##// END OF EJS Templates
patch: display a nice error for invalid base85 data...
Patrick Mezard -
r16522:a8065323 stable
parent child Browse files
Show More
@@ -1,180 +1,180
1 1 /*
2 2 base85 codec
3 3
4 4 Copyright 2006 Brendan Cully <brendan@kublai.com>
5 5
6 6 This software may be used and distributed according to the terms of
7 7 the GNU General Public License, incorporated herein by reference.
8 8
9 9 Largely based on git's implementation
10 10 */
11 11
12 12 #include <Python.h>
13 13
14 14 #include "util.h"
15 15
16 16 static const char b85chars[] = "0123456789ABCDEFGHIJKLMNOPQRSTUVWXYZ"
17 17 "abcdefghijklmnopqrstuvwxyz!#$%&()*+-;<=>?@^_`{|}~";
18 18 static char b85dec[256];
19 19
20 20 static void
21 21 b85prep(void)
22 22 {
23 23 int i;
24 24
25 25 memset(b85dec, 0, sizeof(b85dec));
26 26 for (i = 0; i < sizeof(b85chars); i++)
27 27 b85dec[(int)(b85chars[i])] = i + 1;
28 28 }
29 29
30 30 static PyObject *
31 31 b85encode(PyObject *self, PyObject *args)
32 32 {
33 33 const unsigned char *text;
34 34 PyObject *out;
35 35 char *dst;
36 36 int len, olen, i;
37 37 unsigned int acc, val, ch;
38 38 int pad = 0;
39 39
40 40 if (!PyArg_ParseTuple(args, "s#|i", &text, &len, &pad))
41 41 return NULL;
42 42
43 43 if (pad)
44 44 olen = ((len + 3) / 4 * 5) - 3;
45 45 else {
46 46 olen = len % 4;
47 47 if (olen)
48 48 olen++;
49 49 olen += len / 4 * 5;
50 50 }
51 51 if (!(out = PyBytes_FromStringAndSize(NULL, olen + 3)))
52 52 return NULL;
53 53
54 54 dst = PyBytes_AsString(out);
55 55
56 56 while (len) {
57 57 acc = 0;
58 58 for (i = 24; i >= 0; i -= 8) {
59 59 ch = *text++;
60 60 acc |= ch << i;
61 61 if (--len == 0)
62 62 break;
63 63 }
64 64 for (i = 4; i >= 0; i--) {
65 65 val = acc % 85;
66 66 acc /= 85;
67 67 dst[i] = b85chars[val];
68 68 }
69 69 dst += 5;
70 70 }
71 71
72 72 if (!pad)
73 73 _PyBytes_Resize(&out, olen);
74 74
75 75 return out;
76 76 }
77 77
78 78 static PyObject *
79 79 b85decode(PyObject *self, PyObject *args)
80 80 {
81 81 PyObject *out;
82 82 const char *text;
83 83 char *dst;
84 84 int len, i, j, olen, c, cap;
85 85 unsigned int acc;
86 86
87 87 if (!PyArg_ParseTuple(args, "s#", &text, &len))
88 88 return NULL;
89 89
90 90 olen = len / 5 * 4;
91 91 i = len % 5;
92 92 if (i)
93 93 olen += i - 1;
94 94 if (!(out = PyBytes_FromStringAndSize(NULL, olen)))
95 95 return NULL;
96 96
97 97 dst = PyBytes_AsString(out);
98 98
99 99 i = 0;
100 100 while (i < len)
101 101 {
102 102 acc = 0;
103 103 cap = len - i - 1;
104 104 if (cap > 4)
105 105 cap = 4;
106 106 for (j = 0; j < cap; i++, j++)
107 107 {
108 108 c = b85dec[(int)*text++] - 1;
109 109 if (c < 0)
110 110 return PyErr_Format(
111 111 PyExc_ValueError,
112 "Bad base85 character at position %d", i);
112 "bad base85 character at position %d", i);
113 113 acc = acc * 85 + c;
114 114 }
115 115 if (i++ < len)
116 116 {
117 117 c = b85dec[(int)*text++] - 1;
118 118 if (c < 0)
119 119 return PyErr_Format(
120 120 PyExc_ValueError,
121 "Bad base85 character at position %d", i);
121 "bad base85 character at position %d", i);
122 122 /* overflow detection: 0xffffffff == "|NsC0",
123 123 * "|NsC" == 0x03030303 */
124 124 if (acc > 0x03030303 || (acc *= 85) > 0xffffffff - c)
125 125 return PyErr_Format(
126 126 PyExc_ValueError,
127 "Bad base85 sequence at position %d", i);
127 "bad base85 sequence at position %d", i);
128 128 acc += c;
129 129 }
130 130
131 131 cap = olen < 4 ? olen : 4;
132 132 olen -= cap;
133 133 for (j = 0; j < 4 - cap; j++)
134 134 acc *= 85;
135 135 if (cap && cap < 4)
136 136 acc += 0xffffff >> (cap - 1) * 8;
137 137 for (j = 0; j < cap; j++)
138 138 {
139 139 acc = (acc << 8) | (acc >> 24);
140 140 *dst++ = acc;
141 141 }
142 142 }
143 143
144 144 return out;
145 145 }
146 146
147 147 static char base85_doc[] = "Base85 Data Encoding";
148 148
149 149 static PyMethodDef methods[] = {
150 150 {"b85encode", b85encode, METH_VARARGS,
151 151 "Encode text in base85.\n\n"
152 152 "If the second parameter is true, pad the result to a multiple of "
153 153 "five characters.\n"},
154 154 {"b85decode", b85decode, METH_VARARGS, "Decode base85 text.\n"},
155 155 {NULL, NULL}
156 156 };
157 157
158 158 #ifdef IS_PY3K
159 159 static struct PyModuleDef base85_module = {
160 160 PyModuleDef_HEAD_INIT,
161 161 "base85",
162 162 base85_doc,
163 163 -1,
164 164 methods
165 165 };
166 166
167 167 PyMODINIT_FUNC PyInit_base85(void)
168 168 {
169 169 b85prep();
170 170
171 171 return PyModule_Create(&base85_module);
172 172 }
173 173 #else
174 174 PyMODINIT_FUNC initbase85(void)
175 175 {
176 176 Py_InitModule3("base85", methods, base85_doc);
177 177
178 178 b85prep();
179 179 }
180 180 #endif
@@ -1,1879 +1,1883
1 1 # patch.py - patch file parsing routines
2 2 #
3 3 # Copyright 2006 Brendan Cully <brendan@kublai.com>
4 4 # Copyright 2007 Chris Mason <chris.mason@oracle.com>
5 5 #
6 6 # This software may be used and distributed according to the terms of the
7 7 # GNU General Public License version 2 or any later version.
8 8
9 9 import cStringIO, email.Parser, os, errno, re
10 10 import tempfile, zlib, shutil
11 11
12 12 from i18n import _
13 13 from node import hex, nullid, short
14 14 import base85, mdiff, scmutil, util, diffhelpers, copies, encoding, error
15 15 import context
16 16
17 17 gitre = re.compile('diff --git a/(.*) b/(.*)')
18 18
19 19 class PatchError(Exception):
20 20 pass
21 21
22 22
23 23 # public functions
24 24
25 25 def split(stream):
26 26 '''return an iterator of individual patches from a stream'''
27 27 def isheader(line, inheader):
28 28 if inheader and line[0] in (' ', '\t'):
29 29 # continuation
30 30 return True
31 31 if line[0] in (' ', '-', '+'):
32 32 # diff line - don't check for header pattern in there
33 33 return False
34 34 l = line.split(': ', 1)
35 35 return len(l) == 2 and ' ' not in l[0]
36 36
37 37 def chunk(lines):
38 38 return cStringIO.StringIO(''.join(lines))
39 39
40 40 def hgsplit(stream, cur):
41 41 inheader = True
42 42
43 43 for line in stream:
44 44 if not line.strip():
45 45 inheader = False
46 46 if not inheader and line.startswith('# HG changeset patch'):
47 47 yield chunk(cur)
48 48 cur = []
49 49 inheader = True
50 50
51 51 cur.append(line)
52 52
53 53 if cur:
54 54 yield chunk(cur)
55 55
56 56 def mboxsplit(stream, cur):
57 57 for line in stream:
58 58 if line.startswith('From '):
59 59 for c in split(chunk(cur[1:])):
60 60 yield c
61 61 cur = []
62 62
63 63 cur.append(line)
64 64
65 65 if cur:
66 66 for c in split(chunk(cur[1:])):
67 67 yield c
68 68
69 69 def mimesplit(stream, cur):
70 70 def msgfp(m):
71 71 fp = cStringIO.StringIO()
72 72 g = email.Generator.Generator(fp, mangle_from_=False)
73 73 g.flatten(m)
74 74 fp.seek(0)
75 75 return fp
76 76
77 77 for line in stream:
78 78 cur.append(line)
79 79 c = chunk(cur)
80 80
81 81 m = email.Parser.Parser().parse(c)
82 82 if not m.is_multipart():
83 83 yield msgfp(m)
84 84 else:
85 85 ok_types = ('text/plain', 'text/x-diff', 'text/x-patch')
86 86 for part in m.walk():
87 87 ct = part.get_content_type()
88 88 if ct not in ok_types:
89 89 continue
90 90 yield msgfp(part)
91 91
92 92 def headersplit(stream, cur):
93 93 inheader = False
94 94
95 95 for line in stream:
96 96 if not inheader and isheader(line, inheader):
97 97 yield chunk(cur)
98 98 cur = []
99 99 inheader = True
100 100 if inheader and not isheader(line, inheader):
101 101 inheader = False
102 102
103 103 cur.append(line)
104 104
105 105 if cur:
106 106 yield chunk(cur)
107 107
108 108 def remainder(cur):
109 109 yield chunk(cur)
110 110
111 111 class fiter(object):
112 112 def __init__(self, fp):
113 113 self.fp = fp
114 114
115 115 def __iter__(self):
116 116 return self
117 117
118 118 def next(self):
119 119 l = self.fp.readline()
120 120 if not l:
121 121 raise StopIteration
122 122 return l
123 123
124 124 inheader = False
125 125 cur = []
126 126
127 127 mimeheaders = ['content-type']
128 128
129 129 if not util.safehasattr(stream, 'next'):
130 130 # http responses, for example, have readline but not next
131 131 stream = fiter(stream)
132 132
133 133 for line in stream:
134 134 cur.append(line)
135 135 if line.startswith('# HG changeset patch'):
136 136 return hgsplit(stream, cur)
137 137 elif line.startswith('From '):
138 138 return mboxsplit(stream, cur)
139 139 elif isheader(line, inheader):
140 140 inheader = True
141 141 if line.split(':', 1)[0].lower() in mimeheaders:
142 142 # let email parser handle this
143 143 return mimesplit(stream, cur)
144 144 elif line.startswith('--- ') and inheader:
145 145 # No evil headers seen by diff start, split by hand
146 146 return headersplit(stream, cur)
147 147 # Not enough info, keep reading
148 148
149 149 # if we are here, we have a very plain patch
150 150 return remainder(cur)
151 151
152 152 def extract(ui, fileobj):
153 153 '''extract patch from data read from fileobj.
154 154
155 155 patch can be a normal patch or contained in an email message.
156 156
157 157 return tuple (filename, message, user, date, branch, node, p1, p2).
158 158 Any item in the returned tuple can be None. If filename is None,
159 159 fileobj did not contain a patch. Caller must unlink filename when done.'''
160 160
161 161 # attempt to detect the start of a patch
162 162 # (this heuristic is borrowed from quilt)
163 163 diffre = re.compile(r'^(?:Index:[ \t]|diff[ \t]|RCS file: |'
164 164 r'retrieving revision [0-9]+(\.[0-9]+)*$|'
165 165 r'---[ \t].*?^\+\+\+[ \t]|'
166 166 r'\*\*\*[ \t].*?^---[ \t])', re.MULTILINE|re.DOTALL)
167 167
168 168 fd, tmpname = tempfile.mkstemp(prefix='hg-patch-')
169 169 tmpfp = os.fdopen(fd, 'w')
170 170 try:
171 171 msg = email.Parser.Parser().parse(fileobj)
172 172
173 173 subject = msg['Subject']
174 174 user = msg['From']
175 175 if not subject and not user:
176 176 # Not an email, restore parsed headers if any
177 177 subject = '\n'.join(': '.join(h) for h in msg.items()) + '\n'
178 178
179 179 gitsendmail = 'git-send-email' in msg.get('X-Mailer', '')
180 180 # should try to parse msg['Date']
181 181 date = None
182 182 nodeid = None
183 183 branch = None
184 184 parents = []
185 185
186 186 if subject:
187 187 if subject.startswith('[PATCH'):
188 188 pend = subject.find(']')
189 189 if pend >= 0:
190 190 subject = subject[pend + 1:].lstrip()
191 191 subject = re.sub(r'\n[ \t]+', ' ', subject)
192 192 ui.debug('Subject: %s\n' % subject)
193 193 if user:
194 194 ui.debug('From: %s\n' % user)
195 195 diffs_seen = 0
196 196 ok_types = ('text/plain', 'text/x-diff', 'text/x-patch')
197 197 message = ''
198 198 for part in msg.walk():
199 199 content_type = part.get_content_type()
200 200 ui.debug('Content-Type: %s\n' % content_type)
201 201 if content_type not in ok_types:
202 202 continue
203 203 payload = part.get_payload(decode=True)
204 204 m = diffre.search(payload)
205 205 if m:
206 206 hgpatch = False
207 207 hgpatchheader = False
208 208 ignoretext = False
209 209
210 210 ui.debug('found patch at byte %d\n' % m.start(0))
211 211 diffs_seen += 1
212 212 cfp = cStringIO.StringIO()
213 213 for line in payload[:m.start(0)].splitlines():
214 214 if line.startswith('# HG changeset patch') and not hgpatch:
215 215 ui.debug('patch generated by hg export\n')
216 216 hgpatch = True
217 217 hgpatchheader = True
218 218 # drop earlier commit message content
219 219 cfp.seek(0)
220 220 cfp.truncate()
221 221 subject = None
222 222 elif hgpatchheader:
223 223 if line.startswith('# User '):
224 224 user = line[7:]
225 225 ui.debug('From: %s\n' % user)
226 226 elif line.startswith("# Date "):
227 227 date = line[7:]
228 228 elif line.startswith("# Branch "):
229 229 branch = line[9:]
230 230 elif line.startswith("# Node ID "):
231 231 nodeid = line[10:]
232 232 elif line.startswith("# Parent "):
233 233 parents.append(line[9:].lstrip())
234 234 elif not line.startswith("# "):
235 235 hgpatchheader = False
236 236 elif line == '---' and gitsendmail:
237 237 ignoretext = True
238 238 if not hgpatchheader and not ignoretext:
239 239 cfp.write(line)
240 240 cfp.write('\n')
241 241 message = cfp.getvalue()
242 242 if tmpfp:
243 243 tmpfp.write(payload)
244 244 if not payload.endswith('\n'):
245 245 tmpfp.write('\n')
246 246 elif not diffs_seen and message and content_type == 'text/plain':
247 247 message += '\n' + payload
248 248 except:
249 249 tmpfp.close()
250 250 os.unlink(tmpname)
251 251 raise
252 252
253 253 if subject and not message.startswith(subject):
254 254 message = '%s\n%s' % (subject, message)
255 255 tmpfp.close()
256 256 if not diffs_seen:
257 257 os.unlink(tmpname)
258 258 return None, message, user, date, branch, None, None, None
259 259 p1 = parents and parents.pop(0) or None
260 260 p2 = parents and parents.pop(0) or None
261 261 return tmpname, message, user, date, branch, nodeid, p1, p2
262 262
263 263 class patchmeta(object):
264 264 """Patched file metadata
265 265
266 266 'op' is the performed operation within ADD, DELETE, RENAME, MODIFY
267 267 or COPY. 'path' is patched file path. 'oldpath' is set to the
268 268 origin file when 'op' is either COPY or RENAME, None otherwise. If
269 269 file mode is changed, 'mode' is a tuple (islink, isexec) where
270 270 'islink' is True if the file is a symlink and 'isexec' is True if
271 271 the file is executable. Otherwise, 'mode' is None.
272 272 """
273 273 def __init__(self, path):
274 274 self.path = path
275 275 self.oldpath = None
276 276 self.mode = None
277 277 self.op = 'MODIFY'
278 278 self.binary = False
279 279
280 280 def setmode(self, mode):
281 281 islink = mode & 020000
282 282 isexec = mode & 0100
283 283 self.mode = (islink, isexec)
284 284
285 285 def copy(self):
286 286 other = patchmeta(self.path)
287 287 other.oldpath = self.oldpath
288 288 other.mode = self.mode
289 289 other.op = self.op
290 290 other.binary = self.binary
291 291 return other
292 292
293 293 def _ispatchinga(self, afile):
294 294 if afile == '/dev/null':
295 295 return self.op == 'ADD'
296 296 return afile == 'a/' + (self.oldpath or self.path)
297 297
298 298 def _ispatchingb(self, bfile):
299 299 if bfile == '/dev/null':
300 300 return self.op == 'DELETE'
301 301 return bfile == 'b/' + self.path
302 302
303 303 def ispatching(self, afile, bfile):
304 304 return self._ispatchinga(afile) and self._ispatchingb(bfile)
305 305
306 306 def __repr__(self):
307 307 return "<patchmeta %s %r>" % (self.op, self.path)
308 308
309 309 def readgitpatch(lr):
310 310 """extract git-style metadata about patches from <patchname>"""
311 311
312 312 # Filter patch for git information
313 313 gp = None
314 314 gitpatches = []
315 315 for line in lr:
316 316 line = line.rstrip(' \r\n')
317 317 if line.startswith('diff --git'):
318 318 m = gitre.match(line)
319 319 if m:
320 320 if gp:
321 321 gitpatches.append(gp)
322 322 dst = m.group(2)
323 323 gp = patchmeta(dst)
324 324 elif gp:
325 325 if line.startswith('--- '):
326 326 gitpatches.append(gp)
327 327 gp = None
328 328 continue
329 329 if line.startswith('rename from '):
330 330 gp.op = 'RENAME'
331 331 gp.oldpath = line[12:]
332 332 elif line.startswith('rename to '):
333 333 gp.path = line[10:]
334 334 elif line.startswith('copy from '):
335 335 gp.op = 'COPY'
336 336 gp.oldpath = line[10:]
337 337 elif line.startswith('copy to '):
338 338 gp.path = line[8:]
339 339 elif line.startswith('deleted file'):
340 340 gp.op = 'DELETE'
341 341 elif line.startswith('new file mode '):
342 342 gp.op = 'ADD'
343 343 gp.setmode(int(line[-6:], 8))
344 344 elif line.startswith('new mode '):
345 345 gp.setmode(int(line[-6:], 8))
346 346 elif line.startswith('GIT binary patch'):
347 347 gp.binary = True
348 348 if gp:
349 349 gitpatches.append(gp)
350 350
351 351 return gitpatches
352 352
353 353 class linereader(object):
354 354 # simple class to allow pushing lines back into the input stream
355 355 def __init__(self, fp):
356 356 self.fp = fp
357 357 self.buf = []
358 358
359 359 def push(self, line):
360 360 if line is not None:
361 361 self.buf.append(line)
362 362
363 363 def readline(self):
364 364 if self.buf:
365 365 l = self.buf[0]
366 366 del self.buf[0]
367 367 return l
368 368 return self.fp.readline()
369 369
370 370 def __iter__(self):
371 371 while True:
372 372 l = self.readline()
373 373 if not l:
374 374 break
375 375 yield l
376 376
377 377 class abstractbackend(object):
378 378 def __init__(self, ui):
379 379 self.ui = ui
380 380
381 381 def getfile(self, fname):
382 382 """Return target file data and flags as a (data, (islink,
383 383 isexec)) tuple.
384 384 """
385 385 raise NotImplementedError
386 386
387 387 def setfile(self, fname, data, mode, copysource):
388 388 """Write data to target file fname and set its mode. mode is a
389 389 (islink, isexec) tuple. If data is None, the file content should
390 390 be left unchanged. If the file is modified after being copied,
391 391 copysource is set to the original file name.
392 392 """
393 393 raise NotImplementedError
394 394
395 395 def unlink(self, fname):
396 396 """Unlink target file."""
397 397 raise NotImplementedError
398 398
399 399 def writerej(self, fname, failed, total, lines):
400 400 """Write rejected lines for fname. total is the number of hunks
401 401 which failed to apply and total the total number of hunks for this
402 402 files.
403 403 """
404 404 pass
405 405
406 406 def exists(self, fname):
407 407 raise NotImplementedError
408 408
409 409 class fsbackend(abstractbackend):
410 410 def __init__(self, ui, basedir):
411 411 super(fsbackend, self).__init__(ui)
412 412 self.opener = scmutil.opener(basedir)
413 413
414 414 def _join(self, f):
415 415 return os.path.join(self.opener.base, f)
416 416
417 417 def getfile(self, fname):
418 418 path = self._join(fname)
419 419 if os.path.islink(path):
420 420 return (os.readlink(path), (True, False))
421 421 isexec = False
422 422 try:
423 423 isexec = os.lstat(path).st_mode & 0100 != 0
424 424 except OSError, e:
425 425 if e.errno != errno.ENOENT:
426 426 raise
427 427 return (self.opener.read(fname), (False, isexec))
428 428
429 429 def setfile(self, fname, data, mode, copysource):
430 430 islink, isexec = mode
431 431 if data is None:
432 432 util.setflags(self._join(fname), islink, isexec)
433 433 return
434 434 if islink:
435 435 self.opener.symlink(data, fname)
436 436 else:
437 437 self.opener.write(fname, data)
438 438 if isexec:
439 439 util.setflags(self._join(fname), False, True)
440 440
441 441 def unlink(self, fname):
442 442 try:
443 443 util.unlinkpath(self._join(fname))
444 444 except OSError, inst:
445 445 if inst.errno != errno.ENOENT:
446 446 raise
447 447
448 448 def writerej(self, fname, failed, total, lines):
449 449 fname = fname + ".rej"
450 450 self.ui.warn(
451 451 _("%d out of %d hunks FAILED -- saving rejects to file %s\n") %
452 452 (failed, total, fname))
453 453 fp = self.opener(fname, 'w')
454 454 fp.writelines(lines)
455 455 fp.close()
456 456
457 457 def exists(self, fname):
458 458 return os.path.lexists(self._join(fname))
459 459
460 460 class workingbackend(fsbackend):
461 461 def __init__(self, ui, repo, similarity):
462 462 super(workingbackend, self).__init__(ui, repo.root)
463 463 self.repo = repo
464 464 self.similarity = similarity
465 465 self.removed = set()
466 466 self.changed = set()
467 467 self.copied = []
468 468
469 469 def _checkknown(self, fname):
470 470 if self.repo.dirstate[fname] == '?' and self.exists(fname):
471 471 raise PatchError(_('cannot patch %s: file is not tracked') % fname)
472 472
473 473 def setfile(self, fname, data, mode, copysource):
474 474 self._checkknown(fname)
475 475 super(workingbackend, self).setfile(fname, data, mode, copysource)
476 476 if copysource is not None:
477 477 self.copied.append((copysource, fname))
478 478 self.changed.add(fname)
479 479
480 480 def unlink(self, fname):
481 481 self._checkknown(fname)
482 482 super(workingbackend, self).unlink(fname)
483 483 self.removed.add(fname)
484 484 self.changed.add(fname)
485 485
486 486 def close(self):
487 487 wctx = self.repo[None]
488 488 addremoved = set(self.changed)
489 489 for src, dst in self.copied:
490 490 scmutil.dirstatecopy(self.ui, self.repo, wctx, src, dst)
491 491 if self.removed:
492 492 wctx.forget(sorted(self.removed))
493 493 for f in self.removed:
494 494 if f not in self.repo.dirstate:
495 495 # File was deleted and no longer belongs to the
496 496 # dirstate, it was probably marked added then
497 497 # deleted, and should not be considered by
498 498 # addremove().
499 499 addremoved.discard(f)
500 500 if addremoved:
501 501 cwd = self.repo.getcwd()
502 502 if cwd:
503 503 addremoved = [util.pathto(self.repo.root, cwd, f)
504 504 for f in addremoved]
505 505 scmutil.addremove(self.repo, addremoved, similarity=self.similarity)
506 506 return sorted(self.changed)
507 507
508 508 class filestore(object):
509 509 def __init__(self, maxsize=None):
510 510 self.opener = None
511 511 self.files = {}
512 512 self.created = 0
513 513 self.maxsize = maxsize
514 514 if self.maxsize is None:
515 515 self.maxsize = 4*(2**20)
516 516 self.size = 0
517 517 self.data = {}
518 518
519 519 def setfile(self, fname, data, mode, copied=None):
520 520 if self.maxsize < 0 or (len(data) + self.size) <= self.maxsize:
521 521 self.data[fname] = (data, mode, copied)
522 522 self.size += len(data)
523 523 else:
524 524 if self.opener is None:
525 525 root = tempfile.mkdtemp(prefix='hg-patch-')
526 526 self.opener = scmutil.opener(root)
527 527 # Avoid filename issues with these simple names
528 528 fn = str(self.created)
529 529 self.opener.write(fn, data)
530 530 self.created += 1
531 531 self.files[fname] = (fn, mode, copied)
532 532
533 533 def getfile(self, fname):
534 534 if fname in self.data:
535 535 return self.data[fname]
536 536 if not self.opener or fname not in self.files:
537 537 raise IOError()
538 538 fn, mode, copied = self.files[fname]
539 539 return self.opener.read(fn), mode, copied
540 540
541 541 def close(self):
542 542 if self.opener:
543 543 shutil.rmtree(self.opener.base)
544 544
545 545 class repobackend(abstractbackend):
546 546 def __init__(self, ui, repo, ctx, store):
547 547 super(repobackend, self).__init__(ui)
548 548 self.repo = repo
549 549 self.ctx = ctx
550 550 self.store = store
551 551 self.changed = set()
552 552 self.removed = set()
553 553 self.copied = {}
554 554
555 555 def _checkknown(self, fname):
556 556 if fname not in self.ctx:
557 557 raise PatchError(_('cannot patch %s: file is not tracked') % fname)
558 558
559 559 def getfile(self, fname):
560 560 try:
561 561 fctx = self.ctx[fname]
562 562 except error.LookupError:
563 563 raise IOError()
564 564 flags = fctx.flags()
565 565 return fctx.data(), ('l' in flags, 'x' in flags)
566 566
567 567 def setfile(self, fname, data, mode, copysource):
568 568 if copysource:
569 569 self._checkknown(copysource)
570 570 if data is None:
571 571 data = self.ctx[fname].data()
572 572 self.store.setfile(fname, data, mode, copysource)
573 573 self.changed.add(fname)
574 574 if copysource:
575 575 self.copied[fname] = copysource
576 576
577 577 def unlink(self, fname):
578 578 self._checkknown(fname)
579 579 self.removed.add(fname)
580 580
581 581 def exists(self, fname):
582 582 return fname in self.ctx
583 583
584 584 def close(self):
585 585 return self.changed | self.removed
586 586
587 587 # @@ -start,len +start,len @@ or @@ -start +start @@ if len is 1
588 588 unidesc = re.compile('@@ -(\d+)(?:,(\d+))? \+(\d+)(?:,(\d+))? @@')
589 589 contextdesc = re.compile('(?:---|\*\*\*) (\d+)(?:,(\d+))? (?:---|\*\*\*)')
590 590 eolmodes = ['strict', 'crlf', 'lf', 'auto']
591 591
592 592 class patchfile(object):
593 593 def __init__(self, ui, gp, backend, store, eolmode='strict'):
594 594 self.fname = gp.path
595 595 self.eolmode = eolmode
596 596 self.eol = None
597 597 self.backend = backend
598 598 self.ui = ui
599 599 self.lines = []
600 600 self.exists = False
601 601 self.missing = True
602 602 self.mode = gp.mode
603 603 self.copysource = gp.oldpath
604 604 self.create = gp.op in ('ADD', 'COPY', 'RENAME')
605 605 self.remove = gp.op == 'DELETE'
606 606 try:
607 607 if self.copysource is None:
608 608 data, mode = backend.getfile(self.fname)
609 609 self.exists = True
610 610 else:
611 611 data, mode = store.getfile(self.copysource)[:2]
612 612 self.exists = backend.exists(self.fname)
613 613 self.missing = False
614 614 if data:
615 615 self.lines = mdiff.splitnewlines(data)
616 616 if self.mode is None:
617 617 self.mode = mode
618 618 if self.lines:
619 619 # Normalize line endings
620 620 if self.lines[0].endswith('\r\n'):
621 621 self.eol = '\r\n'
622 622 elif self.lines[0].endswith('\n'):
623 623 self.eol = '\n'
624 624 if eolmode != 'strict':
625 625 nlines = []
626 626 for l in self.lines:
627 627 if l.endswith('\r\n'):
628 628 l = l[:-2] + '\n'
629 629 nlines.append(l)
630 630 self.lines = nlines
631 631 except IOError:
632 632 if self.create:
633 633 self.missing = False
634 634 if self.mode is None:
635 635 self.mode = (False, False)
636 636 if self.missing:
637 637 self.ui.warn(_("unable to find '%s' for patching\n") % self.fname)
638 638
639 639 self.hash = {}
640 640 self.dirty = 0
641 641 self.offset = 0
642 642 self.skew = 0
643 643 self.rej = []
644 644 self.fileprinted = False
645 645 self.printfile(False)
646 646 self.hunks = 0
647 647
648 648 def writelines(self, fname, lines, mode):
649 649 if self.eolmode == 'auto':
650 650 eol = self.eol
651 651 elif self.eolmode == 'crlf':
652 652 eol = '\r\n'
653 653 else:
654 654 eol = '\n'
655 655
656 656 if self.eolmode != 'strict' and eol and eol != '\n':
657 657 rawlines = []
658 658 for l in lines:
659 659 if l and l[-1] == '\n':
660 660 l = l[:-1] + eol
661 661 rawlines.append(l)
662 662 lines = rawlines
663 663
664 664 self.backend.setfile(fname, ''.join(lines), mode, self.copysource)
665 665
666 666 def printfile(self, warn):
667 667 if self.fileprinted:
668 668 return
669 669 if warn or self.ui.verbose:
670 670 self.fileprinted = True
671 671 s = _("patching file %s\n") % self.fname
672 672 if warn:
673 673 self.ui.warn(s)
674 674 else:
675 675 self.ui.note(s)
676 676
677 677
678 678 def findlines(self, l, linenum):
679 679 # looks through the hash and finds candidate lines. The
680 680 # result is a list of line numbers sorted based on distance
681 681 # from linenum
682 682
683 683 cand = self.hash.get(l, [])
684 684 if len(cand) > 1:
685 685 # resort our list of potentials forward then back.
686 686 cand.sort(key=lambda x: abs(x - linenum))
687 687 return cand
688 688
689 689 def write_rej(self):
690 690 # our rejects are a little different from patch(1). This always
691 691 # creates rejects in the same form as the original patch. A file
692 692 # header is inserted so that you can run the reject through patch again
693 693 # without having to type the filename.
694 694 if not self.rej:
695 695 return
696 696 base = os.path.basename(self.fname)
697 697 lines = ["--- %s\n+++ %s\n" % (base, base)]
698 698 for x in self.rej:
699 699 for l in x.hunk:
700 700 lines.append(l)
701 701 if l[-1] != '\n':
702 702 lines.append("\n\ No newline at end of file\n")
703 703 self.backend.writerej(self.fname, len(self.rej), self.hunks, lines)
704 704
705 705 def apply(self, h):
706 706 if not h.complete():
707 707 raise PatchError(_("bad hunk #%d %s (%d %d %d %d)") %
708 708 (h.number, h.desc, len(h.a), h.lena, len(h.b),
709 709 h.lenb))
710 710
711 711 self.hunks += 1
712 712
713 713 if self.missing:
714 714 self.rej.append(h)
715 715 return -1
716 716
717 717 if self.exists and self.create:
718 718 if self.copysource:
719 719 self.ui.warn(_("cannot create %s: destination already "
720 720 "exists\n" % self.fname))
721 721 else:
722 722 self.ui.warn(_("file %s already exists\n") % self.fname)
723 723 self.rej.append(h)
724 724 return -1
725 725
726 726 if isinstance(h, binhunk):
727 727 if self.remove:
728 728 self.backend.unlink(self.fname)
729 729 else:
730 730 self.lines[:] = h.new()
731 731 self.offset += len(h.new())
732 732 self.dirty = True
733 733 return 0
734 734
735 735 horig = h
736 736 if (self.eolmode in ('crlf', 'lf')
737 737 or self.eolmode == 'auto' and self.eol):
738 738 # If new eols are going to be normalized, then normalize
739 739 # hunk data before patching. Otherwise, preserve input
740 740 # line-endings.
741 741 h = h.getnormalized()
742 742
743 743 # fast case first, no offsets, no fuzz
744 744 old, oldstart, new, newstart = h.fuzzit(0, False)
745 745 oldstart += self.offset
746 746 orig_start = oldstart
747 747 # if there's skew we want to emit the "(offset %d lines)" even
748 748 # when the hunk cleanly applies at start + skew, so skip the
749 749 # fast case code
750 750 if (self.skew == 0 and
751 751 diffhelpers.testhunk(old, self.lines, oldstart) == 0):
752 752 if self.remove:
753 753 self.backend.unlink(self.fname)
754 754 else:
755 755 self.lines[oldstart:oldstart + len(old)] = new
756 756 self.offset += len(new) - len(old)
757 757 self.dirty = True
758 758 return 0
759 759
760 760 # ok, we couldn't match the hunk. Lets look for offsets and fuzz it
761 761 self.hash = {}
762 762 for x, s in enumerate(self.lines):
763 763 self.hash.setdefault(s, []).append(x)
764 764
765 765 for fuzzlen in xrange(3):
766 766 for toponly in [True, False]:
767 767 old, oldstart, new, newstart = h.fuzzit(fuzzlen, toponly)
768 768 oldstart = oldstart + self.offset + self.skew
769 769 oldstart = min(oldstart, len(self.lines))
770 770 if old:
771 771 cand = self.findlines(old[0][1:], oldstart)
772 772 else:
773 773 # Only adding lines with no or fuzzed context, just
774 774 # take the skew in account
775 775 cand = [oldstart]
776 776
777 777 for l in cand:
778 778 if not old or diffhelpers.testhunk(old, self.lines, l) == 0:
779 779 self.lines[l : l + len(old)] = new
780 780 self.offset += len(new) - len(old)
781 781 self.skew = l - orig_start
782 782 self.dirty = True
783 783 offset = l - orig_start - fuzzlen
784 784 if fuzzlen:
785 785 msg = _("Hunk #%d succeeded at %d "
786 786 "with fuzz %d "
787 787 "(offset %d lines).\n")
788 788 self.printfile(True)
789 789 self.ui.warn(msg %
790 790 (h.number, l + 1, fuzzlen, offset))
791 791 else:
792 792 msg = _("Hunk #%d succeeded at %d "
793 793 "(offset %d lines).\n")
794 794 self.ui.note(msg % (h.number, l + 1, offset))
795 795 return fuzzlen
796 796 self.printfile(True)
797 797 self.ui.warn(_("Hunk #%d FAILED at %d\n") % (h.number, orig_start))
798 798 self.rej.append(horig)
799 799 return -1
800 800
801 801 def close(self):
802 802 if self.dirty:
803 803 self.writelines(self.fname, self.lines, self.mode)
804 804 self.write_rej()
805 805 return len(self.rej)
806 806
807 807 class hunk(object):
808 808 def __init__(self, desc, num, lr, context):
809 809 self.number = num
810 810 self.desc = desc
811 811 self.hunk = [desc]
812 812 self.a = []
813 813 self.b = []
814 814 self.starta = self.lena = None
815 815 self.startb = self.lenb = None
816 816 if lr is not None:
817 817 if context:
818 818 self.read_context_hunk(lr)
819 819 else:
820 820 self.read_unified_hunk(lr)
821 821
822 822 def getnormalized(self):
823 823 """Return a copy with line endings normalized to LF."""
824 824
825 825 def normalize(lines):
826 826 nlines = []
827 827 for line in lines:
828 828 if line.endswith('\r\n'):
829 829 line = line[:-2] + '\n'
830 830 nlines.append(line)
831 831 return nlines
832 832
833 833 # Dummy object, it is rebuilt manually
834 834 nh = hunk(self.desc, self.number, None, None)
835 835 nh.number = self.number
836 836 nh.desc = self.desc
837 837 nh.hunk = self.hunk
838 838 nh.a = normalize(self.a)
839 839 nh.b = normalize(self.b)
840 840 nh.starta = self.starta
841 841 nh.startb = self.startb
842 842 nh.lena = self.lena
843 843 nh.lenb = self.lenb
844 844 return nh
845 845
846 846 def read_unified_hunk(self, lr):
847 847 m = unidesc.match(self.desc)
848 848 if not m:
849 849 raise PatchError(_("bad hunk #%d") % self.number)
850 850 self.starta, self.lena, self.startb, self.lenb = m.groups()
851 851 if self.lena is None:
852 852 self.lena = 1
853 853 else:
854 854 self.lena = int(self.lena)
855 855 if self.lenb is None:
856 856 self.lenb = 1
857 857 else:
858 858 self.lenb = int(self.lenb)
859 859 self.starta = int(self.starta)
860 860 self.startb = int(self.startb)
861 861 diffhelpers.addlines(lr, self.hunk, self.lena, self.lenb, self.a, self.b)
862 862 # if we hit eof before finishing out the hunk, the last line will
863 863 # be zero length. Lets try to fix it up.
864 864 while len(self.hunk[-1]) == 0:
865 865 del self.hunk[-1]
866 866 del self.a[-1]
867 867 del self.b[-1]
868 868 self.lena -= 1
869 869 self.lenb -= 1
870 870 self._fixnewline(lr)
871 871
872 872 def read_context_hunk(self, lr):
873 873 self.desc = lr.readline()
874 874 m = contextdesc.match(self.desc)
875 875 if not m:
876 876 raise PatchError(_("bad hunk #%d") % self.number)
877 877 self.starta, aend = m.groups()
878 878 self.starta = int(self.starta)
879 879 if aend is None:
880 880 aend = self.starta
881 881 self.lena = int(aend) - self.starta
882 882 if self.starta:
883 883 self.lena += 1
884 884 for x in xrange(self.lena):
885 885 l = lr.readline()
886 886 if l.startswith('---'):
887 887 # lines addition, old block is empty
888 888 lr.push(l)
889 889 break
890 890 s = l[2:]
891 891 if l.startswith('- ') or l.startswith('! '):
892 892 u = '-' + s
893 893 elif l.startswith(' '):
894 894 u = ' ' + s
895 895 else:
896 896 raise PatchError(_("bad hunk #%d old text line %d") %
897 897 (self.number, x))
898 898 self.a.append(u)
899 899 self.hunk.append(u)
900 900
901 901 l = lr.readline()
902 902 if l.startswith('\ '):
903 903 s = self.a[-1][:-1]
904 904 self.a[-1] = s
905 905 self.hunk[-1] = s
906 906 l = lr.readline()
907 907 m = contextdesc.match(l)
908 908 if not m:
909 909 raise PatchError(_("bad hunk #%d") % self.number)
910 910 self.startb, bend = m.groups()
911 911 self.startb = int(self.startb)
912 912 if bend is None:
913 913 bend = self.startb
914 914 self.lenb = int(bend) - self.startb
915 915 if self.startb:
916 916 self.lenb += 1
917 917 hunki = 1
918 918 for x in xrange(self.lenb):
919 919 l = lr.readline()
920 920 if l.startswith('\ '):
921 921 # XXX: the only way to hit this is with an invalid line range.
922 922 # The no-eol marker is not counted in the line range, but I
923 923 # guess there are diff(1) out there which behave differently.
924 924 s = self.b[-1][:-1]
925 925 self.b[-1] = s
926 926 self.hunk[hunki - 1] = s
927 927 continue
928 928 if not l:
929 929 # line deletions, new block is empty and we hit EOF
930 930 lr.push(l)
931 931 break
932 932 s = l[2:]
933 933 if l.startswith('+ ') or l.startswith('! '):
934 934 u = '+' + s
935 935 elif l.startswith(' '):
936 936 u = ' ' + s
937 937 elif len(self.b) == 0:
938 938 # line deletions, new block is empty
939 939 lr.push(l)
940 940 break
941 941 else:
942 942 raise PatchError(_("bad hunk #%d old text line %d") %
943 943 (self.number, x))
944 944 self.b.append(s)
945 945 while True:
946 946 if hunki >= len(self.hunk):
947 947 h = ""
948 948 else:
949 949 h = self.hunk[hunki]
950 950 hunki += 1
951 951 if h == u:
952 952 break
953 953 elif h.startswith('-'):
954 954 continue
955 955 else:
956 956 self.hunk.insert(hunki - 1, u)
957 957 break
958 958
959 959 if not self.a:
960 960 # this happens when lines were only added to the hunk
961 961 for x in self.hunk:
962 962 if x.startswith('-') or x.startswith(' '):
963 963 self.a.append(x)
964 964 if not self.b:
965 965 # this happens when lines were only deleted from the hunk
966 966 for x in self.hunk:
967 967 if x.startswith('+') or x.startswith(' '):
968 968 self.b.append(x[1:])
969 969 # @@ -start,len +start,len @@
970 970 self.desc = "@@ -%d,%d +%d,%d @@\n" % (self.starta, self.lena,
971 971 self.startb, self.lenb)
972 972 self.hunk[0] = self.desc
973 973 self._fixnewline(lr)
974 974
975 975 def _fixnewline(self, lr):
976 976 l = lr.readline()
977 977 if l.startswith('\ '):
978 978 diffhelpers.fix_newline(self.hunk, self.a, self.b)
979 979 else:
980 980 lr.push(l)
981 981
982 982 def complete(self):
983 983 return len(self.a) == self.lena and len(self.b) == self.lenb
984 984
985 985 def _fuzzit(self, old, new, fuzz, toponly):
986 986 # this removes context lines from the top and bottom of list 'l'. It
987 987 # checks the hunk to make sure only context lines are removed, and then
988 988 # returns a new shortened list of lines.
989 989 fuzz = min(fuzz, len(old))
990 990 if fuzz:
991 991 top = 0
992 992 bot = 0
993 993 hlen = len(self.hunk)
994 994 for x in xrange(hlen - 1):
995 995 # the hunk starts with the @@ line, so use x+1
996 996 if self.hunk[x + 1][0] == ' ':
997 997 top += 1
998 998 else:
999 999 break
1000 1000 if not toponly:
1001 1001 for x in xrange(hlen - 1):
1002 1002 if self.hunk[hlen - bot - 1][0] == ' ':
1003 1003 bot += 1
1004 1004 else:
1005 1005 break
1006 1006
1007 1007 bot = min(fuzz, bot)
1008 1008 top = min(fuzz, top)
1009 1009 return old[top:len(old)-bot], new[top:len(new)-bot], top
1010 1010 return old, new, 0
1011 1011
1012 1012 def fuzzit(self, fuzz, toponly):
1013 1013 old, new, top = self._fuzzit(self.a, self.b, fuzz, toponly)
1014 1014 oldstart = self.starta + top
1015 1015 newstart = self.startb + top
1016 1016 # zero length hunk ranges already have their start decremented
1017 1017 if self.lena:
1018 1018 oldstart -= 1
1019 1019 if self.lenb:
1020 1020 newstart -= 1
1021 1021 return old, oldstart, new, newstart
1022 1022
1023 1023 class binhunk(object):
1024 1024 'A binary patch file. Only understands literals so far.'
1025 1025 def __init__(self, lr):
1026 1026 self.text = None
1027 1027 self.hunk = ['GIT binary patch\n']
1028 1028 self._read(lr)
1029 1029
1030 1030 def complete(self):
1031 1031 return self.text is not None
1032 1032
1033 1033 def new(self):
1034 1034 return [self.text]
1035 1035
1036 1036 def _read(self, lr):
1037 1037 line = lr.readline()
1038 1038 self.hunk.append(line)
1039 1039 while line and not line.startswith('literal '):
1040 1040 line = lr.readline()
1041 1041 self.hunk.append(line)
1042 1042 if not line:
1043 1043 raise PatchError(_('could not extract binary patch'))
1044 1044 size = int(line[8:].rstrip())
1045 1045 dec = []
1046 1046 line = lr.readline()
1047 1047 self.hunk.append(line)
1048 1048 while len(line) > 1:
1049 1049 l = line[0]
1050 1050 if l <= 'Z' and l >= 'A':
1051 1051 l = ord(l) - ord('A') + 1
1052 1052 else:
1053 1053 l = ord(l) - ord('a') + 27
1054 dec.append(base85.b85decode(line[1:-1])[:l])
1054 try:
1055 dec.append(base85.b85decode(line[1:-1])[:l])
1056 except ValueError, e:
1057 raise PatchError(_('could not decode binary patch: %s')
1058 % str(e))
1055 1059 line = lr.readline()
1056 1060 self.hunk.append(line)
1057 1061 text = zlib.decompress(''.join(dec))
1058 1062 if len(text) != size:
1059 1063 raise PatchError(_('binary patch is %d bytes, not %d') %
1060 1064 len(text), size)
1061 1065 self.text = text
1062 1066
1063 1067 def parsefilename(str):
1064 1068 # --- filename \t|space stuff
1065 1069 s = str[4:].rstrip('\r\n')
1066 1070 i = s.find('\t')
1067 1071 if i < 0:
1068 1072 i = s.find(' ')
1069 1073 if i < 0:
1070 1074 return s
1071 1075 return s[:i]
1072 1076
1073 1077 def pathstrip(path, strip):
1074 1078 pathlen = len(path)
1075 1079 i = 0
1076 1080 if strip == 0:
1077 1081 return '', path.rstrip()
1078 1082 count = strip
1079 1083 while count > 0:
1080 1084 i = path.find('/', i)
1081 1085 if i == -1:
1082 1086 raise PatchError(_("unable to strip away %d of %d dirs from %s") %
1083 1087 (count, strip, path))
1084 1088 i += 1
1085 1089 # consume '//' in the path
1086 1090 while i < pathlen - 1 and path[i] == '/':
1087 1091 i += 1
1088 1092 count -= 1
1089 1093 return path[:i].lstrip(), path[i:].rstrip()
1090 1094
1091 1095 def makepatchmeta(backend, afile_orig, bfile_orig, hunk, strip):
1092 1096 nulla = afile_orig == "/dev/null"
1093 1097 nullb = bfile_orig == "/dev/null"
1094 1098 create = nulla and hunk.starta == 0 and hunk.lena == 0
1095 1099 remove = nullb and hunk.startb == 0 and hunk.lenb == 0
1096 1100 abase, afile = pathstrip(afile_orig, strip)
1097 1101 gooda = not nulla and backend.exists(afile)
1098 1102 bbase, bfile = pathstrip(bfile_orig, strip)
1099 1103 if afile == bfile:
1100 1104 goodb = gooda
1101 1105 else:
1102 1106 goodb = not nullb and backend.exists(bfile)
1103 1107 missing = not goodb and not gooda and not create
1104 1108
1105 1109 # some diff programs apparently produce patches where the afile is
1106 1110 # not /dev/null, but afile starts with bfile
1107 1111 abasedir = afile[:afile.rfind('/') + 1]
1108 1112 bbasedir = bfile[:bfile.rfind('/') + 1]
1109 1113 if (missing and abasedir == bbasedir and afile.startswith(bfile)
1110 1114 and hunk.starta == 0 and hunk.lena == 0):
1111 1115 create = True
1112 1116 missing = False
1113 1117
1114 1118 # If afile is "a/b/foo" and bfile is "a/b/foo.orig" we assume the
1115 1119 # diff is between a file and its backup. In this case, the original
1116 1120 # file should be patched (see original mpatch code).
1117 1121 isbackup = (abase == bbase and bfile.startswith(afile))
1118 1122 fname = None
1119 1123 if not missing:
1120 1124 if gooda and goodb:
1121 1125 fname = isbackup and afile or bfile
1122 1126 elif gooda:
1123 1127 fname = afile
1124 1128
1125 1129 if not fname:
1126 1130 if not nullb:
1127 1131 fname = isbackup and afile or bfile
1128 1132 elif not nulla:
1129 1133 fname = afile
1130 1134 else:
1131 1135 raise PatchError(_("undefined source and destination files"))
1132 1136
1133 1137 gp = patchmeta(fname)
1134 1138 if create:
1135 1139 gp.op = 'ADD'
1136 1140 elif remove:
1137 1141 gp.op = 'DELETE'
1138 1142 return gp
1139 1143
1140 1144 def scangitpatch(lr, firstline):
1141 1145 """
1142 1146 Git patches can emit:
1143 1147 - rename a to b
1144 1148 - change b
1145 1149 - copy a to c
1146 1150 - change c
1147 1151
1148 1152 We cannot apply this sequence as-is, the renamed 'a' could not be
1149 1153 found for it would have been renamed already. And we cannot copy
1150 1154 from 'b' instead because 'b' would have been changed already. So
1151 1155 we scan the git patch for copy and rename commands so we can
1152 1156 perform the copies ahead of time.
1153 1157 """
1154 1158 pos = 0
1155 1159 try:
1156 1160 pos = lr.fp.tell()
1157 1161 fp = lr.fp
1158 1162 except IOError:
1159 1163 fp = cStringIO.StringIO(lr.fp.read())
1160 1164 gitlr = linereader(fp)
1161 1165 gitlr.push(firstline)
1162 1166 gitpatches = readgitpatch(gitlr)
1163 1167 fp.seek(pos)
1164 1168 return gitpatches
1165 1169
1166 1170 def iterhunks(fp):
1167 1171 """Read a patch and yield the following events:
1168 1172 - ("file", afile, bfile, firsthunk): select a new target file.
1169 1173 - ("hunk", hunk): a new hunk is ready to be applied, follows a
1170 1174 "file" event.
1171 1175 - ("git", gitchanges): current diff is in git format, gitchanges
1172 1176 maps filenames to gitpatch records. Unique event.
1173 1177 """
1174 1178 afile = ""
1175 1179 bfile = ""
1176 1180 state = None
1177 1181 hunknum = 0
1178 1182 emitfile = newfile = False
1179 1183 gitpatches = None
1180 1184
1181 1185 # our states
1182 1186 BFILE = 1
1183 1187 context = None
1184 1188 lr = linereader(fp)
1185 1189
1186 1190 while True:
1187 1191 x = lr.readline()
1188 1192 if not x:
1189 1193 break
1190 1194 if state == BFILE and (
1191 1195 (not context and x[0] == '@')
1192 1196 or (context is not False and x.startswith('***************'))
1193 1197 or x.startswith('GIT binary patch')):
1194 1198 gp = None
1195 1199 if (gitpatches and
1196 1200 gitpatches[-1].ispatching(afile, bfile)):
1197 1201 gp = gitpatches.pop()
1198 1202 if x.startswith('GIT binary patch'):
1199 1203 h = binhunk(lr)
1200 1204 else:
1201 1205 if context is None and x.startswith('***************'):
1202 1206 context = True
1203 1207 h = hunk(x, hunknum + 1, lr, context)
1204 1208 hunknum += 1
1205 1209 if emitfile:
1206 1210 emitfile = False
1207 1211 yield 'file', (afile, bfile, h, gp and gp.copy() or None)
1208 1212 yield 'hunk', h
1209 1213 elif x.startswith('diff --git'):
1210 1214 m = gitre.match(x)
1211 1215 if not m:
1212 1216 continue
1213 1217 if gitpatches is None:
1214 1218 # scan whole input for git metadata
1215 1219 gitpatches = scangitpatch(lr, x)
1216 1220 yield 'git', [g.copy() for g in gitpatches
1217 1221 if g.op in ('COPY', 'RENAME')]
1218 1222 gitpatches.reverse()
1219 1223 afile = 'a/' + m.group(1)
1220 1224 bfile = 'b/' + m.group(2)
1221 1225 while gitpatches and not gitpatches[-1].ispatching(afile, bfile):
1222 1226 gp = gitpatches.pop()
1223 1227 yield 'file', ('a/' + gp.path, 'b/' + gp.path, None, gp.copy())
1224 1228 if not gitpatches:
1225 1229 raise PatchError(_('failed to synchronize metadata for "%s"')
1226 1230 % afile[2:])
1227 1231 gp = gitpatches[-1]
1228 1232 newfile = True
1229 1233 elif x.startswith('---'):
1230 1234 # check for a unified diff
1231 1235 l2 = lr.readline()
1232 1236 if not l2.startswith('+++'):
1233 1237 lr.push(l2)
1234 1238 continue
1235 1239 newfile = True
1236 1240 context = False
1237 1241 afile = parsefilename(x)
1238 1242 bfile = parsefilename(l2)
1239 1243 elif x.startswith('***'):
1240 1244 # check for a context diff
1241 1245 l2 = lr.readline()
1242 1246 if not l2.startswith('---'):
1243 1247 lr.push(l2)
1244 1248 continue
1245 1249 l3 = lr.readline()
1246 1250 lr.push(l3)
1247 1251 if not l3.startswith("***************"):
1248 1252 lr.push(l2)
1249 1253 continue
1250 1254 newfile = True
1251 1255 context = True
1252 1256 afile = parsefilename(x)
1253 1257 bfile = parsefilename(l2)
1254 1258
1255 1259 if newfile:
1256 1260 newfile = False
1257 1261 emitfile = True
1258 1262 state = BFILE
1259 1263 hunknum = 0
1260 1264
1261 1265 while gitpatches:
1262 1266 gp = gitpatches.pop()
1263 1267 yield 'file', ('a/' + gp.path, 'b/' + gp.path, None, gp.copy())
1264 1268
1265 1269 def applydiff(ui, fp, backend, store, strip=1, eolmode='strict'):
1266 1270 """Reads a patch from fp and tries to apply it.
1267 1271
1268 1272 Returns 0 for a clean patch, -1 if any rejects were found and 1 if
1269 1273 there was any fuzz.
1270 1274
1271 1275 If 'eolmode' is 'strict', the patch content and patched file are
1272 1276 read in binary mode. Otherwise, line endings are ignored when
1273 1277 patching then normalized according to 'eolmode'.
1274 1278 """
1275 1279 return _applydiff(ui, fp, patchfile, backend, store, strip=strip,
1276 1280 eolmode=eolmode)
1277 1281
1278 1282 def _applydiff(ui, fp, patcher, backend, store, strip=1,
1279 1283 eolmode='strict'):
1280 1284
1281 1285 def pstrip(p):
1282 1286 return pathstrip(p, strip - 1)[1]
1283 1287
1284 1288 rejects = 0
1285 1289 err = 0
1286 1290 current_file = None
1287 1291
1288 1292 for state, values in iterhunks(fp):
1289 1293 if state == 'hunk':
1290 1294 if not current_file:
1291 1295 continue
1292 1296 ret = current_file.apply(values)
1293 1297 if ret > 0:
1294 1298 err = 1
1295 1299 elif state == 'file':
1296 1300 if current_file:
1297 1301 rejects += current_file.close()
1298 1302 current_file = None
1299 1303 afile, bfile, first_hunk, gp = values
1300 1304 if gp:
1301 1305 gp.path = pstrip(gp.path)
1302 1306 if gp.oldpath:
1303 1307 gp.oldpath = pstrip(gp.oldpath)
1304 1308 else:
1305 1309 gp = makepatchmeta(backend, afile, bfile, first_hunk, strip)
1306 1310 if gp.op == 'RENAME':
1307 1311 backend.unlink(gp.oldpath)
1308 1312 if not first_hunk:
1309 1313 if gp.op == 'DELETE':
1310 1314 backend.unlink(gp.path)
1311 1315 continue
1312 1316 data, mode = None, None
1313 1317 if gp.op in ('RENAME', 'COPY'):
1314 1318 data, mode = store.getfile(gp.oldpath)[:2]
1315 1319 if gp.mode:
1316 1320 mode = gp.mode
1317 1321 if gp.op == 'ADD':
1318 1322 # Added files without content have no hunk and
1319 1323 # must be created
1320 1324 data = ''
1321 1325 if data or mode:
1322 1326 if (gp.op in ('ADD', 'RENAME', 'COPY')
1323 1327 and backend.exists(gp.path)):
1324 1328 raise PatchError(_("cannot create %s: destination "
1325 1329 "already exists") % gp.path)
1326 1330 backend.setfile(gp.path, data, mode, gp.oldpath)
1327 1331 continue
1328 1332 try:
1329 1333 current_file = patcher(ui, gp, backend, store,
1330 1334 eolmode=eolmode)
1331 1335 except PatchError, inst:
1332 1336 ui.warn(str(inst) + '\n')
1333 1337 current_file = None
1334 1338 rejects += 1
1335 1339 continue
1336 1340 elif state == 'git':
1337 1341 for gp in values:
1338 1342 path = pstrip(gp.oldpath)
1339 1343 data, mode = backend.getfile(path)
1340 1344 store.setfile(path, data, mode)
1341 1345 else:
1342 1346 raise util.Abort(_('unsupported parser state: %s') % state)
1343 1347
1344 1348 if current_file:
1345 1349 rejects += current_file.close()
1346 1350
1347 1351 if rejects:
1348 1352 return -1
1349 1353 return err
1350 1354
1351 1355 def _externalpatch(ui, repo, patcher, patchname, strip, files,
1352 1356 similarity):
1353 1357 """use <patcher> to apply <patchname> to the working directory.
1354 1358 returns whether patch was applied with fuzz factor."""
1355 1359
1356 1360 fuzz = False
1357 1361 args = []
1358 1362 cwd = repo.root
1359 1363 if cwd:
1360 1364 args.append('-d %s' % util.shellquote(cwd))
1361 1365 fp = util.popen('%s %s -p%d < %s' % (patcher, ' '.join(args), strip,
1362 1366 util.shellquote(patchname)))
1363 1367 try:
1364 1368 for line in fp:
1365 1369 line = line.rstrip()
1366 1370 ui.note(line + '\n')
1367 1371 if line.startswith('patching file '):
1368 1372 pf = util.parsepatchoutput(line)
1369 1373 printed_file = False
1370 1374 files.add(pf)
1371 1375 elif line.find('with fuzz') >= 0:
1372 1376 fuzz = True
1373 1377 if not printed_file:
1374 1378 ui.warn(pf + '\n')
1375 1379 printed_file = True
1376 1380 ui.warn(line + '\n')
1377 1381 elif line.find('saving rejects to file') >= 0:
1378 1382 ui.warn(line + '\n')
1379 1383 elif line.find('FAILED') >= 0:
1380 1384 if not printed_file:
1381 1385 ui.warn(pf + '\n')
1382 1386 printed_file = True
1383 1387 ui.warn(line + '\n')
1384 1388 finally:
1385 1389 if files:
1386 1390 cfiles = list(files)
1387 1391 cwd = repo.getcwd()
1388 1392 if cwd:
1389 1393 cfiles = [util.pathto(repo.root, cwd, f)
1390 1394 for f in cfiles]
1391 1395 scmutil.addremove(repo, cfiles, similarity=similarity)
1392 1396 code = fp.close()
1393 1397 if code:
1394 1398 raise PatchError(_("patch command failed: %s") %
1395 1399 util.explainexit(code)[0])
1396 1400 return fuzz
1397 1401
1398 1402 def patchbackend(ui, backend, patchobj, strip, files=None, eolmode='strict'):
1399 1403 if files is None:
1400 1404 files = set()
1401 1405 if eolmode is None:
1402 1406 eolmode = ui.config('patch', 'eol', 'strict')
1403 1407 if eolmode.lower() not in eolmodes:
1404 1408 raise util.Abort(_('unsupported line endings type: %s') % eolmode)
1405 1409 eolmode = eolmode.lower()
1406 1410
1407 1411 store = filestore()
1408 1412 try:
1409 1413 fp = open(patchobj, 'rb')
1410 1414 except TypeError:
1411 1415 fp = patchobj
1412 1416 try:
1413 1417 ret = applydiff(ui, fp, backend, store, strip=strip,
1414 1418 eolmode=eolmode)
1415 1419 finally:
1416 1420 if fp != patchobj:
1417 1421 fp.close()
1418 1422 files.update(backend.close())
1419 1423 store.close()
1420 1424 if ret < 0:
1421 1425 raise PatchError(_('patch failed to apply'))
1422 1426 return ret > 0
1423 1427
1424 1428 def internalpatch(ui, repo, patchobj, strip, files=None, eolmode='strict',
1425 1429 similarity=0):
1426 1430 """use builtin patch to apply <patchobj> to the working directory.
1427 1431 returns whether patch was applied with fuzz factor."""
1428 1432 backend = workingbackend(ui, repo, similarity)
1429 1433 return patchbackend(ui, backend, patchobj, strip, files, eolmode)
1430 1434
1431 1435 def patchrepo(ui, repo, ctx, store, patchobj, strip, files=None,
1432 1436 eolmode='strict'):
1433 1437 backend = repobackend(ui, repo, ctx, store)
1434 1438 return patchbackend(ui, backend, patchobj, strip, files, eolmode)
1435 1439
1436 1440 def makememctx(repo, parents, text, user, date, branch, files, store,
1437 1441 editor=None):
1438 1442 def getfilectx(repo, memctx, path):
1439 1443 data, (islink, isexec), copied = store.getfile(path)
1440 1444 return context.memfilectx(path, data, islink=islink, isexec=isexec,
1441 1445 copied=copied)
1442 1446 extra = {}
1443 1447 if branch:
1444 1448 extra['branch'] = encoding.fromlocal(branch)
1445 1449 ctx = context.memctx(repo, parents, text, files, getfilectx, user,
1446 1450 date, extra)
1447 1451 if editor:
1448 1452 ctx._text = editor(repo, ctx, [])
1449 1453 return ctx
1450 1454
1451 1455 def patch(ui, repo, patchname, strip=1, files=None, eolmode='strict',
1452 1456 similarity=0):
1453 1457 """Apply <patchname> to the working directory.
1454 1458
1455 1459 'eolmode' specifies how end of lines should be handled. It can be:
1456 1460 - 'strict': inputs are read in binary mode, EOLs are preserved
1457 1461 - 'crlf': EOLs are ignored when patching and reset to CRLF
1458 1462 - 'lf': EOLs are ignored when patching and reset to LF
1459 1463 - None: get it from user settings, default to 'strict'
1460 1464 'eolmode' is ignored when using an external patcher program.
1461 1465
1462 1466 Returns whether patch was applied with fuzz factor.
1463 1467 """
1464 1468 patcher = ui.config('ui', 'patch')
1465 1469 if files is None:
1466 1470 files = set()
1467 1471 try:
1468 1472 if patcher:
1469 1473 return _externalpatch(ui, repo, patcher, patchname, strip,
1470 1474 files, similarity)
1471 1475 return internalpatch(ui, repo, patchname, strip, files, eolmode,
1472 1476 similarity)
1473 1477 except PatchError, err:
1474 1478 raise util.Abort(str(err))
1475 1479
1476 1480 def changedfiles(ui, repo, patchpath, strip=1):
1477 1481 backend = fsbackend(ui, repo.root)
1478 1482 fp = open(patchpath, 'rb')
1479 1483 try:
1480 1484 changed = set()
1481 1485 for state, values in iterhunks(fp):
1482 1486 if state == 'file':
1483 1487 afile, bfile, first_hunk, gp = values
1484 1488 if gp:
1485 1489 gp.path = pathstrip(gp.path, strip - 1)[1]
1486 1490 if gp.oldpath:
1487 1491 gp.oldpath = pathstrip(gp.oldpath, strip - 1)[1]
1488 1492 else:
1489 1493 gp = makepatchmeta(backend, afile, bfile, first_hunk, strip)
1490 1494 changed.add(gp.path)
1491 1495 if gp.op == 'RENAME':
1492 1496 changed.add(gp.oldpath)
1493 1497 elif state not in ('hunk', 'git'):
1494 1498 raise util.Abort(_('unsupported parser state: %s') % state)
1495 1499 return changed
1496 1500 finally:
1497 1501 fp.close()
1498 1502
1499 1503 def b85diff(to, tn):
1500 1504 '''print base85-encoded binary diff'''
1501 1505 def gitindex(text):
1502 1506 if not text:
1503 1507 return hex(nullid)
1504 1508 l = len(text)
1505 1509 s = util.sha1('blob %d\0' % l)
1506 1510 s.update(text)
1507 1511 return s.hexdigest()
1508 1512
1509 1513 def fmtline(line):
1510 1514 l = len(line)
1511 1515 if l <= 26:
1512 1516 l = chr(ord('A') + l - 1)
1513 1517 else:
1514 1518 l = chr(l - 26 + ord('a') - 1)
1515 1519 return '%c%s\n' % (l, base85.b85encode(line, True))
1516 1520
1517 1521 def chunk(text, csize=52):
1518 1522 l = len(text)
1519 1523 i = 0
1520 1524 while i < l:
1521 1525 yield text[i:i + csize]
1522 1526 i += csize
1523 1527
1524 1528 tohash = gitindex(to)
1525 1529 tnhash = gitindex(tn)
1526 1530 if tohash == tnhash:
1527 1531 return ""
1528 1532
1529 1533 # TODO: deltas
1530 1534 ret = ['index %s..%s\nGIT binary patch\nliteral %s\n' %
1531 1535 (tohash, tnhash, len(tn))]
1532 1536 for l in chunk(zlib.compress(tn)):
1533 1537 ret.append(fmtline(l))
1534 1538 ret.append('\n')
1535 1539 return ''.join(ret)
1536 1540
1537 1541 class GitDiffRequired(Exception):
1538 1542 pass
1539 1543
1540 1544 def diffopts(ui, opts=None, untrusted=False, section='diff'):
1541 1545 def get(key, name=None, getter=ui.configbool):
1542 1546 return ((opts and opts.get(key)) or
1543 1547 getter(section, name or key, None, untrusted=untrusted))
1544 1548 return mdiff.diffopts(
1545 1549 text=opts and opts.get('text'),
1546 1550 git=get('git'),
1547 1551 nodates=get('nodates'),
1548 1552 showfunc=get('show_function', 'showfunc'),
1549 1553 ignorews=get('ignore_all_space', 'ignorews'),
1550 1554 ignorewsamount=get('ignore_space_change', 'ignorewsamount'),
1551 1555 ignoreblanklines=get('ignore_blank_lines', 'ignoreblanklines'),
1552 1556 context=get('unified', getter=ui.config))
1553 1557
1554 1558 def diff(repo, node1=None, node2=None, match=None, changes=None, opts=None,
1555 1559 losedatafn=None, prefix=''):
1556 1560 '''yields diff of changes to files between two nodes, or node and
1557 1561 working directory.
1558 1562
1559 1563 if node1 is None, use first dirstate parent instead.
1560 1564 if node2 is None, compare node1 with working directory.
1561 1565
1562 1566 losedatafn(**kwarg) is a callable run when opts.upgrade=True and
1563 1567 every time some change cannot be represented with the current
1564 1568 patch format. Return False to upgrade to git patch format, True to
1565 1569 accept the loss or raise an exception to abort the diff. It is
1566 1570 called with the name of current file being diffed as 'fn'. If set
1567 1571 to None, patches will always be upgraded to git format when
1568 1572 necessary.
1569 1573
1570 1574 prefix is a filename prefix that is prepended to all filenames on
1571 1575 display (used for subrepos).
1572 1576 '''
1573 1577
1574 1578 if opts is None:
1575 1579 opts = mdiff.defaultopts
1576 1580
1577 1581 if not node1 and not node2:
1578 1582 node1 = repo.dirstate.p1()
1579 1583
1580 1584 def lrugetfilectx():
1581 1585 cache = {}
1582 1586 order = []
1583 1587 def getfilectx(f, ctx):
1584 1588 fctx = ctx.filectx(f, filelog=cache.get(f))
1585 1589 if f not in cache:
1586 1590 if len(cache) > 20:
1587 1591 del cache[order.pop(0)]
1588 1592 cache[f] = fctx.filelog()
1589 1593 else:
1590 1594 order.remove(f)
1591 1595 order.append(f)
1592 1596 return fctx
1593 1597 return getfilectx
1594 1598 getfilectx = lrugetfilectx()
1595 1599
1596 1600 ctx1 = repo[node1]
1597 1601 ctx2 = repo[node2]
1598 1602
1599 1603 if not changes:
1600 1604 changes = repo.status(ctx1, ctx2, match=match)
1601 1605 modified, added, removed = changes[:3]
1602 1606
1603 1607 if not modified and not added and not removed:
1604 1608 return []
1605 1609
1606 1610 revs = None
1607 1611 if not repo.ui.quiet:
1608 1612 hexfunc = repo.ui.debugflag and hex or short
1609 1613 revs = [hexfunc(node) for node in [node1, node2] if node]
1610 1614
1611 1615 copy = {}
1612 1616 if opts.git or opts.upgrade:
1613 1617 copy = copies.pathcopies(ctx1, ctx2)
1614 1618
1615 1619 difffn = lambda opts, losedata: trydiff(repo, revs, ctx1, ctx2,
1616 1620 modified, added, removed, copy, getfilectx, opts, losedata, prefix)
1617 1621 if opts.upgrade and not opts.git:
1618 1622 try:
1619 1623 def losedata(fn):
1620 1624 if not losedatafn or not losedatafn(fn=fn):
1621 1625 raise GitDiffRequired()
1622 1626 # Buffer the whole output until we are sure it can be generated
1623 1627 return list(difffn(opts.copy(git=False), losedata))
1624 1628 except GitDiffRequired:
1625 1629 return difffn(opts.copy(git=True), None)
1626 1630 else:
1627 1631 return difffn(opts, None)
1628 1632
1629 1633 def difflabel(func, *args, **kw):
1630 1634 '''yields 2-tuples of (output, label) based on the output of func()'''
1631 1635 headprefixes = [('diff', 'diff.diffline'),
1632 1636 ('copy', 'diff.extended'),
1633 1637 ('rename', 'diff.extended'),
1634 1638 ('old', 'diff.extended'),
1635 1639 ('new', 'diff.extended'),
1636 1640 ('deleted', 'diff.extended'),
1637 1641 ('---', 'diff.file_a'),
1638 1642 ('+++', 'diff.file_b')]
1639 1643 textprefixes = [('@', 'diff.hunk'),
1640 1644 ('-', 'diff.deleted'),
1641 1645 ('+', 'diff.inserted')]
1642 1646 head = False
1643 1647 for chunk in func(*args, **kw):
1644 1648 lines = chunk.split('\n')
1645 1649 for i, line in enumerate(lines):
1646 1650 if i != 0:
1647 1651 yield ('\n', '')
1648 1652 if head:
1649 1653 if line.startswith('@'):
1650 1654 head = False
1651 1655 else:
1652 1656 if line and not line[0] in ' +-@\\':
1653 1657 head = True
1654 1658 stripline = line
1655 1659 if not head and line and line[0] in '+-':
1656 1660 # highlight trailing whitespace, but only in changed lines
1657 1661 stripline = line.rstrip()
1658 1662 prefixes = textprefixes
1659 1663 if head:
1660 1664 prefixes = headprefixes
1661 1665 for prefix, label in prefixes:
1662 1666 if stripline.startswith(prefix):
1663 1667 yield (stripline, label)
1664 1668 break
1665 1669 else:
1666 1670 yield (line, '')
1667 1671 if line != stripline:
1668 1672 yield (line[len(stripline):], 'diff.trailingwhitespace')
1669 1673
1670 1674 def diffui(*args, **kw):
1671 1675 '''like diff(), but yields 2-tuples of (output, label) for ui.write()'''
1672 1676 return difflabel(diff, *args, **kw)
1673 1677
1674 1678
1675 1679 def _addmodehdr(header, omode, nmode):
1676 1680 if omode != nmode:
1677 1681 header.append('old mode %s\n' % omode)
1678 1682 header.append('new mode %s\n' % nmode)
1679 1683
1680 1684 def trydiff(repo, revs, ctx1, ctx2, modified, added, removed,
1681 1685 copy, getfilectx, opts, losedatafn, prefix):
1682 1686
1683 1687 def join(f):
1684 1688 return os.path.join(prefix, f)
1685 1689
1686 1690 date1 = util.datestr(ctx1.date())
1687 1691 man1 = ctx1.manifest()
1688 1692
1689 1693 gone = set()
1690 1694 gitmode = {'l': '120000', 'x': '100755', '': '100644'}
1691 1695
1692 1696 copyto = dict([(v, k) for k, v in copy.items()])
1693 1697
1694 1698 if opts.git:
1695 1699 revs = None
1696 1700
1697 1701 for f in sorted(modified + added + removed):
1698 1702 to = None
1699 1703 tn = None
1700 1704 dodiff = True
1701 1705 header = []
1702 1706 if f in man1:
1703 1707 to = getfilectx(f, ctx1).data()
1704 1708 if f not in removed:
1705 1709 tn = getfilectx(f, ctx2).data()
1706 1710 a, b = f, f
1707 1711 if opts.git or losedatafn:
1708 1712 if f in added:
1709 1713 mode = gitmode[ctx2.flags(f)]
1710 1714 if f in copy or f in copyto:
1711 1715 if opts.git:
1712 1716 if f in copy:
1713 1717 a = copy[f]
1714 1718 else:
1715 1719 a = copyto[f]
1716 1720 omode = gitmode[man1.flags(a)]
1717 1721 _addmodehdr(header, omode, mode)
1718 1722 if a in removed and a not in gone:
1719 1723 op = 'rename'
1720 1724 gone.add(a)
1721 1725 else:
1722 1726 op = 'copy'
1723 1727 header.append('%s from %s\n' % (op, join(a)))
1724 1728 header.append('%s to %s\n' % (op, join(f)))
1725 1729 to = getfilectx(a, ctx1).data()
1726 1730 else:
1727 1731 losedatafn(f)
1728 1732 else:
1729 1733 if opts.git:
1730 1734 header.append('new file mode %s\n' % mode)
1731 1735 elif ctx2.flags(f):
1732 1736 losedatafn(f)
1733 1737 # In theory, if tn was copied or renamed we should check
1734 1738 # if the source is binary too but the copy record already
1735 1739 # forces git mode.
1736 1740 if util.binary(tn):
1737 1741 if opts.git:
1738 1742 dodiff = 'binary'
1739 1743 else:
1740 1744 losedatafn(f)
1741 1745 if not opts.git and not tn:
1742 1746 # regular diffs cannot represent new empty file
1743 1747 losedatafn(f)
1744 1748 elif f in removed:
1745 1749 if opts.git:
1746 1750 # have we already reported a copy above?
1747 1751 if ((f in copy and copy[f] in added
1748 1752 and copyto[copy[f]] == f) or
1749 1753 (f in copyto and copyto[f] in added
1750 1754 and copy[copyto[f]] == f)):
1751 1755 dodiff = False
1752 1756 else:
1753 1757 header.append('deleted file mode %s\n' %
1754 1758 gitmode[man1.flags(f)])
1755 1759 elif not to or util.binary(to):
1756 1760 # regular diffs cannot represent empty file deletion
1757 1761 losedatafn(f)
1758 1762 else:
1759 1763 oflag = man1.flags(f)
1760 1764 nflag = ctx2.flags(f)
1761 1765 binary = util.binary(to) or util.binary(tn)
1762 1766 if opts.git:
1763 1767 _addmodehdr(header, gitmode[oflag], gitmode[nflag])
1764 1768 if binary:
1765 1769 dodiff = 'binary'
1766 1770 elif binary or nflag != oflag:
1767 1771 losedatafn(f)
1768 1772 if opts.git:
1769 1773 header.insert(0, mdiff.diffline(revs, join(a), join(b), opts))
1770 1774
1771 1775 if dodiff:
1772 1776 if dodiff == 'binary':
1773 1777 text = b85diff(to, tn)
1774 1778 else:
1775 1779 text = mdiff.unidiff(to, date1,
1776 1780 # ctx2 date may be dynamic
1777 1781 tn, util.datestr(ctx2.date()),
1778 1782 join(a), join(b), revs, opts=opts)
1779 1783 if header and (text or len(header) > 1):
1780 1784 yield ''.join(header)
1781 1785 if text:
1782 1786 yield text
1783 1787
1784 1788 def diffstatsum(stats):
1785 1789 maxfile, maxtotal, addtotal, removetotal, binary = 0, 0, 0, 0, False
1786 1790 for f, a, r, b in stats:
1787 1791 maxfile = max(maxfile, encoding.colwidth(f))
1788 1792 maxtotal = max(maxtotal, a + r)
1789 1793 addtotal += a
1790 1794 removetotal += r
1791 1795 binary = binary or b
1792 1796
1793 1797 return maxfile, maxtotal, addtotal, removetotal, binary
1794 1798
1795 1799 def diffstatdata(lines):
1796 1800 diffre = re.compile('^diff .*-r [a-z0-9]+\s(.*)$')
1797 1801
1798 1802 results = []
1799 1803 filename, adds, removes, isbinary = None, 0, 0, False
1800 1804
1801 1805 def addresult():
1802 1806 if filename:
1803 1807 results.append((filename, adds, removes, isbinary))
1804 1808
1805 1809 for line in lines:
1806 1810 if line.startswith('diff'):
1807 1811 addresult()
1808 1812 # set numbers to 0 anyway when starting new file
1809 1813 adds, removes, isbinary = 0, 0, False
1810 1814 if line.startswith('diff --git'):
1811 1815 filename = gitre.search(line).group(1)
1812 1816 elif line.startswith('diff -r'):
1813 1817 # format: "diff -r ... -r ... filename"
1814 1818 filename = diffre.search(line).group(1)
1815 1819 elif line.startswith('+') and not line.startswith('+++ '):
1816 1820 adds += 1
1817 1821 elif line.startswith('-') and not line.startswith('--- '):
1818 1822 removes += 1
1819 1823 elif (line.startswith('GIT binary patch') or
1820 1824 line.startswith('Binary file')):
1821 1825 isbinary = True
1822 1826 addresult()
1823 1827 return results
1824 1828
1825 1829 def diffstat(lines, width=80, git=False):
1826 1830 output = []
1827 1831 stats = diffstatdata(lines)
1828 1832 maxname, maxtotal, totaladds, totalremoves, hasbinary = diffstatsum(stats)
1829 1833
1830 1834 countwidth = len(str(maxtotal))
1831 1835 if hasbinary and countwidth < 3:
1832 1836 countwidth = 3
1833 1837 graphwidth = width - countwidth - maxname - 6
1834 1838 if graphwidth < 10:
1835 1839 graphwidth = 10
1836 1840
1837 1841 def scale(i):
1838 1842 if maxtotal <= graphwidth:
1839 1843 return i
1840 1844 # If diffstat runs out of room it doesn't print anything,
1841 1845 # which isn't very useful, so always print at least one + or -
1842 1846 # if there were at least some changes.
1843 1847 return max(i * graphwidth // maxtotal, int(bool(i)))
1844 1848
1845 1849 for filename, adds, removes, isbinary in stats:
1846 1850 if isbinary:
1847 1851 count = 'Bin'
1848 1852 else:
1849 1853 count = adds + removes
1850 1854 pluses = '+' * scale(adds)
1851 1855 minuses = '-' * scale(removes)
1852 1856 output.append(' %s%s | %*s %s%s\n' %
1853 1857 (filename, ' ' * (maxname - encoding.colwidth(filename)),
1854 1858 countwidth, count, pluses, minuses))
1855 1859
1856 1860 if stats:
1857 1861 output.append(_(' %d files changed, %d insertions(+), %d deletions(-)\n')
1858 1862 % (len(stats), totaladds, totalremoves))
1859 1863
1860 1864 return ''.join(output)
1861 1865
1862 1866 def diffstatui(*args, **kw):
1863 1867 '''like diffstat(), but yields 2-tuples of (output, label) for
1864 1868 ui.write()
1865 1869 '''
1866 1870
1867 1871 for line in diffstat(*args, **kw).splitlines():
1868 1872 if line and line[-1] in '+-':
1869 1873 name, graph = line.rsplit(' ', 1)
1870 1874 yield (name + ' ', '')
1871 1875 m = re.search(r'\++', graph)
1872 1876 if m:
1873 1877 yield (m.group(0), 'diffstat.inserted')
1874 1878 m = re.search(r'-+', graph)
1875 1879 if m:
1876 1880 yield (m.group(0), 'diffstat.deleted')
1877 1881 else:
1878 1882 yield (line, '')
1879 1883 yield ('\n', '')
@@ -1,510 +1,529
1 1 $ "$TESTDIR/hghave" symlink || exit 80
2 2
3 3 $ hg init
4 4
5 5 New file:
6 6
7 7 $ hg import -d "1000000 0" -mnew - <<EOF
8 8 > diff --git a/new b/new
9 9 > new file mode 100644
10 10 > index 0000000..7898192
11 11 > --- /dev/null
12 12 > +++ b/new
13 13 > @@ -0,0 +1 @@
14 14 > +a
15 15 > EOF
16 16 applying patch from stdin
17 17
18 18 $ hg tip -q
19 19 0:ae3ee40d2079
20 20
21 21 New empty file:
22 22
23 23 $ hg import -d "1000000 0" -mempty - <<EOF
24 24 > diff --git a/empty b/empty
25 25 > new file mode 100644
26 26 > EOF
27 27 applying patch from stdin
28 28
29 29 $ hg tip -q
30 30 1:ab199dc869b5
31 31
32 32 $ hg locate empty
33 33 empty
34 34
35 35 chmod +x:
36 36
37 37 $ hg import -d "1000000 0" -msetx - <<EOF
38 38 > diff --git a/new b/new
39 39 > old mode 100644
40 40 > new mode 100755
41 41 > EOF
42 42 applying patch from stdin
43 43
44 44 $ hg tip -q
45 45 2:3a34410f282e
46 46
47 47 $ test -x new
48 48
49 49 Copy:
50 50
51 51 $ hg import -d "1000000 0" -mcopy - <<EOF
52 52 > diff --git a/new b/copy
53 53 > old mode 100755
54 54 > new mode 100644
55 55 > similarity index 100%
56 56 > copy from new
57 57 > copy to copy
58 58 > diff --git a/new b/copyx
59 59 > similarity index 100%
60 60 > copy from new
61 61 > copy to copyx
62 62 > EOF
63 63 applying patch from stdin
64 64
65 65 $ hg tip -q
66 66 3:37bacb7ca14d
67 67
68 68 $ if "$TESTDIR/hghave" -q execbit; then
69 69 > test -f copy -a ! -x copy || echo bad
70 70 > test -x copyx || echo bad
71 71 > else
72 72 > test -f copy || echo bad
73 73 > fi
74 74
75 75 $ cat copy
76 76 a
77 77
78 78 $ hg cat copy
79 79 a
80 80
81 81 Rename:
82 82
83 83 $ hg import -d "1000000 0" -mrename - <<EOF
84 84 > diff --git a/copy b/rename
85 85 > similarity index 100%
86 86 > rename from copy
87 87 > rename to rename
88 88 > EOF
89 89 applying patch from stdin
90 90
91 91 $ hg tip -q
92 92 4:47b81a94361d
93 93
94 94 $ hg locate
95 95 copyx
96 96 empty
97 97 new
98 98 rename
99 99
100 100 Delete:
101 101
102 102 $ hg import -d "1000000 0" -mdelete - <<EOF
103 103 > diff --git a/copyx b/copyx
104 104 > deleted file mode 100755
105 105 > index 7898192..0000000
106 106 > --- a/copyx
107 107 > +++ /dev/null
108 108 > @@ -1 +0,0 @@
109 109 > -a
110 110 > EOF
111 111 applying patch from stdin
112 112
113 113 $ hg tip -q
114 114 5:d9b001d98336
115 115
116 116 $ hg locate
117 117 empty
118 118 new
119 119 rename
120 120
121 121 $ test -f copyx
122 122 [1]
123 123
124 124 Regular diff:
125 125
126 126 $ hg import -d "1000000 0" -mregular - <<EOF
127 127 > diff --git a/rename b/rename
128 128 > index 7898192..72e1fe3 100644
129 129 > --- a/rename
130 130 > +++ b/rename
131 131 > @@ -1 +1,5 @@
132 132 > a
133 133 > +a
134 134 > +a
135 135 > +a
136 136 > +a
137 137 > EOF
138 138 applying patch from stdin
139 139
140 140 $ hg tip -q
141 141 6:ebe901e7576b
142 142
143 143 Copy and modify:
144 144
145 145 $ hg import -d "1000000 0" -mcopymod - <<EOF
146 146 > diff --git a/rename b/copy2
147 147 > similarity index 80%
148 148 > copy from rename
149 149 > copy to copy2
150 150 > index 72e1fe3..b53c148 100644
151 151 > --- a/rename
152 152 > +++ b/copy2
153 153 > @@ -1,5 +1,5 @@
154 154 > a
155 155 > a
156 156 > -a
157 157 > +b
158 158 > a
159 159 > a
160 160 > EOF
161 161 applying patch from stdin
162 162
163 163 $ hg tip -q
164 164 7:18f368958ecd
165 165
166 166 $ hg cat copy2
167 167 a
168 168 a
169 169 b
170 170 a
171 171 a
172 172
173 173 Rename and modify:
174 174
175 175 $ hg import -d "1000000 0" -mrenamemod - <<EOF
176 176 > diff --git a/copy2 b/rename2
177 177 > similarity index 80%
178 178 > rename from copy2
179 179 > rename to rename2
180 180 > index b53c148..8f81e29 100644
181 181 > --- a/copy2
182 182 > +++ b/rename2
183 183 > @@ -1,5 +1,5 @@
184 184 > a
185 185 > a
186 186 > b
187 187 > -a
188 188 > +c
189 189 > a
190 190 > EOF
191 191 applying patch from stdin
192 192
193 193 $ hg tip -q
194 194 8:c32b0d7e6f44
195 195
196 196 $ hg locate copy2
197 197 [1]
198 198 $ hg cat rename2
199 199 a
200 200 a
201 201 b
202 202 c
203 203 a
204 204
205 205 One file renamed multiple times:
206 206
207 207 $ hg import -d "1000000 0" -mmultirenames - <<EOF
208 208 > diff --git a/rename2 b/rename3
209 209 > rename from rename2
210 210 > rename to rename3
211 211 > diff --git a/rename2 b/rename3-2
212 212 > rename from rename2
213 213 > rename to rename3-2
214 214 > EOF
215 215 applying patch from stdin
216 216
217 217 $ hg tip -q
218 218 9:034a6bf95330
219 219
220 220 $ hg log -vr. --template '{rev} {files} / {file_copies}\n'
221 221 9 rename2 rename3 rename3-2 / rename3 (rename2)rename3-2 (rename2)
222 222
223 223 $ hg locate rename2 rename3 rename3-2
224 224 rename3
225 225 rename3-2
226 226
227 227 $ hg cat rename3
228 228 a
229 229 a
230 230 b
231 231 c
232 232 a
233 233
234 234 $ hg cat rename3-2
235 235 a
236 236 a
237 237 b
238 238 c
239 239 a
240 240
241 241 $ echo foo > foo
242 242 $ hg add foo
243 243 $ hg ci -m 'add foo'
244 244
245 245 Binary files and regular patch hunks:
246 246
247 247 $ hg import -d "1000000 0" -m binaryregular - <<EOF
248 248 > diff --git a/binary b/binary
249 249 > new file mode 100644
250 250 > index 0000000000000000000000000000000000000000..593f4708db84ac8fd0f5cc47c634f38c013fe9e4
251 251 > GIT binary patch
252 252 > literal 4
253 253 > Lc\${NkU|;|M00aO5
254 254 >
255 255 > diff --git a/foo b/foo2
256 256 > rename from foo
257 257 > rename to foo2
258 258 > EOF
259 259 applying patch from stdin
260 260
261 261 $ hg tip -q
262 262 11:c39bce63e786
263 263
264 264 $ cat foo2
265 265 foo
266 266
267 267 $ hg manifest --debug | grep binary
268 268 045c85ba38952325e126c70962cc0f9d9077bc67 644 binary
269 269
270 270 Multiple binary files:
271 271
272 272 $ hg import -d "1000000 0" -m multibinary - <<EOF
273 273 > diff --git a/mbinary1 b/mbinary1
274 274 > new file mode 100644
275 275 > index 0000000000000000000000000000000000000000..593f4708db84ac8fd0f5cc47c634f38c013fe9e4
276 276 > GIT binary patch
277 277 > literal 4
278 278 > Lc\${NkU|;|M00aO5
279 279 >
280 280 > diff --git a/mbinary2 b/mbinary2
281 281 > new file mode 100644
282 282 > index 0000000000000000000000000000000000000000..112363ac1917b417ffbd7f376ca786a1e5fa7490
283 283 > GIT binary patch
284 284 > literal 5
285 285 > Mc\${NkU|\`?^000jF3jhEB
286 286 >
287 287 > EOF
288 288 applying patch from stdin
289 289
290 290 $ hg tip -q
291 291 12:30b530085242
292 292
293 293 $ hg manifest --debug | grep mbinary
294 294 045c85ba38952325e126c70962cc0f9d9077bc67 644 mbinary1
295 295 a874b471193996e7cb034bb301cac7bdaf3e3f46 644 mbinary2
296 296
297 297 Filenames with spaces:
298 298
299 299 $ hg import -d "1000000 0" -m spaces - <<EOF
300 300 > diff --git a/foo bar b/foo bar
301 301 > new file mode 100644
302 302 > index 0000000..257cc56
303 303 > --- /dev/null
304 304 > +++ b/foo bar
305 305 > @@ -0,0 +1 @@
306 306 > +foo
307 307 > EOF
308 308 applying patch from stdin
309 309
310 310 $ hg tip -q
311 311 13:04750ef42fb3
312 312
313 313 $ cat "foo bar"
314 314 foo
315 315
316 316 Copy then modify the original file:
317 317
318 318 $ hg import -d "1000000 0" -m copy-mod-orig - <<EOF
319 319 > diff --git a/foo2 b/foo2
320 320 > index 257cc56..fe08ec6 100644
321 321 > --- a/foo2
322 322 > +++ b/foo2
323 323 > @@ -1 +1,2 @@
324 324 > foo
325 325 > +new line
326 326 > diff --git a/foo2 b/foo3
327 327 > similarity index 100%
328 328 > copy from foo2
329 329 > copy to foo3
330 330 > EOF
331 331 applying patch from stdin
332 332
333 333 $ hg tip -q
334 334 14:c4cd9cdeaa74
335 335
336 336 $ cat foo3
337 337 foo
338 338
339 339 Move text file and patch as binary
340 340
341 341 $ echo a > text2
342 342 $ hg ci -Am0
343 343 adding text2
344 344 $ hg import -d "1000000 0" -m rename-as-binary - <<"EOF"
345 345 > diff --git a/text2 b/binary2
346 346 > rename from text2
347 347 > rename to binary2
348 348 > index 78981922613b2afb6025042ff6bd878ac1994e85..10efcb362e9f3b3420fcfbfc0e37f3dc16e29757
349 349 > GIT binary patch
350 350 > literal 5
351 351 > Mc$`b*O5$Pw00T?_*Z=?k
352 352 >
353 353 > EOF
354 354 applying patch from stdin
355 355
356 356 $ cat binary2
357 357 a
358 358 b
359 359 \x00 (no-eol) (esc)
360 360
361 361 $ hg st --copies --change .
362 362 A binary2
363 363 text2
364 364 R text2
365
366 Invalid base85 content
367 $ hg rollback
368 repository tip rolled back to revision 15 (undo import)
369 working directory now based on revision 15
370 $ hg revert -aq
371 $ hg import -d "1000000 0" -m invalid-binary - <<"EOF"
372 > diff --git a/text2 b/binary2
373 > rename from text2
374 > rename to binary2
375 > index 78981922613b2afb6025042ff6bd878ac1994e85..10efcb362e9f3b3420fcfbfc0e37f3dc16e29757
376 > GIT binary patch
377 > literal 5
378 > Mc$`b*O.$Pw00T?_*Z=?k
379 >
380 > EOF
381 applying patch from stdin
382 abort: could not decode binary patch: bad base85 character at position 6
383 [255]
365 384 $ cd ..
366 385
367 386 Consecutive import with renames (issue2459)
368 387
369 388 $ hg init issue2459
370 389 $ cd issue2459
371 390 $ hg import --no-commit --force - <<EOF
372 391 > diff --git a/a b/a
373 392 > new file mode 100644
374 393 > EOF
375 394 applying patch from stdin
376 395 $ hg import --no-commit --force - <<EOF
377 396 > diff --git a/a b/b
378 397 > rename from a
379 398 > rename to b
380 399 > EOF
381 400 applying patch from stdin
382 401 a has not been committed yet, so no copy data will be stored for b.
383 402 $ hg debugstate
384 403 a 0 -1 unset b
385 404 $ hg ci -m done
386 405 $ cd ..
387 406
388 407 Renames and strip
389 408
390 409 $ hg init renameandstrip
391 410 $ cd renameandstrip
392 411 $ echo a > a
393 412 $ hg ci -Am adda
394 413 adding a
395 414 $ hg import --no-commit -p2 - <<EOF
396 415 > diff --git a/foo/a b/foo/b
397 416 > rename from foo/a
398 417 > rename to foo/b
399 418 > EOF
400 419 applying patch from stdin
401 420 $ hg st --copies
402 421 A b
403 422 a
404 423 R a
405 424
406 425 Renames, similarity and git diff
407 426
408 427 $ hg revert -aC
409 428 undeleting a
410 429 forgetting b
411 430 $ rm b
412 431 $ hg import --similarity 90 --no-commit - <<EOF
413 432 > diff --git a/a b/b
414 433 > rename from a
415 434 > rename to b
416 435 > EOF
417 436 applying patch from stdin
418 437 $ hg st --copies
419 438 A b
420 439 a
421 440 R a
422 441 $ cd ..
423 442
424 443 Pure copy with existing destination
425 444
426 445 $ hg init copytoexisting
427 446 $ cd copytoexisting
428 447 $ echo a > a
429 448 $ echo b > b
430 449 $ hg ci -Am add
431 450 adding a
432 451 adding b
433 452 $ hg import --no-commit - <<EOF
434 453 > diff --git a/a b/b
435 454 > copy from a
436 455 > copy to b
437 456 > EOF
438 457 applying patch from stdin
439 458 abort: cannot create b: destination already exists
440 459 [255]
441 460 $ cat b
442 461 b
443 462
444 463 Copy and changes with existing destination
445 464
446 465 $ hg import --no-commit - <<EOF
447 466 > diff --git a/a b/b
448 467 > copy from a
449 468 > copy to b
450 469 > --- a/a
451 470 > +++ b/b
452 471 > @@ -1,1 +1,2 @@
453 472 > a
454 473 > +b
455 474 > EOF
456 475 applying patch from stdin
457 476 cannot create b: destination already exists
458 477 1 out of 1 hunks FAILED -- saving rejects to file b.rej
459 478 abort: patch failed to apply
460 479 [255]
461 480 $ cat b
462 481 b
463 482
464 483 $ ln -s b linkb
465 484 $ hg add linkb
466 485 $ hg ci -m addlinkb
467 486 $ hg import --no-commit - <<EOF
468 487 > diff --git a/linkb b/linkb
469 488 > deleted file mode 120000
470 489 > --- a/linkb
471 490 > +++ /dev/null
472 491 > @@ -1,1 +0,0 @@
473 492 > -badhunk
474 493 > \ No newline at end of file
475 494 > EOF
476 495 applying patch from stdin
477 496 patching file linkb
478 497 Hunk #1 FAILED at 0
479 498 1 out of 1 hunks FAILED -- saving rejects to file linkb.rej
480 499 abort: patch failed to apply
481 500 [255]
482 501 $ hg st
483 502 ? b.rej
484 503 ? linkb.rej
485 504
486 505 Test corner case involving copies and multiple hunks (issue3384)
487 506
488 507 $ hg revert -qa
489 508 $ hg import --no-commit - <<EOF
490 509 > diff --git a/a b/c
491 510 > copy from a
492 511 > copy to c
493 512 > --- a/a
494 513 > +++ b/c
495 514 > @@ -1,1 +1,2 @@
496 515 > a
497 516 > +a
498 517 > @@ -2,1 +2,2 @@
499 518 > a
500 519 > +a
501 520 > diff --git a/a b/a
502 521 > --- a/a
503 522 > +++ b/a
504 523 > @@ -1,1 +1,2 @@
505 524 > a
506 525 > +b
507 526 > EOF
508 527 applying patch from stdin
509 528
510 529 $ cd ..
General Comments 0
You need to be logged in to leave comments. Login now