Show More
@@ -1,413 +1,423 b'' | |||
|
1 | 1 | # match.py - filename matching |
|
2 | 2 | # |
|
3 | 3 | # Copyright 2008, 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 | import re |
|
9 | 9 | import util, pathutil |
|
10 | 10 | from i18n import _ |
|
11 | 11 | |
|
12 | 12 | def _rematcher(regex): |
|
13 | 13 | '''compile the regexp with the best available regexp engine and return a |
|
14 | 14 | matcher function''' |
|
15 | 15 | m = util.re.compile(regex) |
|
16 | 16 | try: |
|
17 | 17 | # slightly faster, provided by facebook's re2 bindings |
|
18 | 18 | return m.test_match |
|
19 | 19 | except AttributeError: |
|
20 | 20 | return m.match |
|
21 | 21 | |
|
22 | 22 | def _expandsets(kindpats, ctx): |
|
23 | 23 | '''Returns the kindpats list with the 'set' patterns expanded.''' |
|
24 | 24 | fset = set() |
|
25 | 25 | other = [] |
|
26 | 26 | |
|
27 | 27 | for kind, pat in kindpats: |
|
28 | 28 | if kind == 'set': |
|
29 | 29 | if not ctx: |
|
30 | 30 | raise util.Abort("fileset expression with no context") |
|
31 | 31 | s = ctx.getfileset(pat) |
|
32 | 32 | fset.update(s) |
|
33 | 33 | continue |
|
34 | 34 | other.append((kind, pat)) |
|
35 | 35 | return fset, other |
|
36 | 36 | |
|
37 | 37 | class match(object): |
|
38 | 38 | def __init__(self, root, cwd, patterns, include=[], exclude=[], |
|
39 | 39 | default='glob', exact=False, auditor=None, ctx=None): |
|
40 | 40 | """build an object to match a set of file patterns |
|
41 | 41 | |
|
42 | 42 | arguments: |
|
43 | 43 | root - the canonical root of the tree you're matching against |
|
44 | 44 | cwd - the current working directory, if relevant |
|
45 | 45 | patterns - patterns to find |
|
46 | 46 | include - patterns to include (unless they are excluded) |
|
47 | 47 | exclude - patterns to exclude (even if they are included) |
|
48 | 48 | default - if a pattern in patterns has no explicit type, assume this one |
|
49 | 49 | exact - patterns are actually filenames (include/exclude still apply) |
|
50 | 50 | |
|
51 | 51 | a pattern is one of: |
|
52 | 52 | 'glob:<glob>' - a glob relative to cwd |
|
53 | 53 | 're:<regexp>' - a regular expression |
|
54 | 54 | 'path:<path>' - a path relative to repository root |
|
55 | 55 | 'relglob:<glob>' - an unrooted glob (*.c matches C files in all dirs) |
|
56 | 56 | 'relpath:<path>' - a path relative to cwd |
|
57 | 57 | 'relre:<regexp>' - a regexp that needn't match the start of a name |
|
58 | 58 | 'set:<fileset>' - a fileset expression |
|
59 | 59 | '<something>' - a pattern of the specified default type |
|
60 | 60 | """ |
|
61 | 61 | |
|
62 | 62 | self._root = root |
|
63 | 63 | self._cwd = cwd |
|
64 | 64 | self._files = [] # exact files and roots of patterns |
|
65 | 65 | self._anypats = bool(include or exclude) |
|
66 | 66 | self._ctx = ctx |
|
67 | 67 | self._always = False |
|
68 | 68 | self._pathrestricted = bool(include or exclude or patterns) |
|
69 | 69 | |
|
70 | 70 | matchfns = [] |
|
71 | 71 | if include: |
|
72 | 72 | kindpats = _normalize(include, 'glob', root, cwd, auditor) |
|
73 | 73 | self.includepat, im = _buildmatch(ctx, kindpats, '(?:/|$)') |
|
74 | 74 | matchfns.append(im) |
|
75 | 75 | if exclude: |
|
76 | 76 | kindpats = _normalize(exclude, 'glob', root, cwd, auditor) |
|
77 | 77 | self.excludepat, em = _buildmatch(ctx, kindpats, '(?:/|$)') |
|
78 | 78 | matchfns.append(lambda f: not em(f)) |
|
79 | 79 | if exact: |
|
80 | 80 | if isinstance(patterns, list): |
|
81 | 81 | self._files = patterns |
|
82 | 82 | else: |
|
83 | 83 | self._files = list(patterns) |
|
84 | 84 | matchfns.append(self.exact) |
|
85 | 85 | elif patterns: |
|
86 | 86 | kindpats = _normalize(patterns, default, root, cwd, auditor) |
|
87 | 87 | self._files = _roots(kindpats) |
|
88 | 88 | self._anypats = self._anypats or _anypats(kindpats) |
|
89 | 89 | self.patternspat, pm = _buildmatch(ctx, kindpats, '$') |
|
90 | 90 | matchfns.append(pm) |
|
91 | 91 | |
|
92 | 92 | if not matchfns: |
|
93 | 93 | m = util.always |
|
94 | 94 | self._always = True |
|
95 | 95 | elif len(matchfns) == 1: |
|
96 | 96 | m = matchfns[0] |
|
97 | 97 | else: |
|
98 | 98 | def m(f): |
|
99 | 99 | for matchfn in matchfns: |
|
100 | 100 | if not matchfn(f): |
|
101 | 101 | return False |
|
102 | 102 | return True |
|
103 | 103 | |
|
104 | 104 | self.matchfn = m |
|
105 | 105 | self._fmap = set(self._files) |
|
106 | 106 | |
|
107 | 107 | def __call__(self, fn): |
|
108 | 108 | return self.matchfn(fn) |
|
109 | 109 | def __iter__(self): |
|
110 | 110 | for f in self._files: |
|
111 | 111 | yield f |
|
112 | 112 | |
|
113 | 113 | # Callbacks related to how the matcher is used by dirstate.walk. |
|
114 | 114 | # Subscribers to these events must monkeypatch the matcher object. |
|
115 | 115 | def bad(self, f, msg): |
|
116 | 116 | '''Callback from dirstate.walk for each explicit file that can't be |
|
117 | 117 | found/accessed, with an error message.''' |
|
118 | 118 | pass |
|
119 | 119 | |
|
120 | 120 | # If an explicitdir is set, it will be called when an explicitly listed |
|
121 | 121 | # directory is visited. |
|
122 | 122 | explicitdir = None |
|
123 | 123 | |
|
124 | 124 | # If an traversedir is set, it will be called when a directory discovered |
|
125 | 125 | # by recursive traversal is visited. |
|
126 | 126 | traversedir = None |
|
127 | 127 | |
|
128 | def abs(self, f): | |
|
129 | '''Convert a repo path back to path that is relative to the root of the | |
|
130 | matcher.''' | |
|
131 | return f | |
|
132 | ||
|
128 | 133 | def rel(self, f): |
|
129 | 134 | '''Convert repo path back to path that is relative to cwd of matcher.''' |
|
130 | 135 | return util.pathto(self._root, self._cwd, f) |
|
131 | 136 | |
|
132 | 137 | def uipath(self, f): |
|
133 | 138 | '''Convert repo path to a display path. If patterns or -I/-X were used |
|
134 | 139 | to create this matcher, the display path will be relative to cwd. |
|
135 | 140 | Otherwise it is relative to the root of the repo.''' |
|
136 | 141 | return (self._pathrestricted and self.rel(f)) or f |
|
137 | 142 | |
|
138 | 143 | def files(self): |
|
139 | 144 | '''Explicitly listed files or patterns or roots: |
|
140 | 145 | if no patterns or .always(): empty list, |
|
141 | 146 | if exact: list exact files, |
|
142 | 147 | if not .anypats(): list all files and dirs, |
|
143 | 148 | else: optimal roots''' |
|
144 | 149 | return self._files |
|
145 | 150 | |
|
146 | 151 | def exact(self, f): |
|
147 | 152 | '''Returns True if f is in .files().''' |
|
148 | 153 | return f in self._fmap |
|
149 | 154 | |
|
150 | 155 | def anypats(self): |
|
151 | 156 | '''Matcher uses patterns or include/exclude.''' |
|
152 | 157 | return self._anypats |
|
153 | 158 | |
|
154 | 159 | def always(self): |
|
155 | 160 | '''Matcher will match everything and .files() will be empty |
|
156 | 161 | - optimization might be possible and necessary.''' |
|
157 | 162 | return self._always |
|
158 | 163 | |
|
159 | 164 | def exact(root, cwd, files): |
|
160 | 165 | return match(root, cwd, files, exact=True) |
|
161 | 166 | |
|
162 | 167 | def always(root, cwd): |
|
163 | 168 | return match(root, cwd, []) |
|
164 | 169 | |
|
165 | 170 | class narrowmatcher(match): |
|
166 | 171 | """Adapt a matcher to work on a subdirectory only. |
|
167 | 172 | |
|
168 | 173 | The paths are remapped to remove/insert the path as needed: |
|
169 | 174 | |
|
170 | 175 | >>> m1 = match('root', '', ['a.txt', 'sub/b.txt']) |
|
171 | 176 | >>> m2 = narrowmatcher('sub', m1) |
|
172 | 177 | >>> bool(m2('a.txt')) |
|
173 | 178 | False |
|
174 | 179 | >>> bool(m2('b.txt')) |
|
175 | 180 | True |
|
176 | 181 | >>> bool(m2.matchfn('a.txt')) |
|
177 | 182 | False |
|
178 | 183 | >>> bool(m2.matchfn('b.txt')) |
|
179 | 184 | True |
|
180 | 185 | >>> m2.files() |
|
181 | 186 | ['b.txt'] |
|
182 | 187 | >>> m2.exact('b.txt') |
|
183 | 188 | True |
|
184 | 189 | >>> m2.rel('b.txt') |
|
185 | 190 | 'b.txt' |
|
186 | 191 | >>> def bad(f, msg): |
|
187 | 192 | ... print "%s: %s" % (f, msg) |
|
188 | 193 | >>> m1.bad = bad |
|
189 | 194 | >>> m2.bad('x.txt', 'No such file') |
|
190 | 195 | sub/x.txt: No such file |
|
196 | >>> m2.abs('c.txt') | |
|
197 | 'sub/c.txt' | |
|
191 | 198 | """ |
|
192 | 199 | |
|
193 | 200 | def __init__(self, path, matcher): |
|
194 | 201 | self._root = matcher._root |
|
195 | 202 | self._cwd = matcher._cwd |
|
196 | 203 | self._path = path |
|
197 | 204 | self._matcher = matcher |
|
198 | 205 | self._always = matcher._always |
|
199 | 206 | self._pathrestricted = matcher._pathrestricted |
|
200 | 207 | |
|
201 | 208 | self._files = [f[len(path) + 1:] for f in matcher._files |
|
202 | 209 | if f.startswith(path + "/")] |
|
203 | 210 | self._anypats = matcher._anypats |
|
204 | 211 | self.matchfn = lambda fn: matcher.matchfn(self._path + "/" + fn) |
|
205 | 212 | self._fmap = set(self._files) |
|
206 | 213 | |
|
214 | def abs(self, f): | |
|
215 | return self._matcher.abs(self._path + "/" + f) | |
|
216 | ||
|
207 | 217 | def bad(self, f, msg): |
|
208 | 218 | self._matcher.bad(self._path + "/" + f, msg) |
|
209 | 219 | |
|
210 | 220 | def patkind(pattern, default=None): |
|
211 | 221 | '''If pattern is 'kind:pat' with a known kind, return kind.''' |
|
212 | 222 | return _patsplit(pattern, default)[0] |
|
213 | 223 | |
|
214 | 224 | def _patsplit(pattern, default): |
|
215 | 225 | """Split a string into the optional pattern kind prefix and the actual |
|
216 | 226 | pattern.""" |
|
217 | 227 | if ':' in pattern: |
|
218 | 228 | kind, pat = pattern.split(':', 1) |
|
219 | 229 | if kind in ('re', 'glob', 'path', 'relglob', 'relpath', 'relre', |
|
220 | 230 | 'listfile', 'listfile0', 'set'): |
|
221 | 231 | return kind, pat |
|
222 | 232 | return default, pattern |
|
223 | 233 | |
|
224 | 234 | def _globre(pat): |
|
225 | 235 | r'''Convert an extended glob string to a regexp string. |
|
226 | 236 | |
|
227 | 237 | >>> print _globre(r'?') |
|
228 | 238 | . |
|
229 | 239 | >>> print _globre(r'*') |
|
230 | 240 | [^/]* |
|
231 | 241 | >>> print _globre(r'**') |
|
232 | 242 | .* |
|
233 | 243 | >>> print _globre(r'**/a') |
|
234 | 244 | (?:.*/)?a |
|
235 | 245 | >>> print _globre(r'a/**/b') |
|
236 | 246 | a\/(?:.*/)?b |
|
237 | 247 | >>> print _globre(r'[a*?!^][^b][!c]') |
|
238 | 248 | [a*?!^][\^b][^c] |
|
239 | 249 | >>> print _globre(r'{a,b}') |
|
240 | 250 | (?:a|b) |
|
241 | 251 | >>> print _globre(r'.\*\?') |
|
242 | 252 | \.\*\? |
|
243 | 253 | ''' |
|
244 | 254 | i, n = 0, len(pat) |
|
245 | 255 | res = '' |
|
246 | 256 | group = 0 |
|
247 | 257 | escape = util.re.escape |
|
248 | 258 | def peek(): |
|
249 | 259 | return i < n and pat[i] |
|
250 | 260 | while i < n: |
|
251 | 261 | c = pat[i] |
|
252 | 262 | i += 1 |
|
253 | 263 | if c not in '*?[{},\\': |
|
254 | 264 | res += escape(c) |
|
255 | 265 | elif c == '*': |
|
256 | 266 | if peek() == '*': |
|
257 | 267 | i += 1 |
|
258 | 268 | if peek() == '/': |
|
259 | 269 | i += 1 |
|
260 | 270 | res += '(?:.*/)?' |
|
261 | 271 | else: |
|
262 | 272 | res += '.*' |
|
263 | 273 | else: |
|
264 | 274 | res += '[^/]*' |
|
265 | 275 | elif c == '?': |
|
266 | 276 | res += '.' |
|
267 | 277 | elif c == '[': |
|
268 | 278 | j = i |
|
269 | 279 | if j < n and pat[j] in '!]': |
|
270 | 280 | j += 1 |
|
271 | 281 | while j < n and pat[j] != ']': |
|
272 | 282 | j += 1 |
|
273 | 283 | if j >= n: |
|
274 | 284 | res += '\\[' |
|
275 | 285 | else: |
|
276 | 286 | stuff = pat[i:j].replace('\\','\\\\') |
|
277 | 287 | i = j + 1 |
|
278 | 288 | if stuff[0] == '!': |
|
279 | 289 | stuff = '^' + stuff[1:] |
|
280 | 290 | elif stuff[0] == '^': |
|
281 | 291 | stuff = '\\' + stuff |
|
282 | 292 | res = '%s[%s]' % (res, stuff) |
|
283 | 293 | elif c == '{': |
|
284 | 294 | group += 1 |
|
285 | 295 | res += '(?:' |
|
286 | 296 | elif c == '}' and group: |
|
287 | 297 | res += ')' |
|
288 | 298 | group -= 1 |
|
289 | 299 | elif c == ',' and group: |
|
290 | 300 | res += '|' |
|
291 | 301 | elif c == '\\': |
|
292 | 302 | p = peek() |
|
293 | 303 | if p: |
|
294 | 304 | i += 1 |
|
295 | 305 | res += escape(p) |
|
296 | 306 | else: |
|
297 | 307 | res += escape(c) |
|
298 | 308 | else: |
|
299 | 309 | res += escape(c) |
|
300 | 310 | return res |
|
301 | 311 | |
|
302 | 312 | def _regex(kind, pat, globsuffix): |
|
303 | 313 | '''Convert a (normalized) pattern of any kind into a regular expression. |
|
304 | 314 | globsuffix is appended to the regexp of globs.''' |
|
305 | 315 | if not pat: |
|
306 | 316 | return '' |
|
307 | 317 | if kind == 're': |
|
308 | 318 | return pat |
|
309 | 319 | if kind == 'path': |
|
310 | 320 | return '^' + util.re.escape(pat) + '(?:/|$)' |
|
311 | 321 | if kind == 'relglob': |
|
312 | 322 | return '(?:|.*/)' + _globre(pat) + globsuffix |
|
313 | 323 | if kind == 'relpath': |
|
314 | 324 | return util.re.escape(pat) + '(?:/|$)' |
|
315 | 325 | if kind == 'relre': |
|
316 | 326 | if pat.startswith('^'): |
|
317 | 327 | return pat |
|
318 | 328 | return '.*' + pat |
|
319 | 329 | return _globre(pat) + globsuffix |
|
320 | 330 | |
|
321 | 331 | def _buildmatch(ctx, kindpats, globsuffix): |
|
322 | 332 | '''Return regexp string and a matcher function for kindpats. |
|
323 | 333 | globsuffix is appended to the regexp of globs.''' |
|
324 | 334 | fset, kindpats = _expandsets(kindpats, ctx) |
|
325 | 335 | if not kindpats: |
|
326 | 336 | return "", fset.__contains__ |
|
327 | 337 | |
|
328 | 338 | regex, mf = _buildregexmatch(kindpats, globsuffix) |
|
329 | 339 | if fset: |
|
330 | 340 | return regex, lambda f: f in fset or mf(f) |
|
331 | 341 | return regex, mf |
|
332 | 342 | |
|
333 | 343 | def _buildregexmatch(kindpats, globsuffix): |
|
334 | 344 | """Build a match function from a list of kinds and kindpats, |
|
335 | 345 | return regexp string and a matcher function.""" |
|
336 | 346 | try: |
|
337 | 347 | regex = '(?:%s)' % '|'.join([_regex(k, p, globsuffix) |
|
338 | 348 | for (k, p) in kindpats]) |
|
339 | 349 | if len(regex) > 20000: |
|
340 | 350 | raise OverflowError |
|
341 | 351 | return regex, _rematcher(regex) |
|
342 | 352 | except OverflowError: |
|
343 | 353 | # We're using a Python with a tiny regex engine and we |
|
344 | 354 | # made it explode, so we'll divide the pattern list in two |
|
345 | 355 | # until it works |
|
346 | 356 | l = len(kindpats) |
|
347 | 357 | if l < 2: |
|
348 | 358 | raise |
|
349 | 359 | regexa, a = _buildregexmatch(kindpats[:l//2], globsuffix) |
|
350 | 360 | regexb, b = _buildregexmatch(kindpats[l//2:], globsuffix) |
|
351 | 361 | return regex, lambda s: a(s) or b(s) |
|
352 | 362 | except re.error: |
|
353 | 363 | for k, p in kindpats: |
|
354 | 364 | try: |
|
355 | 365 | _rematcher('(?:%s)' % _regex(k, p, globsuffix)) |
|
356 | 366 | except re.error: |
|
357 | 367 | raise util.Abort(_("invalid pattern (%s): %s") % (k, p)) |
|
358 | 368 | raise util.Abort(_("invalid pattern")) |
|
359 | 369 | |
|
360 | 370 | def _normalize(patterns, default, root, cwd, auditor): |
|
361 | 371 | '''Convert 'kind:pat' from the patterns list to tuples with kind and |
|
362 | 372 | normalized and rooted patterns and with listfiles expanded.''' |
|
363 | 373 | kindpats = [] |
|
364 | 374 | for kind, pat in [_patsplit(p, default) for p in patterns]: |
|
365 | 375 | if kind in ('glob', 'relpath'): |
|
366 | 376 | pat = pathutil.canonpath(root, cwd, pat, auditor) |
|
367 | 377 | elif kind in ('relglob', 'path'): |
|
368 | 378 | pat = util.normpath(pat) |
|
369 | 379 | elif kind in ('listfile', 'listfile0'): |
|
370 | 380 | try: |
|
371 | 381 | files = util.readfile(pat) |
|
372 | 382 | if kind == 'listfile0': |
|
373 | 383 | files = files.split('\0') |
|
374 | 384 | else: |
|
375 | 385 | files = files.splitlines() |
|
376 | 386 | files = [f for f in files if f] |
|
377 | 387 | except EnvironmentError: |
|
378 | 388 | raise util.Abort(_("unable to read file list (%s)") % pat) |
|
379 | 389 | kindpats += _normalize(files, default, root, cwd, auditor) |
|
380 | 390 | continue |
|
381 | 391 | # else: re or relre - which cannot be normalized |
|
382 | 392 | kindpats.append((kind, pat)) |
|
383 | 393 | return kindpats |
|
384 | 394 | |
|
385 | 395 | def _roots(kindpats): |
|
386 | 396 | '''return roots and exact explicitly listed files from patterns |
|
387 | 397 | |
|
388 | 398 | >>> _roots([('glob', 'g/*'), ('glob', 'g'), ('glob', 'g*')]) |
|
389 | 399 | ['g', 'g', '.'] |
|
390 | 400 | >>> _roots([('relpath', 'r'), ('path', 'p/p'), ('path', '')]) |
|
391 | 401 | ['r', 'p/p', '.'] |
|
392 | 402 | >>> _roots([('relglob', 'rg*'), ('re', 're/'), ('relre', 'rr')]) |
|
393 | 403 | ['.', '.', '.'] |
|
394 | 404 | ''' |
|
395 | 405 | r = [] |
|
396 | 406 | for kind, pat in kindpats: |
|
397 | 407 | if kind == 'glob': # find the non-glob prefix |
|
398 | 408 | root = [] |
|
399 | 409 | for p in pat.split('/'): |
|
400 | 410 | if '[' in p or '{' in p or '*' in p or '?' in p: |
|
401 | 411 | break |
|
402 | 412 | root.append(p) |
|
403 | 413 | r.append('/'.join(root) or '.') |
|
404 | 414 | elif kind in ('relpath', 'path'): |
|
405 | 415 | r.append(pat or '.') |
|
406 | 416 | else: # relglob, re, relre |
|
407 | 417 | r.append('.') |
|
408 | 418 | return r |
|
409 | 419 | |
|
410 | 420 | def _anypats(kindpats): |
|
411 | 421 | for kind, pat in kindpats: |
|
412 | 422 | if kind in ('glob', 're', 'relglob', 'relre', 'set'): |
|
413 | 423 | return True |
General Comments 0
You need to be logged in to leave comments.
Login now