Show More
@@ -1,300 +1,298 b'' | |||
|
1 | 1 | # manifest.py - manifest revision 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 i18n import _ |
|
9 | 9 | import mdiff, parsers, error, revlog, util |
|
10 | 10 | import array, struct |
|
11 | 11 | |
|
12 | 12 | class manifestdict(dict): |
|
13 |
def __init__(self |
|
|
14 | if mapping is None: | |
|
15 | mapping = {} | |
|
16 | if flags is None: | |
|
17 | flags = {} | |
|
18 | dict.__init__(self, mapping) | |
|
19 | self._flags = flags | |
|
13 | def __init__(self): | |
|
14 | self._flags = {} | |
|
20 | 15 | def __setitem__(self, k, v): |
|
21 | 16 | assert v is not None |
|
22 | 17 | dict.__setitem__(self, k, v) |
|
23 | 18 | def flags(self, f): |
|
24 | 19 | return self._flags.get(f, "") |
|
25 | 20 | def setflag(self, f, flags): |
|
26 | 21 | """Set the flags (symlink, executable) for path f.""" |
|
27 | 22 | self._flags[f] = flags |
|
28 | 23 | def copy(self): |
|
29 | return manifestdict(self, dict.copy(self._flags)) | |
|
24 | copy = manifestdict() | |
|
25 | dict.__init__(copy, self) | |
|
26 | copy._flags = dict.copy(self._flags) | |
|
27 | return copy | |
|
30 | 28 | def intersectfiles(self, files): |
|
31 | 29 | '''make a new manifestdict with the intersection of self with files |
|
32 | 30 | |
|
33 | 31 | The algorithm assumes that files is much smaller than self.''' |
|
34 | 32 | ret = manifestdict() |
|
35 | 33 | for fn in files: |
|
36 | 34 | if fn in self: |
|
37 | 35 | ret[fn] = self[fn] |
|
38 | 36 | flags = self._flags.get(fn, None) |
|
39 | 37 | if flags: |
|
40 | 38 | ret._flags[fn] = flags |
|
41 | 39 | return ret |
|
42 | 40 | |
|
43 | 41 | def matches(self, match): |
|
44 | 42 | '''generate a new manifest filtered by the match argument''' |
|
45 | 43 | if match.always(): |
|
46 | 44 | return self.copy() |
|
47 | 45 | |
|
48 | 46 | files = match.files() |
|
49 | 47 | if (match.matchfn == match.exact or |
|
50 | 48 | (not match.anypats() and util.all(fn in self for fn in files))): |
|
51 | 49 | return self.intersectfiles(files) |
|
52 | 50 | |
|
53 | 51 | mf = self.copy() |
|
54 | 52 | for fn in mf.keys(): |
|
55 | 53 | if not match(fn): |
|
56 | 54 | del mf[fn] |
|
57 | 55 | return mf |
|
58 | 56 | |
|
59 | 57 | def diff(self, m2, clean=False): |
|
60 | 58 | '''Finds changes between the current manifest and m2. |
|
61 | 59 | |
|
62 | 60 | Args: |
|
63 | 61 | m2: the manifest to which this manifest should be compared. |
|
64 | 62 | clean: if true, include files unchanged between these manifests |
|
65 | 63 | with a None value in the returned dictionary. |
|
66 | 64 | |
|
67 | 65 | The result is returned as a dict with filename as key and |
|
68 | 66 | values of the form ((n1,fl1),(n2,fl2)), where n1/n2 is the |
|
69 | 67 | nodeid in the current/other manifest and fl1/fl2 is the flag |
|
70 | 68 | in the current/other manifest. Where the file does not exist, |
|
71 | 69 | the nodeid will be None and the flags will be the empty |
|
72 | 70 | string. |
|
73 | 71 | ''' |
|
74 | 72 | diff = {} |
|
75 | 73 | |
|
76 | 74 | for fn, n1 in self.iteritems(): |
|
77 | 75 | fl1 = self._flags.get(fn, '') |
|
78 | 76 | n2 = m2.get(fn, None) |
|
79 | 77 | fl2 = m2._flags.get(fn, '') |
|
80 | 78 | if n2 is None: |
|
81 | 79 | fl2 = '' |
|
82 | 80 | if n1 != n2 or fl1 != fl2: |
|
83 | 81 | diff[fn] = ((n1, fl1), (n2, fl2)) |
|
84 | 82 | elif clean: |
|
85 | 83 | diff[fn] = None |
|
86 | 84 | |
|
87 | 85 | for fn, n2 in m2.iteritems(): |
|
88 | 86 | if fn not in self: |
|
89 | 87 | fl2 = m2._flags.get(fn, '') |
|
90 | 88 | diff[fn] = ((None, ''), (n2, fl2)) |
|
91 | 89 | |
|
92 | 90 | return diff |
|
93 | 91 | |
|
94 | 92 | def text(self): |
|
95 | 93 | """Get the full data of this manifest as a bytestring.""" |
|
96 | 94 | fl = sorted(self) |
|
97 | 95 | _checkforbidden(fl) |
|
98 | 96 | |
|
99 | 97 | hex, flags = revlog.hex, self.flags |
|
100 | 98 | # if this is changed to support newlines in filenames, |
|
101 | 99 | # be sure to check the templates/ dir again (especially *-raw.tmpl) |
|
102 | 100 | return ''.join("%s\0%s%s\n" % (f, hex(self[f]), flags(f)) for f in fl) |
|
103 | 101 | |
|
104 | 102 | def fastdelta(self, base, changes): |
|
105 | 103 | """Given a base manifest text as an array.array and a list of changes |
|
106 | 104 | relative to that text, compute a delta that can be used by revlog. |
|
107 | 105 | """ |
|
108 | 106 | delta = [] |
|
109 | 107 | dstart = None |
|
110 | 108 | dend = None |
|
111 | 109 | dline = [""] |
|
112 | 110 | start = 0 |
|
113 | 111 | # zero copy representation of base as a buffer |
|
114 | 112 | addbuf = util.buffer(base) |
|
115 | 113 | |
|
116 | 114 | # start with a readonly loop that finds the offset of |
|
117 | 115 | # each line and creates the deltas |
|
118 | 116 | for f, todelete in changes: |
|
119 | 117 | # bs will either be the index of the item or the insert point |
|
120 | 118 | start, end = _msearch(addbuf, f, start) |
|
121 | 119 | if not todelete: |
|
122 | 120 | l = "%s\0%s%s\n" % (f, revlog.hex(self[f]), self.flags(f)) |
|
123 | 121 | else: |
|
124 | 122 | if start == end: |
|
125 | 123 | # item we want to delete was not found, error out |
|
126 | 124 | raise AssertionError( |
|
127 | 125 | _("failed to remove %s from manifest") % f) |
|
128 | 126 | l = "" |
|
129 | 127 | if dstart is not None and dstart <= start and dend >= start: |
|
130 | 128 | if dend < end: |
|
131 | 129 | dend = end |
|
132 | 130 | if l: |
|
133 | 131 | dline.append(l) |
|
134 | 132 | else: |
|
135 | 133 | if dstart is not None: |
|
136 | 134 | delta.append([dstart, dend, "".join(dline)]) |
|
137 | 135 | dstart = start |
|
138 | 136 | dend = end |
|
139 | 137 | dline = [l] |
|
140 | 138 | |
|
141 | 139 | if dstart is not None: |
|
142 | 140 | delta.append([dstart, dend, "".join(dline)]) |
|
143 | 141 | # apply the delta to the base, and get a delta for addrevision |
|
144 | 142 | deltatext, arraytext = _addlistdelta(base, delta) |
|
145 | 143 | return arraytext, deltatext |
|
146 | 144 | |
|
147 | 145 | def _msearch(m, s, lo=0, hi=None): |
|
148 | 146 | '''return a tuple (start, end) that says where to find s within m. |
|
149 | 147 | |
|
150 | 148 | If the string is found m[start:end] are the line containing |
|
151 | 149 | that string. If start == end the string was not found and |
|
152 | 150 | they indicate the proper sorted insertion point. |
|
153 | 151 | |
|
154 | 152 | m should be a buffer or a string |
|
155 | 153 | s is a string''' |
|
156 | 154 | def advance(i, c): |
|
157 | 155 | while i < lenm and m[i] != c: |
|
158 | 156 | i += 1 |
|
159 | 157 | return i |
|
160 | 158 | if not s: |
|
161 | 159 | return (lo, lo) |
|
162 | 160 | lenm = len(m) |
|
163 | 161 | if not hi: |
|
164 | 162 | hi = lenm |
|
165 | 163 | while lo < hi: |
|
166 | 164 | mid = (lo + hi) // 2 |
|
167 | 165 | start = mid |
|
168 | 166 | while start > 0 and m[start - 1] != '\n': |
|
169 | 167 | start -= 1 |
|
170 | 168 | end = advance(start, '\0') |
|
171 | 169 | if m[start:end] < s: |
|
172 | 170 | # we know that after the null there are 40 bytes of sha1 |
|
173 | 171 | # this translates to the bisect lo = mid + 1 |
|
174 | 172 | lo = advance(end + 40, '\n') + 1 |
|
175 | 173 | else: |
|
176 | 174 | # this translates to the bisect hi = mid |
|
177 | 175 | hi = start |
|
178 | 176 | end = advance(lo, '\0') |
|
179 | 177 | found = m[lo:end] |
|
180 | 178 | if s == found: |
|
181 | 179 | # we know that after the null there are 40 bytes of sha1 |
|
182 | 180 | end = advance(end + 40, '\n') |
|
183 | 181 | return (lo, end + 1) |
|
184 | 182 | else: |
|
185 | 183 | return (lo, lo) |
|
186 | 184 | |
|
187 | 185 | def _checkforbidden(l): |
|
188 | 186 | """Check filenames for illegal characters.""" |
|
189 | 187 | for f in l: |
|
190 | 188 | if '\n' in f or '\r' in f: |
|
191 | 189 | raise error.RevlogError( |
|
192 | 190 | _("'\\n' and '\\r' disallowed in filenames: %r") % f) |
|
193 | 191 | |
|
194 | 192 | |
|
195 | 193 | # apply the changes collected during the bisect loop to our addlist |
|
196 | 194 | # return a delta suitable for addrevision |
|
197 | 195 | def _addlistdelta(addlist, x): |
|
198 | 196 | # for large addlist arrays, building a new array is cheaper |
|
199 | 197 | # than repeatedly modifying the existing one |
|
200 | 198 | currentposition = 0 |
|
201 | 199 | newaddlist = array.array('c') |
|
202 | 200 | |
|
203 | 201 | for start, end, content in x: |
|
204 | 202 | newaddlist += addlist[currentposition:start] |
|
205 | 203 | if content: |
|
206 | 204 | newaddlist += array.array('c', content) |
|
207 | 205 | |
|
208 | 206 | currentposition = end |
|
209 | 207 | |
|
210 | 208 | newaddlist += addlist[currentposition:] |
|
211 | 209 | |
|
212 | 210 | deltatext = "".join(struct.pack(">lll", start, end, len(content)) |
|
213 | 211 | + content for start, end, content in x) |
|
214 | 212 | return deltatext, newaddlist |
|
215 | 213 | |
|
216 | 214 | def _parse(lines): |
|
217 | 215 | mfdict = manifestdict() |
|
218 | 216 | parsers.parse_manifest(mfdict, mfdict._flags, lines) |
|
219 | 217 | return mfdict |
|
220 | 218 | |
|
221 | 219 | class manifest(revlog.revlog): |
|
222 | 220 | def __init__(self, opener): |
|
223 | 221 | # During normal operations, we expect to deal with not more than four |
|
224 | 222 | # revs at a time (such as during commit --amend). When rebasing large |
|
225 | 223 | # stacks of commits, the number can go up, hence the config knob below. |
|
226 | 224 | cachesize = 4 |
|
227 | 225 | opts = getattr(opener, 'options', None) |
|
228 | 226 | if opts is not None: |
|
229 | 227 | cachesize = opts.get('manifestcachesize', cachesize) |
|
230 | 228 | self._mancache = util.lrucachedict(cachesize) |
|
231 | 229 | revlog.revlog.__init__(self, opener, "00manifest.i") |
|
232 | 230 | |
|
233 | 231 | def readdelta(self, node): |
|
234 | 232 | r = self.rev(node) |
|
235 | 233 | return _parse(mdiff.patchtext(self.revdiff(self.deltaparent(r), r))) |
|
236 | 234 | |
|
237 | 235 | def readfast(self, node): |
|
238 | 236 | '''use the faster of readdelta or read''' |
|
239 | 237 | r = self.rev(node) |
|
240 | 238 | deltaparent = self.deltaparent(r) |
|
241 | 239 | if deltaparent != revlog.nullrev and deltaparent in self.parentrevs(r): |
|
242 | 240 | return self.readdelta(node) |
|
243 | 241 | return self.read(node) |
|
244 | 242 | |
|
245 | 243 | def read(self, node): |
|
246 | 244 | if node == revlog.nullid: |
|
247 | 245 | return manifestdict() # don't upset local cache |
|
248 | 246 | if node in self._mancache: |
|
249 | 247 | return self._mancache[node][0] |
|
250 | 248 | text = self.revision(node) |
|
251 | 249 | arraytext = array.array('c', text) |
|
252 | 250 | mapping = _parse(text) |
|
253 | 251 | self._mancache[node] = (mapping, arraytext) |
|
254 | 252 | return mapping |
|
255 | 253 | |
|
256 | 254 | def find(self, node, f): |
|
257 | 255 | '''look up entry for a single file efficiently. |
|
258 | 256 | return (node, flags) pair if found, (None, None) if not.''' |
|
259 | 257 | if node in self._mancache: |
|
260 | 258 | mapping = self._mancache[node][0] |
|
261 | 259 | return mapping.get(f), mapping.flags(f) |
|
262 | 260 | text = self.revision(node) |
|
263 | 261 | start, end = _msearch(text, f) |
|
264 | 262 | if start == end: |
|
265 | 263 | return None, None |
|
266 | 264 | l = text[start:end] |
|
267 | 265 | f, n = l.split('\0') |
|
268 | 266 | return revlog.bin(n[:40]), n[40:-1] |
|
269 | 267 | |
|
270 | 268 | def add(self, map, transaction, link, p1, p2, added, removed): |
|
271 | 269 | if p1 in self._mancache: |
|
272 | 270 | # If our first parent is in the manifest cache, we can |
|
273 | 271 | # compute a delta here using properties we know about the |
|
274 | 272 | # manifest up-front, which may save time later for the |
|
275 | 273 | # revlog layer. |
|
276 | 274 | |
|
277 | 275 | _checkforbidden(added) |
|
278 | 276 | # combine the changed lists into one list for sorting |
|
279 | 277 | work = [(x, False) for x in added] |
|
280 | 278 | work.extend((x, True) for x in removed) |
|
281 | 279 | # this could use heapq.merge() (from Python 2.6+) or equivalent |
|
282 | 280 | # since the lists are already sorted |
|
283 | 281 | work.sort() |
|
284 | 282 | |
|
285 | 283 | arraytext, deltatext = map.fastdelta(self._mancache[p1][1], work) |
|
286 | 284 | cachedelta = self.rev(p1), deltatext |
|
287 | 285 | text = util.buffer(arraytext) |
|
288 | 286 | else: |
|
289 | 287 | # The first parent manifest isn't already loaded, so we'll |
|
290 | 288 | # just encode a fulltext of the manifest and pass that |
|
291 | 289 | # through to the revlog layer, and let it handle the delta |
|
292 | 290 | # process. |
|
293 | 291 | text = map.text() |
|
294 | 292 | arraytext = array.array('c', text) |
|
295 | 293 | cachedelta = None |
|
296 | 294 | |
|
297 | 295 | n = self.addrevision(text, transaction, link, p1, p2, cachedelta) |
|
298 | 296 | self._mancache[n] = (map, arraytext) |
|
299 | 297 | |
|
300 | 298 | return n |
General Comments 0
You need to be logged in to leave comments.
Login now