/brz/remove-bazaar

To get this branch, use:
bzr branch http://gegoxaren.bato24.eu/bzr/brz/remove-bazaar

« back to all changes in this revision

Viewing changes to bzrlib/diff.py

  • Committer: John Arbash Meinel
  • Date: 2010-01-13 23:06:42 UTC
  • mfrom: (4634.119.7 2.0)
  • mto: This revision was merged to the branch mainline in revision 4959.
  • Revision ID: john@arbash-meinel.com-20100113230642-gr0vg8u3qyzz3p2b
Merge bzr.stable, bringing in bug fixes #175839, #504390

Show diffs side-by-side

added added

removed removed

Lines of Context:
bzrlib/diff.py
1
 
# Copyright (C) 2005-2014 Canonical Ltd.
 
1
# Copyright (C) 2004, 2005, 2006 Canonical Ltd.
2
2
#
3
3
# This program is free software; you can redistribute it and/or modify
4
4
# it under the terms of the GNU General Public License as published by
14
14
# along with this program; if not, write to the Free Software
15
15
# Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
16
16
 
17
 
from __future__ import absolute_import
18
 
 
19
17
import difflib
20
18
import os
21
19
import re
 
20
import shutil
22
21
import string
23
22
import sys
24
23
 
25
 
from .lazy_import import lazy_import
 
24
from bzrlib.lazy_import import lazy_import
26
25
lazy_import(globals(), """
27
26
import errno
28
27
import subprocess
29
28
import tempfile
 
29
import time
30
30
 
31
 
from breezy import (
32
 
    cmdline,
33
 
    controldir,
 
31
from bzrlib import (
 
32
    branch as _mod_branch,
 
33
    bzrdir,
 
34
    commands,
34
35
    errors,
35
36
    osutils,
36
37
    patiencediff,
39
40
    views,
40
41
    )
41
42
 
42
 
from breezy.workingtree import WorkingTree
43
 
from breezy.i18n import gettext
 
43
from bzrlib.workingtree import WorkingTree
44
44
""")
45
45
 
46
 
from .registry import (
47
 
    Registry,
 
46
from bzrlib.symbol_versioning import (
 
47
    deprecated_function,
48
48
    )
49
 
from .sixish import text_type
50
 
from .trace import mutter, note, warning
51
 
from .tree import FileTimestampUnavailable
52
 
 
53
 
 
54
 
DEFAULT_CONTEXT_AMOUNT = 3
 
49
from bzrlib.trace import mutter, note, warning
55
50
 
56
51
 
57
52
class AtTemplate(string.Template):
74
69
        self.opcodes = None
75
70
 
76
71
 
77
 
def internal_diff(old_label, oldlines, new_label, newlines, to_file,
 
72
def internal_diff(old_filename, oldlines, new_filename, newlines, to_file,
78
73
                  allow_binary=False, sequence_matcher=None,
79
 
                  path_encoding='utf8', context_lines=DEFAULT_CONTEXT_AMOUNT):
 
74
                  path_encoding='utf8'):
80
75
    # FIXME: difflib is wrong if there is no trailing newline.
81
76
    # The syntax used by patch seems to be "\ No newline at
82
77
    # end of file" following the last diff line from that
87
82
    # In the meantime we at least make sure the patch isn't
88
83
    # mangled.
89
84
 
 
85
 
 
86
    # Special workaround for Python2.3, where difflib fails if
 
87
    # both sequences are empty.
 
88
    if not oldlines and not newlines:
 
89
        return
 
90
 
90
91
    if allow_binary is False:
91
92
        textfile.check_text_lines(oldlines)
92
93
        textfile.check_text_lines(newlines)
93
94
 
94
95
    if sequence_matcher is None:
95
96
        sequence_matcher = patiencediff.PatienceSequenceMatcher
96
 
    ud = patiencediff.unified_diff_bytes(oldlines, newlines,
97
 
                                         fromfile=old_label.encode(
98
 
                                             path_encoding, 'replace'),
99
 
                                         tofile=new_label.encode(
100
 
                                             path_encoding, 'replace'),
101
 
                                         n=context_lines, sequencematcher=sequence_matcher)
 
97
    ud = patiencediff.unified_diff(oldlines, newlines,
 
98
                      fromfile=old_filename.encode(path_encoding),
 
99
                      tofile=new_filename.encode(path_encoding),
 
100
                      sequencematcher=sequence_matcher)
102
101
 
103
102
    ud = list(ud)
104
 
    if len(ud) == 0:  # Identical contents, nothing to do
 
103
    if len(ud) == 0: # Identical contents, nothing to do
105
104
        return
106
105
    # work-around for difflib being too smart for its own good
107
106
    # if /dev/null is "1,0", patch won't recognize it as /dev/null
108
107
    if not oldlines:
109
 
        ud[2] = ud[2].replace(b'-1,0', b'-0,0')
 
108
        ud[2] = ud[2].replace('-1,0', '-0,0')
110
109
    elif not newlines:
111
 
        ud[2] = ud[2].replace(b'+1,0', b'+0,0')
 
110
        ud[2] = ud[2].replace('+1,0', '+0,0')
112
111
 
113
112
    for line in ud:
114
113
        to_file.write(line)
115
 
        if not line.endswith(b'\n'):
116
 
            to_file.write(b"\n\\ No newline at end of file\n")
117
 
    to_file.write(b'\n')
 
114
        if not line.endswith('\n'):
 
115
            to_file.write("\n\\ No newline at end of file\n")
 
116
    to_file.write('\n')
118
117
 
119
118
 
120
119
def _spawn_external_diff(diffcmd, capture_errors=True):
121
 
    """Spawn the external diff process, and return the child handle.
 
120
    """Spawn the externall diff process, and return the child handle.
122
121
 
123
122
    :param diffcmd: The command list to spawn
124
123
    :param capture_errors: Capture stderr as well as setting LANG=C
146
145
                                stdout=subprocess.PIPE,
147
146
                                stderr=stderr,
148
147
                                env=env)
149
 
    except OSError as e:
 
148
    except OSError, e:
150
149
        if e.errno == errno.ENOENT:
151
150
            raise errors.NoDiff(str(e))
152
151
        raise
154
153
    return pipe
155
154
 
156
155
 
157
 
# diff style options as of GNU diff v3.2
158
 
style_option_list = ['-c', '-C', '--context',
159
 
                     '-e', '--ed',
160
 
                     '-f', '--forward-ed',
161
 
                     '-q', '--brief',
162
 
                     '--normal',
163
 
                     '-n', '--rcs',
164
 
                     '-u', '-U', '--unified',
165
 
                     '-y', '--side-by-side',
166
 
                     '-D', '--ifdef']
167
 
 
168
 
 
169
 
def default_style_unified(diff_opts):
170
 
    """Default to unified diff style if alternative not specified in diff_opts.
171
 
 
172
 
        diff only allows one style to be specified; they don't override.
173
 
        Note that some of these take optargs, and the optargs can be
174
 
        directly appended to the options.
175
 
        This is only an approximate parser; it doesn't properly understand
176
 
        the grammar.
177
 
 
178
 
    :param diff_opts: List of options for external (GNU) diff.
179
 
    :return: List of options with default style=='unified'.
180
 
    """
181
 
    for s in style_option_list:
182
 
        for j in diff_opts:
183
 
            if j.startswith(s):
184
 
                break
185
 
        else:
186
 
            continue
187
 
        break
188
 
    else:
189
 
        diff_opts.append('-u')
190
 
    return diff_opts
191
 
 
192
 
 
193
 
def external_diff(old_label, oldlines, new_label, newlines, to_file,
 
156
def external_diff(old_filename, oldlines, new_filename, newlines, to_file,
194
157
                  diff_opts):
195
158
    """Display a diff by calling out to the external diff program."""
196
159
    # make sure our own output is properly ordered before the diff
197
160
    to_file.flush()
198
161
 
199
 
    oldtmp_fd, old_abspath = tempfile.mkstemp(prefix='brz-diff-old-')
200
 
    newtmp_fd, new_abspath = tempfile.mkstemp(prefix='brz-diff-new-')
 
162
    oldtmp_fd, old_abspath = tempfile.mkstemp(prefix='bzr-diff-old-')
 
163
    newtmp_fd, new_abspath = tempfile.mkstemp(prefix='bzr-diff-new-')
201
164
    oldtmpf = os.fdopen(oldtmp_fd, 'wb')
202
165
    newtmpf = os.fdopen(newtmp_fd, 'wb')
203
166
 
220
183
        if sys.platform == 'win32':
221
184
            # Popen doesn't do the proper encoding for external commands
222
185
            # Since we are dealing with an ANSI api, use mbcs encoding
223
 
            old_label = old_label.encode('mbcs')
224
 
            new_label = new_label.encode('mbcs')
 
186
            old_filename = old_filename.encode('mbcs')
 
187
            new_filename = new_filename.encode('mbcs')
225
188
        diffcmd = ['diff',
226
 
                   '--label', old_label,
 
189
                   '--label', old_filename,
227
190
                   old_abspath,
228
 
                   '--label', new_label,
 
191
                   '--label', new_filename,
229
192
                   new_abspath,
230
193
                   '--binary',
231
 
                   ]
 
194
                  ]
232
195
 
233
 
        diff_opts = default_style_unified(diff_opts)
 
196
        # diff only allows one style to be specified; they don't override.
 
197
        # note that some of these take optargs, and the optargs can be
 
198
        # directly appended to the options.
 
199
        # this is only an approximate parser; it doesn't properly understand
 
200
        # the grammar.
 
201
        for s in ['-c', '-u', '-C', '-U',
 
202
                  '-e', '--ed',
 
203
                  '-q', '--brief',
 
204
                  '--normal',
 
205
                  '-n', '--rcs',
 
206
                  '-y', '--side-by-side',
 
207
                  '-D', '--ifdef']:
 
208
            for j in diff_opts:
 
209
                if j.startswith(s):
 
210
                    break
 
211
            else:
 
212
                continue
 
213
            break
 
214
        else:
 
215
            diffcmd.append('-u')
234
216
 
235
217
        if diff_opts:
236
218
            diffcmd.extend(diff_opts)
237
219
 
238
220
        pipe = _spawn_external_diff(diffcmd, capture_errors=True)
239
 
        out, err = pipe.communicate()
 
221
        out,err = pipe.communicate()
240
222
        rc = pipe.returncode
241
223
 
242
224
        # internal_diff() adds a trailing newline, add one here for consistency
243
 
        out += b'\n'
 
225
        out += '\n'
244
226
        if rc == 2:
245
227
            # 'diff' gives retcode == 2 for all sorts of errors
246
228
            # one of those is 'Binary files differ'.
253
235
            out, err = pipe.communicate()
254
236
 
255
237
            # Write out the new i18n diff response
256
 
            to_file.write(out + b'\n')
 
238
            to_file.write(out+'\n')
257
239
            if pipe.returncode != 2:
258
240
                raise errors.BzrError(
259
 
                    'external diff failed with exit code 2'
260
 
                    ' when run with LANG=C and LC_ALL=C,'
261
 
                    ' but not when run natively: %r' % (diffcmd,))
 
241
                               'external diff failed with exit code 2'
 
242
                               ' when run with LANG=C and LC_ALL=C,'
 
243
                               ' but not when run natively: %r' % (diffcmd,))
262
244
 
263
 
            first_line = lang_c_out.split(b'\n', 1)[0]
 
245
            first_line = lang_c_out.split('\n', 1)[0]
264
246
            # Starting with diffutils 2.8.4 the word "binary" was dropped.
265
 
            m = re.match(b'^(binary )?files.*differ$', first_line, re.I)
 
247
            m = re.match('^(binary )?files.*differ$', first_line, re.I)
266
248
            if m is None:
267
249
                raise errors.BzrError('external diff failed with exit code 2;'
268
250
                                      ' command: %r' % (diffcmd,))
281
263
                msg = 'exit code %d' % rc
282
264
 
283
265
            raise errors.BzrError('external diff failed with %s; command: %r'
284
 
                                  % (msg, diffcmd))
 
266
                                  % (rc, diffcmd))
 
267
 
285
268
 
286
269
    finally:
287
270
        oldtmpf.close()                 # and delete
288
271
        newtmpf.close()
289
 
 
290
 
        def cleanup(path):
291
 
            # Warn in case the file couldn't be deleted (in case windows still
292
 
            # holds the file open, but not if the files have already been
293
 
            # deleted)
294
 
            try:
295
 
                os.remove(path)
296
 
            except OSError as e:
297
 
                if e.errno not in (errno.ENOENT,):
298
 
                    warning('Failed to delete temporary file: %s %s', path, e)
299
 
 
300
 
        cleanup(old_abspath)
301
 
        cleanup(new_abspath)
302
 
 
303
 
 
304
 
def get_trees_and_branches_to_diff_locked(
305
 
        path_list, revision_specs, old_url, new_url, add_cleanup, apply_view=True):
 
272
        # Clean up. Warn in case the files couldn't be deleted
 
273
        # (in case windows still holds the file open, but not
 
274
        # if the files have already been deleted)
 
275
        try:
 
276
            os.remove(old_abspath)
 
277
        except OSError, e:
 
278
            if e.errno not in (errno.ENOENT,):
 
279
                warning('Failed to delete temporary file: %s %s',
 
280
                        old_abspath, e)
 
281
        try:
 
282
            os.remove(new_abspath)
 
283
        except OSError:
 
284
            if e.errno not in (errno.ENOENT,):
 
285
                warning('Failed to delete temporary file: %s %s',
 
286
                        new_abspath, e)
 
287
 
 
288
 
 
289
def get_trees_and_branches_to_diff(path_list, revision_specs, old_url, new_url,
 
290
                                   apply_view=True):
306
291
    """Get the trees and specific files to diff given a list of paths.
307
292
 
308
293
    This method works out the trees to be diff'ed and the files of
319
304
    :param new_url:
320
305
        The url of the new branch or tree. If None, the tree to use is
321
306
        taken from the first path, if any, or the current working tree.
322
 
    :param add_cleanup:
323
 
        a callable like Command.add_cleanup.  get_trees_and_branches_to_diff
324
 
        will register cleanups that must be run to unlock the trees, etc.
325
307
    :param apply_view:
326
308
        if True and a view is set, apply the view or check that the paths
327
309
        are within it
328
310
    :returns:
329
311
        a tuple of (old_tree, new_tree, old_branch, new_branch,
330
312
        specific_files, extra_trees) where extra_trees is a sequence of
331
 
        additional trees to search in for file-ids.  The trees and branches
332
 
        will be read-locked until the cleanups registered via the add_cleanup
333
 
        param are run.
 
313
        additional trees to search in for file-ids.
334
314
    """
335
315
    # Get the old and new revision specs
336
316
    old_revision_spec = None
359
339
        default_location = path_list[0]
360
340
        other_paths = path_list[1:]
361
341
 
362
 
    def lock_tree_or_branch(wt, br):
363
 
        if wt is not None:
364
 
            wt.lock_read()
365
 
            add_cleanup(wt.unlock)
366
 
        elif br is not None:
367
 
            br.lock_read()
368
 
            add_cleanup(br.unlock)
369
 
 
370
342
    # Get the old location
371
343
    specific_files = []
372
344
    if old_url is None:
373
345
        old_url = default_location
374
346
    working_tree, branch, relpath = \
375
 
        controldir.ControlDir.open_containing_tree_or_branch(old_url)
376
 
    lock_tree_or_branch(working_tree, branch)
 
347
        bzrdir.BzrDir.open_containing_tree_or_branch(old_url)
377
348
    if consider_relpath and relpath != '':
378
349
        if working_tree is not None and apply_view:
379
350
            views.check_path_in_view(working_tree, relpath)
386
357
        new_url = default_location
387
358
    if new_url != old_url:
388
359
        working_tree, branch, relpath = \
389
 
            controldir.ControlDir.open_containing_tree_or_branch(new_url)
390
 
        lock_tree_or_branch(working_tree, branch)
 
360
            bzrdir.BzrDir.open_containing_tree_or_branch(new_url)
391
361
        if consider_relpath and relpath != '':
392
362
            if working_tree is not None and apply_view:
393
363
                views.check_path_in_view(working_tree, relpath)
394
364
            specific_files.append(relpath)
395
365
    new_tree = _get_tree_to_diff(new_revision_spec, working_tree, branch,
396
 
                                 basis_is_default=working_tree is None)
 
366
        basis_is_default=working_tree is None)
397
367
    new_branch = branch
398
368
 
399
369
    # Get the specific files (all files is None, no files is [])
400
370
    if make_paths_wt_relative and working_tree is not None:
401
 
        other_paths = working_tree.safe_relpath_files(
402
 
            other_paths,
 
371
        try:
 
372
            from bzrlib.builtins import safe_relpath_files
 
373
            other_paths = safe_relpath_files(working_tree, other_paths,
403
374
            apply_view=apply_view)
 
375
        except errors.FileInWrongBranch:
 
376
            raise errors.BzrCommandError("Files are in different branches")
404
377
    specific_files.extend(other_paths)
405
378
    if len(specific_files) == 0:
406
379
        specific_files = None
407
 
        if (working_tree is not None and working_tree.supports_views() and
408
 
                apply_view):
 
380
        if (working_tree is not None and working_tree.supports_views()
 
381
            and apply_view):
409
382
            view_files = working_tree.views.lookup_view()
410
383
            if view_files:
411
384
                specific_files = view_files
412
385
                view_str = views.view_display_str(view_files)
413
 
                note(gettext("*** Ignoring files outside view. View is %s") % view_str)
 
386
                note("*** Ignoring files outside view. View is %s" % view_str)
414
387
 
415
388
    # Get extra trees that ought to be searched for file-ids
416
389
    extra_trees = None
417
390
    if working_tree is not None and working_tree not in (old_tree, new_tree):
418
391
        extra_trees = (working_tree,)
419
 
    return (old_tree, new_tree, old_branch, new_branch,
420
 
            specific_files, extra_trees)
 
392
    return old_tree, new_tree, old_branch, new_branch, specific_files, extra_trees
421
393
 
422
394
 
423
395
def _get_tree_to_diff(spec, tree=None, branch=None, basis_is_default=True):
439
411
                    old_label='a/', new_label='b/',
440
412
                    extra_trees=None,
441
413
                    path_encoding='utf8',
442
 
                    using=None,
443
 
                    format_cls=None,
444
 
                    context=DEFAULT_CONTEXT_AMOUNT):
 
414
                    using=None):
445
415
    """Show in text form the changes from one tree to another.
446
416
 
447
 
    :param to_file: The output stream.
448
 
    :param specific_files: Include only changes to these files - None for all
449
 
        changes.
450
 
    :param external_diff_options: If set, use an external GNU diff and pass 
451
 
        these options.
452
 
    :param extra_trees: If set, more Trees to use for looking up file ids
453
 
    :param path_encoding: If set, the path will be encoded as specified, 
454
 
        otherwise is supposed to be utf8
455
 
    :param format_cls: Formatter class (DiffTree subclass)
 
417
    to_file
 
418
        The output stream.
 
419
 
 
420
    specific_files
 
421
        Include only changes to these files - None for all changes.
 
422
 
 
423
    external_diff_options
 
424
        If set, use an external GNU diff and pass these options.
 
425
 
 
426
    extra_trees
 
427
        If set, more Trees to use for looking up file ids
 
428
 
 
429
    path_encoding
 
430
        If set, the path will be encoded as specified, otherwise is supposed
 
431
        to be utf8
456
432
    """
457
 
    if context is None:
458
 
        context = DEFAULT_CONTEXT_AMOUNT
459
 
    if format_cls is None:
460
 
        format_cls = DiffTree
461
 
    with old_tree.lock_read():
 
433
    old_tree.lock_read()
 
434
    try:
462
435
        if extra_trees is not None:
463
436
            for tree in extra_trees:
464
437
                tree.lock_read()
465
438
        new_tree.lock_read()
466
439
        try:
467
 
            differ = format_cls.from_trees_options(old_tree, new_tree, to_file,
468
 
                                                   path_encoding,
469
 
                                                   external_diff_options,
470
 
                                                   old_label, new_label, using,
471
 
                                                   context_lines=context)
 
440
            differ = DiffTree.from_trees_options(old_tree, new_tree, to_file,
 
441
                                                 path_encoding,
 
442
                                                 external_diff_options,
 
443
                                                 old_label, new_label, using)
472
444
            return differ.show_diff(specific_files, extra_trees)
473
445
        finally:
474
446
            new_tree.unlock()
475
447
            if extra_trees is not None:
476
448
                for tree in extra_trees:
477
449
                    tree.unlock()
 
450
    finally:
 
451
        old_tree.unlock()
478
452
 
479
453
 
480
454
def _patch_header_date(tree, file_id, path):
481
455
    """Returns a timestamp suitable for use in a patch header."""
482
 
    try:
483
 
        mtime = tree.get_file_mtime(path)
484
 
    except FileTimestampUnavailable:
485
 
        mtime = 0
 
456
    mtime = tree.get_file_mtime(file_id, path)
486
457
    return timestamp.format_patch_date(mtime)
487
458
 
488
459
 
489
460
def get_executable_change(old_is_x, new_is_x):
490
 
    descr = {True: b"+x", False: b"-x", None: b"??"}
 
461
    descr = { True:"+x", False:"-x", None:"??" }
491
462
    if old_is_x != new_is_x:
492
 
        return [b"%s to %s" % (descr[old_is_x], descr[new_is_x],)]
 
463
        return ["%s to %s" % (descr[old_is_x], descr[new_is_x],)]
493
464
    else:
494
465
        return []
495
466
 
542
513
    Represents kind change as deletion + creation.  Uses the other differs
543
514
    to do this.
544
515
    """
545
 
 
546
516
    def __init__(self, differs):
547
517
        self.differs = differs
548
518
 
565
535
        if None in (old_kind, new_kind):
566
536
            return DiffPath.CANNOT_DIFF
567
537
        result = DiffPath._diff_many(self.differs, file_id, old_path,
568
 
                                     new_path, old_kind, None)
 
538
                                       new_path, old_kind, None)
569
539
        if result is DiffPath.CANNOT_DIFF:
570
540
            return result
571
541
        return DiffPath._diff_many(self.differs, file_id, old_path, new_path,
572
 
                                   None, new_kind)
 
542
                                     None, new_kind)
573
543
 
574
544
 
575
545
class DiffDirectory(DiffPath):
601
571
        if 'symlink' not in (old_kind, new_kind):
602
572
            return self.CANNOT_DIFF
603
573
        if old_kind == 'symlink':
604
 
            old_target = self.old_tree.get_symlink_target(old_path)
 
574
            old_target = self.old_tree.get_symlink_target(file_id)
605
575
        elif old_kind is None:
606
576
            old_target = None
607
577
        else:
608
578
            return self.CANNOT_DIFF
609
579
        if new_kind == 'symlink':
610
 
            new_target = self.new_tree.get_symlink_target(new_path)
 
580
            new_target = self.new_tree.get_symlink_target(file_id)
611
581
        elif new_kind is None:
612
582
            new_target = None
613
583
        else:
616
586
 
617
587
    def diff_symlink(self, old_target, new_target):
618
588
        if old_target is None:
619
 
            self.to_file.write(b'=== target is \'%s\'\n' %
620
 
                               new_target.encode(self.path_encoding, 'replace'))
 
589
            self.to_file.write('=== target is %r\n' % new_target)
621
590
        elif new_target is None:
622
 
            self.to_file.write(b'=== target was \'%s\'\n' %
623
 
                               old_target.encode(self.path_encoding, 'replace'))
 
591
            self.to_file.write('=== target was %r\n' % old_target)
624
592
        else:
625
 
            self.to_file.write(b'=== target changed \'%s\' => \'%s\'\n' %
626
 
                               (old_target.encode(self.path_encoding, 'replace'),
627
 
                                new_target.encode(self.path_encoding, 'replace')))
 
593
            self.to_file.write('=== target changed %r => %r\n' %
 
594
                              (old_target, new_target))
628
595
        return self.CHANGED
629
596
 
630
597
 
635
602
    EPOCH_DATE = '1970-01-01 00:00:00 +0000'
636
603
 
637
604
    def __init__(self, old_tree, new_tree, to_file, path_encoding='utf-8',
638
 
                 old_label='', new_label='', text_differ=internal_diff,
639
 
                 context_lines=DEFAULT_CONTEXT_AMOUNT):
 
605
                 old_label='', new_label='', text_differ=internal_diff):
640
606
        DiffPath.__init__(self, old_tree, new_tree, to_file, path_encoding)
641
607
        self.text_differ = text_differ
642
608
        self.old_label = old_label
643
609
        self.new_label = new_label
644
610
        self.path_encoding = path_encoding
645
 
        self.context_lines = context_lines
646
611
 
647
612
    def diff(self, file_id, old_path, new_path, old_kind, new_kind):
648
613
        """Compare two files in unified diff format
670
635
            to_file_id = None
671
636
        else:
672
637
            return self.CANNOT_DIFF
673
 
        from_label = '%s%s\t%s' % (self.old_label, old_path,
674
 
                                   old_date)
675
 
        to_label = '%s%s\t%s' % (self.new_label, new_path,
676
 
                                 new_date)
677
 
        return self.diff_text(old_path, new_path, from_label, to_label,
678
 
                              from_file_id, to_file_id)
 
638
        from_label = '%s%s\t%s' % (self.old_label, old_path, old_date)
 
639
        to_label = '%s%s\t%s' % (self.new_label, new_path, new_date)
 
640
        return self.diff_text(from_file_id, to_file_id, from_label, to_label,
 
641
            old_path, new_path)
679
642
 
680
 
    def diff_text(self, from_path, to_path, from_label, to_label,
681
 
                  from_file_id=None, to_file_id=None):
 
643
    def diff_text(self, from_file_id, to_file_id, from_label, to_label,
 
644
        from_path=None, to_path=None):
682
645
        """Diff the content of given files in two trees
683
646
 
684
 
        :param from_path: The path in the from tree. If None,
 
647
        :param from_file_id: The id of the file in the from tree.  If None,
685
648
            the file is not present in the from tree.
686
 
        :param to_path: The path in the to tree. This may refer
687
 
            to a different file from from_path.  If None,
 
649
        :param to_file_id: The id of the file in the to tree.  This may refer
 
650
            to a different file from from_file_id.  If None,
688
651
            the file is not present in the to tree.
689
 
        :param from_file_id: The id of the file in the from tree or None if
690
 
            unknown.
691
 
        :param to_file_id: The id of the file in the to tree or None if
692
 
            unknown.
 
652
        :param from_path: The path in the from tree or None if unknown.
 
653
        :param to_path: The path in the to tree or None if unknown.
693
654
        """
694
655
        def _get_text(tree, file_id, path):
695
 
            if file_id is None:
 
656
            if file_id is not None:
 
657
                return tree.get_file(file_id, path).readlines()
 
658
            else:
696
659
                return []
697
 
            return tree.get_file_lines(path)
698
660
        try:
699
661
            from_text = _get_text(self.old_tree, from_file_id, from_path)
700
662
            to_text = _get_text(self.new_tree, to_file_id, to_path)
701
663
            self.text_differ(from_label, from_text, to_label, to_text,
702
 
                             self.to_file, path_encoding=self.path_encoding,
703
 
                             context_lines=self.context_lines)
 
664
                             self.to_file)
704
665
        except errors.BinaryFile:
705
666
            self.to_file.write(
706
 
                ("Binary files %s and %s differ\n" %
707
 
                 (from_label, to_label)).encode(self.path_encoding, 'replace'))
 
667
                  ("Binary files %s and %s differ\n" %
 
668
                  (from_label, to_label)).encode(self.path_encoding))
708
669
        return self.CHANGED
709
670
 
710
671
 
714
675
                 path_encoding='utf-8'):
715
676
        DiffPath.__init__(self, old_tree, new_tree, to_file, path_encoding)
716
677
        self.command_template = command_template
717
 
        self._root = osutils.mkdtemp(prefix='brz-diff-')
 
678
        self._root = osutils.mkdtemp(prefix='bzr-diff-')
718
679
 
719
680
    @classmethod
720
681
    def from_string(klass, command_string, old_tree, new_tree, to_file,
721
682
                    path_encoding='utf-8'):
722
 
        command_template = cmdline.split(command_string)
 
683
        command_template = commands.shlex_split_unicode(command_string)
723
684
        if '@' not in command_string:
724
685
            command_template.extend(['@old_path', '@new_path'])
725
686
        return klass(command_template, old_tree, new_tree, to_file,
726
687
                     path_encoding)
727
688
 
728
689
    @classmethod
729
 
    def make_from_diff_tree(klass, command_string, external_diff_options=None):
 
690
    def make_from_diff_tree(klass, command_string):
730
691
        def from_diff_tree(diff_tree):
731
 
            full_command_string = [command_string]
732
 
            if external_diff_options is not None:
733
 
                full_command_string += ' ' + external_diff_options
734
 
            return klass.from_string(full_command_string, diff_tree.old_tree,
 
692
            return klass.from_string(command_string, diff_tree.old_tree,
735
693
                                     diff_tree.new_tree, diff_tree.to_file)
736
694
        return from_diff_tree
737
695
 
738
696
    def _get_command(self, old_path, new_path):
739
697
        my_map = {'old_path': old_path, 'new_path': new_path}
740
 
        command = [AtTemplate(t).substitute(my_map) for t in
741
 
                   self.command_template]
742
 
        if sys.platform == 'win32':  # Popen doesn't accept unicode on win32
743
 
            command_encoded = []
744
 
            for c in command:
745
 
                if isinstance(c, text_type):
746
 
                    command_encoded.append(c.encode('mbcs'))
747
 
                else:
748
 
                    command_encoded.append(c)
749
 
            return command_encoded
750
 
        else:
751
 
            return command
 
698
        return [AtTemplate(t).substitute(my_map) for t in
 
699
                self.command_template]
752
700
 
753
701
    def _execute(self, old_path, new_path):
754
702
        command = self._get_command(old_path, new_path)
755
703
        try:
756
704
            proc = subprocess.Popen(command, stdout=subprocess.PIPE,
757
705
                                    cwd=self._root)
758
 
        except OSError as e:
 
706
        except OSError, e:
759
707
            if e.errno == errno.ENOENT:
760
708
                raise errors.ExecutableMissing(command[0])
761
709
            else:
762
710
                raise
763
711
        self.to_file.write(proc.stdout.read())
764
 
        proc.stdout.close()
765
712
        return proc.wait()
766
713
 
767
714
    def _try_symlink_root(self, tree, prefix):
768
 
        if (getattr(tree, 'abspath', None) is None or
769
 
                not osutils.host_os_dereferences_symlinks()):
 
715
        if (getattr(tree, 'abspath', None) is None
 
716
            or not osutils.host_os_dereferences_symlinks()):
770
717
            return False
771
718
        try:
772
719
            os.symlink(tree.abspath(''), osutils.pathjoin(self._root, prefix))
773
 
        except OSError as e:
 
720
        except OSError, e:
774
721
            if e.errno != errno.EEXIST:
775
722
                raise
776
723
        return True
777
724
 
778
 
    @staticmethod
779
 
    def _fenc():
780
 
        """Returns safe encoding for passing file path to diff tool"""
781
 
        if sys.platform == 'win32':
782
 
            return 'mbcs'
783
 
        else:
784
 
            # Don't fallback to 'utf-8' because subprocess may not be able to
785
 
            # handle utf-8 correctly when locale is not utf-8.
786
 
            return sys.getfilesystemencoding() or 'ascii'
787
 
 
788
 
    def _is_safepath(self, path):
789
 
        """Return true if `path` may be able to pass to subprocess."""
790
 
        fenc = self._fenc()
791
 
        try:
792
 
            return path == path.encode(fenc).decode(fenc)
793
 
        except UnicodeError:
794
 
            return False
795
 
 
796
 
    def _safe_filename(self, prefix, relpath):
797
 
        """Replace unsafe character in `relpath` then join `self._root`,
798
 
        `prefix` and `relpath`."""
799
 
        fenc = self._fenc()
800
 
        # encoded_str.replace('?', '_') may break multibyte char.
801
 
        # So we should encode, decode, then replace(u'?', u'_')
802
 
        relpath_tmp = relpath.encode(fenc, 'replace').decode(fenc, 'replace')
803
 
        relpath_tmp = relpath_tmp.replace(u'?', u'_')
804
 
        return osutils.pathjoin(self._root, prefix, relpath_tmp)
805
 
 
806
 
    def _write_file(self, relpath, tree, prefix, force_temp=False,
807
 
                    allow_write=False, file_id=None):
 
725
    def _write_file(self, file_id, tree, prefix, relpath, force_temp=False,
 
726
                    allow_write=False):
808
727
        if not force_temp and isinstance(tree, WorkingTree):
809
 
            full_path = tree.abspath(relpath)
810
 
            if self._is_safepath(full_path):
811
 
                return full_path
812
 
 
813
 
        full_path = self._safe_filename(prefix, relpath)
 
728
            return tree.abspath(tree.id2path(file_id))
 
729
        
 
730
        full_path = osutils.pathjoin(self._root, prefix, relpath)
814
731
        if not force_temp and self._try_symlink_root(tree, prefix):
815
732
            return full_path
816
733
        parent_dir = osutils.dirname(full_path)
817
734
        try:
818
735
            os.makedirs(parent_dir)
819
 
        except OSError as e:
 
736
        except OSError, e:
820
737
            if e.errno != errno.EEXIST:
821
738
                raise
822
 
        source = tree.get_file(relpath)
 
739
        source = tree.get_file(file_id, relpath)
823
740
        try:
824
 
            with open(full_path, 'wb') as target:
 
741
            target = open(full_path, 'wb')
 
742
            try:
825
743
                osutils.pumpfile(source, target)
 
744
            finally:
 
745
                target.close()
826
746
        finally:
827
747
            source.close()
828
 
        try:
829
 
            mtime = tree.get_file_mtime(relpath)
830
 
        except FileTimestampUnavailable:
831
 
            pass
832
 
        else:
833
 
            os.utime(full_path, (mtime, mtime))
834
748
        if not allow_write:
835
749
            osutils.make_readonly(full_path)
 
750
        mtime = tree.get_file_mtime(file_id)
 
751
        os.utime(full_path, (mtime, mtime))
836
752
        return full_path
837
753
 
838
 
    def _prepare_files(self, old_path, new_path, force_temp=False,
839
 
                       allow_write_new=False, file_id=None):
840
 
        old_disk_path = self._write_file(old_path, self.old_tree, 'old',
841
 
                                         force_temp, file_id=file_id)
842
 
        new_disk_path = self._write_file(new_path, self.new_tree, 'new',
843
 
                                         force_temp, file_id=file_id,
 
754
    def _prepare_files(self, file_id, old_path, new_path, force_temp=False,
 
755
                       allow_write_new=False):
 
756
        old_disk_path = self._write_file(file_id, self.old_tree, 'old',
 
757
                                         old_path, force_temp)
 
758
        new_disk_path = self._write_file(file_id, self.new_tree, 'new',
 
759
                                         new_path, force_temp,
844
760
                                         allow_write=allow_write_new)
845
761
        return old_disk_path, new_disk_path
846
762
 
847
763
    def finish(self):
848
764
        try:
849
765
            osutils.rmtree(self._root)
850
 
        except OSError as e:
 
766
        except OSError, e:
851
767
            if e.errno != errno.ENOENT:
852
768
                mutter("The temporary directory \"%s\" was not "
853
 
                       "cleanly removed: %s." % (self._root, e))
 
769
                        "cleanly removed: %s." % (self._root, e))
854
770
 
855
771
    def diff(self, file_id, old_path, new_path, old_kind, new_kind):
856
772
        if (old_kind, new_kind) != ('file', 'file'):
857
773
            return DiffPath.CANNOT_DIFF
858
774
        (old_disk_path, new_disk_path) = self._prepare_files(
859
 
            old_path, new_path, file_id=file_id)
 
775
                                                file_id, old_path, new_path)
860
776
        self._execute(old_disk_path, new_disk_path)
861
777
 
862
 
    def edit_file(self, old_path, new_path, file_id=None):
 
778
    def edit_file(self, file_id):
863
779
        """Use this tool to edit a file.
864
780
 
865
781
        A temporary copy will be edited, and the new contents will be
868
784
        :param file_id: The id of the file to edit.
869
785
        :return: The new contents of the file.
870
786
        """
871
 
        old_abs_path, new_abs_path = self._prepare_files(
872
 
            old_path, new_path, allow_write_new=True, force_temp=True,
873
 
            file_id=file_id)
874
 
        command = self._get_command(old_abs_path, new_abs_path)
 
787
        old_path = self.old_tree.id2path(file_id)
 
788
        new_path = self.new_tree.id2path(file_id)
 
789
        new_abs_path = self._prepare_files(file_id, old_path, new_path,
 
790
                                           allow_write_new=True,
 
791
                                           force_temp=True)[1]
 
792
        command = self._get_command(osutils.pathjoin('old', old_path),
 
793
                                    osutils.pathjoin('new', new_path))
875
794
        subprocess.call(command, cwd=self._root)
876
 
        with open(new_abs_path, 'rb') as new_file:
 
795
        new_file = open(new_abs_path, 'r')
 
796
        try:
877
797
            return new_file.read()
 
798
        finally:
 
799
            new_file.close()
878
800
 
879
801
 
880
802
class DiffTree(object):
908
830
            DiffPaths"""
909
831
        if diff_text is None:
910
832
            diff_text = DiffText(old_tree, new_tree, to_file, path_encoding,
911
 
                                 '', '', internal_diff)
 
833
                                 '', '',  internal_diff)
912
834
        self.old_tree = old_tree
913
835
        self.new_tree = new_tree
914
836
        self.to_file = to_file
922
844
    @classmethod
923
845
    def from_trees_options(klass, old_tree, new_tree, to_file,
924
846
                           path_encoding, external_diff_options, old_label,
925
 
                           new_label, using, context_lines):
 
847
                           new_label, using):
926
848
        """Factory for producing a DiffTree.
927
849
 
928
850
        Designed to accept options used by show_diff_trees.
929
 
 
930
851
        :param old_tree: The tree to show as old in the comparison
931
852
        :param new_tree: The tree to show as new in the comparison
932
853
        :param to_file: File to write comparisons to
938
859
        :param using: Commandline to use to invoke an external diff tool
939
860
        """
940
861
        if using is not None:
941
 
            extra_factories = [DiffFromTool.make_from_diff_tree(
942
 
                using, external_diff_options)]
 
862
            extra_factories = [DiffFromTool.make_from_diff_tree(using)]
943
863
        else:
944
864
            extra_factories = []
945
865
        if external_diff_options:
946
866
            opts = external_diff_options.split()
947
 
 
948
 
            def diff_file(olab, olines, nlab, nlines, to_file, path_encoding=None, context_lines=None):
949
 
                """:param path_encoding: not used but required
950
 
                        to match the signature of internal_diff.
951
 
                """
 
867
            def diff_file(olab, olines, nlab, nlines, to_file):
952
868
                external_diff(olab, olines, nlab, nlines, to_file, opts)
953
869
        else:
954
870
            diff_file = internal_diff
955
871
        diff_text = DiffText(old_tree, new_tree, to_file, path_encoding,
956
 
                             old_label, new_label, diff_file, context_lines=context_lines)
 
872
                             old_label, new_label, diff_file)
957
873
        return klass(old_tree, new_tree, to_file, path_encoding, diff_text,
958
874
                     extra_factories)
959
875
 
960
876
    def show_diff(self, specific_files, extra_trees=None):
961
877
        """Write tree diff to self.to_file
962
878
 
963
 
        :param specific_files: the specific files to compare (recursive)
 
879
        :param sepecific_files: the specific files to compare (recursive)
964
880
        :param extra_trees: extra trees to use for mapping paths to file_ids
965
881
        """
966
882
        try:
973
889
        # TODO: Generation of pseudo-diffs for added/deleted files could
974
890
        # be usefully made into a much faster special case.
975
891
        iterator = self.new_tree.iter_changes(self.old_tree,
976
 
                                              specific_files=specific_files,
977
 
                                              extra_trees=extra_trees,
978
 
                                              require_versioned=True)
 
892
                                               specific_files=specific_files,
 
893
                                               extra_trees=extra_trees,
 
894
                                               require_versioned=True)
979
895
        has_changes = 0
980
 
 
981
896
        def changes_key(change):
982
897
            old_path, new_path = change[1]
983
898
            path = new_path
984
899
            if path is None:
985
900
                path = old_path
986
901
            return path
987
 
 
988
902
        def get_encoded_path(path):
989
903
            if path is not None:
990
904
                return path.encode(self.path_encoding, "replace")
1002
916
            renamed = (parent[0], name[0]) != (parent[1], name[1])
1003
917
 
1004
918
            properties_changed = []
1005
 
            properties_changed.extend(
1006
 
                get_executable_change(executable[0], executable[1]))
 
919
            properties_changed.extend(get_executable_change(executable[0], executable[1]))
1007
920
 
1008
921
            if properties_changed:
1009
 
                prop_str = b" (properties changed: %s)" % (
1010
 
                    b", ".join(properties_changed),)
 
922
                prop_str = " (properties changed: %s)" % (", ".join(properties_changed),)
1011
923
            else:
1012
 
                prop_str = b""
 
924
                prop_str = ""
1013
925
 
1014
926
            if (old_present, new_present) == (True, False):
1015
 
                self.to_file.write(b"=== removed %s '%s'\n" %
1016
 
                                   (kind[0].encode('ascii'), oldpath_encoded))
 
927
                self.to_file.write("=== removed %s '%s'\n" %
 
928
                                   (kind[0], oldpath_encoded))
1017
929
                newpath = oldpath
1018
930
            elif (old_present, new_present) == (False, True):
1019
 
                self.to_file.write(b"=== added %s '%s'\n" %
1020
 
                                   (kind[1].encode('ascii'), newpath_encoded))
 
931
                self.to_file.write("=== added %s '%s'\n" %
 
932
                                   (kind[1], newpath_encoded))
1021
933
                oldpath = newpath
1022
934
            elif renamed:
1023
 
                self.to_file.write(b"=== renamed %s '%s' => '%s'%s\n" %
1024
 
                                   (kind[0].encode('ascii'), oldpath_encoded, newpath_encoded, prop_str))
 
935
                self.to_file.write("=== renamed %s '%s' => '%s'%s\n" %
 
936
                    (kind[0], oldpath_encoded, newpath_encoded, prop_str))
1025
937
            else:
1026
938
                # if it was produced by iter_changes, it must be
1027
939
                # modified *somehow*, either content or execute bit.
1028
 
                self.to_file.write(b"=== modified %s '%s'%s\n" % (kind[0].encode('ascii'),
1029
 
                                                                  newpath_encoded, prop_str))
 
940
                self.to_file.write("=== modified %s '%s'%s\n" % (kind[0],
 
941
                                   newpath_encoded, prop_str))
1030
942
            if changed_content:
1031
 
                self._diff(oldpath, newpath, kind[0], kind[1], file_id=file_id)
 
943
                self._diff(file_id, oldpath, newpath, kind[0], kind[1])
1032
944
                has_changes = 1
1033
945
            if renamed:
1034
946
                has_changes = 1
1041
953
        :param old_path: The path of the file in the old tree
1042
954
        :param new_path: The path of the file in the new tree
1043
955
        """
1044
 
        if old_path is None:
 
956
        try:
 
957
            old_kind = self.old_tree.kind(file_id)
 
958
        except (errors.NoSuchId, errors.NoSuchFile):
1045
959
            old_kind = None
1046
 
        else:
1047
 
            old_kind = self.old_tree.kind(old_path)
1048
 
        if new_path is None:
 
960
        try:
 
961
            new_kind = self.new_tree.kind(file_id)
 
962
        except (errors.NoSuchId, errors.NoSuchFile):
1049
963
            new_kind = None
1050
 
        else:
1051
 
            new_kind = self.new_tree.kind(new_path)
1052
 
        self._diff(old_path, new_path, old_kind, new_kind, file_id=file_id)
1053
 
 
1054
 
    def _diff(self, old_path, new_path, old_kind, new_kind, file_id):
 
964
        self._diff(file_id, old_path, new_path, old_kind, new_kind)
 
965
 
 
966
 
 
967
    def _diff(self, file_id, old_path, new_path, old_kind, new_kind):
1055
968
        result = DiffPath._diff_many(self.differs, file_id, old_path,
1056
 
                                     new_path, old_kind, new_kind)
 
969
                                       new_path, old_kind, new_kind)
1057
970
        if result is DiffPath.CANNOT_DIFF:
1058
971
            error_path = new_path
1059
972
            if error_path is None:
1060
973
                error_path = old_path
1061
974
            raise errors.NoDiffFound(error_path)
1062
 
 
1063
 
 
1064
 
format_registry = Registry()
1065
 
format_registry.register('default', DiffTree)