogv1', 'parentdelta')) |
|
25 | 25 | supported = supportedformats | set(('store', 'fncache', 'shared', |
|
26 | 26 | 'dotencode')) |
|
27 | 27 | |
|
28 | 28 | def __init__(self, baseui, path=None, create=0): |
|
29 | 29 | repo.repository.__init__(self) |
|
30 | 30 | self.root = os.path.realpath(util.expandpath(path)) |
|
31 | 31 | self.path = os.path.join(self.root, ".hg") |
|
32 | 32 | self.origroot = path |
|
33 | 33 | self.auditor = util.path_auditor(self.root, self._checknested) |
|
34 | 34 | self.opener = util.opener(self.path) |
|
35 | 35 | self.wopener = util.opener(self.root) |
|
36 | 36 | self.baseui = baseui |
|
37 | 37 | self.ui = baseui.copy() |
|
38 | 38 | |
|
39 | 39 | try: |
|
40 | 40 | self.ui.readconfig(self.join("hgrc"), self.root) |
|
41 | 41 | extensions.loadall(self.ui) |
|
42 | 42 | except IOError: |
|
43 | 43 | pass |
|
44 | 44 | |
|
45 | 45 | if not os.path.isdir(self.path): |
|
46 | 46 | if create: |
|
47 | 47 | if not os.path.exists(path): |
|
48 | 48 | util.makedirs(path) |
|
49 | 49 | os.mkdir(self.path) |
|
50 | 50 | requirements = ["revlogv1"] |
|
51 | 51 | if self.ui.configbool('format', 'usestore', True): |
|
52 | 52 | os.mkdir(os.path.join(self.path, "store")) |
|
53 | 53 | requirements.append("store") |
|
54 | 54 | if self.ui.configbool('format', 'usefncache', True): |
|
55 | 55 | requirements.append("fncache") |
|
56 | 56 | if self.ui.configbool('format', 'dotencode', True): |
|
57 | 57 | requirements.append('dotencode') |
|
58 | 58 | # create an invalid changelog |
|
59 | 59 | self.opener("00changelog.i", "a").write( |
|
60 | 60 | '\0\0\0\2' # represents revlogv2 |
|
61 | 61 | ' dummy changelog to prevent using the old repo layout' |
|
62 | 62 | ) |
|
63 | 63 | if self.ui.configbool('format', 'parentdelta', False): |
|
64 | 64 | requirements.append("parentdelta") |
|
65 | 65 | else: |
|
66 | 66 | raise error.RepoError(_("repository %s not found") % path) |
|
67 | 67 | elif create: |
|
68 | 68 | raise error.RepoError(_("repository %s already exists") % path) |
|
69 | 69 | else: |
|
70 | 70 | # find requirements |
|
71 | 71 | requirements = set() |
|
72 | 72 | try: |
|
73 | 73 | requirements = set(self.opener("requires").read().splitlines()) |
|
74 | 74 | except IOError, inst: |
|
75 | 75 | if inst.errno != errno.ENOENT: |
|
76 | 76 | raise |
|
77 | 77 | for r in requirements - self.supported: |
|
78 | 78 | raise error.RepoError(_("requirement '%s' not supported") % r) |
|
79 | 79 | |
|
80 | 80 | self.sharedpath = self.path |
|
81 | 81 | try: |
|
82 | 82 | s = os.path.realpath(self.opener("sharedpath").read()) |
|
83 | 83 | if not os.path.exists(s): |
|
84 | 84 | raise error.RepoError( |
|
85 | 85 | _('.hg/sharedpath points to nonexistent directory %s') % s) |
|
86 | 86 | self.sharedpath = s |
|
87 | 87 | except IOError, inst: |
|
88 | 88 | if inst.errno != errno.ENOENT: |
|
89 | 89 | raise |
|
90 | 90 | |
|
91 | 91 | self.store = store.store(requirements, self.sharedpath, util.opener) |
|
92 | 92 | self.spath = self.store.path |
|
93 | 93 | self.sopener = self.store.opener |
|
94 | 94 | self.sjoin = self.store.join |
|
95 | 95 | self.opener.createmode = self.store.createmode |
|
96 | 96 | self._applyrequirements(requirements) |
|
97 | 97 | if create: |
|
98 | 98 | self._writerequirements() |
|
99 | 99 | |
|
100 | 100 | # These two define the set of tags for this repository. _tags |
|
101 | 101 | # maps tag name to node; _tagtypes maps tag name to 'global' or |
|
102 | 102 | # 'local'. (Global tags are defined by .hgtags across all |
|
103 | 103 | # heads, and local tags are defined in .hg/localtags.) They |
|
104 | 104 | # constitute the in-memory cache of tags. |
|
105 | 105 | self._tags = None |
|
106 | 106 | self._tagtypes = None |
|
107 | 107 | |
|
108 | 108 | self._branchcache = None |
|
109 | 109 | self._branchcachetip = None |
|
110 | 110 | self.nodetagscache = None |
|
111 | 111 | self.filterpats = {} |
|
112 | 112 | self._datafilters = {} |
|
113 | 113 | self._transref = self._lockref = self._wlockref = None |
|
114 | 114 | |
|
115 | 115 | def _applyrequirements(self, requirements): |
|
116 | 116 | self.requirements = requirements |
|
117 | 117 | self.sopener.options = {} |
|
118 | 118 | if 'parentdelta' in requirements: |
|
119 | 119 | self.sopener.options['parentdelta'] = 1 |
|
120 | 120 | |
|
121 | 121 | def _writerequirements(self): |
|
122 | 122 | reqfile = self.opener("requires", "w") |
|
123 | 123 | for r in self.requirements: |
|
124 | 124 | reqfile.write("%s\n" % r) |
|
125 | 125 | reqfile.close() |
|
126 | 126 | |
|
127 | 127 | def _checknested(self, path): |
|
128 | 128 | """Determine if path is a legal nested repository.""" |
|
129 | 129 | if not path.startswith(self.root): |
|
130 | 130 | return False |
|
131 | 131 | subpath = path[len(self.root) + 1:] |
|
132 | 132 | |
|
133 | 133 | # XXX: Checking against the current working copy is wrong in |
|
134 | 134 | # the sense that it can reject things like |
|
135 | 135 | # |
|
136 | 136 | # $ hg cat -r 10 sub/x.txt |
|
137 | 137 | # |
|
138 | 138 | # if sub/ is no longer a subrepository in the working copy |
|
139 | 139 | # parent revision. |
|
140 | 140 | # |
|
141 | 141 | # However, it can of course also allow things that would have |
|
142 | 142 | # been rejected before, such as the above cat command if sub/ |
|
143 | 143 | # is a subrepository now, but was a normal directory before. |
|
144 | 144 | # The old path auditor would have rejected by mistake since it |
|
145 | 145 | # panics when it sees sub/.hg/. |
|
146 | 146 | # |
|
147 | 147 | # All in all, checking against the working copy seems sensible |
|
148 | 148 | # since we want to prevent access to nested repositories on |
|
149 | 149 | # the filesystem *now*. |
|
150 | 150 | ctx = self[None] |
|
151 | 151 | parts = util.splitpath(subpath) |
|
152 | 152 | while parts: |
|
153 | 153 | prefix = os.sep.join(parts) |
|
154 | 154 | if prefix in ctx.substate: |
|
155 | 155 | if prefix == subpath: |
|
156 | 156 | return True |
|
157 | 157 | else: |
|
158 | 158 | sub = ctx.sub(prefix) |
|
159 | 159 | return sub.checknested(subpath[len(prefix) + 1:]) |
|
160 | 160 | else: |
|
161 | 161 | parts.pop() |
|
162 | 162 | return False |
|
163 | 163 | |
|
164 | 164 | |
|
165 | 165 | @propertycache |
|
166 | 166 | def changelog(self): |
|
167 | 167 | c = changelog.changelog(self.sopener) |
|
168 | 168 | if 'HG_PENDING' in os.environ: |
|
169 | 169 | p = os.environ['HG_PENDING'] |
|
170 | 170 | if p.startswith(self.root): |
|
171 | 171 | c.readpending('00changelog.i.a') |
|
172 | 172 | self.sopener.options['defversion'] = c.version |
|
173 | 173 | return c |
|
174 | 174 | |
|
175 | 175 | @propertycache |
|
176 | 176 | def manifest(self): |
|
177 | 177 | return manifest.manifest(self.sopener) |
|
178 | 178 | |
|
179 | 179 | @propertycache |
|
180 | 180 | def dirstate(self): |
|
181 | 181 | warned = [0] |
|
182 | 182 | def validate(node): |
|
183 | 183 | try: |
|
184 | 184 | r = self.changelog.rev(node) |
|
185 | 185 | return node |
|
186 | 186 | except error.LookupError: |
|
187 | 187 | if not warned[0]: |
|
188 | 188 | warned[0] = True |
|
189 | 189 | self.ui.warn(_("warning: ignoring unknown" |
|
190 | 190 | " working parent %s!\n") % short(node)) |
|
191 | 191 | return nullid |
|
192 | 192 | |
|
193 | 193 | return dirstate.dirstate(self.opener, self.ui, self.root, validate) |
|
194 | 194 | |
|
195 | 195 | def __getitem__(self, changeid): |
|
196 | 196 | if changeid is None: |
|
197 | 197 | return context.workingctx(self) |
|
198 | 198 | return context.changectx(self, changeid) |
|
199 | 199 | |
|
200 | 200 | def __contains__(self, changeid): |
|
201 | 201 | try: |
|
202 | 202 | return bool(self.lookup(changeid)) |
|
203 | 203 | except error.RepoLookupError: |
|
204 | 204 | return False |
|
205 | 205 | |
|
206 | 206 | def __nonzero__(self): |
|
207 | 207 | return True |
|
208 | 208 | |
|
209 | 209 | def __len__(self): |
|
210 | 210 | return len(self.changelog) |
|
211 | 211 | |
|
212 | 212 | def __iter__(self): |
|
213 | 213 | for i in xrange(len(self)): |
|
214 | 214 | yield i |
|
215 | 215 | |
|
216 | 216 | def url(self): |
|
217 | 217 | return 'file:' + self.root |
|
218 | 218 | |
|
219 | 219 | def hook(self, name, throw=False, **args): |
|
220 | 220 | return hook.hook(self.ui, self, name, throw, **args) |
|
221 | 221 | |
|
222 | 222 | tag_disallowed = ':\r\n' |
|
223 | 223 | |
|
224 | 224 | def _tag(self, names, node, message, local, user, date, extra={}): |
|
225 | 225 | if isinstance(names, str): |
|
226 | 226 | allchars = names |
|
227 | 227 | names = (names,) |
|
228 | 228 | else: |
|
229 | 229 | allchars = ''.join(names) |
|
230 | 230 | for c in self.tag_disallowed: |
|
231 | 231 | if c in allchars: |
|
232 | 232 | raise util.Abort(_('%r cannot be used in a tag name') % c) |
|
233 | 233 | |
|
234 | 234 | branches = self.branchmap() |
|
235 | 235 | for name in names: |
|
236 | 236 | self.hook('pretag', throw=True, node=hex(node), tag=name, |
|
237 | 237 | local=local) |
|
238 | 238 | if name in branches: |
|
239 | 239 | self.ui.warn(_("warning: tag %s conflicts with existing" |
|
240 | 240 | " branch name\n") % name) |
|
241 | 241 | |
|
242 | 242 | def writetags(fp, names, munge, prevtags): |
|
243 | 243 | fp.seek(0, 2) |
|
244 | 244 | if prevtags and prevtags[-1] != '\n': |
|
245 | 245 | fp.write('\n') |
|
246 | 246 | for name in names: |
|
247 | 247 | m = munge and munge(name) or name |
|
248 | 248 | if self._tagtypes and name in self._tagtypes: |
|
249 | 249 | old = self._tags.get(name, nullid) |
|
250 | 250 | fp.write('%s %s\n' % (hex(old), m)) |
|
251 | 251 | fp.write('%s %s\n' % (hex(node), m)) |
|
252 | 252 | fp.close() |
|
253 | 253 | |
|
254 | 254 | prevtags = '' |
|
255 | 255 | if local: |
|
256 | 256 | try: |
|
257 | 257 | fp = self.opener('localtags', 'r+') |
|
258 | 258 | except IOError: |
|
259 | 259 | fp = self.opener('localtags', 'a') |
|
260 | 260 | else: |
|
261 | 261 | prevtags = fp.read() |
|
262 | 262 | |
|
263 | 263 | # local tags are stored in the current charset |
|
264 | 264 | writetags(fp, names, None, prevtags) |
|
265 | 265 | for name in names: |
|
266 | 266 | self.hook('tag', node=hex(node), tag=name, local=local) |
|
267 | 267 | return |
|
268 | 268 | |
|
269 | 269 | try: |
|
270 | 270 | fp = self.wfile('.hgtags', 'rb+') |
|
271 | 271 | except IOError: |
|
272 | 272 | fp = self.wfile('.hgtags', 'ab') |
|
273 | 273 | else: |
|
274 | 274 | prevtags = fp.read() |
|
275 | 275 | |
|
276 | 276 | # committed tags are stored in UTF-8 |
|
277 | 277 | writetags(fp, names, encoding.fromlocal, prevtags) |
|
278 | 278 | |
|
279 | 279 | if '.hgtags' not in self.dirstate: |
|
280 | 280 | self[None].add(['.hgtags']) |
|
281 | 281 | |
|
282 | 282 | m = matchmod.exact(self.root, '', ['.hgtags']) |
|
283 | 283 | tagnode = self.commit(message, user, date, extra=extra, match=m) |
|
284 | 284 | |
|
285 | 285 | for name in names: |
|
286 | 286 | self.hook('tag', node=hex(node), tag=name, local=local) |
|
287 | 287 | |
|
288 | 288 | return tagnode |
|
289 | 289 | |
|
290 | 290 | def tag(self, names, node, message, local, user, date): |
|
291 | 291 | '''tag a revision with one or more symbolic names. |
|
292 | 292 | |
|
293 | 293 | names is a list of strings or, when adding a single tag, names may be a |
|
294 | 294 | string. |
|
295 | 295 | |
|
296 | 296 | if local is True, the tags are stored in a per-repository file. |
|
297 | 297 | otherwise, they are stored in the .hgtags file, and a new |
|
298 | 298 | changeset is committed with the change. |
|
299 | 299 | |
|
300 | 300 | keyword arguments: |
|
301 | 301 | |
|
302 | 302 | local: whether to store tags in non-version-controlled file |
|
303 | 303 | (default False) |
|
304 | 304 | |
|
305 | 305 | message: commit message to use if committing |
|
306 | 306 | |
|
307 | 307 | user: name of user to use if committing |
|
308 | 308 | |
|
309 | 309 | date: date tuple to use if committing''' |
|
310 | 310 | |
|
311 | 311 | if not local: |
|
312 | 312 | for x in self.status()[:5]: |
|
313 | 313 | if '.hgtags' in x: |
|
314 | 314 | raise util.Abort(_('working copy of .hgtags is changed ' |
|
315 | 315 | '(please commit .hgtags manually)')) |
|
316 | 316 | |
|
317 | 317 | self.tags() # instantiate the cache |
|
318 | 318 | self._tag(names, node, message, local, user, date) |
|
319 | 319 | |
|
320 | 320 | def tags(self): |
|
321 | 321 | '''return a mapping of tag to node''' |
|
322 | 322 | if self._tags is None: |
|
323 | 323 | (self._tags, self._tagtypes) = self._findtags() |
|
324 | 324 | |
|
325 | 325 | return self._tags |
|
326 | 326 | |
|
327 | 327 | def _findtags(self): |
|
328 | 328 | '''Do the hard work of finding tags. Return a pair of dicts |
|
329 | 329 | (tags, tagtypes) where tags maps tag name to node, and tagtypes |
|
330 | 330 | maps tag name to a string like \'global\' or \'local\'. |
|
331 | 331 | Subclasses or extensions are free to add their own tags, but |
|
332 | 332 | should be aware that the returned dicts will be retained for the |
|
333 | 333 | duration of the localrepo object.''' |
|
334 | 334 | |
|
335 | 335 | # XXX what tagtype should subclasses/extensions use? Currently |
|
336 | 336 | # mq and bookmarks add tags, but do not set the tagtype at all. |
|
337 | 337 | # Should each extension invent its own tag type? Should there |
|
338 | 338 | # be one tagtype for all such "virtual" tags? Or is the status |
|
339 | 339 | # quo fine? |
|
340 | 340 | |
|
341 | 341 | alltags = {} # map tag name to (node, hist) |
|
342 | 342 | tagtypes = {} |
|
343 | 343 | |
|
344 | 344 | tagsmod.findglobaltags(self.ui, self, alltags, tagtypes) |
|
345 | 345 | tagsmod.readlocaltags(self.ui, self, alltags, tagtypes) |
|
346 | 346 | |
|
347 | 347 | # Build the return dicts. Have to re-encode tag names because |
|
348 | 348 | # the tags module always uses UTF-8 (in order not to lose info |
|
349 | 349 | # writing to the cache), but the rest of Mercurial wants them in |
|
350 | 350 | # local encoding. |
|
351 | 351 | tags = {} |
|
352 | 352 | for (name, (node, hist)) in alltags.iteritems(): |
|
353 | 353 | if node != nullid: |
|
354 | 354 | tags[encoding.tolocal(name)] = node |
|
355 | 355 | tags['tip'] = self.changelog.tip() |
|
356 | 356 | tagtypes = dict([(encoding.tolocal(name), value) |
|
357 | 357 | for (name, value) in tagtypes.iteritems()]) |
|
358 | 358 | return (tags, tagtypes) |
|
359 | 359 | |
|
360 | 360 | def tagtype(self, tagname): |
|
361 | 361 | ''' |
|
362 | 362 | return the type of the given tag. result can be: |
|
363 | 363 | |
|
364 | 364 | 'local' : a local tag |
|
365 | 365 | 'global' : a global tag |
|
366 | 366 | None : tag does not exist |
|
367 | 367 | ''' |
|
368 | 368 | |
|
369 | 369 | self.tags() |
|
370 | 370 | |
|
371 | 371 | return self._tagtypes.get(tagname) |
|
372 | 372 | |
|
373 | 373 | def tagslist(self): |
|
374 | 374 | '''return a list of tags ordered by revision''' |
|
375 | 375 | l = [] |
|
376 | 376 | for t, n in self.tags().iteritems(): |
|
377 | 377 | try: |
|
378 | 378 | r = self.changelog.rev(n) |
|
379 | 379 | except: |
|
380 | 380 | r = -2 # sort to the beginning of the list if unknown |
|
381 | 381 | l.append((r, t, n)) |
|
382 | 382 | return [(t, n) for r, t, n in sorted(l)] |
|
383 | 383 | |
|
384 | 384 | def nodetags(self, node): |
|
385 | 385 | '''return the tags associated with a node''' |
|
386 | 386 | if not self.nodetagscache: |
|
387 | 387 | self.nodetagscache = {} |
|
388 | 388 | for t, n in self.tags().iteritems(): |
|
389 | 389 | self.nodetagscache.setdefault(n, []).append(t) |
|
390 | 390 | for tags in self.nodetagscache.itervalues(): |
|
391 | 391 | tags.sort() |
|
392 | 392 | return self.nodetagscache.get(node, []) |
|
393 | 393 | |
|
394 | 394 | def _branchtags(self, partial, lrev): |
|
395 | 395 | # TODO: rename this function? |
|
396 | 396 | tiprev = len(self) - 1 |
|
397 | 397 | if lrev != tiprev: |
|
398 | 398 | ctxgen = (self[r] for r in xrange(lrev + 1, tiprev + 1)) |
|
399 | 399 | self._updatebranchcache(partial, ctxgen) |
|
400 | 400 | self._writebranchcache(partial, self.changelog.tip(), tiprev) |
|
401 | 401 | |
|
402 | 402 | return partial |
|
403 | 403 | |
|
404 | 404 | def updatebranchcache(self): |
|
405 | 405 | tip = self.changelog.tip() |
|
406 | 406 | if self._branchcache is not None and self._branchcachetip == tip: |
|
407 | 407 | return self._branchcache |
|
408 | 408 | |
|
409 | 409 | oldtip = self._branchcachetip |
|
410 | 410 | self._branchcachetip = tip |
|
411 | 411 | if oldtip is None or oldtip not in self.changelog.nodemap: |
|
412 | 412 | partial, last, lrev = self._readbranchcache() |
|
413 | 413 | else: |
|
414 | 414 | lrev = self.changelog.rev(oldtip) |
|
415 | 415 | partial = self._branchcache |
|
416 | 416 | |
|
417 | 417 | self._branchtags(partial, lrev) |
|
418 | 418 | # this private cache holds all heads (not just tips) |
|
419 | 419 | self._branchcache = partial |
|
420 | 420 | |
|
421 | 421 | def branchmap(self): |
|
422 | 422 | '''returns a dictionary {branch: [branchheads]}''' |
|
423 | 423 | self.updatebranchcache() |
|
424 | 424 | return self._branchcache |
|
425 | 425 | |
|
426 | 426 | def branchtags(self): |
|
427 | 427 | '''return a dict where branch names map to the tipmost head of |
|
428 | 428 | the branch, open heads come before closed''' |
|
429 | 429 | bt = {} |
|
430 | 430 | for bn, heads in self.branchmap().iteritems(): |
|
431 | 431 | tip = heads[-1] |
|
432 | 432 | for h in reversed(heads): |
|
433 | 433 | if 'close' not in self.changelog.read(h)[5]: |
|
434 | 434 | tip = h |
|
435 | 435 | break |
|
436 | 436 | bt[bn] = tip |
|
437 | 437 | return bt |
|
438 | 438 | |
|
439 | 439 | def _readbranchcache(self): |
|
440 | 440 | partial = {} |
|
441 | 441 | try: |
|
442 | 442 | f = self.opener("branchheads.cache") |
|
443 | 443 | lines = f.read().split('\n') |
|
444 | 444 | f.close() |
|
445 | 445 | except (IOError, OSError): |
|
446 | 446 | return {}, nullid, nullrev |
|
447 | 447 | |
|
448 | 448 | try: |
|
449 | 449 | last, lrev = lines.pop(0).split(" ", 1) |
|
450 | 450 | last, lrev = bin(last), int(lrev) |
|
451 | 451 | if lrev >= len(self) or self[lrev].node() != last: |
|
452 | 452 | # invalidate the cache |
|
453 | 453 | raise ValueError('invalidating branch cache (tip differs)') |
|
454 | 454 | for l in lines: |
|
455 | 455 | if not l: |
|
456 | 456 | continue |
|
457 | 457 | node, label = l.split(" ", 1) |
|
458 | 458 | label = encoding.tolocal(label.strip()) |
|
459 | 459 | partial.setdefault(label, []).append(bin(node)) |
|
460 | 460 | except KeyboardInterrupt: |
|
461 | 461 | raise |
|
462 | 462 | except Exception, inst: |
|
463 | 463 | if self.ui.debugflag: |
|
464 | 464 | self.ui.warn(str(inst), '\n') |
|
465 | 465 | partial, last, lrev = {}, nullid, nullrev |
|
466 | 466 | return partial, last, lrev |
|
467 | 467 | |
|
468 | 468 | def _writebranchcache(self, branches, tip, tiprev): |
|
469 | 469 | try: |
|
470 | 470 | f = self.opener("branchheads.cache", "w", atomictemp=True) |
|
471 | 471 | f.write("%s %s\n" % (hex(tip), tiprev)) |
|
472 | 472 | for label, nodes in branches.iteritems(): |
|
473 | 473 | for node in nodes: |
|
474 | 474 | f.write("%s %s\n" % (hex(node), encoding.fromlocal(label))) |
|
475 | 475 | f.rename() |
|
476 | 476 | except (IOError, OSError): |
|
477 | 477 | pass |
|
478 | 478 | |
|
479 | 479 | def _updatebranchcache(self, partial, ctxgen): |
|
480 | 480 | # collect new branch entries |
|
481 | 481 | newbranches = {} |
|
482 | 482 | for c in ctxgen: |
|
483 | 483 | newbranches.setdefault(c.branch(), []).append(c.node()) |
|
484 | 484 | # if older branchheads are reachable from new ones, they aren't |
|
485 | 485 | # really branchheads. Note checking parents is insufficient: |
|
486 | 486 | # 1 (branch a) -> 2 (branch b) -> 3 (branch a) |
|
487 | 487 | for branch, newnodes in newbranches.iteritems(): |
|
488 | 488 | bheads = partial.setdefault(branch, []) |
|
489 | 489 | bheads.extend(newnodes) |
|
490 | 490 | if len(bheads) <= 1: |
|
491 | 491 | continue |
|
492 | 492 | # starting from tip means fewer passes over reachable |
|
493 | 493 | while newnodes: |
|
494 | 494 | latest = newnodes.pop() |
|
495 | 495 | if latest not in bheads: |
|
496 | 496 | continue |
|
497 | 497 | minbhrev = self[min([self[bh].rev() for bh in bheads])].node() |
|
498 | 498 | reachable = self.changelog.reachable(latest, minbhrev) |
|
499 | 499 | reachable.remove(latest) |
|
500 | 500 | bheads = [b for b in bheads if b not in reachable] |
|
501 | 501 | partial[branch] = bheads |
|
502 | 502 | |
|
503 | 503 | def lookup(self, key): |
|
504 | 504 | if isinstance(key, int): |
|
505 | 505 | return self.changelog.node(key) |
|
506 | 506 | elif key == '.': |
|
507 | 507 | return self.dirstate.parents()[0] |
|
508 | 508 | elif key == 'null': |
|
509 | 509 | return nullid |
|
510 | 510 | elif key == 'tip': |
|
511 | 511 | return self.changelog.tip() |
|
512 | 512 | n = self.changelog._match(key) |
|
513 | 513 | if n: |
|
514 | 514 | return n |
|
515 | 515 | if key in self.tags(): |
|
516 | 516 | return self.tags()[key] |
|
517 | 517 | if key in self.branchtags(): |
|
518 | 518 | return self.branchtags()[key] |
|
519 | 519 | n = self.changelog._partialmatch(key) |
|
520 | 520 | if n: |
|
521 | 521 | return n |
|
522 | 522 | |
|
523 | 523 | # can't find key, check if it might have come from damaged dirstate |
|
524 | 524 | if key in self.dirstate.parents(): |
|
525 | 525 | raise error.Abort(_("working directory has unknown parent '%s'!") |
|
526 | 526 | % short(key)) |
|
527 | 527 | try: |
|
528 | 528 | if len(key) == 20: |
|
529 | 529 | key = hex(key) |
|
530 | 530 | except: |
|
531 | 531 | pass |
|
532 | 532 | raise error.RepoLookupError(_("unknown revision '%s'") % key) |
|
533 | 533 | |
|
534 | 534 | def lookupbranch(self, key, remote=None): |
|
535 | 535 | repo = remote or self |
|
536 | 536 | if key in repo.branchmap(): |
|
537 | 537 | return key |
|
538 | 538 | |
|
539 | 539 | repo = (remote and remote.local()) and remote or self |
|
540 | 540 | return repo[key].branch() |
|
541 | 541 | |
|
542 | 542 | def local(self): |
|
543 | 543 | return True |
|
544 | 544 | |
|
545 | 545 | def join(self, f): |
|
546 | 546 | return os.path.join(self.path, f) |
|
547 | 547 | |
|
548 | 548 | def wjoin(self, f): |
|
549 | 549 | return os.path.join(self.root, f) |
|
550 | 550 | |
|
551 | 551 | def file(self, f): |
|
552 | 552 | if f[0] == '/': |
|
553 | 553 | f = f[1:] |
|
554 | 554 | return filelog.filelog(self.sopener, f) |
|
555 | 555 | |
|
556 | 556 | def changectx(self, changeid): |
|
557 | 557 | return self[changeid] |
|
558 | 558 | |
|
559 | 559 | def parents(self, changeid=None): |
|
560 | 560 | '''get list of changectxs for parents of changeid''' |
|
561 | 561 | return self[changeid].parents() |
|
562 | 562 | |
|
563 | 563 | def filectx(self, path, changeid=None, fileid=None): |
|
564 | 564 | """changeid can be a changeset revision, node, or tag. |
|
565 | 565 | fileid can be a file revision or node.""" |
|
566 | 566 | return context.filectx(self, path, changeid, fileid) |
|
567 | 567 | |
|
568 | 568 | def getcwd(self): |
|
569 | 569 | return self.dirstate.getcwd() |
|
570 | 570 | |
|
571 | 571 | def pathto(self, f, cwd=None): |
|
572 | 572 | return self.dirstate.pathto(f, cwd) |
|
573 | 573 | |
|
574 | 574 | def wfile(self, f, mode='r'): |
|
575 | 575 | return self.wopener(f, mode) |
|
576 | 576 | |
|
577 | 577 | def _link(self, f): |
|
578 | 578 | return os.path.islink(self.wjoin(f)) |
|
579 | 579 | |
|
580 | 580 | def _loadfilter(self, filter): |
|
581 | 581 | if filter not in self.filterpats: |
|
582 | 582 | l = [] |
|
583 | 583 | for pat, cmd in self.ui.configitems(filter): |
|
584 | 584 | if cmd == '!': |
|
585 | 585 | continue |
|
586 | 586 | mf = matchmod.match(self.root, '', [pat]) |
|
587 | 587 | fn = None |
|
588 | 588 | params = cmd |
|
589 | 589 | for name, filterfn in self._datafilters.iteritems(): |
|
590 | 590 | if cmd.startswith(name): |
|
591 | 591 | fn = filterfn |
|
592 | 592 | params = cmd[len(name):].lstrip() |
|
593 | 593 | break |
|
594 | 594 | if not fn: |
|
595 | 595 | fn = lambda s, c, **kwargs: util.filter(s, c) |
|
596 | 596 | # Wrap old filters not supporting keyword arguments |
|
597 | 597 | if not inspect.getargspec(fn)[2]: |
|
598 | 598 | oldfn = fn |
|
599 | 599 | fn = lambda s, c, **kwargs: oldfn(s, c) |
|
600 | 600 | l.append((mf, fn, params)) |
|
601 | 601 | self.filterpats[filter] = l |
|
602 | 602 | return self.filterpats[filter] |
|
603 | 603 | |
|
604 | 604 | def _filter(self, filterpats, filename, data): |
|
605 | 605 | for mf, fn, cmd in filterpats: |
|
606 | 606 | if mf(filename): |
|
607 | 607 | self.ui.debug("filtering %s through %s\n" % (filename, cmd)) |
|
608 | 608 | data = fn(data, cmd, ui=self.ui, repo=self, filename=filename) |
|
609 | 609 | break |
|
610 | 610 | |
|
611 | 611 | return data |
|
612 | 612 | |
|
613 | 613 | @propertycache |
|
614 | 614 | def _encodefilterpats(self): |
|
615 | 615 | return self._loadfilter('encode') |
|
616 | 616 | |
|
617 | 617 | @propertycache |
|
618 | 618 | def _decodefilterpats(self): |
|
619 | 619 | return self._loadfilter('decode') |
|
620 | 620 | |
|
621 | 621 | def adddatafilter(self, name, filter): |
|
622 | 622 | self._datafilters[name] = filter |
|
623 | 623 | |
|
624 | 624 | def wread(self, filename): |
|
625 | 625 | if self._link(filename): |
|
626 | 626 | data = os.readlink(self.wjoin(filename)) |
|
627 | 627 | else: |
|
628 | 628 | data = self.wopener(filename, 'r').read() |
|
629 | 629 | return self._filter(self._encodefilterpats, filename, data) |
|
630 | 630 | |
|
631 | 631 | def wwrite(self, filename, data, flags): |
|
632 | 632 | data = self._filter(self._decodefilterpats, filename, data) |
|
633 | 633 | if 'l' in flags: |
|
634 | 634 | self.wopener.symlink(data, filename) |
|
635 | 635 | else: |
|
636 | 636 | self.wopener(filename, 'w').write(data) |
|
637 | 637 | if 'x' in flags: |
|
638 | 638 | util.set_flags(self.wjoin(filename), False, True) |
|
639 | 639 | |
|
640 | 640 | def wwritedata(self, filename, data): |
|
641 | 641 | return self._filter(self._decodefilterpats, filename, data) |
|
642 | 642 | |
|
643 | 643 | def transaction(self, desc): |
|
644 | 644 | tr = self._transref and self._transref() or None |
|
645 | 645 | if tr and tr.running(): |
|
646 | 646 | return tr.nest() |
|
647 | 647 | |
|
648 | 648 | # abort here if the journal already exists |
|
649 | 649 | if os.path.exists(self.sjoin("journal")): |
|
650 | 650 | raise error.RepoError( |
|
651 | 651 | _("abandoned transaction found - run hg recover")) |
|
652 | 652 | |
|
653 | 653 | # save dirstate for rollback |
|
654 | 654 | try: |
|
655 | 655 | ds = self.opener("dirstate").read() |
|
656 | 656 | except IOError: |
|
657 | 657 | ds = "" |
|
658 | 658 | self.opener("journal.dirstate", "w").write(ds) |
|
659 | 659 | self.opener("journal.branch", "w").write( |
|
660 | 660 | encoding.fromlocal(self.dirstate.branch())) |
|
661 | 661 | self.opener("journal.desc", "w").write("%d\n%s\n" % (len(self), desc)) |
|
662 | 662 | |
|
663 | 663 | renames = [(self.sjoin("journal"), self.sjoin("undo")), |
|
664 | 664 | (self.join("journal.dirstate"), self.join("undo.dirstate")), |
|
665 | 665 | (self.join("journal.branch"), self.join("undo.branch")), |
|
666 | 666 | (self.join("journal.desc"), self.join("undo.desc"))] |
|
667 | 667 | tr = transaction.transaction(self.ui.warn, self.sopener, |
|
668 | 668 | self.sjoin("journal"), |
|
669 | 669 | aftertrans(renames), |
|
670 | 670 | self.store.createmode) |
|
671 | 671 | self._transref = weakref.ref(tr) |
|
672 | 672 | return tr |
|
673 | 673 | |
|
674 | 674 | def recover(self): |
|
675 | 675 | lock = self.lock() |
|
676 | 676 | try: |
|
677 | 677 | if os.path.exists(self.sjoin("journal")): |
|
678 | 678 | self.ui.status(_("rolling back interrupted transaction\n")) |
|
679 | 679 | transaction.rollback(self.sopener, self.sjoin("journal"), |
|
680 | 680 | self.ui.warn) |
|
681 | 681 | self.invalidate() |
|
682 | 682 | return True |
|
683 | 683 | else: |
|
684 | 684 | self.ui.warn(_("no interrupted transaction available\n")) |
|
685 | 685 | return False |
|
686 | 686 | finally: |
|
687 | 687 | lock.release() |
|
688 | 688 | |
|
689 | 689 | def rollback(self, dryrun=False): |
|
690 | 690 | wlock = lock = None |
|
691 | 691 | try: |
|
692 | 692 | wlock = self.wlock() |
|
693 | 693 | lock = self.lock() |
|
694 | 694 | if os.path.exists(self.sjoin("undo")): |
|
695 | 695 | try: |
|
696 | 696 | args = self.opener("undo.desc", "r").read().splitlines() |
|
697 | 697 | if len(args) >= 3 and self.ui.verbose: |
|
698 | 698 | desc = _("rolling back to revision %s" |
|
699 | 699 | " (undo %s: %s)\n") % ( |
|
700 | 700 | int(args[0]) - 1, args[1], args[2]) |
|
701 | 701 | elif len(args) >= 2: |
|
702 | 702 | desc = _("rolling back to revision %s (undo %s)\n") % ( |
|
703 | 703 | int(args[0]) - 1, args[1]) |
|
704 | 704 | except IOError: |
|
705 | 705 | desc = _("rolling back unknown transaction\n") |
|
706 | 706 | self.ui.status(desc) |
|
707 | 707 | if dryrun: |
|
708 | 708 | return |
|
709 | 709 | transaction.rollback(self.sopener, self.sjoin("undo"), |
|
710 | 710 | self.ui.warn) |
|
711 | 711 | util.rename(self.join("undo.dirstate"), self.join("dirstate")) |
|
712 | 712 | try: |
|
713 | 713 | branch = self.opener("undo.branch").read() |
|
714 | 714 | self.dirstate.setbranch(branch) |
|
715 | 715 | except IOError: |
|
716 | 716 | self.ui.warn(_("Named branch could not be reset, " |
|
717 | 717 | "current branch still is: %s\n") |
|
718 | 718 | % self.dirstate.branch()) |
|
719 | 719 | self.invalidate() |
|
720 | 720 | self.dirstate.invalidate() |
|
721 | 721 | self.destroyed() |
|
722 | 722 | else: |
|
723 | 723 | self.ui.warn(_("no rollback information available\n")) |
|
724 | 724 | return 1 |
|
725 | 725 | finally: |
|
726 | 726 | release(lock, wlock) |
|
727 | 727 | |
|
728 | 728 | def invalidatecaches(self): |
|
729 | 729 | self._tags = None |
|
730 | 730 | self._tagtypes = None |
|
731 | 731 | self.nodetagscache = None |
|
732 | 732 | self._branchcache = None # in UTF-8 |
|
733 | 733 | self._branchcachetip = None |
|
734 | 734 | |
|
735 | 735 | def invalidate(self): |
|
736 | 736 | for a in ("changelog", "manifest"): |
|
737 | 737 | if a in self.__dict__: |
|
738 | 738 | delattr(self, a) |
|
739 | 739 | self.invalidatecaches() |
|
740 | 740 | |
|
741 | 741 | def _lock(self, lockname, wait, releasefn, acquirefn, desc): |
|
742 | 742 | try: |
|
743 | 743 | l = lock.lock(lockname, 0, releasefn, desc=desc) |
|
744 | 744 | except error.LockHeld, inst: |
|
745 | 745 | if not wait: |
|
746 | 746 | raise |
|
747 | 747 | self.ui.warn(_("waiting for lock on %s held by %r\n") % |
|
748 | 748 | (desc, inst.locker)) |
|
749 | 749 | # default to 600 seconds timeout |
|
750 | 750 | l = lock.lock(lockname, int(self.ui.config("ui", "timeout", "600")), |
|
751 | 751 | releasefn, desc=desc) |
|
752 | 752 | if acquirefn: |
|
753 | 753 | acquirefn() |
|
754 | 754 | return l |
|
755 | 755 | |
|
756 | 756 | def lock(self, wait=True): |
|
757 | 757 | '''Lock the repository store (.hg/store) and return a weak reference |
|
758 | 758 | to the lock. Use this before modifying the store (e.g. committing or |
|
759 | 759 | stripping). If you are opening a transaction, get a lock as well.)''' |
|
760 | 760 | l = self._lockref and self._lockref() |
|
761 | 761 | if l is not None and l.held: |
|
762 | 762 | l.lock() |
|
763 | 763 | return l |
|
764 | 764 | |
|
765 | 765 | l = self._lock(self.sjoin("lock"), wait, None, self.invalidate, |
|
766 | 766 | _('repository %s') % self.origroot) |
|
767 | 767 | self._lockref = weakref.ref(l) |
|
768 | 768 | return l |
|
769 | 769 | |
|
770 | 770 | def wlock(self, wait=True): |
|
771 | 771 | '''Lock the non-store parts of the repository (everything under |
|
772 | 772 | .hg except .hg/store) and return a weak reference to the lock. |
|
773 | 773 | Use this before modifying files in .hg.''' |
|
774 | 774 | l = self._wlockref and self._wlockref() |
|
775 | 775 | if l is not None and l.held: |
|
776 | 776 | l.lock() |
|
777 | 777 | return l |
|
778 | 778 | |
|
779 | 779 | l = self._lock(self.join("wlock"), wait, self.dirstate.write, |
|
780 | 780 | self.dirstate.invalidate, _('working directory of %s') % |
|
781 | 781 | self.origroot) |
|
782 | 782 | self._wlockref = weakref.ref(l) |
|
783 | 783 | return l |
|
784 | 784 | |
|
785 | 785 | def _filecommit(self, fctx, manifest1, manifest2, linkrev, tr, changelist): |
|
786 | 786 | """ |
|
787 | 787 | commit an individual file as part of a larger transaction |
|
788 | 788 | """ |
|
789 | 789 | |
|
790 | 790 | fname = fctx.path() |
|
791 | 791 | text = fctx.data() |
|
792 | 792 | flog = self.file(fname) |
|
793 | 793 | fparent1 = manifest1.get(fname, nullid) |
|
794 | 794 | fparent2 = fparent2o = manifest2.get(fname, nullid) |
|
795 | 795 | |
|
796 | 796 | meta = {} |
|
797 | 797 | copy = fctx.renamed() |
|
798 | 798 | if copy and copy[0] != fname: |
|
799 | 799 | # Mark the new revision of this file as a copy of another |
|
800 | 800 | # file. This copy data will effectively act as a parent |
|
801 | 801 | # of this new revision. If this is a merge, the first |
|
802 | 802 | # parent will be the nullid (meaning "look up the copy data") |
|
803 | 803 | # and the second one will be the other parent. For example: |
|
804 | 804 | # |
|
805 | 805 | # 0 --- 1 --- 3 rev1 changes file foo |
|
806 | 806 | # \ / rev2 renames foo to bar and changes it |
|
807 | 807 | # \- 2 -/ rev3 should have bar with all changes and |
|
808 | 808 | # should record that bar descends from |
|
809 | 809 | # bar in rev2 and foo in rev1 |
|
810 | 810 | # |
|
811 | 811 | # this allows this merge to succeed: |
|
812 | 812 | # |
|
813 | 813 | # 0 --- 1 --- 3 rev4 reverts the content change from rev2 |
|
814 | 814 | # \ / merging rev3 and rev4 should use bar@rev2 |
|
815 | 815 | # \- 2 --- 4 as the merge base |
|
816 | 816 | # |
|
817 | 817 | |
|
818 | 818 | cfname = copy[0] |
|
819 | 819 | crev = manifest1.get(cfname) |
|
820 | 820 | newfparent = fparent2 |
|
821 | 821 | |
|
822 | 822 | if manifest2: # branch merge |
|
823 | 823 | if fparent2 == nullid or crev is None: # copied on remote side |
|
824 | 824 | if cfname in manifest2: |
|
825 | 825 | crev = manifest2[cfname] |
|
826 | 826 | newfparent = fparent1 |
|
827 | 827 | |
|
828 | 828 | # find source in nearest ancestor if we've lost track |
|
829 | 829 | if not crev: |
|
830 | 830 | self.ui.debug(" %s: searching for copy revision for %s\n" % |
|
831 | 831 | (fname, cfname)) |
|
832 | 832 | for ancestor in self[None].ancestors(): |
|
833 | 833 | if cfname in ancestor: |
|
834 | 834 | crev = ancestor[cfname].filenode() |
|
835 | 835 | break |
|
836 | 836 | |
|
837 | 837 | if crev: |
|
838 | 838 | self.ui.debug(" %s: copy %s:%s\n" % (fname, cfname, hex(crev))) |
|
839 | 839 | meta["copy"] = cfname |
|
840 | 840 | meta["copyrev"] = hex(crev) |
|
841 | 841 | fparent1, fparent2 = nullid, newfparent |
|
842 | 842 | else: |
|
843 | 843 | self.ui.warn(_("warning: can't find ancestor for '%s' " |
|
844 | 844 | "copied from '%s'!\n") % (fname, cfname)) |
|
845 | 845 | |
|
846 | 846 | elif fparent2 != nullid: |
|
847 | 847 | # is one parent an ancestor of the other? |
|
848 | 848 | fparentancestor = flog.ancestor(fparent1, fparent2) |
|
849 | 849 | if fparentancestor == fparent1: |
|
850 | 850 | fparent1, fparent2 = fparent2, nullid |
|
851 | 851 | elif fparentancestor == fparent2: |
|
852 | 852 | fparent2 = nullid |
|
853 | 853 | |
|
854 | 854 | # is the file changed? |
|
855 | 855 | if fparent2 != nullid or flog.cmp(fparent1, text) or meta: |
|
856 | 856 | changelist.append(fname) |
|
857 | 857 | return flog.add(text, meta, tr, linkrev, fparent1, fparent2) |
|
858 | 858 | |
|
859 | 859 | # are just the flags changed during merge? |
|
860 | 860 | if fparent1 != fparent2o and manifest1.flags(fname) != fctx.flags(): |
|
861 | 861 | changelist.append(fname) |
|
862 | 862 | |
|
863 | 863 | return fparent1 |
|
864 | 864 | |
|
865 | 865 | def commit(self, text="", user=None, date=None, match=None, force=False, |
|
866 | 866 | editor=False, extra={}): |
|
867 | 867 | """Add a new revision to current repository. |
|
868 | 868 | |
|
869 | 869 | Revision information is gathered from the working directory, |
|
870 | 870 | match can be used to filter the committed files. If editor is |
|
871 | 871 | supplied, it is called to get a commit message. |
|
872 | 872 | """ |
|
873 | 873 | |
|
874 | 874 | def fail(f, msg): |
|
875 | 875 | raise util.Abort('%s: %s' % (f, msg)) |
|
876 | 876 | |
|
877 | 877 | if not match: |
|
878 | 878 | match = matchmod.always(self.root, '') |
|
879 | 879 | |
|
880 | 880 | if not force: |
|
881 | 881 | vdirs = [] |
|
882 | 882 | match.dir = vdirs.append |
|
883 | 883 | match.bad = fail |
|
884 | 884 | |
|
885 | 885 | wlock = self.wlock() |
|
886 | 886 | try: |
|
887 | 887 | wctx = self[None] |
|
888 | 888 | merge = len(wctx.parents()) > 1 |
|
889 | 889 | |
|
890 | 890 | if (not force and merge and match and |
|
891 | 891 | (match.files() or match.anypats())): |
|
892 | 892 | raise util.Abort(_('cannot partially commit a merge ' |
|
893 | 893 | '(do not specify files or patterns)')) |
|
894 | 894 | |
|
895 | 895 | changes = self.status(match=match, clean=force) |
|
896 | 896 | if force: |
|
897 | 897 | changes[0].extend(changes[6]) # mq may commit unchanged files |
|
898 | 898 | |
|
899 | 899 | # check subrepos |
|
900 | 900 | subs = [] |
|
901 | 901 | removedsubs = set() |
|
902 | 902 | for p in wctx.parents(): |
|
903 | 903 | removedsubs.update(s for s in p.substate if match(s)) |
|
904 | 904 | for s in wctx.substate: |
|
905 | 905 | removedsubs.discard(s) |
|
906 | 906 | if match(s) and wctx.sub(s).dirty(): |
|
907 | 907 | subs.append(s) |
|
908 | 908 | if (subs or removedsubs): |
|
909 | 909 | if (not match('.hgsub') and |
|
910 | 910 | '.hgsub' in (wctx.modified() + wctx.added())): |
|
911 | 911 | raise util.Abort(_("can't commit subrepos without .hgsub")) |
|
912 | 912 | if '.hgsubstate' not in changes[0]: |
|
913 | 913 | changes[0].insert(0, '.hgsubstate') |
|
914 | 914 | |
|
915 | 915 | # make sure all explicit patterns are matched |
|
916 | 916 | if not force and match.files(): |
|
917 | 917 | matched = set(changes[0] + changes[1] + changes[2]) |
|
918 | 918 | |
|
919 | 919 | for f in match.files(): |
|
920 | 920 | if f == '.' or f in matched or f in wctx.substate: |
|
921 | 921 | continue |
|
922 | 922 | if f in changes[3]: # missing |
|
923 | 923 | fail(f, _('file not found!')) |
|
924 | 924 | if f in vdirs: # visited directory |
|
925 | 925 | d = f + '/' |
|
926 | 926 | for mf in matched: |
|
927 | 927 | if mf.startswith(d): |
|
928 | 928 | break |
|
929 | 929 | else: |
|
930 | 930 | fail(f, _("no match under directory!")) |
|
931 | 931 | elif f not in self.dirstate: |
|
932 | 932 | fail(f, _("file not tracked!")) |
|
933 | 933 | |
|
934 | 934 | if (not force and not extra.get("close") and not merge |
|
935 | 935 | and not (changes[0] or changes[1] or changes[2]) |
|
936 | 936 | and wctx.branch() == wctx.p1().branch()): |
|
937 | 937 | return None |
|
938 | 938 | |
|
939 | 939 | ms = mergemod.mergestate(self) |
|
940 | 940 | for f in changes[0]: |
|
941 | 941 | if f in ms and ms[f] == 'u': |
|
942 | 942 | raise util.Abort(_("unresolved merge conflicts " |
|
943 | 943 | "(see hg resolve)")) |
|
944 | 944 | |
|
945 | 945 | cctx = context.workingctx(self, text, user, date, extra, changes) |
|
946 | 946 | if editor: |
|
947 | 947 | cctx._text = editor(self, cctx, subs) |
|
948 | 948 | edited = (text != cctx._text) |
|
949 | 949 | |
|
950 | 950 | # commit subs |
|
951 | 951 | if subs or removedsubs: |
|
952 | 952 | state = wctx.substate.copy() |
|
953 | 953 | for s in sorted(subs): |
|
954 | 954 | sub = wctx.sub(s) |
|
955 | 955 | self.ui.status(_('committing subrepository %s\n') % |
|
956 | 956 | subrepo.subrelpath(sub)) |
|
957 | 957 | sr = sub.commit(cctx._text, user, date) |
|
958 | 958 | state[s] = (state[s][0], sr) |
|
959 | 959 | subrepo.writestate(self, state) |
|
960 | 960 | |
|
961 | 961 | # Save commit message in case this transaction gets rolled back |
|
962 | 962 | # (e.g. by a pretxncommit hook). Leave the content alone on |
|
963 | 963 | # the assumption that the user will use the same editor again. |
|
964 | 964 | msgfile = self.opener('last-message.txt', 'wb') |
|
965 | 965 | msgfile.write(cctx._text) |
|
966 | 966 | msgfile.close() |
|
967 | 967 | |
|
968 | 968 | p1, p2 = self.dirstate.parents() |
|
969 | 969 | hookp1, hookp2 = hex(p1), (p2 != nullid and hex(p2) or '') |
|
970 | 970 | try: |
|
971 | 971 | self.hook("precommit", throw=True, parent1=hookp1, parent2=hookp2) |
|
972 | 972 | ret = self.commitctx(cctx, True) |
|
973 | 973 | except: |
|
974 | 974 | if edited: |
|
975 | 975 | msgfn = self.pathto(msgfile.name[len(self.root)+1:]) |
|
976 | 976 | self.ui.write( |
|
977 | 977 | _('note: commit message saved in %s\n') % msgfn) |
|
978 | 978 | raise |
|
979 | 979 | |
|
980 | 980 | # update dirstate and mergestate |
|
981 | 981 | for f in changes[0] + changes[1]: |
|
982 | 982 | self.dirstate.normal(f) |
|
983 | 983 | for f in changes[2]: |
|
984 | 984 | self.dirstate.forget(f) |
|
985 | 985 | self.dirstate.setparents(ret) |
|
986 | 986 | ms.reset() |
|
987 | 987 | finally: |
|
988 | 988 | wlock.release() |
|
989 | 989 | |
|
990 | 990 | self.hook("commit", node=hex(ret), parent1=hookp1, parent2=hookp2) |
|
991 | 991 | return ret |
|
992 | 992 | |
|
993 | 993 | def commitctx(self, ctx, error=False): |
|
994 | 994 | """Add a new revision to current repository. |
|
995 | 995 | Revision information is passed via the context argument. |
|
996 | 996 | """ |
|
997 | 997 | |
|
998 | 998 | tr = lock = None |
|
999 | 999 | removed = list(ctx.removed()) |
|
1000 | 1000 | p1, p2 = ctx.p1(), ctx.p2() |
|
1001 | 1001 | m1 = p1.manifest().copy() |
|
1002 | 1002 | m2 = p2.manifest() |
|
1003 | 1003 | user = ctx.user() |
|
1004 | 1004 | |
|
1005 | 1005 | lock = self.lock() |
|
1006 | 1006 | try: |
|
1007 | 1007 | tr = self.transaction("commit") |
|
1008 | 1008 | trp = weakref.proxy(tr) |
|
1009 | 1009 | |
|
1010 | 1010 | # check in files |
|
1011 | 1011 | new = {} |
|
1012 | 1012 | changed = [] |
|
1013 | 1013 | linkrev = len(self) |
|
1014 | 1014 | for f in sorted(ctx.modified() + ctx.added()): |
|
1015 | 1015 | self.ui.note(f + "\n") |
|
1016 | 1016 | try: |
|
1017 | 1017 | fctx = ctx[f] |
|
1018 | 1018 | new[f] = self._filecommit(fctx, m1, m2, linkrev, trp, |
|
1019 | 1019 | changed) |
|
1020 | 1020 | m1.set(f, fctx.flags()) |
|
1021 | 1021 | except OSError, inst: |
|
1022 | 1022 | self.ui.warn(_("trouble committing %s!\n") % f) |
|
1023 | 1023 | raise |
|
1024 | 1024 | except IOError, inst: |
|
1025 | 1025 | errcode = getattr(inst, 'errno', errno.ENOENT) |
|
1026 | 1026 | if error or errcode and errcode != errno.ENOENT: |
|
1027 | 1027 | self.ui.warn(_("trouble committing %s!\n") % f) |
|
1028 | 1028 | raise |
|
1029 | 1029 | else: |
|
1030 | 1030 | removed.append(f) |
|
1031 | 1031 | |
|
1032 | 1032 | # update manifest |
|
1033 | 1033 | m1.update(new) |
|
1034 | 1034 | removed = [f for f in sorted(removed) if f in m1 or f in m2] |
|
1035 | 1035 | drop = [f for f in removed if f in m1] |
|
1036 | 1036 | for f in drop: |
|
1037 | 1037 | del m1[f] |
|
1038 | 1038 | mn = self.manifest.add(m1, trp, linkrev, p1.manifestnode(), |
|
1039 | 1039 | p2.manifestnode(), (new, drop)) |
|
1040 | 1040 | |
|
1041 | 1041 | # update changelog |
|
1042 | 1042 | self.changelog.delayupdate() |
|
1043 | 1043 | n = self.changelog.add(mn, changed + removed, ctx.description(), |
|
1044 | 1044 | trp, p1.node(), p2.node(), |
|
1045 | 1045 | user, ctx.date(), ctx.extra().copy()) |
|
1046 | 1046 | p = lambda: self.changelog.writepending() and self.root or "" |
|
1047 | 1047 | xp1, xp2 = p1.hex(), p2 and p2.hex() or '' |
|
1048 | 1048 | self.hook('pretxncommit', throw=True, node=hex(n), parent1=xp1, |
|
1049 | 1049 | parent2=xp2, pending=p) |
|
1050 | 1050 | self.changelog.finalize(trp) |
|
1051 | 1051 | tr.close() |
|
1052 | 1052 | |
|
1053 | 1053 | if self._branchcache: |
|
1054 | 1054 | self.updatebranchcache() |
|
1055 | 1055 | return n |
|
1056 | 1056 | finally: |
|
1057 | 1057 | if tr: |
|
1058 | 1058 | tr.release() |
|
1059 | 1059 | lock.release() |
|
1060 | 1060 | |
|
1061 | 1061 | def destroyed(self): |
|
1062 | 1062 | '''Inform the repository that nodes have been destroyed. |
|
1063 | 1063 | Intended for use by strip and rollback, so there's a common |
|
1064 | 1064 | place for anything that has to be done after destroying history.''' |
|
1065 | 1065 | # XXX it might be nice if we could take the list of destroyed |
|
1066 | 1066 | # nodes, but I don't see an easy way for rollback() to do that |
|
1067 | 1067 | |
|
1068 | 1068 | # Ensure the persistent tag cache is updated. Doing it now |
|
1069 | 1069 | # means that the tag cache only has to worry about destroyed |
|
1070 | 1070 | # heads immediately after a strip/rollback. That in turn |
|
1071 | 1071 | # guarantees that "cachetip == currenttip" (comparing both rev |
|
1072 | 1072 | # and node) always means no nodes have been added or destroyed. |
|
1073 | 1073 | |
|
1074 | 1074 | # XXX this is suboptimal when qrefresh'ing: we strip the current |
|
1075 | 1075 | # head, refresh the tag cache, then immediately add a new head. |
|
1076 | 1076 | # But I think doing it this way is necessary for the "instant |
|
1077 | 1077 | # tag cache retrieval" case to work. |
|
1078 | 1078 | self.invalidatecaches() |
|
1079 | 1079 | |
|
1080 | 1080 | def walk(self, match, node=None): |
|
1081 | 1081 | ''' |
|
1082 | 1082 | walk recursively through the directory tree or a given |
|
1083 | 1083 | changeset, finding all files matched by the match |
|
1084 | 1084 | function |
|
1085 | 1085 | ''' |
|
1086 | 1086 | return self[node].walk(match) |
|
1087 | 1087 | |
|
1088 | 1088 | def status(self, node1='.', node2=None, match=None, |
|
1089 | 1089 | ignored=False, clean=False, unknown=False, |
|
1090 | 1090 | listsubrepos=False): |
|
1091 | 1091 | """return status of files between two nodes or node and working directory |
|
1092 | 1092 | |
|
1093 | 1093 | If node1 is None, use the first dirstate parent instead. |
|
1094 | 1094 | If node2 is None, compare node1 with working directory. |
|
1095 | 1095 | """ |
|
1096 | 1096 | |
|
1097 | 1097 | def mfmatches(ctx): |
|
1098 | 1098 | mf = ctx.manifest().copy() |
|
1099 | 1099 | for fn in mf.keys(): |
|
1100 | 1100 | if not match(fn): |
|
1101 | 1101 | del mf[fn] |
|
1102 | 1102 | return mf |
|
1103 | 1103 | |
|
1104 | 1104 | if isinstance(node1, context.changectx): |
|
1105 | 1105 | ctx1 = node1 |
|
1106 | 1106 | else: |
|
1107 | 1107 | ctx1 = self[node1] |
|
1108 | 1108 | if isinstance(node2, context.changectx): |
|
1109 | 1109 | ctx2 = node2 |
|
1110 | 1110 | else: |
|
1111 | 1111 | ctx2 = self[node2] |
|
1112 | 1112 | |
|
1113 | 1113 | working = ctx2.rev() is None |
|
1114 | 1114 | parentworking = working and ctx1 == self['.'] |
|
1115 | 1115 | match = match or matchmod.always(self.root, self.getcwd()) |
|
1116 | 1116 | listignored, listclean, listunknown = ignored, clean, unknown |
|
1117 | 1117 | |
|
1118 | 1118 | # load earliest manifest first for caching reasons |
|
1119 | 1119 | if not working and ctx2.rev() < ctx1.rev(): |
|
1120 | 1120 | ctx2.manifest() |
|
1121 | 1121 | |
|
1122 | 1122 | if not parentworking: |
|
1123 | 1123 | def bad(f, msg): |
|
1124 | 1124 | if f not in ctx1: |
|
1125 | 1125 | self.ui.warn('%s: %s\n' % (self.dirstate.pathto(f), msg)) |
|
1126 | 1126 | match.bad = bad |
|
1127 | 1127 | |
|
1128 | 1128 | if working: # we need to scan the working dir |
|
1129 | 1129 | subrepos = [] |
|
1130 | 1130 | if '.hgsub' in self.dirstate: |
|
1131 | 1131 | subrepos = ctx1.substate.keys() |
|
1132 | 1132 | s = self.dirstate.status(match, subrepos, listignored, |
|
1133 | 1133 | listclean, listunknown) |
|
1134 | 1134 | cmp, modified, added, removed, deleted, unknown, ignored, clean = s |
|
1135 | 1135 | |
|
1136 | 1136 | # check for any possibly clean files |
|
1137 | 1137 | if parentworking and cmp: |
|
1138 | 1138 | fixup = [] |
|
1139 | 1139 | # do a full compare of any files that might have changed |
|
1140 | 1140 | for f in sorted(cmp): |
|
1141 | 1141 | if (f not in ctx1 or ctx2.flags(f) != ctx1.flags(f) |
|
1142 | 1142 | or ctx1[f].cmp(ctx2[f])): |
|
1143 | 1143 | modified.append(f) |
|
1144 | 1144 | else: |
|
1145 | 1145 | fixup.append(f) |
|
1146 | 1146 | |
|
1147 | 1147 | # update dirstate for files that are actually clean |
|
1148 | 1148 | if fixup: |
|
1149 | 1149 | if listclean: |
|
1150 | 1150 | clean += fixup |
|
1151 | 1151 | |
|
1152 | 1152 | try: |
|
1153 | 1153 | # updating the dirstate is optional |
|
1154 | 1154 | # so we don't wait on the lock |
|
1155 | 1155 | wlock = self.wlock(False) |
|
1156 | 1156 | try: |
|
1157 | 1157 | for f in fixup: |
|
1158 | 1158 | self.dirstate.normal(f) |
|
1159 | 1159 | finally: |
|
1160 | 1160 | wlock.release() |
|
1161 | 1161 | except error.LockError: |
|
1162 | 1162 | pass |
|
1163 | 1163 | |
|
1164 | 1164 | if not parentworking: |
|
1165 | 1165 | mf1 = mfmatches(ctx1) |
|
1166 | 1166 | if working: |
|
1167 | 1167 | # we are comparing working dir against non-parent |
|
1168 | 1168 | # generate a pseudo-manifest for the working dir |
|
1169 | 1169 | mf2 = mfmatches(self['.']) |
|
1170 | 1170 | for f in cmp + modified + added: |
|
1171 | 1171 | mf2[f] = None |
|
1172 | 1172 | mf2.set(f, ctx2.flags(f)) |
|
1173 | 1173 | for f in removed: |
|
1174 | 1174 | if f in mf2: |
|
1175 | 1175 | del mf2[f] |
|
1176 | 1176 | else: |
|
1177 | 1177 | # we are comparing two revisions |
|
1178 | 1178 | deleted, unknown, ignored = [], [], [] |
|
1179 | 1179 | mf2 = mfmatches(ctx2) |
|
1180 | 1180 | |
|
1181 | 1181 | modified, added, clean = [], [], [] |
|
1182 | 1182 | for fn in mf2: |
|
1183 | 1183 | if fn in mf1: |
|
1184 | 1184 | if (mf1.flags(fn) != mf2.flags(fn) or |
|
1185 | 1185 | (mf1[fn] != mf2[fn] and |
|
1186 | 1186 | (mf2[fn] or ctx1[fn].cmp(ctx2[fn])))): |
|
1187 | 1187 | modified.append(fn) |
|
1188 | 1188 | elif listclean: |
|
1189 | 1189 | clean.append(fn) |
|
1190 | 1190 | del mf1[fn] |
|
1191 | 1191 | else: |
|
1192 | 1192 | added.append(fn) |
|
1193 | 1193 | removed = mf1.keys() |
|
1194 | 1194 | |
|
1195 | 1195 | r = modified, added, removed, deleted, unknown, ignored, clean |
|
1196 | 1196 | |
|
1197 | 1197 | if listsubrepos: |
|
1198 | 1198 | for subpath, sub in subrepo.itersubrepos(ctx1, ctx2): |
|
1199 | 1199 | if working: |
|
1200 | 1200 | rev2 = None |
|
1201 | 1201 | else: |
|
1202 | 1202 | rev2 = ctx2.substate[subpath][1] |
|
1203 | 1203 | try: |
|
1204 | 1204 | submatch = matchmod.narrowmatcher(subpath, match) |
|
1205 | 1205 | s = sub.status(rev2, match=submatch, ignored=listignored, |
|
1206 | 1206 | clean=listclean, unknown=listunknown, |
|
1207 | 1207 | listsubrepos=True) |
|
1208 | 1208 | for rfiles, sfiles in zip(r, s): |
|
1209 | 1209 | rfiles.extend("%s/%s" % (subpath, f) for f in sfiles) |
|
1210 | 1210 | except error.LookupError: |
|
1211 | 1211 | self.ui.status(_("skipping missing subrepository: %s\n") |
|
1212 | 1212 | % subpath) |
|
1213 | 1213 | |
|
1214 | 1214 | [l.sort() for l in r] |
|
1215 | 1215 | return r |
|
1216 | 1216 | |
|
1217 | 1217 | def heads(self, start=None): |
|
1218 | 1218 | heads = self.changelog.heads(start) |
|
1219 | 1219 | # sort the output in rev descending order |
|
1220 | 1220 | return sorted(heads, key=self.changelog.rev, reverse=True) |
|
1221 | 1221 | |
|
1222 | 1222 | def branchheads(self, branch=None, start=None, closed=False): |
|
1223 | 1223 | '''return a (possibly filtered) list of heads for the given branch |
|
1224 | 1224 | |
|
1225 | 1225 | Heads are returned in topological order, from newest to oldest. |
|
1226 | 1226 | If branch is None, use the dirstate branch. |
|
1227 | 1227 | If start is not None, return only heads reachable from start. |
|
1228 | 1228 | If closed is True, return heads that are marked as closed as well. |
|
1229 | 1229 | ''' |
|
1230 | 1230 | if branch is None: |
|
1231 | 1231 | branch = self[None].branch() |
|
1232 | 1232 | branches = self.branchmap() |
|
1233 | 1233 | if branch not in branches: |
|
1234 | 1234 | return [] |
|
1235 | 1235 | # the cache returns heads ordered lowest to highest |
|
1236 | 1236 | bheads = list(reversed(branches[branch])) |
|
1237 | 1237 | if start is not None: |
|
1238 | 1238 | # filter out the heads that cannot be reached from startrev |
|
1239 | 1239 | fbheads = set(self.changelog.nodesbetween([start], bheads)[2]) |
|
1240 | 1240 | bheads = [h for h in bheads if h in fbheads] |
|
1241 | 1241 | if not closed: |
|
1242 | 1242 | bheads = [h for h in bheads if |
|
1243 | 1243 | ('close' not in self.changelog.read(h)[5])] |
|
1244 | 1244 | return bheads |
|
1245 | 1245 | |
|
1246 | 1246 | def branches(self, nodes): |
|
1247 | 1247 | if not nodes: |
|
1248 | 1248 | nodes = [self.changelog.tip()] |
|
1249 | 1249 | b = [] |
|
1250 | 1250 | for n in nodes: |
|
1251 | 1251 | t = n |
|
1252 | 1252 | while 1: |
|
1253 | 1253 | p = self.changelog.parents(n) |
|
1254 | 1254 | if p[1] != nullid or p[0] == nullid: |
|
1255 | 1255 | b.append((t, n, p[0], p[1])) |
|
1256 | 1256 | break |
|
1257 | 1257 | n = p[0] |
|
1258 | 1258 | return b |
|
1259 | 1259 | |
|
1260 | 1260 | def between(self, pairs): |
|
1261 | 1261 | r = [] |
|
1262 | 1262 | |
|
1263 | 1263 | for top, bottom in pairs: |
|
1264 | 1264 | n, l, i = top, [], 0 |
|
1265 | 1265 | f = 1 |
|
1266 | 1266 | |
|
1267 | 1267 | while n != bottom and n != nullid: |
|
1268 | 1268 | p = self.changelog.parents(n)[0] |
|
1269 | 1269 | if i == f: |
|
1270 | 1270 | l.append(n) |
|
1271 | 1271 | f = f * 2 |
|
1272 | 1272 | n = p |
|
1273 | 1273 | i += 1 |
|
1274 | 1274 | |
|
1275 | 1275 | r.append(l) |
|
1276 | 1276 | |
|
1277 | 1277 | return r |
|
1278 | 1278 | |
|
1279 | 1279 | def pull(self, remote, heads=None, force=False): |
|
1280 | 1280 | lock = self.lock() |
|
1281 | 1281 | try: |
|
1282 | 1282 | tmp = discovery.findcommonincoming(self, remote, heads=heads, |
|
1283 | 1283 | force=force) |
|
1284 | 1284 | common, fetch, rheads = tmp |
|
1285 | 1285 | if not fetch: |
|
1286 | 1286 | self.ui.status(_("no changes found\n")) |
|
1287 | 1287 | return 0 |
|
1288 | 1288 | |
|
1289 | 1289 | if heads is None and fetch == [nullid]: |
|
1290 | 1290 | self.ui.status(_("requesting all changes\n")) |
|
1291 | 1291 | elif heads is None and remote.capable('changegroupsubset'): |
|
1292 | 1292 | # issue1320, avoid a race if remote changed after discovery |
|
1293 | 1293 | heads = rheads |
|
1294 | 1294 | |
|
1295 | 1295 | if heads is None: |
|
1296 | 1296 | cg = remote.changegroup(fetch, 'pull') |
|
1297 | 1297 | else: |
|
1298 | 1298 | if not remote.capable('changegroupsubset'): |
|
1299 | 1299 | raise util.Abort(_("partial pull cannot be done because " |
|
1300 | 1300 | "other repository doesn't support " |
|
1301 | 1301 | "changegroupsubset.")) |
|
1302 | 1302 | cg = remote.changegroupsubset(fetch, heads, 'pull') |
|
1303 | 1303 | return self.addchangegroup(cg, 'pull', remote.url(), lock=lock) |
|
1304 | 1304 | finally: |
|
1305 | 1305 | lock.release() |
|
1306 | 1306 | |
|
1307 | 1307 | def push(self, remote, force=False, revs=None, newbranch=False): |
|
1308 | 1308 | '''Push outgoing changesets (limited by revs) from the current |
|
1309 | 1309 | repository to remote. Return an integer: |
|
1310 | 1310 | - 0 means HTTP error *or* nothing to push |
|
1311 | 1311 | - 1 means we pushed and remote head count is unchanged *or* |
|
1312 | 1312 | we have outgoing changesets but refused to push |
|
1313 | 1313 | - other values as described by addchangegroup() |
|
1314 | 1314 | ''' |
|
1315 | 1315 | # there are two ways to push to remote repo: |
|
1316 | 1316 | # |
|
1317 | 1317 | # addchangegroup assumes local user can lock remote |
|
1318 | 1318 | # repo (local filesystem, old ssh servers). |
|
1319 | 1319 | # |
|
1320 | 1320 | # unbundle assumes local user cannot lock remote repo (new ssh |
|
1321 | 1321 | # servers, http servers). |
|
1322 | 1322 | |
|
1323 | 1323 | lock = None |
|
1324 | 1324 | unbundle = remote.capable('unbundle') |
|
1325 | 1325 | if not unbundle: |
|
1326 | 1326 | lock = remote.lock() |
|
1327 | 1327 | try: |
|
1328 | 1328 | ret = discovery.prepush(self, remote, force, revs, newbranch) |
|
1329 | 1329 | if ret[0] is None: |
|
1330 | 1330 | # and here we return 0 for "nothing to push" or 1 for |
|
1331 | 1331 | # "something to push but I refuse" |
|
1332 | 1332 | return ret[1] |
|
1333 | 1333 | |
|
1334 | 1334 | cg, remote_heads = ret |
|
1335 | 1335 | if unbundle: |
|
1336 | 1336 | # local repo finds heads on server, finds out what revs it must |
|
1337 | 1337 | # push. once revs transferred, if server finds it has |
|
1338 | 1338 | # different heads (someone else won commit/push race), server |
|
1339 | 1339 | # aborts. |
|
1340 | 1340 | if force: |
|
1341 | 1341 | remote_heads = ['force'] |
|
1342 | 1342 | # ssh: return remote's addchangegroup() |
|
1343 | 1343 | # http: return remote's addchangegroup() or 0 for error |
|
1344 | 1344 | return remote.unbundle(cg, remote_heads, 'push') |
|
1345 | 1345 | else: |
|
1346 | 1346 | # we return an integer indicating remote head count change |
|
1347 | 1347 | return remote.addchangegroup(cg, 'push', self.url(), lock=lock) |
|
1348 | 1348 | finally: |
|
1349 | 1349 | if lock is not None: |
|
1350 | 1350 | lock.release() |
|
1351 | 1351 | |
|
1352 | 1352 | def changegroupinfo(self, nodes, source): |
|
1353 | 1353 | if self.ui.verbose or source == 'bundle': |
|
1354 | 1354 | self.ui.status(_("%d changesets found\n") % len(nodes)) |
|
1355 | 1355 | if self.ui.debugflag: |
|
1356 | 1356 | self.ui.debug("list of changesets:\n") |
|
1357 | 1357 | for node in nodes: |
|
1358 | 1358 | self.ui.debug("%s\n" % hex(node)) |
|
1359 | 1359 | |
|
1360 | 1360 | def changegroupsubset(self, bases, heads, source, extranodes=None): |
|
1361 | 1361 | """Compute a changegroup consisting of all the nodes that are |
|
1362 | 1362 | descendents of any of the bases and ancestors of any of the heads. |
|
1363 | 1363 | Return a chunkbuffer object whose read() method will return |
|
1364 | 1364 | successive changegroup chunks. |
|
1365 | 1365 | |
|
1366 | 1366 | It is fairly complex as determining which filenodes and which |
|
1367 | 1367 | manifest nodes need to be included for the changeset to be complete |
|
1368 | 1368 | is non-trivial. |
|
1369 | 1369 | |
|
1370 | 1370 | Another wrinkle is doing the reverse, figuring out which changeset in |
|
1371 | 1371 | the changegroup a particular filenode or manifestnode belongs to. |
|
1372 | 1372 | |
|
1373 | 1373 | The caller can specify some nodes that must be included in the |
|
1374 | 1374 | changegroup using the extranodes argument. It should be a dict |
|
1375 | 1375 | where the keys are the filenames (or 1 for the manifest), and the |
|
1376 | 1376 | values are lists of (node, linknode) tuples, where node is a wanted |
|
1377 | 1377 | node and linknode is the changelog node that should be transmitted as |
|
1378 | 1378 | the linkrev. |
|
1379 | 1379 | """ |
|
1380 | 1380 | |
|
1381 | 1381 | # Set up some initial variables |
|
1382 | 1382 | # Make it easy to refer to self.changelog |
|
1383 | 1383 | cl = self.changelog |
|
1384 | 1384 | # Compute the list of changesets in this changegroup. |
|
1385 | 1385 | # Some bases may turn out to be superfluous, and some heads may be |
|
1386 | 1386 | # too. nodesbetween will return the minimal set of bases and heads |
|
1387 | 1387 | # necessary to re-create the changegroup. |
|
1388 | 1388 | if not bases: |
|
1389 | 1389 | bases = [nullid] |
|
1390 | 1390 | msng_cl_lst, bases, heads = cl.nodesbetween(bases, heads) |
|
1391 | 1391 | |
|
1392 | 1392 | if extranodes is None: |
|
1393 | 1393 | # can we go through the fast path ? |
|
1394 | 1394 | heads.sort() |
|
1395 | 1395 | allheads = self.heads() |
|
1396 | 1396 | allheads.sort() |
|
1397 | 1397 | if heads == allheads: |
|
1398 | 1398 | return self._changegroup(msng_cl_lst, source) |
|
1399 | 1399 | |
|
1400 | 1400 | # slow path |
|
1401 | 1401 | self.hook('preoutgoing', throw=True, source=source) |
|
1402 | 1402 | |
|
1403 | 1403 | self.changegroupinfo(msng_cl_lst, source) |
|
1404 | 1404 | |
|
1405 | 1405 | # We assume that all ancestors of bases are known |
|
1406 | 1406 | commonrevs = set(cl.ancestors(*[cl.rev(n) for n in bases])) |
|
1407 | 1407 | |
|
1408 | 1408 | # Make it easy to refer to self.manifest |
|
1409 | 1409 | mnfst = self.manifest |
|
1410 | 1410 | # We don't know which manifests are missing yet |
|
1411 | 1411 | msng_mnfst_set = {} |
|
1412 | 1412 | # Nor do we know which filenodes are missing. |
|
1413 | 1413 | msng_filenode_set = {} |
|
1414 | 1414 | |
|
1415 | junk = mnfst.index[len(mnfst) - 1] # Get around a bug in lazyindex | |
|
1416 | junk = None | |
|
1417 | ||
|
1418 | 1415 | # A changeset always belongs to itself, so the changenode lookup |
|
1419 | 1416 | # function for a changenode is identity. |
|
1420 | 1417 | def identity(x): |
|
1421 | 1418 | return x |
|
1422 | 1419 | |
|
1423 | 1420 | # A function generating function that sets up the initial environment |
|
1424 | 1421 | # the inner function. |
|
1425 | 1422 | def filenode_collector(changedfiles): |
|
1426 | 1423 | # This gathers information from each manifestnode included in the |
|
1427 | 1424 | # changegroup about which filenodes the manifest node references |
|
1428 | 1425 | # so we can include those in the changegroup too. |
|
1429 | 1426 | # |
|
1430 | 1427 | # It also remembers which changenode each filenode belongs to. It |
|
1431 | 1428 | # does this by assuming the a filenode belongs to the changenode |
|
1432 | 1429 | # the first manifest that references it belongs to. |
|
1433 | 1430 | def collect_msng_filenodes(mnfstnode): |
|
1434 | 1431 | r = mnfst.rev(mnfstnode) |
|
1435 | 1432 | if mnfst.deltaparent(r) in mnfst.parentrevs(r): |
|
1436 | 1433 | # If the previous rev is one of the parents, |
|
1437 | 1434 | # we only need to see a diff. |
|
1438 | 1435 | deltamf = mnfst.readdelta(mnfstnode) |
|
1439 | 1436 | # For each line in the delta |
|
1440 | 1437 | for f, fnode in deltamf.iteritems(): |
|
1441 | 1438 | # And if the file is in the list of files we care |
|
1442 | 1439 | # about. |
|
1443 | 1440 | if f in changedfiles: |
|
1444 | 1441 | # Get the changenode this manifest belongs to |
|
1445 | 1442 | clnode = msng_mnfst_set[mnfstnode] |
|
1446 | 1443 | # Create the set of filenodes for the file if |
|
1447 | 1444 | # there isn't one already. |
|
1448 | 1445 | ndset = msng_filenode_set.setdefault(f, {}) |
|
1449 | 1446 | # And set the filenode's changelog node to the |
|
1450 | 1447 | # manifest's if it hasn't been set already. |
|
1451 | 1448 | ndset.setdefault(fnode, clnode) |
|
1452 | 1449 | else: |
|
1453 | 1450 | # Otherwise we need a full manifest. |
|
1454 | 1451 | m = mnfst.read(mnfstnode) |
|
1455 | 1452 | # For every file in we care about. |
|
1456 | 1453 | for f in changedfiles: |
|
1457 | 1454 | fnode = m.get(f, None) |
|
1458 | 1455 | # If it's in the manifest |
|
1459 | 1456 | if fnode is not None: |
|
1460 | 1457 | # See comments above. |
|
1461 | 1458 | clnode = msng_mnfst_set[mnfstnode] |
|
1462 | 1459 | ndset = msng_filenode_set.setdefault(f, {}) |
|
1463 | 1460 | ndset.setdefault(fnode, clnode) |
|
1464 | 1461 | return collect_msng_filenodes |
|
1465 | 1462 | |
|
1466 | 1463 | # If we determine that a particular file or manifest node must be a |
|
1467 | 1464 | # node that the recipient of the changegroup will already have, we can |
|
1468 | 1465 | # also assume the recipient will have all the parents. This function |
|
1469 | 1466 | # prunes them from the set of missing nodes. |
|
1470 | 1467 | def prune(revlog, missingnodes): |
|
1471 | 1468 | hasset = set() |
|
1472 | 1469 | # If a 'missing' filenode thinks it belongs to a changenode we |
|
1473 | 1470 | # assume the recipient must have, then the recipient must have |
|
1474 | 1471 | # that filenode. |
|
1475 | 1472 | for n in missingnodes: |
|
1476 | 1473 | clrev = revlog.linkrev(revlog.rev(n)) |
|
1477 | 1474 | if clrev in commonrevs: |
|
1478 | 1475 | hasset.add(n) |
|
1479 | 1476 | for n in hasset: |
|
1480 | 1477 | missingnodes.pop(n, None) |
|
1481 | 1478 | for r in revlog.ancestors(*[revlog.rev(n) for n in hasset]): |
|
1482 | 1479 | missingnodes.pop(revlog.node(r), None) |
|
1483 | 1480 | |
|
1484 | 1481 | # Add the nodes that were explicitly requested. |
|
1485 | 1482 | def add_extra_nodes(name, nodes): |
|
1486 | 1483 | if not extranodes or name not in extranodes: |
|
1487 | 1484 | return |
|
1488 | 1485 | |
|
1489 | 1486 | for node, linknode in extranodes[name]: |
|
1490 | 1487 | if node not in nodes: |
|
1491 | 1488 | nodes[node] = linknode |
|
1492 | 1489 | |
|
1493 | 1490 | # Now that we have all theses utility functions to help out and |
|
1494 | 1491 | # logically divide up the task, generate the group. |
|
1495 | 1492 | def gengroup(): |
|
1496 | 1493 | # The set of changed files starts empty. |
|
1497 | 1494 | changedfiles = set() |
|
1498 | 1495 | collect = changegroup.collector(cl, msng_mnfst_set, changedfiles) |
|
1499 | 1496 | |
|
1500 | 1497 | # Create a changenode group generator that will call our functions |
|
1501 | 1498 | # back to lookup the owning changenode and collect information. |
|
1502 | 1499 | group = cl.group(msng_cl_lst, identity, collect) |
|
1503 | 1500 | for cnt, chnk in enumerate(group): |
|
1504 | 1501 | yield chnk |
|
1505 | 1502 | # revlog.group yields three entries per node, so |
|
1506 | 1503 | # dividing by 3 gives an approximation of how many |
|
1507 | 1504 | # nodes have been processed. |
|
1508 | 1505 | self.ui.progress(_('bundling'), cnt / 3, |
|
1509 | 1506 | unit=_('changesets')) |
|
1510 | 1507 | changecount = cnt / 3 |
|
1511 | 1508 | self.ui.progress(_('bundling'), None) |
|
1512 | 1509 | |
|
1513 | 1510 | prune(mnfst, msng_mnfst_set) |
|
1514 | 1511 | add_extra_nodes(1, msng_mnfst_set) |
|
1515 | 1512 | msng_mnfst_lst = msng_mnfst_set.keys() |
|
1516 | 1513 | # Sort the manifestnodes by revision number. |
|
1517 | 1514 | msng_mnfst_lst.sort(key=mnfst.rev) |
|
1518 | 1515 | # Create a generator for the manifestnodes that calls our lookup |
|
1519 | 1516 | # and data collection functions back. |
|
1520 | 1517 | group = mnfst.group(msng_mnfst_lst, |
|
1521 | 1518 | lambda mnode: msng_mnfst_set[mnode], |
|
1522 | 1519 | filenode_collector(changedfiles)) |
|
1523 | 1520 | efiles = {} |
|
1524 | 1521 | for cnt, chnk in enumerate(group): |
|
1525 | 1522 | if cnt % 3 == 1: |
|
1526 | 1523 | mnode = chnk[:20] |
|
1527 | 1524 | efiles.update(mnfst.readdelta(mnode)) |
|
1528 | 1525 | yield chnk |
|
1529 | 1526 | # see above comment for why we divide by 3 |
|
1530 | 1527 | self.ui.progress(_('bundling'), cnt / 3, |
|
1531 | 1528 | unit=_('manifests'), total=changecount) |
|
1532 | 1529 | self.ui.progress(_('bundling'), None) |
|
1533 | 1530 | efiles = len(efiles) |
|
1534 | 1531 | |
|
1535 | 1532 | # These are no longer needed, dereference and toss the memory for |
|
1536 | 1533 | # them. |
|
1537 | 1534 | msng_mnfst_lst = None |
|
1538 | 1535 | msng_mnfst_set.clear() |
|
1539 | 1536 | |
|
1540 | 1537 | if extranodes: |
|
1541 | 1538 | for fname in extranodes: |
|
1542 | 1539 | if isinstance(fname, int): |
|
1543 | 1540 | continue |
|
1544 | 1541 | msng_filenode_set.setdefault(fname, {}) |
|
1545 | 1542 | changedfiles.add(fname) |
|
1546 | 1543 | # Go through all our files in order sorted by name. |
|
1547 | 1544 | for idx, fname in enumerate(sorted(changedfiles)): |
|
1548 | 1545 | filerevlog = self.file(fname) |
|
1549 | 1546 | if not len(filerevlog): |
|
1550 | 1547 | raise util.Abort(_("empty or missing revlog for %s") % fname) |
|
1551 | 1548 | # Toss out the filenodes that the recipient isn't really |
|
1552 | 1549 | # missing. |
|
1553 | 1550 | missingfnodes = msng_filenode_set.pop(fname, {}) |
|
1554 | 1551 | prune(filerevlog, missingfnodes) |
|
1555 | 1552 | add_extra_nodes(fname, missingfnodes) |
|
1556 | 1553 | # If any filenodes are left, generate the group for them, |
|
1557 | 1554 | # otherwise don't bother. |
|
1558 | 1555 | if missingfnodes: |
|
1559 | 1556 | yield changegroup.chunkheader(len(fname)) |
|
1560 | 1557 | yield fname |
|
1561 | 1558 | # Sort the filenodes by their revision # (topological order) |
|
1562 | 1559 | nodeiter = list(missingfnodes) |
|
1563 | 1560 | nodeiter.sort(key=filerevlog.rev) |
|
1564 | 1561 | # Create a group generator and only pass in a changenode |
|
1565 | 1562 | # lookup function as we need to collect no information |
|
1566 | 1563 | # from filenodes. |
|
1567 | 1564 | group = filerevlog.group(nodeiter, |
|
1568 | 1565 | lambda fnode: missingfnodes[fnode]) |
|
1569 | 1566 | for chnk in group: |
|
1570 | 1567 | # even though we print the same progress on |
|
1571 | 1568 | # most loop iterations, put the progress call |
|
1572 | 1569 | # here so that time estimates (if any) can be updated |
|
1573 | 1570 | self.ui.progress( |
|
1574 | 1571 | _('bundling'), idx, item=fname, |
|
1575 | 1572 | unit=_('files'), total=efiles) |
|
1576 | 1573 | yield chnk |
|
1577 | 1574 | # Signal that no more groups are left. |
|
1578 | 1575 | yield changegroup.closechunk() |
|
1579 | 1576 | self.ui.progress(_('bundling'), None) |
|
1580 | 1577 | |
|
1581 | 1578 | if msng_cl_lst: |
|
1582 | 1579 | self.hook('outgoing', node=hex(msng_cl_lst[0]), source=source) |
|
1583 | 1580 | |
|
1584 | 1581 | return changegroup.unbundle10(util.chunkbuffer(gengroup()), 'UN') |
|
1585 | 1582 | |
|
1586 | 1583 | def changegroup(self, basenodes, source): |
|
1587 | 1584 | # to avoid a race we use changegroupsubset() (issue1320) |
|
1588 | 1585 | return self.changegroupsubset(basenodes, self.heads(), source) |
|
1589 | 1586 | |
|
1590 | 1587 | def _changegroup(self, nodes, source): |
|
1591 | 1588 | """Compute the changegroup of all nodes that we have that a recipient |
|
1592 | 1589 | doesn't. Return a chunkbuffer object whose read() method will return |
|
1593 | 1590 | successive changegroup chunks. |
|
1594 | 1591 | |
|
1595 | 1592 | This is much easier than the previous function as we can assume that |
|
1596 | 1593 | the recipient has any changenode we aren't sending them. |
|
1597 | 1594 | |
|
1598 | 1595 | nodes is the set of nodes to send""" |
|
1599 | 1596 | |
|
1600 | 1597 | self.hook('preoutgoing', throw=True, source=source) |
|
1601 | 1598 | |
|
1602 | 1599 | cl = self.changelog |
|
1603 | 1600 | revset = set([cl.rev(n) for n in nodes]) |
|
1604 | 1601 | self.changegroupinfo(nodes, source) |
|
1605 | 1602 | |
|
1606 | 1603 | def identity(x): |
|
1607 | 1604 | return x |
|
1608 | 1605 | |
|
1609 | 1606 | def gennodelst(log): |
|
1610 | 1607 | for r in log: |
|
1611 | 1608 | if log.linkrev(r) in revset: |
|
1612 | 1609 | yield log.node(r) |
|
1613 | 1610 | |
|
1614 | 1611 | def lookuplinkrev_func(revlog): |
|
1615 | 1612 | def lookuplinkrev(n): |
|
1616 | 1613 | return cl.node(revlog.linkrev(revlog.rev(n))) |
|
1617 | 1614 | return lookuplinkrev |
|
1618 | 1615 | |
|
1619 | 1616 | def gengroup(): |
|
1620 | 1617 | '''yield a sequence of changegroup chunks (strings)''' |
|
1621 | 1618 | # construct a list of all changed files |
|
1622 | 1619 | changedfiles = set() |
|
1623 | 1620 | mmfs = {} |
|
1624 | 1621 | collect = changegroup.collector(cl, mmfs, changedfiles) |
|
1625 | 1622 | |
|
1626 | 1623 | for cnt, chnk in enumerate(cl.group(nodes, identity, collect)): |
|
1627 | 1624 | # revlog.group yields three entries per node, so |
|
1628 | 1625 | # dividing by 3 gives an approximation of how many |
|
1629 | 1626 | # nodes have been processed. |
|
1630 | 1627 | self.ui.progress(_('bundling'), cnt / 3, unit=_('changesets')) |
|
1631 | 1628 | yield chnk |
|
1632 | 1629 | changecount = cnt / 3 |
|
1633 | 1630 | self.ui.progress(_('bundling'), None) |
|
1634 | 1631 | |
|
1635 | 1632 | mnfst = self.manifest |
|
1636 | 1633 | nodeiter = gennodelst(mnfst) |
|
1637 | 1634 | efiles = {} |
|
1638 | 1635 | for cnt, chnk in enumerate(mnfst.group(nodeiter, |
|
1639 | 1636 | lookuplinkrev_func(mnfst))): |
|
1640 | 1637 | if cnt % 3 == 1: |
|
1641 | 1638 | mnode = chnk[:20] |
|
1642 | 1639 | efiles.update(mnfst.readdelta(mnode)) |
|
1643 | 1640 | # see above comment for why we divide by 3 |
|
1644 | 1641 | self.ui.progress(_('bundling'), cnt / 3, |
|
1645 | 1642 | unit=_('manifests'), total=changecount) |
|
1646 | 1643 | yield chnk |
|
1647 | 1644 | efiles = len(efiles) |
|
1648 | 1645 | self.ui.progress(_('bundling'), None) |
|
1649 | 1646 | |
|
1650 | 1647 | for idx, fname in enumerate(sorted(changedfiles)): |
|
1651 | 1648 | filerevlog = self.file(fname) |
|
1652 | 1649 | if not len(filerevlog): |
|
1653 | 1650 | raise util.Abort(_("empty or missing revlog for %s") % fname) |
|
1654 | 1651 | nodeiter = gennodelst(filerevlog) |
|
1655 | 1652 | nodeiter = list(nodeiter) |
|
1656 | 1653 | if nodeiter: |
|
1657 | 1654 | yield changegroup.chunkheader(len(fname)) |
|
1658 | 1655 | yield fname |
|
1659 | 1656 | lookup = lookuplinkrev_func(filerevlog) |
|
1660 | 1657 | for chnk in filerevlog.group(nodeiter, lookup): |
|
1661 | 1658 | self.ui.progress( |
|
1662 | 1659 | _('bundling'), idx, item=fname, |
|
1663 | 1660 | total=efiles, unit=_('files')) |
|
1664 | 1661 | yield chnk |
|
1665 | 1662 | self.ui.progress(_('bundling'), None) |
|
1666 | 1663 | |
|
1667 | 1664 | yield changegroup.closechunk() |
|
1668 | 1665 | |
|
1669 | 1666 | if nodes: |
|
1670 | 1667 | self.hook('outgoing', node=hex(nodes[0]), source=source) |
|
1671 | 1668 | |
|
1672 | 1669 | return changegroup.unbundle10(util.chunkbuffer(gengroup()), 'UN') |
|
1673 | 1670 | |
|
1674 | 1671 | def addchangegroup(self, source, srctype, url, emptyok=False, lock=None): |
|
1675 | 1672 | """Add the changegroup returned by source.read() to this repo. |
|
1676 | 1673 | srctype is a string like 'push', 'pull', or 'unbundle'. url is |
|
1677 | 1674 | the URL of the repo where this changegroup is coming from. |
|
1678 | 1675 | |
|
1679 | 1676 | Return an integer summarizing the change to this repo: |
|
1680 | 1677 | - nothing changed or no source: 0 |
|
1681 | 1678 | - more heads than before: 1+added heads (2..n) |
|
1682 | 1679 | - fewer heads than before: -1-removed heads (-2..-n) |
|
1683 | 1680 | - number of heads stays the same: 1 |
|
1684 | 1681 | """ |
|
1685 | 1682 | def csmap(x): |
|
1686 | 1683 | self.ui.debug("add changeset %s\n" % short(x)) |
|
1687 | 1684 | return len(cl) |
|
1688 | 1685 | |
|
1689 | 1686 | def revmap(x): |
|
1690 | 1687 | return cl.rev(x) |
|
1691 | 1688 | |
|
1692 | 1689 | if not source: |
|
1693 | 1690 | return 0 |
|
1694 | 1691 | |
|
1695 | 1692 | self.hook('prechangegroup', throw=True, source=srctype, url=url) |
|
1696 | 1693 | |
|
1697 | 1694 | changesets = files = revisions = 0 |
|
1698 | 1695 | efiles = set() |
|
1699 | 1696 | |
|
1700 | 1697 | # write changelog data to temp files so concurrent readers will not see |
|
1701 | 1698 | # inconsistent view |
|
1702 | 1699 | cl = self.changelog |
|
1703 | 1700 | cl.delayupdate() |
|
1704 | 1701 | oldheads = len(cl.heads()) |
|
1705 | 1702 | |
|
1706 | 1703 | tr = self.transaction("\n".join([srctype, urlmod.hidepassword(url)])) |
|
1707 | 1704 | try: |
|
1708 | 1705 | trp = weakref.proxy(tr) |
|
1709 | 1706 | # pull off the changeset group |
|
1710 | 1707 | self.ui.status(_("adding changesets\n")) |
|
1711 | 1708 | clstart = len(cl) |
|
1712 | 1709 | class prog(object): |
|
1713 | 1710 | step = _('changesets') |
|
1714 | 1711 | count = 1 |
|
1715 | 1712 | ui = self.ui |
|
1716 | 1713 | total = None |
|
1717 | 1714 | def __call__(self): |
|
1718 | 1715 | self.ui.progress(self.step, self.count, unit=_('chunks'), |
|
1719 | 1716 | total=self.total) |
|
1720 | 1717 | self.count += 1 |
|
1721 | 1718 | pr = prog() |
|
1722 | 1719 | source.callback = pr |
|
1723 | 1720 | |
|
1724 | 1721 | if (cl.addgroup(source, csmap, trp) is None |
|
1725 | 1722 | and not emptyok): |
|
1726 | 1723 | raise util.Abort(_("received changelog group is empty")) |
|
1727 | 1724 | clend = len(cl) |
|
1728 | 1725 | changesets = clend - clstart |
|
1729 | 1726 | for c in xrange(clstart, clend): |
|
1730 | 1727 | efiles.update(self[c].files()) |
|
1731 | 1728 | efiles = len(efiles) |
|
1732 | 1729 | self.ui.progress(_('changesets'), None) |
|
1733 | 1730 | |
|
1734 | 1731 | # pull off the manifest group |
|
1735 | 1732 | self.ui.status(_("adding manifests\n")) |
|
1736 | 1733 | pr.step = _('manifests') |
|
1737 | 1734 | pr.count = 1 |
|
1738 | 1735 | pr.total = changesets # manifests <= changesets |
|
1739 | 1736 | # no need to check for empty manifest group here: |
|
1740 | 1737 | # if the result of the merge of 1 and 2 is the same in 3 and 4, |
|
1741 | 1738 | # no new manifest will be created and the manifest group will |
|
1742 | 1739 | # be empty during the pull |
|
1743 | 1740 | self.manifest.addgroup(source, revmap, trp) |
|
1744 | 1741 | self.ui.progress(_('manifests'), None) |
|
1745 | 1742 | |
|
1746 | 1743 | needfiles = {} |
|
1747 | 1744 | if self.ui.configbool('server', 'validate', default=False): |
|
1748 | 1745 | # validate incoming csets have their manifests |
|
1749 | 1746 | for cset in xrange(clstart, clend): |
|
1750 | 1747 | mfest = self.changelog.read(self.changelog.node(cset))[0] |
|
1751 | 1748 | mfest = self.manifest.readdelta(mfest) |
|
1752 | 1749 | # store file nodes we must see |
|
1753 | 1750 | for f, n in mfest.iteritems(): |
|
1754 | 1751 | needfiles.setdefault(f, set()).add(n) |
|
1755 | 1752 | |
|
1756 | 1753 | # process the files |
|
1757 | 1754 | self.ui.status(_("adding file changes\n")) |
|
1758 | 1755 | pr.step = 'files' |
|
1759 | 1756 | pr.count = 1 |
|
1760 | 1757 | pr.total = efiles |
|
1761 | 1758 | source.callback = None |
|
1762 | 1759 | |
|
1763 | 1760 | while 1: |
|
1764 | 1761 | f = source.chunk() |
|
1765 | 1762 | if not f: |
|
1766 | 1763 | break |
|
1767 | 1764 | self.ui.debug("adding %s revisions\n" % f) |
|
1768 | 1765 | pr() |
|
1769 | 1766 | fl = self.file(f) |
|
1770 | 1767 | o = len(fl) |
|
1771 | 1768 | if fl.addgroup(source, revmap, trp) is None: |
|
1772 | 1769 | raise util.Abort(_("received file revlog group is empty")) |
|
1773 | 1770 | revisions += len(fl) - o |
|
1774 | 1771 | files += 1 |
|
1775 | 1772 | if f in needfiles: |
|
1776 | 1773 | needs = needfiles[f] |
|
1777 | 1774 | for new in xrange(o, len(fl)): |
|
1778 | 1775 | n = fl.node(new) |
|
1779 | 1776 | if n in needs: |
|
1780 | 1777 | needs.remove(n) |
|
1781 | 1778 | if not needs: |
|
1782 | 1779 | del needfiles[f] |
|
1783 | 1780 | self.ui.progress(_('files'), None) |
|
1784 | 1781 | |
|
1785 | 1782 | for f, needs in needfiles.iteritems(): |
|
1786 | 1783 | fl = self.file(f) |
|
1787 | 1784 | for n in needs: |
|
1788 | 1785 | try: |
|
1789 | 1786 | fl.rev(n) |
|
1790 | 1787 | except error.LookupError: |
|
1791 | 1788 | raise util.Abort( |
|
1792 | 1789 | _('missing file data for %s:%s - run hg verify') % |
|
1793 | 1790 | (f, hex(n))) |
|
1794 | 1791 | |
|
1795 | 1792 | newheads = len(cl.heads()) |
|
1796 | 1793 | heads = "" |
|
1797 | 1794 | if oldheads and newheads != oldheads: |
|
1798 | 1795 | heads = _(" (%+d heads)") % (newheads - oldheads) |
|
1799 | 1796 | |
|
1800 | 1797 | self.ui.status(_("added %d changesets" |
|
1801 | 1798 | " with %d changes to %d files%s\n") |
|
1802 | 1799 | % (changesets, revisions, files, heads)) |
|
1803 | 1800 | |
|
1804 | 1801 | if changesets > 0: |
|
1805 | 1802 | p = lambda: cl.writepending() and self.root or "" |
|
1806 | 1803 | self.hook('pretxnchangegroup', throw=True, |
|
1807 | 1804 | node=hex(cl.node(clstart)), source=srctype, |
|
1808 | 1805 | url=url, pending=p) |
|
1809 | 1806 | |
|
1810 | 1807 | # make changelog see real files again |
|
1811 | 1808 | cl.finalize(trp) |
|
1812 | 1809 | |
|
1813 | 1810 | tr.close() |
|
1814 | 1811 | finally: |
|
1815 | 1812 | tr.release() |
|
1816 | 1813 | if lock: |
|
1817 | 1814 | lock.release() |
|
1818 | 1815 | |
|
1819 | 1816 | if changesets > 0: |
|
1820 | 1817 | # forcefully update the on-disk branch cache |
|
1821 | 1818 | self.ui.debug("updating the branch cache\n") |
|
1822 | 1819 | self.updatebranchcache() |
|
1823 | 1820 | self.hook("changegroup", node=hex(cl.node(clstart)), |
|
1824 | 1821 | source=srctype, url=url) |
|
1825 | 1822 | |
|
1826 | 1823 | for i in xrange(clstart, clend): |
|
1827 | 1824 | self.hook("incoming", node=hex(cl.node(i)), |
|
1828 | 1825 | source=srctype, url=url) |
|
1829 | 1826 | |
|
1830 | 1827 | # never return 0 here: |
|
1831 | 1828 | if newheads < oldheads: |
|
1832 | 1829 | return newheads - oldheads - 1 |
|
1833 | 1830 | else: |
|
1834 | 1831 | return newheads - oldheads + 1 |
|
1835 | 1832 | |
|
1836 | 1833 | |
|
1837 | 1834 | def stream_in(self, remote, requirements): |
|
1838 | 1835 | fp = remote.stream_out() |
|
1839 | 1836 | l = fp.readline() |
|
1840 | 1837 | try: |
|
1841 | 1838 | resp = int(l) |
|
1842 | 1839 | except ValueError: |
|
1843 | 1840 | raise error.ResponseError( |
|
1844 | 1841 | _('Unexpected response from remote server:'), l) |
|
1845 | 1842 | if resp == 1: |
|
1846 | 1843 | raise util.Abort(_('operation forbidden by server')) |
|
1847 | 1844 | elif resp == 2: |
|
1848 | 1845 | raise util.Abort(_('locking the remote repository failed')) |
|
1849 | 1846 | elif resp != 0: |
|
1850 | 1847 | raise util.Abort(_('the server sent an unknown error code')) |
|
1851 | 1848 | self.ui.status(_('streaming all changes\n')) |
|
1852 | 1849 | l = fp.readline() |
|
1853 | 1850 | try: |
|
1854 | 1851 | total_files, total_bytes = map(int, l.split(' ', 1)) |
|
1855 | 1852 | except (ValueError, TypeError): |
|
1856 | 1853 | raise error.ResponseError( |
|
1857 | 1854 | _('Unexpected response from remote server:'), l) |
|
1858 | 1855 | self.ui.status(_('%d files to transfer, %s of data\n') % |
|
1859 | 1856 | (total_files, util.bytecount(total_bytes))) |
|
1860 | 1857 | start = time.time() |
|
1861 | 1858 | for i in xrange(total_files): |
|
1862 | 1859 | # XXX doesn't support '\n' or '\r' in filenames |
|
1863 | 1860 | l = fp.readline() |
|
1864 | 1861 | try: |
|
1865 | 1862 | name, size = l.split('\0', 1) |
|
1866 | 1863 | size = int(size) |
|
1867 | 1864 | except (ValueError, TypeError): |
|
1868 | 1865 | raise error.ResponseError( |
|
1869 | 1866 | _('Unexpected response from remote server:'), l) |
|
1870 | 1867 | self.ui.debug('adding %s (%s)\n' % (name, util.bytecount(size))) |
|
1871 | 1868 | # for backwards compat, name was partially encoded |
|
1872 | 1869 | ofp = self.sopener(store.decodedir(name), 'w') |
|
1873 | 1870 | for chunk in util.filechunkiter(fp, limit=size): |
|
1874 | 1871 | ofp.write(chunk) |
|
1875 | 1872 | ofp.close() |
|
1876 | 1873 | elapsed = time.time() - start |
|
1877 | 1874 | if elapsed <= 0: |
|
1878 | 1875 | elapsed = 0.001 |
|
1879 | 1876 | self.ui.status(_('transferred %s in %.1f seconds (%s/sec)\n') % |
|
1880 | 1877 | (util.bytecount(total_bytes), elapsed, |
|
1881 | 1878 | util.bytecount(total_bytes / elapsed))) |
|
1882 | 1879 | |
|
1883 | 1880 | # new requirements = old non-format requirements + new format-related |
|
1884 | 1881 | # requirements from the streamed-in repository |
|
1885 | 1882 | requirements.update(set(self.requirements) - self.supportedformats) |
|
1886 | 1883 | self._applyrequirements(requirements) |
|
1887 | 1884 | self._writerequirements() |
|
1888 | 1885 | |
|
1889 | 1886 | self.invalidate() |
|
1890 | 1887 | return len(self.heads()) + 1 |
|
1891 | 1888 | |
|
1892 | 1889 | def clone(self, remote, heads=[], stream=False): |
|
1893 | 1890 | '''clone remote repository. |
|
1894 | 1891 | |
|
1895 | 1892 | keyword arguments: |
|
1896 | 1893 | heads: list of revs to clone (forces use of pull) |
|
1897 | 1894 | stream: use streaming clone if possible''' |
|
1898 | 1895 | |
|
1899 | 1896 | # now, all clients that can request uncompressed clones can |
|
1900 | 1897 | # read repo formats supported by all servers that can serve |
|
1901 | 1898 | # them. |
|
1902 | 1899 | |
|
1903 | 1900 | # if revlog format changes, client will have to check version |
|
1904 | 1901 | # and format flags on "stream" capability, and use |
|
1905 | 1902 | # uncompressed only if compatible. |
|
1906 | 1903 | |
|
1907 | 1904 | if stream and not heads: |
|
1908 | 1905 | # 'stream' means remote revlog format is revlogv1 only |
|
1909 | 1906 | if remote.capable('stream'): |
|
1910 | 1907 | return self.stream_in(remote, set(('revlogv1',))) |
|
1911 | 1908 | # otherwise, 'streamreqs' contains the remote revlog format |
|
1912 | 1909 | streamreqs = remote.capable('streamreqs') |
|
1913 | 1910 | if streamreqs: |
|
1914 | 1911 | streamreqs = set(streamreqs.split(',')) |
|
1915 | 1912 | # if we support it, stream in and adjust our requirements |
|
1916 | 1913 | if not streamreqs - self.supportedformats: |
|
1917 | 1914 | return self.stream_in(remote, streamreqs) |
|
1918 | 1915 | return self.pull(remote, heads) |
|
1919 | 1916 | |
|
1920 | 1917 | def pushkey(self, namespace, key, old, new): |
|
1921 | 1918 | return pushkey.push(self, namespace, key, old, new) |
|
1922 | 1919 | |
|
1923 | 1920 | def listkeys(self, namespace): |
|
1924 | 1921 | return pushkey.list(self, namespace) |
|
1925 | 1922 | |
|
1926 | 1923 | # used to avoid circular references so destructors work |
|
1927 | 1924 | def aftertrans(files): |
|
1928 | 1925 | renamefiles = [tuple(t) for t in files] |
|
1929 | 1926 | def a(): |
|
1930 | 1927 | for src, dest in renamefiles: |
|
1931 | 1928 | util.rename(src, dest) |
|
1932 | 1929 | return a |
|
1933 | 1930 | |
|
1934 | 1931 | def instance(ui, path, create): |
|
1935 | 1932 | return localrepository(ui, util.drop_scheme('file', path), create) |
|
1936 | 1933 | |
|
1937 | 1934 | def islocal(path): |
|
1938 | 1935 | return True |
@@ -1,90 +1,90 b'' | |||
|
1 | 1 | # parsers.py - Python implementation of parsers.c |
|
2 | 2 | # |
|
3 | 3 | # Copyright 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 mercurial.node import bin, nullid, nullrev |
|
9 | 9 | from mercurial import util |
|
10 | 10 | import struct, zlib |
|
11 | 11 | |
|
12 | 12 | _pack = struct.pack |
|
13 | 13 | _unpack = struct.unpack |
|
14 | 14 | _compress = zlib.compress |
|
15 | 15 | _decompress = zlib.decompress |
|
16 | 16 | _sha = util.sha1 |
|
17 | 17 | |
|
18 | 18 | def parse_manifest(mfdict, fdict, lines): |
|
19 | 19 | for l in lines.splitlines(): |
|
20 | 20 | f, n = l.split('\0') |
|
21 | 21 | if len(n) > 40: |
|
22 | 22 | fdict[f] = n[40:] |
|
23 | 23 | mfdict[f] = bin(n[:40]) |
|
24 | 24 | else: |
|
25 | 25 | mfdict[f] = bin(n) |
|
26 | 26 | |
|
27 | 27 | def parse_index(data, inline): |
|
28 | 28 | def gettype(q): |
|
29 | 29 | return int(q & 0xFFFF) |
|
30 | 30 | |
|
31 | 31 | def offset_type(offset, type): |
|
32 | 32 | return long(long(offset) << 16 | type) |
|
33 | 33 | |
|
34 | 34 | indexformatng = ">Qiiiiii20s12x" |
|
35 | 35 | |
|
36 | 36 | s = struct.calcsize(indexformatng) |
|
37 | 37 | index = [] |
|
38 | 38 | cache = None |
|
39 | 39 | nodemap = {nullid: nullrev} |
|
40 | 40 | n = off = 0 |
|
41 | # if we're not using lazymap, always read the whole index | |
|
41 | ||
|
42 | 42 | l = len(data) - s |
|
43 | 43 | append = index.append |
|
44 | 44 | if inline: |
|
45 | 45 | cache = (0, data) |
|
46 | 46 | while off <= l: |
|
47 | 47 | e = _unpack(indexformatng, data[off:off + s]) |
|
48 | 48 | nodemap[e[7]] = n |
|
49 | 49 | append(e) |
|
50 | 50 | n += 1 |
|
51 | 51 | if e[1] < 0: |
|
52 | 52 | break |
|
53 | 53 | off += e[1] + s |
|
54 | 54 | else: |
|
55 | 55 | while off <= l: |
|
56 | 56 | e = _unpack(indexformatng, data[off:off + s]) |
|
57 | 57 | nodemap[e[7]] = n |
|
58 | 58 | append(e) |
|
59 | 59 | n += 1 |
|
60 | 60 | off += s |
|
61 | 61 | |
|
62 | 62 | e = list(index[0]) |
|
63 | 63 | type = gettype(e[0]) |
|
64 | 64 | e[0] = offset_type(0, type) |
|
65 | 65 | index[0] = tuple(e) |
|
66 | 66 | |
|
67 | 67 | # add the magic null revision at -1 |
|
68 | 68 | index.append((0, 0, 0, -1, -1, -1, -1, nullid)) |
|
69 | 69 | |
|
70 | 70 | return index, nodemap, cache |
|
71 | 71 | |
|
72 | 72 | def parse_dirstate(dmap, copymap, st): |
|
73 | 73 | parents = [st[:20], st[20: 40]] |
|
74 | 74 | # deref fields so they will be local in loop |
|
75 | 75 | format = ">cllll" |
|
76 | 76 | e_size = struct.calcsize(format) |
|
77 | 77 | pos1 = 40 |
|
78 | 78 | l = len(st) |
|
79 | 79 | |
|
80 | 80 | # the inner loop |
|
81 | 81 | while pos1 < l: |
|
82 | 82 | pos2 = pos1 + e_size |
|
83 | 83 | e = _unpack(">cllll", st[pos1:pos2]) # a literal here is faster |
|
84 | 84 | pos1 = pos2 + e[4] |
|
85 | 85 | f = st[pos2:pos1] |
|
86 | 86 | if '\0' in f: |
|
87 | 87 | f, c = f.split('\0') |
|
88 | 88 | copymap[f] = c |
|
89 | 89 | dmap[f] = e[:4] |
|
90 | 90 | return parents |
@@ -1,1482 +1,1235 b'' | |||
|
1 | 1 | # revlog.py - storage back-end 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 | """Storage back-end for Mercurial. |
|
9 | 9 | |
|
10 | 10 | This provides efficient delta storage with O(1) retrieve and append |
|
11 | 11 | and O(changes) merge between branches. |
|
12 | 12 | """ |
|
13 | 13 | |
|
14 | 14 | # import stuff from node for others to import from revlog |
|
15 | 15 | from node import bin, hex, nullid, nullrev, short #@UnusedImport |
|
16 | 16 | from i18n import _ |
|
17 | 17 | import changegroup, ancestor, mdiff, parsers, error, util |
|
18 | 18 | import struct, zlib, errno |
|
19 | 19 | |
|
20 | 20 | _pack = struct.pack |
|
21 | 21 | _unpack = struct.unpack |
|
22 | 22 | _compress = zlib.compress |
|
23 | 23 | _decompress = zlib.decompress |
|
24 | 24 | _sha = util.sha1 |
|
25 | 25 | |
|
26 | 26 | # revlog header flags |
|
27 | 27 | REVLOGV0 = 0 |
|
28 | 28 | REVLOGNG = 1 |
|
29 | 29 | REVLOGNGINLINEDATA = (1 << 16) |
|
30 | 30 | REVLOGSHALLOW = (1 << 17) |
|
31 | 31 | REVLOG_DEFAULT_FLAGS = REVLOGNGINLINEDATA |
|
32 | 32 | REVLOG_DEFAULT_FORMAT = REVLOGNG |
|
33 | 33 | REVLOG_DEFAULT_VERSION = REVLOG_DEFAULT_FORMAT | REVLOG_DEFAULT_FLAGS |
|
34 | 34 | REVLOGNG_FLAGS = REVLOGNGINLINEDATA | REVLOGSHALLOW |
|
35 | 35 | |
|
36 | 36 | # revlog index flags |
|
37 | 37 | REVIDX_PARENTDELTA = 1 |
|
38 | 38 | REVIDX_PUNCHED_FLAG = 2 |
|
39 | 39 | REVIDX_KNOWN_FLAGS = REVIDX_PUNCHED_FLAG | REVIDX_PARENTDELTA |
|
40 | 40 | |
|
41 | # amount of data read unconditionally, should be >= 4 | |
|
42 | # when not inline: threshold for using lazy index | |
|
43 | _prereadsize = 1048576 | |
|
44 | 41 | # max size of revlog with inline data |
|
45 | 42 | _maxinline = 131072 |
|
43 | _chunksize = 1048576 | |
|
46 | 44 | |
|
47 | 45 | RevlogError = error.RevlogError |
|
48 | 46 | LookupError = error.LookupError |
|
49 | 47 | |
|
50 | 48 | def getoffset(q): |
|
51 | 49 | return int(q >> 16) |
|
52 | 50 | |
|
53 | 51 | def gettype(q): |
|
54 | 52 | return int(q & 0xFFFF) |
|
55 | 53 | |
|
56 | 54 | def offset_type(offset, type): |
|
57 | 55 | return long(long(offset) << 16 | type) |
|
58 | 56 | |
|
59 | 57 | nullhash = _sha(nullid) |
|
60 | 58 | |
|
61 | 59 | def hash(text, p1, p2): |
|
62 | 60 | """generate a hash from the given text and its parent hashes |
|
63 | 61 | |
|
64 | 62 | This hash combines both the current file contents and its history |
|
65 | 63 | in a manner that makes it easy to distinguish nodes with the same |
|
66 | 64 | content in the revision graph. |
|
67 | 65 | """ |
|
68 | 66 | # As of now, if one of the parent node is null, p2 is null |
|
69 | 67 | if p2 == nullid: |
|
70 | 68 | # deep copy of a hash is faster than creating one |
|
71 | 69 | s = nullhash.copy() |
|
72 | 70 | s.update(p1) |
|
73 | 71 | else: |
|
74 | 72 | # none of the parent nodes are nullid |
|
75 | 73 | l = [p1, p2] |
|
76 | 74 | l.sort() |
|
77 | 75 | s = _sha(l[0]) |
|
78 | 76 | s.update(l[1]) |
|
79 | 77 | s.update(text) |
|
80 | 78 | return s.digest() |
|
81 | 79 | |
|
82 | 80 | def compress(text): |
|
83 | 81 | """ generate a possibly-compressed representation of text """ |
|
84 | 82 | if not text: |
|
85 | 83 | return ("", text) |
|
86 | 84 | l = len(text) |
|
87 | 85 | bin = None |
|
88 | 86 | if l < 44: |
|
89 | 87 | pass |
|
90 | 88 | elif l > 1000000: |
|
91 | 89 | # zlib makes an internal copy, thus doubling memory usage for |
|
92 | 90 | # large files, so lets do this in pieces |
|
93 | 91 | z = zlib.compressobj() |
|
94 | 92 | p = [] |
|
95 | 93 | pos = 0 |
|
96 | 94 | while pos < l: |
|
97 | 95 | pos2 = pos + 2**20 |
|
98 | 96 | p.append(z.compress(text[pos:pos2])) |
|
99 | 97 | pos = pos2 |
|
100 | 98 | p.append(z.flush()) |
|
101 | 99 | if sum(map(len, p)) < l: |
|
102 | 100 | bin = "".join(p) |
|
103 | 101 | else: |
|
104 | 102 | bin = _compress(text) |
|
105 | 103 | if bin is None or len(bin) > l: |
|
106 | 104 | if text[0] == '\0': |
|
107 | 105 | return ("", text) |
|
108 | 106 | return ('u', text) |
|
109 | 107 | return ("", bin) |
|
110 | 108 | |
|
111 | 109 | def decompress(bin): |
|
112 | 110 | """ decompress the given input """ |
|
113 | 111 | if not bin: |
|
114 | 112 | return bin |
|
115 | 113 | t = bin[0] |
|
116 | 114 | if t == '\0': |
|
117 | 115 | return bin |
|
118 | 116 | if t == 'x': |
|
119 | 117 | return _decompress(bin) |
|
120 | 118 | if t == 'u': |
|
121 | 119 | return bin[1:] |
|
122 | 120 | raise RevlogError(_("unknown compression type %r") % t) |
|
123 | 121 | |
|
124 | class lazyparser(object): | |
|
125 | """ | |
|
126 | this class avoids the need to parse the entirety of large indices | |
|
127 | """ | |
|
128 | ||
|
129 | # lazyparser is not safe to use on windows if win32 extensions not | |
|
130 | # available. it keeps file handle open, which make it not possible | |
|
131 | # to break hardlinks on local cloned repos. | |
|
132 | ||
|
133 | def __init__(self, dataf): | |
|
134 | try: | |
|
135 | size = util.fstat(dataf).st_size | |
|
136 | except AttributeError: | |
|
137 | size = 0 | |
|
138 | self.dataf = dataf | |
|
139 | self.s = struct.calcsize(indexformatng) | |
|
140 | self.datasize = size | |
|
141 | self.l = size // self.s | |
|
142 | self.index = [None] * self.l | |
|
143 | self.map = {nullid: nullrev} | |
|
144 | self.allmap = 0 | |
|
145 | self.all = 0 | |
|
146 | self.mapfind_count = 0 | |
|
147 | ||
|
148 | def loadmap(self): | |
|
149 | """ | |
|
150 | during a commit, we need to make sure the rev being added is | |
|
151 | not a duplicate. This requires loading the entire index, | |
|
152 | which is fairly slow. loadmap can load up just the node map, | |
|
153 | which takes much less time. | |
|
154 | """ | |
|
155 | if self.allmap: | |
|
156 | return | |
|
157 | end = self.datasize | |
|
158 | self.allmap = 1 | |
|
159 | cur = 0 | |
|
160 | count = 0 | |
|
161 | blocksize = self.s * 256 | |
|
162 | self.dataf.seek(0) | |
|
163 | while cur < end: | |
|
164 | data = self.dataf.read(blocksize) | |
|
165 | off = 0 | |
|
166 | for x in xrange(256): | |
|
167 | n = data[off + ngshaoffset:off + ngshaoffset + 20] | |
|
168 | self.map[n] = count | |
|
169 | count += 1 | |
|
170 | if count >= self.l: | |
|
171 | break | |
|
172 | off += self.s | |
|
173 | cur += blocksize | |
|
174 | ||
|
175 | def loadblock(self, blockstart, blocksize, data=None): | |
|
176 | if self.all: | |
|
177 | return | |
|
178 | if data is None: | |
|
179 | self.dataf.seek(blockstart) | |
|
180 | if blockstart + blocksize > self.datasize: | |
|
181 | # the revlog may have grown since we've started running, | |
|
182 | # but we don't have space in self.index for more entries. | |
|
183 | # limit blocksize so that we don't get too much data. | |
|
184 | blocksize = max(self.datasize - blockstart, 0) | |
|
185 | data = self.dataf.read(blocksize) | |
|
186 | lend = len(data) // self.s | |
|
187 | i = blockstart // self.s | |
|
188 | off = 0 | |
|
189 | # lazyindex supports __delitem__ | |
|
190 | if lend > len(self.index) - i: | |
|
191 | lend = len(self.index) - i | |
|
192 | for x in xrange(lend): | |
|
193 | if self.index[i + x] is None: | |
|
194 | b = data[off : off + self.s] | |
|
195 | self.index[i + x] = b | |
|
196 | n = b[ngshaoffset:ngshaoffset + 20] | |
|
197 | self.map[n] = i + x | |
|
198 | off += self.s | |
|
199 | ||
|
200 | def findnode(self, node): | |
|
201 | """search backwards through the index file for a specific node""" | |
|
202 | if self.allmap: | |
|
203 | return None | |
|
204 | ||
|
205 | # hg log will cause many many searches for the manifest | |
|
206 | # nodes. After we get called a few times, just load the whole | |
|
207 | # thing. | |
|
208 | if self.mapfind_count > 8: | |
|
209 | self.loadmap() | |
|
210 | if node in self.map: | |
|
211 | return node | |
|
212 | return None | |
|
213 | self.mapfind_count += 1 | |
|
214 | last = self.l - 1 | |
|
215 | while self.index[last] is not None: | |
|
216 | if last == 0: | |
|
217 | self.all = 1 | |
|
218 | self.allmap = 1 | |
|
219 | return None | |
|
220 | last -= 1 | |
|
221 | end = (last + 1) * self.s | |
|
222 | blocksize = self.s * 256 | |
|
223 | while end >= 0: | |
|
224 | start = max(end - blocksize, 0) | |
|
225 | self.dataf.seek(start) | |
|
226 | data = self.dataf.read(end - start) | |
|
227 | findend = end - start | |
|
228 | while True: | |
|
229 | # we're searching backwards, so we have to make sure | |
|
230 | # we don't find a changeset where this node is a parent | |
|
231 | off = data.find(node, 0, findend) | |
|
232 | findend = off | |
|
233 | if off >= 0: | |
|
234 | i = off / self.s | |
|
235 | off = i * self.s | |
|
236 | n = data[off + ngshaoffset:off + ngshaoffset + 20] | |
|
237 | if n == node: | |
|
238 | self.map[n] = i + start / self.s | |
|
239 | return node | |
|
240 | else: | |
|
241 | break | |
|
242 | end -= blocksize | |
|
243 | return None | |
|
244 | ||
|
245 | def loadindex(self, i=None, end=None): | |
|
246 | if self.all: | |
|
247 | return | |
|
248 | all = False | |
|
249 | if i is None: | |
|
250 | blockstart = 0 | |
|
251 | blocksize = (65536 / self.s) * self.s | |
|
252 | end = self.datasize | |
|
253 | all = True | |
|
254 | else: | |
|
255 | if end: | |
|
256 | blockstart = i * self.s | |
|
257 | end = end * self.s | |
|
258 | blocksize = end - blockstart | |
|
259 | else: | |
|
260 | blockstart = (i & ~1023) * self.s | |
|
261 | blocksize = self.s * 1024 | |
|
262 | end = blockstart + blocksize | |
|
263 | while blockstart < end: | |
|
264 | self.loadblock(blockstart, blocksize) | |
|
265 | blockstart += blocksize | |
|
266 | if all: | |
|
267 | self.all = True | |
|
268 | ||
|
269 | class lazyindex(object): | |
|
270 | """a lazy version of the index array""" | |
|
271 | def __init__(self, parser): | |
|
272 | self.p = parser | |
|
273 | def __len__(self): | |
|
274 | return len(self.p.index) | |
|
275 | def load(self, pos): | |
|
276 | if pos < 0: | |
|
277 | pos += len(self.p.index) | |
|
278 | self.p.loadindex(pos) | |
|
279 | return self.p.index[pos] | |
|
280 | def __getitem__(self, pos): | |
|
281 | return _unpack(indexformatng, self.p.index[pos] or self.load(pos)) | |
|
282 | def __setitem__(self, pos, item): | |
|
283 | self.p.index[pos] = _pack(indexformatng, *item) | |
|
284 | def __delitem__(self, pos): | |
|
285 | del self.p.index[pos] | |
|
286 | def insert(self, pos, e): | |
|
287 | self.p.index.insert(pos, _pack(indexformatng, *e)) | |
|
288 | def append(self, e): | |
|
289 | self.p.index.append(_pack(indexformatng, *e)) | |
|
290 | ||
|
291 | class lazymap(object): | |
|
292 | """a lazy version of the node map""" | |
|
293 | def __init__(self, parser): | |
|
294 | self.p = parser | |
|
295 | def load(self, key): | |
|
296 | n = self.p.findnode(key) | |
|
297 | if n is None: | |
|
298 | raise KeyError(key) | |
|
299 | def __contains__(self, key): | |
|
300 | if key in self.p.map: | |
|
301 | return True | |
|
302 | self.p.loadmap() | |
|
303 | return key in self.p.map | |
|
304 | def __iter__(self): | |
|
305 | yield nullid | |
|
306 | for i, ret in enumerate(self.p.index): | |
|
307 | if not ret: | |
|
308 | self.p.loadindex(i) | |
|
309 | ret = self.p.index[i] | |
|
310 | if isinstance(ret, str): | |
|
311 | ret = _unpack(indexformatng, ret) | |
|
312 | yield ret[7] | |
|
313 | def __getitem__(self, key): | |
|
314 | try: | |
|
315 | return self.p.map[key] | |
|
316 | except KeyError: | |
|
317 | try: | |
|
318 | self.load(key) | |
|
319 | return self.p.map[key] | |
|
320 | except KeyError: | |
|
321 | raise KeyError("node " + hex(key)) | |
|
322 | def __setitem__(self, key, val): | |
|
323 | self.p.map[key] = val | |
|
324 | def __delitem__(self, key): | |
|
325 | del self.p.map[key] | |
|
326 | ||
|
327 | 122 | indexformatv0 = ">4l20s20s20s" |
|
328 | 123 | v0shaoffset = 56 |
|
329 | 124 | |
|
330 | 125 | class revlogoldio(object): |
|
331 | 126 | def __init__(self): |
|
332 | 127 | self.size = struct.calcsize(indexformatv0) |
|
333 | 128 | |
|
334 | 129 | def parseindex(self, fp, data, inline): |
|
335 | 130 | s = self.size |
|
336 | 131 | index = [] |
|
337 | 132 | nodemap = {nullid: nullrev} |
|
338 | 133 | n = off = 0 |
|
339 | if len(data) == _prereadsize: | |
|
340 | data += fp.read() # read the rest | |
|
341 | 134 | l = len(data) |
|
342 | 135 | while off + s <= l: |
|
343 | 136 | cur = data[off:off + s] |
|
344 | 137 | off += s |
|
345 | 138 | e = _unpack(indexformatv0, cur) |
|
346 | 139 | # transform to revlogv1 format |
|
347 | 140 | e2 = (offset_type(e[0], 0), e[1], -1, e[2], e[3], |
|
348 | 141 | nodemap.get(e[4], nullrev), nodemap.get(e[5], nullrev), e[6]) |
|
349 | 142 | index.append(e2) |
|
350 | 143 | nodemap[e[6]] = n |
|
351 | 144 | n += 1 |
|
352 | 145 | |
|
353 | 146 | return index, nodemap, None |
|
354 | 147 | |
|
355 | 148 | def packentry(self, entry, node, version, rev): |
|
356 | 149 | if gettype(entry[0]): |
|
357 | 150 | raise RevlogError(_("index entry flags need RevlogNG")) |
|
358 | 151 | e2 = (getoffset(entry[0]), entry[1], entry[3], entry[4], |
|
359 | 152 | node(entry[5]), node(entry[6]), entry[7]) |
|
360 | 153 | return _pack(indexformatv0, *e2) |
|
361 | 154 | |
|
362 | 155 | # index ng: |
|
363 | 156 | # 6 bytes: offset |
|
364 | 157 | # 2 bytes: flags |
|
365 | 158 | # 4 bytes: compressed length |
|
366 | 159 | # 4 bytes: uncompressed length |
|
367 | 160 | # 4 bytes: base rev |
|
368 | 161 | # 4 bytes: link rev |
|
369 | 162 | # 4 bytes: parent 1 rev |
|
370 | 163 | # 4 bytes: parent 2 rev |
|
371 | 164 | # 32 bytes: nodeid |
|
372 | 165 | indexformatng = ">Qiiiiii20s12x" |
|
373 | 166 | ngshaoffset = 32 |
|
374 | 167 | versionformat = ">I" |
|
375 | 168 | |
|
376 | 169 | class revlogio(object): |
|
377 | 170 | def __init__(self): |
|
378 | 171 | self.size = struct.calcsize(indexformatng) |
|
379 | 172 | |
|
380 | 173 | def parseindex(self, fp, data, inline): |
|
381 | if len(data) == _prereadsize: | |
|
382 | if util.openhardlinks() and not inline: | |
|
383 | # big index, let's parse it on demand | |
|
384 | parser = lazyparser(fp) | |
|
385 | index = lazyindex(parser) | |
|
386 | nodemap = lazymap(parser) | |
|
387 | e = list(index[0]) | |
|
388 | type = gettype(e[0]) | |
|
389 | e[0] = offset_type(0, type) | |
|
390 | index[0] = e | |
|
391 | return index, nodemap, None | |
|
392 | else: | |
|
393 | data += fp.read() | |
|
394 | ||
|
395 | 174 | # call the C implementation to parse the index data |
|
396 | 175 | index, nodemap, cache = parsers.parse_index(data, inline) |
|
397 | 176 | return index, nodemap, cache |
|
398 | 177 | |
|
399 | 178 | def packentry(self, entry, node, version, rev): |
|
400 | 179 | p = _pack(indexformatng, *entry) |
|
401 | 180 | if rev == 0: |
|
402 | 181 | p = _pack(versionformat, version) + p[4:] |
|
403 | 182 | return p |
|
404 | 183 | |
|
405 | 184 | class revlog(object): |
|
406 | 185 | """ |
|
407 | 186 | the underlying revision storage object |
|
408 | 187 | |
|
409 | 188 | A revlog consists of two parts, an index and the revision data. |
|
410 | 189 | |
|
411 | 190 | The index is a file with a fixed record size containing |
|
412 | 191 | information on each revision, including its nodeid (hash), the |
|
413 | 192 | nodeids of its parents, the position and offset of its data within |
|
414 | 193 | the data file, and the revision it's based on. Finally, each entry |
|
415 | 194 | contains a linkrev entry that can serve as a pointer to external |
|
416 | 195 | data. |
|
417 | 196 | |
|
418 | 197 | The revision data itself is a linear collection of data chunks. |
|
419 | 198 | Each chunk represents a revision and is usually represented as a |
|
420 | 199 | delta against the previous chunk. To bound lookup time, runs of |
|
421 | 200 | deltas are limited to about 2 times the length of the original |
|
422 | 201 | version data. This makes retrieval of a version proportional to |
|
423 | 202 | its size, or O(1) relative to the number of revisions. |
|
424 | 203 | |
|
425 | 204 | Both pieces of the revlog are written to in an append-only |
|
426 | 205 | fashion, which means we never need to rewrite a file to insert or |
|
427 | 206 | remove data, and can use some simple techniques to avoid the need |
|
428 | 207 | for locking while reading. |
|
429 | 208 | """ |
|
430 | 209 | def __init__(self, opener, indexfile, shallowroot=None): |
|
431 | 210 | """ |
|
432 | 211 | create a revlog object |
|
433 | 212 | |
|
434 | 213 | opener is a function that abstracts the file opening operation |
|
435 | 214 | and can be used to implement COW semantics or the like. |
|
436 | 215 | """ |
|
437 | 216 | self.indexfile = indexfile |
|
438 | 217 | self.datafile = indexfile[:-2] + ".d" |
|
439 | 218 | self.opener = opener |
|
440 | 219 | self._cache = None |
|
441 | 220 | self._chunkcache = (0, '') |
|
442 | 221 | self.nodemap = {nullid: nullrev} |
|
443 | 222 | self.index = [] |
|
444 | 223 | self._shallowroot = shallowroot |
|
445 | 224 | self._parentdelta = 0 |
|
446 | 225 | |
|
447 | 226 | v = REVLOG_DEFAULT_VERSION |
|
448 | 227 | if hasattr(opener, 'options') and 'defversion' in opener.options: |
|
449 | 228 | v = opener.options['defversion'] |
|
450 | 229 | if v & REVLOGNG: |
|
451 | 230 | v |= REVLOGNGINLINEDATA |
|
452 | 231 | if v & REVLOGNG and 'parentdelta' in opener.options: |
|
453 | 232 | self._parentdelta = 1 |
|
454 | 233 | |
|
455 | 234 | if shallowroot: |
|
456 | 235 | v |= REVLOGSHALLOW |
|
457 | 236 | |
|
458 | 237 | i = '' |
|
459 | 238 | try: |
|
460 | 239 | f = self.opener(self.indexfile) |
|
461 | if "nonlazy" in getattr(self.opener, 'options', {}): | |
|
462 | 240 |
|
|
463 | else: | |
|
464 | i = f.read(_prereadsize) | |
|
465 | 241 | if len(i) > 0: |
|
466 | 242 | v = struct.unpack(versionformat, i[:4])[0] |
|
467 | 243 | except IOError, inst: |
|
468 | 244 | if inst.errno != errno.ENOENT: |
|
469 | 245 | raise |
|
470 | 246 | |
|
471 | 247 | self.version = v |
|
472 | 248 | self._inline = v & REVLOGNGINLINEDATA |
|
473 | 249 | self._shallow = v & REVLOGSHALLOW |
|
474 | 250 | flags = v & ~0xFFFF |
|
475 | 251 | fmt = v & 0xFFFF |
|
476 | 252 | if fmt == REVLOGV0 and flags: |
|
477 | 253 | raise RevlogError(_("index %s unknown flags %#04x for format v0") |
|
478 | 254 | % (self.indexfile, flags >> 16)) |
|
479 | 255 | elif fmt == REVLOGNG and flags & ~REVLOGNG_FLAGS: |
|
480 | 256 | raise RevlogError(_("index %s unknown flags %#04x for revlogng") |
|
481 | 257 | % (self.indexfile, flags >> 16)) |
|
482 | 258 | elif fmt > REVLOGNG: |
|
483 | 259 | raise RevlogError(_("index %s unknown format %d") |
|
484 | 260 | % (self.indexfile, fmt)) |
|
485 | 261 | |
|
486 | 262 | self._io = revlogio() |
|
487 | 263 | if self.version == REVLOGV0: |
|
488 | 264 | self._io = revlogoldio() |
|
489 | 265 | if i: |
|
490 | 266 | try: |
|
491 | 267 | d = self._io.parseindex(f, i, self._inline) |
|
492 | 268 | except (ValueError, IndexError): |
|
493 | 269 | raise RevlogError(_("index %s is corrupted") % (self.indexfile)) |
|
494 | 270 | self.index, self.nodemap, self._chunkcache = d |
|
495 | 271 | if not self._chunkcache: |
|
496 | 272 | self._chunkclear() |
|
497 | 273 | |
|
498 | 274 | # add the magic null revision at -1 (if it hasn't been done already) |
|
499 |
if |
|
|
500 | self.index[-1][7] != nullid) : | |
|
275 | if self.index == [] or self.index[-1][7] != nullid: | |
|
501 | 276 | self.index.append((0, 0, 0, -1, -1, -1, -1, nullid)) |
|
502 | 277 | |
|
503 | def _loadindex(self, start, end): | |
|
504 | """load a block of indexes all at once from the lazy parser""" | |
|
505 | if isinstance(self.index, lazyindex): | |
|
506 | self.index.p.loadindex(start, end) | |
|
507 | ||
|
508 | def _loadindexmap(self): | |
|
509 | """loads both the map and the index from the lazy parser""" | |
|
510 | if isinstance(self.index, lazyindex): | |
|
511 | p = self.index.p | |
|
512 | p.loadindex() | |
|
513 | self.nodemap = p.map | |
|
514 | ||
|
515 | def _loadmap(self): | |
|
516 | """loads the map from the lazy parser""" | |
|
517 | if isinstance(self.nodemap, lazymap): | |
|
518 | self.nodemap.p.loadmap() | |
|
519 | self.nodemap = self.nodemap.p.map | |
|
520 | ||
|
521 | 278 | def tip(self): |
|
522 | 279 | return self.node(len(self.index) - 2) |
|
523 | 280 | def __len__(self): |
|
524 | 281 | return len(self.index) - 1 |
|
525 | 282 | def __iter__(self): |
|
526 | 283 | for i in xrange(len(self)): |
|
527 | 284 | yield i |
|
528 | 285 | def rev(self, node): |
|
529 | 286 | try: |
|
530 | 287 | return self.nodemap[node] |
|
531 | 288 | except KeyError: |
|
532 | 289 | raise LookupError(node, self.indexfile, _('no node')) |
|
533 | 290 | def node(self, rev): |
|
534 | 291 | return self.index[rev][7] |
|
535 | 292 | def linkrev(self, rev): |
|
536 | 293 | return self.index[rev][4] |
|
537 | 294 | def parents(self, node): |
|
538 | 295 | i = self.index |
|
539 | 296 | d = i[self.rev(node)] |
|
540 | 297 | return i[d[5]][7], i[d[6]][7] # map revisions to nodes inline |
|
541 | 298 | def parentrevs(self, rev): |
|
542 | 299 | return self.index[rev][5:7] |
|
543 | 300 | def start(self, rev): |
|
544 | 301 | return int(self.index[rev][0] >> 16) |
|
545 | 302 | def end(self, rev): |
|
546 | 303 | return self.start(rev) + self.length(rev) |
|
547 | 304 | def length(self, rev): |
|
548 | 305 | return self.index[rev][1] |
|
549 | 306 | def base(self, rev): |
|
550 | 307 | return self.index[rev][3] |
|
551 | 308 | def flags(self, rev): |
|
552 | 309 | return self.index[rev][0] & 0xFFFF |
|
553 | 310 | def rawsize(self, rev): |
|
554 | 311 | """return the length of the uncompressed text for a given revision""" |
|
555 | 312 | l = self.index[rev][2] |
|
556 | 313 | if l >= 0: |
|
557 | 314 | return l |
|
558 | 315 | |
|
559 | 316 | t = self.revision(self.node(rev)) |
|
560 | 317 | return len(t) |
|
561 | 318 | size = rawsize |
|
562 | 319 | |
|
563 | 320 | def reachable(self, node, stop=None): |
|
564 | 321 | """return the set of all nodes ancestral to a given node, including |
|
565 | 322 | the node itself, stopping when stop is matched""" |
|
566 | 323 | reachable = set((node,)) |
|
567 | 324 | visit = [node] |
|
568 | 325 | if stop: |
|
569 | 326 | stopn = self.rev(stop) |
|
570 | 327 | else: |
|
571 | 328 | stopn = 0 |
|
572 | 329 | while visit: |
|
573 | 330 | n = visit.pop(0) |
|
574 | 331 | if n == stop: |
|
575 | 332 | continue |
|
576 | 333 | if n == nullid: |
|
577 | 334 | continue |
|
578 | 335 | for p in self.parents(n): |
|
579 | 336 | if self.rev(p) < stopn: |
|
580 | 337 | continue |
|
581 | 338 | if p not in reachable: |
|
582 | 339 | reachable.add(p) |
|
583 | 340 | visit.append(p) |
|
584 | 341 | return reachable |
|
585 | 342 | |
|
586 | 343 | def ancestors(self, *revs): |
|
587 | 344 | """Generate the ancestors of 'revs' in reverse topological order. |
|
588 | 345 | |
|
589 | 346 | Yield a sequence of revision numbers starting with the parents |
|
590 | 347 | of each revision in revs, i.e., each revision is *not* considered |
|
591 | 348 | an ancestor of itself. Results are in breadth-first order: |
|
592 | 349 | parents of each rev in revs, then parents of those, etc. Result |
|
593 | 350 | does not include the null revision.""" |
|
594 | 351 | visit = list(revs) |
|
595 | 352 | seen = set([nullrev]) |
|
596 | 353 | while visit: |
|
597 | 354 | for parent in self.parentrevs(visit.pop(0)): |
|
598 | 355 | if parent not in seen: |
|
599 | 356 | visit.append(parent) |
|
600 | 357 | seen.add(parent) |
|
601 | 358 | yield parent |
|
602 | 359 | |
|
603 | 360 | def descendants(self, *revs): |
|
604 | 361 | """Generate the descendants of 'revs' in revision order. |
|
605 | 362 | |
|
606 | 363 | Yield a sequence of revision numbers starting with a child of |
|
607 | 364 | some rev in revs, i.e., each revision is *not* considered a |
|
608 | 365 | descendant of itself. Results are ordered by revision number (a |
|
609 | 366 | topological sort).""" |
|
610 | 367 | first = min(revs) |
|
611 | 368 | if first == nullrev: |
|
612 | 369 | for i in self: |
|
613 | 370 | yield i |
|
614 | 371 | return |
|
615 | 372 | |
|
616 | 373 | seen = set(revs) |
|
617 | 374 | for i in xrange(first + 1, len(self)): |
|
618 | 375 | for x in self.parentrevs(i): |
|
619 | 376 | if x != nullrev and x in seen: |
|
620 | 377 | seen.add(i) |
|
621 | 378 | yield i |
|
622 | 379 | break |
|
623 | 380 | |
|
624 | 381 | def findmissing(self, common=None, heads=None): |
|
625 | 382 | """Return the ancestors of heads that are not ancestors of common. |
|
626 | 383 | |
|
627 | 384 | More specifically, return a list of nodes N such that every N |
|
628 | 385 | satisfies the following constraints: |
|
629 | 386 | |
|
630 | 387 | 1. N is an ancestor of some node in 'heads' |
|
631 | 388 | 2. N is not an ancestor of any node in 'common' |
|
632 | 389 | |
|
633 | 390 | The list is sorted by revision number, meaning it is |
|
634 | 391 | topologically sorted. |
|
635 | 392 | |
|
636 | 393 | 'heads' and 'common' are both lists of node IDs. If heads is |
|
637 | 394 | not supplied, uses all of the revlog's heads. If common is not |
|
638 | 395 | supplied, uses nullid.""" |
|
639 | 396 | if common is None: |
|
640 | 397 | common = [nullid] |
|
641 | 398 | if heads is None: |
|
642 | 399 | heads = self.heads() |
|
643 | 400 | |
|
644 | 401 | common = [self.rev(n) for n in common] |
|
645 | 402 | heads = [self.rev(n) for n in heads] |
|
646 | 403 | |
|
647 | 404 | # we want the ancestors, but inclusive |
|
648 | 405 | has = set(self.ancestors(*common)) |
|
649 | 406 | has.add(nullrev) |
|
650 | 407 | has.update(common) |
|
651 | 408 | |
|
652 | 409 | # take all ancestors from heads that aren't in has |
|
653 | 410 | missing = set() |
|
654 | 411 | visit = [r for r in heads if r not in has] |
|
655 | 412 | while visit: |
|
656 | 413 | r = visit.pop(0) |
|
657 | 414 | if r in missing: |
|
658 | 415 | continue |
|
659 | 416 | else: |
|
660 | 417 | missing.add(r) |
|
661 | 418 | for p in self.parentrevs(r): |
|
662 | 419 | if p not in has: |
|
663 | 420 | visit.append(p) |
|
664 | 421 | missing = list(missing) |
|
665 | 422 | missing.sort() |
|
666 | 423 | return [self.node(r) for r in missing] |
|
667 | 424 | |
|
668 | 425 | def nodesbetween(self, roots=None, heads=None): |
|
669 | 426 | """Return a topological path from 'roots' to 'heads'. |
|
670 | 427 | |
|
671 | 428 | Return a tuple (nodes, outroots, outheads) where 'nodes' is a |
|
672 | 429 | topologically sorted list of all nodes N that satisfy both of |
|
673 | 430 | these constraints: |
|
674 | 431 | |
|
675 | 432 | 1. N is a descendant of some node in 'roots' |
|
676 | 433 | 2. N is an ancestor of some node in 'heads' |
|
677 | 434 | |
|
678 | 435 | Every node is considered to be both a descendant and an ancestor |
|
679 | 436 | of itself, so every reachable node in 'roots' and 'heads' will be |
|
680 | 437 | included in 'nodes'. |
|
681 | 438 | |
|
682 | 439 | 'outroots' is the list of reachable nodes in 'roots', i.e., the |
|
683 | 440 | subset of 'roots' that is returned in 'nodes'. Likewise, |
|
684 | 441 | 'outheads' is the subset of 'heads' that is also in 'nodes'. |
|
685 | 442 | |
|
686 | 443 | 'roots' and 'heads' are both lists of node IDs. If 'roots' is |
|
687 | 444 | unspecified, uses nullid as the only root. If 'heads' is |
|
688 | 445 | unspecified, uses list of all of the revlog's heads.""" |
|
689 | 446 | nonodes = ([], [], []) |
|
690 | 447 | if roots is not None: |
|
691 | 448 | roots = list(roots) |
|
692 | 449 | if not roots: |
|
693 | 450 | return nonodes |
|
694 | 451 | lowestrev = min([self.rev(n) for n in roots]) |
|
695 | 452 | else: |
|
696 | 453 | roots = [nullid] # Everybody's a descendent of nullid |
|
697 | 454 | lowestrev = nullrev |
|
698 | 455 | if (lowestrev == nullrev) and (heads is None): |
|
699 | 456 | # We want _all_ the nodes! |
|
700 | 457 | return ([self.node(r) for r in self], [nullid], list(self.heads())) |
|
701 | 458 | if heads is None: |
|
702 | 459 | # All nodes are ancestors, so the latest ancestor is the last |
|
703 | 460 | # node. |
|
704 | 461 | highestrev = len(self) - 1 |
|
705 | 462 | # Set ancestors to None to signal that every node is an ancestor. |
|
706 | 463 | ancestors = None |
|
707 | 464 | # Set heads to an empty dictionary for later discovery of heads |
|
708 | 465 | heads = {} |
|
709 | 466 | else: |
|
710 | 467 | heads = list(heads) |
|
711 | 468 | if not heads: |
|
712 | 469 | return nonodes |
|
713 | 470 | ancestors = set() |
|
714 | 471 | # Turn heads into a dictionary so we can remove 'fake' heads. |
|
715 | 472 | # Also, later we will be using it to filter out the heads we can't |
|
716 | 473 | # find from roots. |
|
717 | 474 | heads = dict.fromkeys(heads, 0) |
|
718 | 475 | # Start at the top and keep marking parents until we're done. |
|
719 | 476 | nodestotag = set(heads) |
|
720 | 477 | # Remember where the top was so we can use it as a limit later. |
|
721 | 478 | highestrev = max([self.rev(n) for n in nodestotag]) |
|
722 | 479 | while nodestotag: |
|
723 | 480 | # grab a node to tag |
|
724 | 481 | n = nodestotag.pop() |
|
725 | 482 | # Never tag nullid |
|
726 | 483 | if n == nullid: |
|
727 | 484 | continue |
|
728 | 485 | # A node's revision number represents its place in a |
|
729 | 486 | # topologically sorted list of nodes. |
|
730 | 487 | r = self.rev(n) |
|
731 | 488 | if r >= lowestrev: |
|
732 | 489 | if n not in ancestors: |
|
733 | 490 | # If we are possibly a descendent of one of the roots |
|
734 | 491 | # and we haven't already been marked as an ancestor |
|
735 | 492 | ancestors.add(n) # Mark as ancestor |
|
736 | 493 | # Add non-nullid parents to list of nodes to tag. |
|
737 | 494 | nodestotag.update([p for p in self.parents(n) if |
|
738 | 495 | p != nullid]) |
|
739 | 496 | elif n in heads: # We've seen it before, is it a fake head? |
|
740 | 497 | # So it is, real heads should not be the ancestors of |
|
741 | 498 | # any other heads. |
|
742 | 499 | heads.pop(n) |
|
743 | 500 | if not ancestors: |
|
744 | 501 | return nonodes |
|
745 | 502 | # Now that we have our set of ancestors, we want to remove any |
|
746 | 503 | # roots that are not ancestors. |
|
747 | 504 | |
|
748 | 505 | # If one of the roots was nullid, everything is included anyway. |
|
749 | 506 | if lowestrev > nullrev: |
|
750 | 507 | # But, since we weren't, let's recompute the lowest rev to not |
|
751 | 508 | # include roots that aren't ancestors. |
|
752 | 509 | |
|
753 | 510 | # Filter out roots that aren't ancestors of heads |
|
754 | 511 | roots = [n for n in roots if n in ancestors] |
|
755 | 512 | # Recompute the lowest revision |
|
756 | 513 | if roots: |
|
757 | 514 | lowestrev = min([self.rev(n) for n in roots]) |
|
758 | 515 | else: |
|
759 | 516 | # No more roots? Return empty list |
|
760 | 517 | return nonodes |
|
761 | 518 | else: |
|
762 | 519 | # We are descending from nullid, and don't need to care about |
|
763 | 520 | # any other roots. |
|
764 | 521 | lowestrev = nullrev |
|
765 | 522 | roots = [nullid] |
|
766 | 523 | # Transform our roots list into a set. |
|
767 | 524 | descendents = set(roots) |
|
768 | 525 | # Also, keep the original roots so we can filter out roots that aren't |
|
769 | 526 | # 'real' roots (i.e. are descended from other roots). |
|
770 | 527 | roots = descendents.copy() |
|
771 | 528 | # Our topologically sorted list of output nodes. |
|
772 | 529 | orderedout = [] |
|
773 | 530 | # Don't start at nullid since we don't want nullid in our output list, |
|
774 | 531 | # and if nullid shows up in descedents, empty parents will look like |
|
775 | 532 | # they're descendents. |
|
776 | 533 | for r in xrange(max(lowestrev, 0), highestrev + 1): |
|
777 | 534 | n = self.node(r) |
|
778 | 535 | isdescendent = False |
|
779 | 536 | if lowestrev == nullrev: # Everybody is a descendent of nullid |
|
780 | 537 | isdescendent = True |
|
781 | 538 | elif n in descendents: |
|
782 | 539 | # n is already a descendent |
|
783 | 540 | isdescendent = True |
|
784 | 541 | # This check only needs to be done here because all the roots |
|
785 | 542 | # will start being marked is descendents before the loop. |
|
786 | 543 | if n in roots: |
|
787 | 544 | # If n was a root, check if it's a 'real' root. |
|
788 | 545 | p = tuple(self.parents(n)) |
|
789 | 546 | # If any of its parents are descendents, it's not a root. |
|
790 | 547 | if (p[0] in descendents) or (p[1] in descendents): |
|
791 | 548 | roots.remove(n) |
|
792 | 549 | else: |
|
793 | 550 | p = tuple(self.parents(n)) |
|
794 | 551 | # A node is a descendent if either of its parents are |
|
795 | 552 | # descendents. (We seeded the dependents list with the roots |
|
796 | 553 | # up there, remember?) |
|
797 | 554 | if (p[0] in descendents) or (p[1] in descendents): |
|
798 | 555 | descendents.add(n) |
|
799 | 556 | isdescendent = True |
|
800 | 557 | if isdescendent and ((ancestors is None) or (n in ancestors)): |
|
801 | 558 | # Only include nodes that are both descendents and ancestors. |
|
802 | 559 | orderedout.append(n) |
|
803 | 560 | if (ancestors is not None) and (n in heads): |
|
804 | 561 | # We're trying to figure out which heads are reachable |
|
805 | 562 | # from roots. |
|
806 | 563 | # Mark this head as having been reached |
|
807 | 564 | heads[n] = 1 |
|
808 | 565 | elif ancestors is None: |
|
809 | 566 | # Otherwise, we're trying to discover the heads. |
|
810 | 567 | # Assume this is a head because if it isn't, the next step |
|
811 | 568 | # will eventually remove it. |
|
812 | 569 | heads[n] = 1 |
|
813 | 570 | # But, obviously its parents aren't. |
|
814 | 571 | for p in self.parents(n): |
|
815 | 572 | heads.pop(p, None) |
|
816 | 573 | heads = [n for n in heads.iterkeys() if heads[n] != 0] |
|
817 | 574 | roots = list(roots) |
|
818 | 575 | assert orderedout |
|
819 | 576 | assert roots |
|
820 | 577 | assert heads |
|
821 | 578 | return (orderedout, roots, heads) |
|
822 | 579 | |
|
823 | 580 | def heads(self, start=None, stop=None): |
|
824 | 581 | """return the list of all nodes that have no children |
|
825 | 582 | |
|
826 | 583 | if start is specified, only heads that are descendants of |
|
827 | 584 | start will be returned |
|
828 | 585 | if stop is specified, it will consider all the revs from stop |
|
829 | 586 | as if they had no children |
|
830 | 587 | """ |
|
831 | 588 | if start is None and stop is None: |
|
832 | 589 | count = len(self) |
|
833 | 590 | if not count: |
|
834 | 591 | return [nullid] |
|
835 | 592 | ishead = [1] * (count + 1) |
|
836 | 593 | index = self.index |
|
837 | 594 | for r in xrange(count): |
|
838 | 595 | e = index[r] |
|
839 | 596 | ishead[e[5]] = ishead[e[6]] = 0 |
|
840 | 597 | return [self.node(r) for r in xrange(count) if ishead[r]] |
|
841 | 598 | |
|
842 | 599 | if start is None: |
|
843 | 600 | start = nullid |
|
844 | 601 | if stop is None: |
|
845 | 602 | stop = [] |
|
846 | 603 | stoprevs = set([self.rev(n) for n in stop]) |
|
847 | 604 | startrev = self.rev(start) |
|
848 | 605 | reachable = set((startrev,)) |
|
849 | 606 | heads = set((startrev,)) |
|
850 | 607 | |
|
851 | 608 | parentrevs = self.parentrevs |
|
852 | 609 | for r in xrange(startrev + 1, len(self)): |
|
853 | 610 | for p in parentrevs(r): |
|
854 | 611 | if p in reachable: |
|
855 | 612 | if r not in stoprevs: |
|
856 | 613 | reachable.add(r) |
|
857 | 614 | heads.add(r) |
|
858 | 615 | if p in heads and p not in stoprevs: |
|
859 | 616 | heads.remove(p) |
|
860 | 617 | |
|
861 | 618 | return [self.node(r) for r in heads] |
|
862 | 619 | |
|
863 | 620 | def children(self, node): |
|
864 | 621 | """find the children of a given node""" |
|
865 | 622 | c = [] |
|
866 | 623 | p = self.rev(node) |
|
867 | 624 | for r in range(p + 1, len(self)): |
|
868 | 625 | prevs = [pr for pr in self.parentrevs(r) if pr != nullrev] |
|
869 | 626 | if prevs: |
|
870 | 627 | for pr in prevs: |
|
871 | 628 | if pr == p: |
|
872 | 629 | c.append(self.node(r)) |
|
873 | 630 | elif p == nullrev: |
|
874 | 631 | c.append(self.node(r)) |
|
875 | 632 | return c |
|
876 | 633 | |
|
877 | 634 | def descendant(self, start, end): |
|
878 | 635 | if start == nullrev: |
|
879 | 636 | return True |
|
880 | 637 | for i in self.descendants(start): |
|
881 | 638 | if i == end: |
|
882 | 639 | return True |
|
883 | 640 | elif i > end: |
|
884 | 641 | break |
|
885 | 642 | return False |
|
886 | 643 | |
|
887 | 644 | def ancestor(self, a, b): |
|
888 | 645 | """calculate the least common ancestor of nodes a and b""" |
|
889 | 646 | |
|
890 | 647 | # fast path, check if it is a descendant |
|
891 | 648 | a, b = self.rev(a), self.rev(b) |
|
892 | 649 | start, end = sorted((a, b)) |
|
893 | 650 | if self.descendant(start, end): |
|
894 | 651 | return self.node(start) |
|
895 | 652 | |
|
896 | 653 | def parents(rev): |
|
897 | 654 | return [p for p in self.parentrevs(rev) if p != nullrev] |
|
898 | 655 | |
|
899 | 656 | c = ancestor.ancestor(a, b, parents) |
|
900 | 657 | if c is None: |
|
901 | 658 | return nullid |
|
902 | 659 | |
|
903 | 660 | return self.node(c) |
|
904 | 661 | |
|
905 | 662 | def _match(self, id): |
|
906 | 663 | if isinstance(id, (long, int)): |
|
907 | 664 | # rev |
|
908 | 665 | return self.node(id) |
|
909 | 666 | if len(id) == 20: |
|
910 | 667 | # possibly a binary node |
|
911 | 668 | # odds of a binary node being all hex in ASCII are 1 in 10**25 |
|
912 | 669 | try: |
|
913 | 670 | node = id |
|
914 | 671 | self.rev(node) # quick search the index |
|
915 | 672 | return node |
|
916 | 673 | except LookupError: |
|
917 | 674 | pass # may be partial hex id |
|
918 | 675 | try: |
|
919 | 676 | # str(rev) |
|
920 | 677 | rev = int(id) |
|
921 | 678 | if str(rev) != id: |
|
922 | 679 | raise ValueError |
|
923 | 680 | if rev < 0: |
|
924 | 681 | rev = len(self) + rev |
|
925 | 682 | if rev < 0 or rev >= len(self): |
|
926 | 683 | raise ValueError |
|
927 | 684 | return self.node(rev) |
|
928 | 685 | except (ValueError, OverflowError): |
|
929 | 686 | pass |
|
930 | 687 | if len(id) == 40: |
|
931 | 688 | try: |
|
932 | 689 | # a full hex nodeid? |
|
933 | 690 | node = bin(id) |
|
934 | 691 | self.rev(node) |
|
935 | 692 | return node |
|
936 | 693 | except (TypeError, LookupError): |
|
937 | 694 | pass |
|
938 | 695 | |
|
939 | 696 | def _partialmatch(self, id): |
|
940 | 697 | if len(id) < 40: |
|
941 | 698 | try: |
|
942 | 699 | # hex(node)[:...] |
|
943 | 700 | l = len(id) // 2 # grab an even number of digits |
|
944 | 701 | bin_id = bin(id[:l * 2]) |
|
945 | 702 | nl = [n for n in self.nodemap if n[:l] == bin_id] |
|
946 | 703 | nl = [n for n in nl if hex(n).startswith(id)] |
|
947 | 704 | if len(nl) > 0: |
|
948 | 705 | if len(nl) == 1: |
|
949 | 706 | return nl[0] |
|
950 | 707 | raise LookupError(id, self.indexfile, |
|
951 | 708 | _('ambiguous identifier')) |
|
952 | 709 | return None |
|
953 | 710 | except TypeError: |
|
954 | 711 | pass |
|
955 | 712 | |
|
956 | 713 | def lookup(self, id): |
|
957 | 714 | """locate a node based on: |
|
958 | 715 | - revision number or str(revision number) |
|
959 | 716 | - nodeid or subset of hex nodeid |
|
960 | 717 | """ |
|
961 | 718 | n = self._match(id) |
|
962 | 719 | if n is not None: |
|
963 | 720 | return n |
|
964 | 721 | n = self._partialmatch(id) |
|
965 | 722 | if n: |
|
966 | 723 | return n |
|
967 | 724 | |
|
968 | 725 | raise LookupError(id, self.indexfile, _('no match found')) |
|
969 | 726 | |
|
970 | 727 | def cmp(self, node, text): |
|
971 | 728 | """compare text with a given file revision |
|
972 | 729 | |
|
973 | 730 | returns True if text is different than what is stored. |
|
974 | 731 | """ |
|
975 | 732 | p1, p2 = self.parents(node) |
|
976 | 733 | return hash(text, p1, p2) != node |
|
977 | 734 | |
|
978 | 735 | def _addchunk(self, offset, data): |
|
979 | 736 | o, d = self._chunkcache |
|
980 | 737 | # try to add to existing cache |
|
981 |
if o + len(d) == offset and len(d) + len(data) < _ |
|
|
738 | if o + len(d) == offset and len(d) + len(data) < _chunksize: | |
|
982 | 739 | self._chunkcache = o, d + data |
|
983 | 740 | else: |
|
984 | 741 | self._chunkcache = offset, data |
|
985 | 742 | |
|
986 | 743 | def _loadchunk(self, offset, length): |
|
987 | 744 | if self._inline: |
|
988 | 745 | df = self.opener(self.indexfile) |
|
989 | 746 | else: |
|
990 | 747 | df = self.opener(self.datafile) |
|
991 | 748 | |
|
992 | 749 | readahead = max(65536, length) |
|
993 | 750 | df.seek(offset) |
|
994 | 751 | d = df.read(readahead) |
|
995 | 752 | self._addchunk(offset, d) |
|
996 | 753 | if readahead > length: |
|
997 | 754 | return d[:length] |
|
998 | 755 | return d |
|
999 | 756 | |
|
1000 | 757 | def _getchunk(self, offset, length): |
|
1001 | 758 | o, d = self._chunkcache |
|
1002 | 759 | l = len(d) |
|
1003 | 760 | |
|
1004 | 761 | # is it in the cache? |
|
1005 | 762 | cachestart = offset - o |
|
1006 | 763 | cacheend = cachestart + length |
|
1007 | 764 | if cachestart >= 0 and cacheend <= l: |
|
1008 | 765 | if cachestart == 0 and cacheend == l: |
|
1009 | 766 | return d # avoid a copy |
|
1010 | 767 | return d[cachestart:cacheend] |
|
1011 | 768 | |
|
1012 | 769 | return self._loadchunk(offset, length) |
|
1013 | 770 | |
|
1014 | 771 | def _chunkraw(self, startrev, endrev): |
|
1015 | 772 | start = self.start(startrev) |
|
1016 | 773 | length = self.end(endrev) - start |
|
1017 | 774 | if self._inline: |
|
1018 | 775 | start += (startrev + 1) * self._io.size |
|
1019 | 776 | return self._getchunk(start, length) |
|
1020 | 777 | |
|
1021 | 778 | def _chunk(self, rev): |
|
1022 | 779 | return decompress(self._chunkraw(rev, rev)) |
|
1023 | 780 | |
|
1024 | 781 | def _chunkclear(self): |
|
1025 | 782 | self._chunkcache = (0, '') |
|
1026 | 783 | |
|
1027 | 784 | def deltaparent(self, rev): |
|
1028 | 785 | """return previous revision or parentrev according to flags""" |
|
1029 | 786 | if self.flags(rev) & REVIDX_PARENTDELTA: |
|
1030 | 787 | return self.parentrevs(rev)[0] |
|
1031 | 788 | else: |
|
1032 | 789 | return rev - 1 |
|
1033 | 790 | |
|
1034 | 791 | def revdiff(self, rev1, rev2): |
|
1035 | 792 | """return or calculate a delta between two revisions""" |
|
1036 | 793 | if self.base(rev2) != rev2 and self.deltaparent(rev2) == rev1: |
|
1037 | 794 | return self._chunk(rev2) |
|
1038 | 795 | |
|
1039 | 796 | return mdiff.textdiff(self.revision(self.node(rev1)), |
|
1040 | 797 | self.revision(self.node(rev2))) |
|
1041 | 798 | |
|
1042 | 799 | def revision(self, node): |
|
1043 | 800 | """return an uncompressed revision of a given node""" |
|
1044 | 801 | cachedrev = None |
|
1045 | 802 | if node == nullid: |
|
1046 | 803 | return "" |
|
1047 | 804 | if self._cache: |
|
1048 | 805 | if self._cache[0] == node: |
|
1049 | 806 | return self._cache[2] |
|
1050 | 807 | cachedrev = self._cache[1] |
|
1051 | 808 | |
|
1052 | 809 | # look up what we need to read |
|
1053 | 810 | text = None |
|
1054 | 811 | rev = self.rev(node) |
|
1055 | 812 | base = self.base(rev) |
|
1056 | 813 | |
|
1057 | 814 | # check rev flags |
|
1058 | 815 | if self.flags(rev) & ~REVIDX_KNOWN_FLAGS: |
|
1059 | 816 | raise RevlogError(_('incompatible revision flag %x') % |
|
1060 | 817 | (self.flags(rev) & ~REVIDX_KNOWN_FLAGS)) |
|
1061 | 818 | |
|
1062 | 819 | # build delta chain |
|
1063 | self._loadindex(base, rev + 1) | |
|
1064 | 820 | chain = [] |
|
1065 | 821 | index = self.index # for performance |
|
1066 | 822 | iterrev = rev |
|
1067 | 823 | e = index[iterrev] |
|
1068 | 824 | while iterrev != base and iterrev != cachedrev: |
|
1069 | 825 | chain.append(iterrev) |
|
1070 | 826 | if e[0] & REVIDX_PARENTDELTA: |
|
1071 | 827 | iterrev = e[5] |
|
1072 | 828 | else: |
|
1073 | 829 | iterrev -= 1 |
|
1074 | 830 | e = index[iterrev] |
|
1075 | 831 | chain.reverse() |
|
1076 | 832 | base = iterrev |
|
1077 | 833 | |
|
1078 | 834 | if iterrev == cachedrev: |
|
1079 | 835 | # cache hit |
|
1080 | 836 | text = self._cache[2] |
|
1081 | 837 | |
|
1082 | 838 | # drop cache to save memory |
|
1083 | 839 | self._cache = None |
|
1084 | 840 | |
|
1085 | 841 | self._chunkraw(base, rev) |
|
1086 | 842 | if text is None: |
|
1087 | 843 | text = self._chunk(base) |
|
1088 | 844 | |
|
1089 | 845 | bins = [self._chunk(r) for r in chain] |
|
1090 | 846 | text = mdiff.patches(text, bins) |
|
1091 | 847 | |
|
1092 | 848 | text = self._checkhash(text, node) |
|
1093 | 849 | |
|
1094 | 850 | self._cache = (node, rev, text) |
|
1095 | 851 | return text |
|
1096 | 852 | |
|
1097 | 853 | def _checkhash(self, text, node): |
|
1098 | 854 | p1, p2 = self.parents(node) |
|
1099 | 855 | if (node != hash(text, p1, p2) and |
|
1100 | 856 | not (self.flags(rev) & REVIDX_PUNCHED_FLAG)): |
|
1101 | 857 | raise RevlogError(_("integrity check failed on %s:%d") |
|
1102 | 858 | % (self.indexfile, rev)) |
|
1103 | 859 | return text |
|
1104 | 860 | |
|
1105 | 861 | def checkinlinesize(self, tr, fp=None): |
|
1106 | 862 | if not self._inline or (self.start(-2) + self.length(-2)) < _maxinline: |
|
1107 | 863 | return |
|
1108 | 864 | |
|
1109 | 865 | trinfo = tr.find(self.indexfile) |
|
1110 | 866 | if trinfo is None: |
|
1111 | 867 | raise RevlogError(_("%s not found in the transaction") |
|
1112 | 868 | % self.indexfile) |
|
1113 | 869 | |
|
1114 | 870 | trindex = trinfo[2] |
|
1115 | 871 | dataoff = self.start(trindex) |
|
1116 | 872 | |
|
1117 | 873 | tr.add(self.datafile, dataoff) |
|
1118 | 874 | |
|
1119 | 875 | if fp: |
|
1120 | 876 | fp.flush() |
|
1121 | 877 | fp.close() |
|
1122 | 878 | |
|
1123 | 879 | df = self.opener(self.datafile, 'w') |
|
1124 | 880 | try: |
|
1125 | 881 | for r in self: |
|
1126 | 882 | df.write(self._chunkraw(r, r)) |
|
1127 | 883 | finally: |
|
1128 | 884 | df.close() |
|
1129 | 885 | |
|
1130 | 886 | fp = self.opener(self.indexfile, 'w', atomictemp=True) |
|
1131 | 887 | self.version &= ~(REVLOGNGINLINEDATA) |
|
1132 | 888 | self._inline = False |
|
1133 | 889 | for i in self: |
|
1134 | 890 | e = self._io.packentry(self.index[i], self.node, self.version, i) |
|
1135 | 891 | fp.write(e) |
|
1136 | 892 | |
|
1137 | 893 | # if we don't call rename, the temp file will never replace the |
|
1138 | 894 | # real index |
|
1139 | 895 | fp.rename() |
|
1140 | 896 | |
|
1141 | 897 | tr.replace(self.indexfile, trindex * self._io.size) |
|
1142 | 898 | self._chunkclear() |
|
1143 | 899 | |
|
1144 | 900 | def addrevision(self, text, transaction, link, p1, p2, cachedelta=None): |
|
1145 | 901 | """add a revision to the log |
|
1146 | 902 | |
|
1147 | 903 | text - the revision data to add |
|
1148 | 904 | transaction - the transaction object used for rollback |
|
1149 | 905 | link - the linkrev data to add |
|
1150 | 906 | p1, p2 - the parent nodeids of the revision |
|
1151 | 907 | cachedelta - an optional precomputed delta |
|
1152 | 908 | """ |
|
1153 | 909 | node = hash(text, p1, p2) |
|
1154 | 910 | if (node in self.nodemap and |
|
1155 | 911 | (not self.flags(self.rev(node)) & REVIDX_PUNCHED_FLAG)): |
|
1156 | 912 | return node |
|
1157 | 913 | |
|
1158 | 914 | dfh = None |
|
1159 | 915 | if not self._inline: |
|
1160 | 916 | dfh = self.opener(self.datafile, "a") |
|
1161 | 917 | ifh = self.opener(self.indexfile, "a+") |
|
1162 | 918 | try: |
|
1163 | 919 | return self._addrevision(node, text, transaction, link, p1, p2, |
|
1164 | 920 | cachedelta, ifh, dfh) |
|
1165 | 921 | finally: |
|
1166 | 922 | if dfh: |
|
1167 | 923 | dfh.close() |
|
1168 | 924 | ifh.close() |
|
1169 | 925 | |
|
1170 | 926 | def _addrevision(self, node, text, transaction, link, p1, p2, |
|
1171 | 927 | cachedelta, ifh, dfh): |
|
1172 | 928 | |
|
1173 | 929 | btext = [text] |
|
1174 | 930 | def buildtext(): |
|
1175 | 931 | if btext[0] is not None: |
|
1176 | 932 | return btext[0] |
|
1177 | 933 | # flush any pending writes here so we can read it in revision |
|
1178 | 934 | if dfh: |
|
1179 | 935 | dfh.flush() |
|
1180 | 936 | ifh.flush() |
|
1181 | 937 | basetext = self.revision(self.node(cachedelta[0])) |
|
1182 | 938 | btext[0] = mdiff.patch(basetext, cachedelta[1]) |
|
1183 | 939 | chk = hash(btext[0], p1, p2) |
|
1184 | 940 | if chk != node: |
|
1185 | 941 | raise RevlogError(_("consistency error in delta")) |
|
1186 | 942 | return btext[0] |
|
1187 | 943 | |
|
1188 | 944 | def builddelta(rev): |
|
1189 | 945 | # can we use the cached delta? |
|
1190 | 946 | if cachedelta and cachedelta[0] == rev: |
|
1191 | 947 | delta = cachedelta[1] |
|
1192 | 948 | else: |
|
1193 | 949 | t = buildtext() |
|
1194 | 950 | ptext = self.revision(self.node(rev)) |
|
1195 | 951 | delta = mdiff.textdiff(ptext, t) |
|
1196 | 952 | data = compress(delta) |
|
1197 | 953 | l = len(data[1]) + len(data[0]) |
|
1198 | 954 | base = self.base(rev) |
|
1199 | 955 | dist = l + offset - self.start(base) |
|
1200 | 956 | return dist, l, data, base |
|
1201 | 957 | |
|
1202 | 958 | curr = len(self) |
|
1203 | 959 | prev = curr - 1 |
|
1204 | 960 | base = curr |
|
1205 | 961 | offset = self.end(prev) |
|
1206 | 962 | flags = 0 |
|
1207 | 963 | d = None |
|
1208 | 964 | p1r, p2r = self.rev(p1), self.rev(p2) |
|
1209 | 965 | |
|
1210 | 966 | # should we try to build a delta? |
|
1211 | 967 | if prev != nullrev: |
|
1212 | 968 | d = builddelta(prev) |
|
1213 | 969 | if self._parentdelta and prev != p1r: |
|
1214 | 970 | d2 = builddelta(p1r) |
|
1215 | 971 | if d2 < d: |
|
1216 | 972 | d = d2 |
|
1217 | 973 | flags = REVIDX_PARENTDELTA |
|
1218 | 974 | dist, l, data, base = d |
|
1219 | 975 | |
|
1220 | 976 | # full versions are inserted when the needed deltas |
|
1221 | 977 | # become comparable to the uncompressed text |
|
1222 | 978 | # or the base revision is punched |
|
1223 | 979 | if text is None: |
|
1224 | 980 | textlen = mdiff.patchedsize(self.rawsize(cachedelta[0]), |
|
1225 | 981 | cachedelta[1]) |
|
1226 | 982 | else: |
|
1227 | 983 | textlen = len(text) |
|
1228 | 984 | if (d is None or dist > textlen * 2 or |
|
1229 | 985 | (self.flags(base) & REVIDX_PUNCHED_FLAG)): |
|
1230 | 986 | text = buildtext() |
|
1231 | 987 | data = compress(text) |
|
1232 | 988 | l = len(data[1]) + len(data[0]) |
|
1233 | 989 | base = curr |
|
1234 | 990 | |
|
1235 | 991 | e = (offset_type(offset, flags), l, textlen, |
|
1236 | 992 | base, link, p1r, p2r, node) |
|
1237 | 993 | self.index.insert(-1, e) |
|
1238 | 994 | self.nodemap[node] = curr |
|
1239 | 995 | |
|
1240 | 996 | entry = self._io.packentry(e, self.node, self.version, curr) |
|
1241 | 997 | if not self._inline: |
|
1242 | 998 | transaction.add(self.datafile, offset) |
|
1243 | 999 | transaction.add(self.indexfile, curr * len(entry)) |
|
1244 | 1000 | if data[0]: |
|
1245 | 1001 | dfh.write(data[0]) |
|
1246 | 1002 | dfh.write(data[1]) |
|
1247 | 1003 | dfh.flush() |
|
1248 | 1004 | ifh.write(entry) |
|
1249 | 1005 | else: |
|
1250 | 1006 | offset += curr * self._io.size |
|
1251 | 1007 | transaction.add(self.indexfile, offset, curr) |
|
1252 | 1008 | ifh.write(entry) |
|
1253 | 1009 | ifh.write(data[0]) |
|
1254 | 1010 | ifh.write(data[1]) |
|
1255 | 1011 | self.checkinlinesize(transaction, ifh) |
|
1256 | 1012 | |
|
1257 | 1013 | if type(text) == str: # only accept immutable objects |
|
1258 | 1014 | self._cache = (node, curr, text) |
|
1259 | 1015 | return node |
|
1260 | 1016 | |
|
1261 | 1017 | def group(self, nodelist, lookup, infocollect=None, fullrev=False): |
|
1262 | 1018 | """Calculate a delta group, yielding a sequence of changegroup chunks |
|
1263 | 1019 | (strings). |
|
1264 | 1020 | |
|
1265 | 1021 | Given a list of changeset revs, return a set of deltas and |
|
1266 | 1022 | metadata corresponding to nodes. The first delta is |
|
1267 | 1023 | first parent(nodelist[0]) -> nodelist[0], the receiver is |
|
1268 | 1024 | guaranteed to have this parent as it has all history before |
|
1269 | 1025 | these changesets. In the case firstparent is nullrev the |
|
1270 | 1026 | changegroup starts with a full revision. |
|
1271 | 1027 | fullrev forces the insertion of the full revision, necessary |
|
1272 | 1028 | in the case of shallow clones where the first parent might |
|
1273 | 1029 | not exist at the reciever. |
|
1274 | 1030 | """ |
|
1275 | 1031 | |
|
1276 | 1032 | revs = [self.rev(n) for n in nodelist] |
|
1277 | 1033 | |
|
1278 | 1034 | # if we don't have any revisions touched by these changesets, bail |
|
1279 | 1035 | if not revs: |
|
1280 | 1036 | yield changegroup.closechunk() |
|
1281 | 1037 | return |
|
1282 | 1038 | |
|
1283 | 1039 | # add the parent of the first rev |
|
1284 | 1040 | p = self.parentrevs(revs[0])[0] |
|
1285 | 1041 | revs.insert(0, p) |
|
1286 | 1042 | if p == nullrev: |
|
1287 | 1043 | fullrev = True |
|
1288 | 1044 | |
|
1289 | 1045 | # build deltas |
|
1290 | 1046 | for d in xrange(len(revs) - 1): |
|
1291 | 1047 | a, b = revs[d], revs[d + 1] |
|
1292 | 1048 | nb = self.node(b) |
|
1293 | 1049 | |
|
1294 | 1050 | if infocollect is not None: |
|
1295 | 1051 | infocollect(nb) |
|
1296 | 1052 | |
|
1297 | 1053 | p = self.parents(nb) |
|
1298 | 1054 | meta = nb + p[0] + p[1] + lookup(nb) |
|
1299 | 1055 | if fullrev: |
|
1300 | 1056 | d = self.revision(nb) |
|
1301 | 1057 | meta += mdiff.trivialdiffheader(len(d)) |
|
1302 | 1058 | fullrev = False |
|
1303 | 1059 | else: |
|
1304 | 1060 | d = self.revdiff(a, b) |
|
1305 | 1061 | yield changegroup.chunkheader(len(meta) + len(d)) |
|
1306 | 1062 | yield meta |
|
1307 | 1063 | yield d |
|
1308 | 1064 | |
|
1309 | 1065 | yield changegroup.closechunk() |
|
1310 | 1066 | |
|
1311 | 1067 | def addgroup(self, bundle, linkmapper, transaction): |
|
1312 | 1068 | """ |
|
1313 | 1069 | add a delta group |
|
1314 | 1070 | |
|
1315 | 1071 | given a set of deltas, add them to the revision log. the |
|
1316 | 1072 | first delta is against its parent, which should be in our |
|
1317 | 1073 | log, the rest are against the previous delta. |
|
1318 | 1074 | """ |
|
1319 | 1075 | |
|
1320 | 1076 | # track the base of the current delta log |
|
1321 | 1077 | node = None |
|
1322 | 1078 | |
|
1323 | 1079 | r = len(self) |
|
1324 | 1080 | end = 0 |
|
1325 | 1081 | if r: |
|
1326 | 1082 | end = self.end(r - 1) |
|
1327 | 1083 | ifh = self.opener(self.indexfile, "a+") |
|
1328 | 1084 | isize = r * self._io.size |
|
1329 | 1085 | if self._inline: |
|
1330 | 1086 | transaction.add(self.indexfile, end + isize, r) |
|
1331 | 1087 | dfh = None |
|
1332 | 1088 | else: |
|
1333 | 1089 | transaction.add(self.indexfile, isize, r) |
|
1334 | 1090 | transaction.add(self.datafile, end) |
|
1335 | 1091 | dfh = self.opener(self.datafile, "a") |
|
1336 | 1092 | |
|
1337 | 1093 | try: |
|
1338 | 1094 | # loop through our set of deltas |
|
1339 | 1095 | chain = None |
|
1340 | 1096 | while 1: |
|
1341 | 1097 | chunkdata = bundle.parsechunk() |
|
1342 | 1098 | if not chunkdata: |
|
1343 | 1099 | break |
|
1344 | 1100 | node = chunkdata['node'] |
|
1345 | 1101 | p1 = chunkdata['p1'] |
|
1346 | 1102 | p2 = chunkdata['p2'] |
|
1347 | 1103 | cs = chunkdata['cs'] |
|
1348 | 1104 | delta = chunkdata['data'] |
|
1349 | 1105 | |
|
1350 | 1106 | link = linkmapper(cs) |
|
1351 | 1107 | if (node in self.nodemap and |
|
1352 | 1108 | (not self.flags(self.rev(node)) & REVIDX_PUNCHED_FLAG)): |
|
1353 | 1109 | # this can happen if two branches make the same change |
|
1354 | 1110 | chain = node |
|
1355 | 1111 | continue |
|
1356 | 1112 | |
|
1357 | 1113 | for p in (p1, p2): |
|
1358 | 1114 | if not p in self.nodemap: |
|
1359 | 1115 | if self._shallow: |
|
1360 | 1116 | # add null entries for missing parents |
|
1361 | 1117 | # XXX FIXME |
|
1362 | 1118 | #if base == nullrev: |
|
1363 | 1119 | # base = len(self) |
|
1364 | 1120 | #e = (offset_type(end, REVIDX_PUNCHED_FLAG), |
|
1365 | 1121 | # 0, 0, base, nullrev, nullrev, nullrev, p) |
|
1366 | 1122 | #self.index.insert(-1, e) |
|
1367 | 1123 | #self.nodemap[p] = r |
|
1368 | 1124 | #entry = self._io.packentry(e, self.node, |
|
1369 | 1125 | # self.version, r) |
|
1370 | 1126 | #ifh.write(entry) |
|
1371 | 1127 | #t, r = r, r + 1 |
|
1372 | 1128 | raise LookupError(p, self.indexfile, |
|
1373 | 1129 | _('unknown parent')) |
|
1374 | 1130 | else: |
|
1375 | 1131 | raise LookupError(p, self.indexfile, |
|
1376 | 1132 | _('unknown parent')) |
|
1377 | 1133 | |
|
1378 | 1134 | if not chain: |
|
1379 | 1135 | # retrieve the parent revision of the delta chain |
|
1380 | 1136 | chain = p1 |
|
1381 | 1137 | if not chain in self.nodemap: |
|
1382 | 1138 | raise LookupError(chain, self.indexfile, _('unknown base')) |
|
1383 | 1139 | |
|
1384 | 1140 | chainrev = self.rev(chain) |
|
1385 | 1141 | chain = self._addrevision(node, None, transaction, link, |
|
1386 | 1142 | p1, p2, (chainrev, delta), ifh, dfh) |
|
1387 | 1143 | if not dfh and not self._inline: |
|
1388 | 1144 | # addrevision switched from inline to conventional |
|
1389 | 1145 | # reopen the index |
|
1390 | 1146 | dfh = self.opener(self.datafile, "a") |
|
1391 | 1147 | ifh = self.opener(self.indexfile, "a") |
|
1392 | 1148 | finally: |
|
1393 | 1149 | if dfh: |
|
1394 | 1150 | dfh.close() |
|
1395 | 1151 | ifh.close() |
|
1396 | 1152 | |
|
1397 | 1153 | return node |
|
1398 | 1154 | |
|
1399 | 1155 | def strip(self, minlink, transaction): |
|
1400 | 1156 | """truncate the revlog on the first revision with a linkrev >= minlink |
|
1401 | 1157 | |
|
1402 | 1158 | This function is called when we're stripping revision minlink and |
|
1403 | 1159 | its descendants from the repository. |
|
1404 | 1160 | |
|
1405 | 1161 | We have to remove all revisions with linkrev >= minlink, because |
|
1406 | 1162 | the equivalent changelog revisions will be renumbered after the |
|
1407 | 1163 | strip. |
|
1408 | 1164 | |
|
1409 | 1165 | So we truncate the revlog on the first of these revisions, and |
|
1410 | 1166 | trust that the caller has saved the revisions that shouldn't be |
|
1411 | 1167 | removed and that it'll readd them after this truncation. |
|
1412 | 1168 | """ |
|
1413 | 1169 | if len(self) == 0: |
|
1414 | 1170 | return |
|
1415 | 1171 | |
|
1416 | if isinstance(self.index, lazyindex): | |
|
1417 | self._loadindexmap() | |
|
1418 | ||
|
1419 | 1172 | for rev in self: |
|
1420 | 1173 | if self.index[rev][4] >= minlink: |
|
1421 | 1174 | break |
|
1422 | 1175 | else: |
|
1423 | 1176 | return |
|
1424 | 1177 | |
|
1425 | 1178 | # first truncate the files on disk |
|
1426 | 1179 | end = self.start(rev) |
|
1427 | 1180 | if not self._inline: |
|
1428 | 1181 | transaction.add(self.datafile, end) |
|
1429 | 1182 | end = rev * self._io.size |
|
1430 | 1183 | else: |
|
1431 | 1184 | end += rev * self._io.size |
|
1432 | 1185 | |
|
1433 | 1186 | transaction.add(self.indexfile, end) |
|
1434 | 1187 | |
|
1435 | 1188 | # then reset internal state in memory to forget those revisions |
|
1436 | 1189 | self._cache = None |
|
1437 | 1190 | self._chunkclear() |
|
1438 | 1191 | for x in xrange(rev, len(self)): |
|
1439 | 1192 | del self.nodemap[self.node(x)] |
|
1440 | 1193 | |
|
1441 | 1194 | del self.index[rev:-1] |
|
1442 | 1195 | |
|
1443 | 1196 | def checksize(self): |
|
1444 | 1197 | expected = 0 |
|
1445 | 1198 | if len(self): |
|
1446 | 1199 | expected = max(0, self.end(len(self) - 1)) |
|
1447 | 1200 | |
|
1448 | 1201 | try: |
|
1449 | 1202 | f = self.opener(self.datafile) |
|
1450 | 1203 | f.seek(0, 2) |
|
1451 | 1204 | actual = f.tell() |
|
1452 | 1205 | dd = actual - expected |
|
1453 | 1206 | except IOError, inst: |
|
1454 | 1207 | if inst.errno != errno.ENOENT: |
|
1455 | 1208 | raise |
|
1456 | 1209 | dd = 0 |
|
1457 | 1210 | |
|
1458 | 1211 | try: |
|
1459 | 1212 | f = self.opener(self.indexfile) |
|
1460 | 1213 | f.seek(0, 2) |
|
1461 | 1214 | actual = f.tell() |
|
1462 | 1215 | s = self._io.size |
|
1463 | 1216 | i = max(0, actual // s) |
|
1464 | 1217 | di = actual - (i * s) |
|
1465 | 1218 | if self._inline: |
|
1466 | 1219 | databytes = 0 |
|
1467 | 1220 | for r in self: |
|
1468 | 1221 | databytes += max(0, self.length(r)) |
|
1469 | 1222 | dd = 0 |
|
1470 | 1223 | di = actual - len(self) * s - databytes |
|
1471 | 1224 | except IOError, inst: |
|
1472 | 1225 | if inst.errno != errno.ENOENT: |
|
1473 | 1226 | raise |
|
1474 | 1227 | di = 0 |
|
1475 | 1228 | |
|
1476 | 1229 | return (dd, di) |
|
1477 | 1230 | |
|
1478 | 1231 | def files(self): |
|
1479 | 1232 | res = [self.indexfile] |
|
1480 | 1233 | if not self._inline: |
|
1481 | 1234 | res.append(self.datafile) |
|
1482 | 1235 | return res |
@@ -1,146 +1,145 b'' | |||
|
1 | 1 | # statichttprepo.py - simple http repository class for mercurial |
|
2 | 2 | # |
|
3 | 3 | # This provides read-only repo access to repositories exported via static http |
|
4 | 4 | # |
|
5 | 5 | # Copyright 2005-2007 Matt Mackall <mpm@selenic.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 | from i18n import _ |
|
11 | 11 | import changelog, byterange, url, error |
|
12 | 12 | import localrepo, manifest, util, store |
|
13 | 13 | import urllib, urllib2, errno |
|
14 | 14 | |
|
15 | 15 | class httprangereader(object): |
|
16 | 16 | def __init__(self, url, opener): |
|
17 | 17 | # we assume opener has HTTPRangeHandler |
|
18 | 18 | self.url = url |
|
19 | 19 | self.pos = 0 |
|
20 | 20 | self.opener = opener |
|
21 | 21 | self.name = url |
|
22 | 22 | def seek(self, pos): |
|
23 | 23 | self.pos = pos |
|
24 | 24 | def read(self, bytes=None): |
|
25 | 25 | req = urllib2.Request(self.url) |
|
26 | 26 | end = '' |
|
27 | 27 | if bytes: |
|
28 | 28 | end = self.pos + bytes - 1 |
|
29 | 29 | req.add_header('Range', 'bytes=%d-%s' % (self.pos, end)) |
|
30 | 30 | |
|
31 | 31 | try: |
|
32 | 32 | f = self.opener.open(req) |
|
33 | 33 | data = f.read() |
|
34 | 34 | if hasattr(f, 'getcode'): |
|
35 | 35 | # python 2.6+ |
|
36 | 36 | code = f.getcode() |
|
37 | 37 | elif hasattr(f, 'code'): |
|
38 | 38 | # undocumented attribute, seems to be set in 2.4 and 2.5 |
|
39 | 39 | code = f.code |
|
40 | 40 | else: |
|
41 | 41 | # Don't know how to check, hope for the best. |
|
42 | 42 | code = 206 |
|
43 | 43 | except urllib2.HTTPError, inst: |
|
44 | 44 | num = inst.code == 404 and errno.ENOENT or None |
|
45 | 45 | raise IOError(num, inst) |
|
46 | 46 | except urllib2.URLError, inst: |
|
47 | 47 | raise IOError(None, inst.reason[1]) |
|
48 | 48 | |
|
49 | 49 | if code == 200: |
|
50 | 50 | # HTTPRangeHandler does nothing if remote does not support |
|
51 | 51 | # Range headers and returns the full entity. Let's slice it. |
|
52 | 52 | if bytes: |
|
53 | 53 | data = data[self.pos:self.pos + bytes] |
|
54 | 54 | else: |
|
55 | 55 | data = data[self.pos:] |
|
56 | 56 | elif bytes: |
|
57 | 57 | data = data[:bytes] |
|
58 | 58 | self.pos += len(data) |
|
59 | 59 | return data |
|
60 | 60 | def __iter__(self): |
|
61 | 61 | return iter(self.read().splitlines(1)) |
|
62 | 62 | def close(self): |
|
63 | 63 | pass |
|
64 | 64 | |
|
65 | 65 | def build_opener(ui, authinfo): |
|
66 | 66 | # urllib cannot handle URLs with embedded user or passwd |
|
67 | 67 | urlopener = url.opener(ui, authinfo) |
|
68 | 68 | urlopener.add_handler(byterange.HTTPRangeHandler()) |
|
69 | 69 | |
|
70 | 70 | def opener(base): |
|
71 | 71 | """return a function that opens files over http""" |
|
72 | 72 | p = base |
|
73 | 73 | def o(path, mode="r", atomictemp=None): |
|
74 | 74 | if 'a' in mode or 'w' in mode: |
|
75 | 75 | raise IOError('Permission denied') |
|
76 | 76 | f = "/".join((p, urllib.quote(path))) |
|
77 | 77 | return httprangereader(f, urlopener) |
|
78 | 78 | return o |
|
79 | 79 | |
|
80 | opener.options = {'nonlazy': 1} | |
|
81 | 80 | return opener |
|
82 | 81 | |
|
83 | 82 | class statichttprepository(localrepo.localrepository): |
|
84 | 83 | def __init__(self, ui, path): |
|
85 | 84 | self._url = path |
|
86 | 85 | self.ui = ui |
|
87 | 86 | |
|
88 | 87 | self.root = path |
|
89 | 88 | self.path, authinfo = url.getauthinfo(path.rstrip('/') + "/.hg") |
|
90 | 89 | |
|
91 | 90 | opener = build_opener(ui, authinfo) |
|
92 | 91 | self.opener = opener(self.path) |
|
93 | 92 | |
|
94 | 93 | # find requirements |
|
95 | 94 | try: |
|
96 | 95 | requirements = self.opener("requires").read().splitlines() |
|
97 | 96 | except IOError, inst: |
|
98 | 97 | if inst.errno != errno.ENOENT: |
|
99 | 98 | raise |
|
100 | 99 | # check if it is a non-empty old-style repository |
|
101 | 100 | try: |
|
102 | 101 | self.opener("00changelog.i").read(1) |
|
103 | 102 | except IOError, inst: |
|
104 | 103 | if inst.errno != errno.ENOENT: |
|
105 | 104 | raise |
|
106 | 105 | # we do not care about empty old-style repositories here |
|
107 | 106 | msg = _("'%s' does not appear to be an hg repository") % path |
|
108 | 107 | raise error.RepoError(msg) |
|
109 | 108 | requirements = [] |
|
110 | 109 | |
|
111 | 110 | # check them |
|
112 | 111 | for r in requirements: |
|
113 | 112 | if r not in self.supported: |
|
114 | 113 | raise error.RepoError(_("requirement '%s' not supported") % r) |
|
115 | 114 | |
|
116 | 115 | # setup store |
|
117 | 116 | def pjoin(a, b): |
|
118 | 117 | return a + '/' + b |
|
119 | 118 | self.store = store.store(requirements, self.path, opener, pjoin) |
|
120 | 119 | self.spath = self.store.path |
|
121 | 120 | self.sopener = self.store.opener |
|
122 | 121 | self.sjoin = self.store.join |
|
123 | 122 | |
|
124 | 123 | self.manifest = manifest.manifest(self.sopener) |
|
125 | 124 | self.changelog = changelog.changelog(self.sopener) |
|
126 | 125 | self._tags = None |
|
127 | 126 | self.nodetagscache = None |
|
128 | 127 | self._branchcache = None |
|
129 | 128 | self._branchcachetip = None |
|
130 | 129 | self.encodepats = None |
|
131 | 130 | self.decodepats = None |
|
132 | 131 | self.capabilities.remove("pushkey") |
|
133 | 132 | |
|
134 | 133 | def url(self): |
|
135 | 134 | return self._url |
|
136 | 135 | |
|
137 | 136 | def local(self): |
|
138 | 137 | return False |
|
139 | 138 | |
|
140 | 139 | def lock(self, wait=True): |
|
141 | 140 | raise util.Abort(_('cannot lock static-http repository')) |
|
142 | 141 | |
|
143 | 142 | def instance(ui, path, create): |
|
144 | 143 | if create: |
|
145 | 144 | raise util.Abort(_('cannot create new static-http repository')) |
|
146 | 145 | return statichttprepository(ui, path[7:]) |
@@ -1,113 +1,113 b'' | |||
|
1 | 1 | from mercurial import parsers |
|
2 | 2 | from mercurial.node import nullid, nullrev |
|
3 | 3 | import struct |
|
4 | 4 | |
|
5 | 5 | # This unit test compares the return value of the original Python |
|
6 | 6 | # implementation of parseindex and the new C implementation for |
|
7 | 7 | # an index file with and without inlined data |
|
8 | 8 | |
|
9 | 9 | # original python implementation |
|
10 | 10 | def gettype(q): |
|
11 | 11 | return int(q & 0xFFFF) |
|
12 | 12 | |
|
13 | 13 | def offset_type(offset, type): |
|
14 | 14 | return long(long(offset) << 16 | type) |
|
15 | 15 | |
|
16 | 16 | indexformatng = ">Qiiiiii20s12x" |
|
17 | 17 | |
|
18 | 18 | def py_parseindex(data, inline) : |
|
19 | 19 | s = 64 |
|
20 | 20 | cache = None |
|
21 | 21 | index = [] |
|
22 | 22 | nodemap = {nullid: nullrev} |
|
23 | 23 | n = off = 0 |
|
24 | # if we're not using lazymap, always read the whole index | |
|
24 | ||
|
25 | 25 | l = len(data) - s |
|
26 | 26 | append = index.append |
|
27 | 27 | if inline: |
|
28 | 28 | cache = (0, data) |
|
29 | 29 | while off <= l: |
|
30 | 30 | e = struct.unpack(indexformatng, data[off:off + s]) |
|
31 | 31 | nodemap[e[7]] = n |
|
32 | 32 | append(e) |
|
33 | 33 | n += 1 |
|
34 | 34 | if e[1] < 0: |
|
35 | 35 | break |
|
36 | 36 | off += e[1] + s |
|
37 | 37 | else: |
|
38 | 38 | while off <= l: |
|
39 | 39 | e = struct.unpack(indexformatng, data[off:off + s]) |
|
40 | 40 | nodemap[e[7]] = n |
|
41 | 41 | append(e) |
|
42 | 42 | n += 1 |
|
43 | 43 | off += s |
|
44 | 44 | |
|
45 | 45 | e = list(index[0]) |
|
46 | 46 | type = gettype(e[0]) |
|
47 | 47 | e[0] = offset_type(0, type) |
|
48 | 48 | index[0] = tuple(e) |
|
49 | 49 | |
|
50 | 50 | # add the magic null revision at -1 |
|
51 | 51 | index.append((0, 0, 0, -1, -1, -1, -1, nullid)) |
|
52 | 52 | |
|
53 | 53 | return index, nodemap, cache |
|
54 | 54 | |
|
55 | 55 | |
|
56 | 56 | data_inlined = '\x00\x01\x00\x01\x00\x00\x00\x00\x00\x00\x01\x8c' \ |
|
57 | 57 | '\x00\x00\x04\x07\x00\x00\x00\x00\x00\x00\x15\x15\xff\xff\xff' \ |
|
58 | 58 | '\xff\xff\xff\xff\xff\xebG\x97\xb7\x1fB\x04\xcf\x13V\x81\tw\x1b' \ |
|
59 | 59 | 'w\xdduR\xda\xc6\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00' \ |
|
60 | 60 | 'x\x9c\x9d\x93?O\xc30\x10\xc5\xf7|\x8a\xdb\x9a\xa8m\x06\xd8*\x95' \ |
|
61 | 61 | '\x81B\xa1\xa2\xa2R\xcb\x86Pd\x9a\x0b5$vd_\x04\xfd\xf6\x9c\xff@' \ |
|
62 | 62 | '\x11!\x0b\xd9\xec\xf7\xbbw\xe7gG6\xad6\x04\xdaN\xc0\x92\xa0$)' \ |
|
63 | 63 | '\xb1\x82\xa2\xd1%\x16\xa4\x8b7\xa9\xca\xd4-\xb2Y\x02\xfc\xc9' \ |
|
64 | 64 | '\xcaS\xf9\xaeX\xed\xb6\xd77Q\x02\x83\xd4\x19\xf5--Y\xea\xe1W' \ |
|
65 | 65 | '\xab\xed\x10\xceR\x0f_\xdf\xdf\r\xe1,\xf5\xf0\xcb\xf5 \xceR\x0f' \ |
|
66 | 66 | '_\xdc\x0e\x0e\xc3R\x0f_\xae\x96\x9b!\x9e\xa5\x1e\xbf\xdb,\x06' \ |
|
67 | 67 | '\xc7q\x9a/\x88\x82\xc3B\xea\xb5\xb4TJ\x93\xb6\x82\x0e\xe16\xe6' \ |
|
68 | 68 | 'KQ\xdb\xaf\xecG\xa3\xd1 \x01\xd3\x0b_^\xe8\xaa\xa0\xae\xad\xd1' \ |
|
69 | 69 | '&\xbef\x1bz\x08\xb0|\xc9Xz\x06\xf6Z\x91\x90J\xaa\x17\x90\xaa' \ |
|
70 | 70 | '\xd2\xa6\x11$5C\xcf\xba#\xa0\x03\x02*2\x92-\xfc\xb1\x94\xdf\xe2' \ |
|
71 | 71 | '\xae\xb8\'m\x8ey0^\x85\xd3\x82\xb4\xf0`:\x9c\x00\x8a\xfd\x01' \ |
|
72 | 72 | '\xb0\xc6\x86\x8b\xdd\xae\x80\xf3\xa9\x9fd\x16\n\x00R%\x1a\x06' \ |
|
73 | 73 | '\xe9\xd8b\x98\x1d\xf4\xf3+\x9bf\x01\xd8p\x1b\xf3.\xed\x9f^g\xc3' \ |
|
74 | 74 | '^\xd9W81T\xdb\xd5\x04sx|\xf2\xeb\xd6`%?x\xed"\x831\xbf\xf3\xdc' \ |
|
75 | 75 | 'b\xeb%gaY\xe1\xad\x9f\xb9f\'1w\xa9\xa5a\x83s\x82J\xb98\xbc4\x8b' \ |
|
76 | 76 | '\x83\x00\x9f$z\xb8#\xa5\xb1\xdf\x98\xd9\xec\x1b\x89O\xe3Ts\x9a4' \ |
|
77 | 77 | '\x17m\x8b\xfc\x8f\xa5\x95\x9a\xfc\xfa\xed,\xe5|\xa1\xfe\x15\xb9' \ |
|
78 | 78 | '\xbc\xb2\x93\x1f\xf2\x95\xff\xdf,\x1a\xc5\xe7\x17*\x93Oz:>\x0e' |
|
79 | 79 | |
|
80 | 80 | data_non_inlined = '\x00\x00\x00\x01\x00\x00\x00\x00\x00\x01D\x19' \ |
|
81 | 81 | '\x00\x07e\x12\x00\x00\x00\x00\x00\x00\x00\x00\xff\xff\xff\xff' \ |
|
82 | 82 | '\xff\xff\xff\xff\xd1\xf4\xbb\xb0\xbe\xfc\x13\xbd\x8c\xd3\x9d' \ |
|
83 | 83 | '\x0f\xcd\xd9;\x8c\x07\x8cJ/\x00\x00\x00\x00\x00\x00\x00\x00\x00' \ |
|
84 | 84 | '\x00\x00\x00\x00\x00\x00\x01D\x19\x00\x00\x00\x00\x00\xdf\x00' \ |
|
85 | 85 | '\x00\x01q\x00\x00\x00\x01\x00\x00\x00\x01\x00\x00\x00\x00\xff' \ |
|
86 | 86 | '\xff\xff\xff\xc1\x12\xb9\x04\x96\xa4Z1t\x91\xdfsJ\x90\xf0\x9bh' \ |
|
87 | 87 | '\x07l&\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00' \ |
|
88 | 88 | '\x00\x01D\xf8\x00\x00\x00\x00\x01\x1b\x00\x00\x01\xb8\x00\x00' \ |
|
89 | 89 | '\x00\x01\x00\x00\x00\x02\x00\x00\x00\x01\xff\xff\xff\xff\x02\n' \ |
|
90 | 90 | '\x0e\xc6&\xa1\x92\xae6\x0b\x02i\xfe-\xe5\xbao\x05\xd1\xe7\x00' \ |
|
91 | 91 | '\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x01F' \ |
|
92 | 92 | '\x13\x00\x00\x00\x00\x01\xec\x00\x00\x03\x06\x00\x00\x00\x01' \ |
|
93 | 93 | '\x00\x00\x00\x03\x00\x00\x00\x02\xff\xff\xff\xff\x12\xcb\xeby1' \ |
|
94 | 94 | '\xb6\r\x98B\xcb\x07\xbd`\x8f\x92\xd9\xc4\x84\xbdK\x00\x00\x00' \ |
|
95 | 95 | '\x00\x00\x00\x00\x00\x00\x00\x00\x00' |
|
96 | 96 | |
|
97 | 97 | def runtest() : |
|
98 | 98 | |
|
99 | 99 | py_res_1 = py_parseindex(data_inlined, True) |
|
100 | 100 | c_res_1 = parsers.parse_index(data_inlined, True) |
|
101 | 101 | |
|
102 | 102 | py_res_2 = py_parseindex(data_non_inlined, False) |
|
103 | 103 | c_res_2 = parsers.parse_index(data_non_inlined, False) |
|
104 | 104 | |
|
105 | 105 | if py_res_1 != c_res_1: |
|
106 | 106 | print "Parse index result (with inlined data) differs!" |
|
107 | 107 | |
|
108 | 108 | if py_res_2 != c_res_2: |
|
109 | 109 | print "Parse index result (no inlined data) differs!" |
|
110 | 110 | |
|
111 | 111 | print "done" |
|
112 | 112 | |
|
113 | 113 | runtest() |
General Comments 0
You need to be logged in to leave comments.
Login now