/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: Sabin Iacob
  • Date: 2009-03-23 14:59:43 UTC
  • mto: (4189.1.1 integration)
  • mto: This revision was merged to the branch mainline in revision 4193.
  • Revision ID: iacobs@m0n5t3r.info-20090323145943-3s3p1px5q1rkh2e5
update FSF mailing address

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
22
 
import string
 
20
import shutil
23
21
import sys
24
22
 
25
 
from .lazy_import import lazy_import
 
23
from bzrlib.lazy_import import lazy_import
26
24
lazy_import(globals(), """
27
25
import errno
28
26
import subprocess
29
27
import tempfile
 
28
import time
30
29
 
31
 
from breezy import (
32
 
    cmdline,
33
 
    controldir,
 
30
from bzrlib import (
 
31
    branch as _mod_branch,
 
32
    bzrdir,
 
33
    commands,
34
34
    errors,
35
35
    osutils,
36
36
    patiencediff,
38
38
    timestamp,
39
39
    views,
40
40
    )
41
 
 
42
 
from breezy.workingtree import WorkingTree
43
 
from breezy.i18n import gettext
44
41
""")
45
42
 
46
 
from .registry import (
47
 
    Registry,
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
55
 
 
56
 
 
57
 
class AtTemplate(string.Template):
58
 
    """Templating class that uses @ instead of $."""
59
 
 
60
 
    delimiter = '@'
 
43
from bzrlib.symbol_versioning import (
 
44
        deprecated_function,
 
45
        one_three
 
46
        )
 
47
from bzrlib.trace import mutter, note, warning
61
48
 
62
49
 
63
50
# TODO: Rather than building a changeset object, we should probably
74
61
        self.opcodes = None
75
62
 
76
63
 
77
 
def internal_diff(old_label, oldlines, new_label, newlines, to_file,
 
64
def internal_diff(old_filename, oldlines, new_filename, newlines, to_file,
78
65
                  allow_binary=False, sequence_matcher=None,
79
 
                  path_encoding='utf8', context_lines=DEFAULT_CONTEXT_AMOUNT):
 
66
                  path_encoding='utf8'):
80
67
    # FIXME: difflib is wrong if there is no trailing newline.
81
68
    # The syntax used by patch seems to be "\ No newline at
82
69
    # end of file" following the last diff line from that
87
74
    # In the meantime we at least make sure the patch isn't
88
75
    # mangled.
89
76
 
 
77
 
 
78
    # Special workaround for Python2.3, where difflib fails if
 
79
    # both sequences are empty.
 
80
    if not oldlines and not newlines:
 
81
        return
 
82
 
90
83
    if allow_binary is False:
91
84
        textfile.check_text_lines(oldlines)
92
85
        textfile.check_text_lines(newlines)
93
86
 
94
87
    if sequence_matcher is None:
95
88
        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)
 
89
    ud = patiencediff.unified_diff(oldlines, newlines,
 
90
                      fromfile=old_filename.encode(path_encoding),
 
91
                      tofile=new_filename.encode(path_encoding),
 
92
                      sequencematcher=sequence_matcher)
102
93
 
103
94
    ud = list(ud)
104
 
    if len(ud) == 0:  # Identical contents, nothing to do
 
95
    if len(ud) == 0: # Identical contents, nothing to do
105
96
        return
106
97
    # work-around for difflib being too smart for its own good
107
98
    # if /dev/null is "1,0", patch won't recognize it as /dev/null
108
99
    if not oldlines:
109
 
        ud[2] = ud[2].replace(b'-1,0', b'-0,0')
 
100
        ud[2] = ud[2].replace('-1,0', '-0,0')
110
101
    elif not newlines:
111
 
        ud[2] = ud[2].replace(b'+1,0', b'+0,0')
 
102
        ud[2] = ud[2].replace('+1,0', '+0,0')
112
103
 
113
104
    for line in ud:
114
105
        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')
 
106
        if not line.endswith('\n'):
 
107
            to_file.write("\n\\ No newline at end of file\n")
 
108
    to_file.write('\n')
118
109
 
119
110
 
120
111
def _spawn_external_diff(diffcmd, capture_errors=True):
121
 
    """Spawn the external diff process, and return the child handle.
 
112
    """Spawn the externall diff process, and return the child handle.
122
113
 
123
114
    :param diffcmd: The command list to spawn
124
115
    :param capture_errors: Capture stderr as well as setting LANG=C
146
137
                                stdout=subprocess.PIPE,
147
138
                                stderr=stderr,
148
139
                                env=env)
149
 
    except OSError as e:
 
140
    except OSError, e:
150
141
        if e.errno == errno.ENOENT:
151
142
            raise errors.NoDiff(str(e))
152
143
        raise
154
145
    return pipe
155
146
 
156
147
 
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,
 
148
def external_diff(old_filename, oldlines, new_filename, newlines, to_file,
194
149
                  diff_opts):
195
150
    """Display a diff by calling out to the external diff program."""
196
151
    # make sure our own output is properly ordered before the diff
197
152
    to_file.flush()
198
153
 
199
 
    oldtmp_fd, old_abspath = tempfile.mkstemp(prefix='brz-diff-old-')
200
 
    newtmp_fd, new_abspath = tempfile.mkstemp(prefix='brz-diff-new-')
 
154
    oldtmp_fd, old_abspath = tempfile.mkstemp(prefix='bzr-diff-old-')
 
155
    newtmp_fd, new_abspath = tempfile.mkstemp(prefix='bzr-diff-new-')
201
156
    oldtmpf = os.fdopen(oldtmp_fd, 'wb')
202
157
    newtmpf = os.fdopen(newtmp_fd, 'wb')
203
158
 
217
172
 
218
173
        if not diff_opts:
219
174
            diff_opts = []
220
 
        if sys.platform == 'win32':
221
 
            # Popen doesn't do the proper encoding for external commands
222
 
            # 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')
225
175
        diffcmd = ['diff',
226
 
                   '--label', old_label,
 
176
                   '--label', old_filename,
227
177
                   old_abspath,
228
 
                   '--label', new_label,
 
178
                   '--label', new_filename,
229
179
                   new_abspath,
230
180
                   '--binary',
231
 
                   ]
 
181
                  ]
232
182
 
233
 
        diff_opts = default_style_unified(diff_opts)
 
183
        # diff only allows one style to be specified; they don't override.
 
184
        # note that some of these take optargs, and the optargs can be
 
185
        # directly appended to the options.
 
186
        # this is only an approximate parser; it doesn't properly understand
 
187
        # the grammar.
 
188
        for s in ['-c', '-u', '-C', '-U',
 
189
                  '-e', '--ed',
 
190
                  '-q', '--brief',
 
191
                  '--normal',
 
192
                  '-n', '--rcs',
 
193
                  '-y', '--side-by-side',
 
194
                  '-D', '--ifdef']:
 
195
            for j in diff_opts:
 
196
                if j.startswith(s):
 
197
                    break
 
198
            else:
 
199
                continue
 
200
            break
 
201
        else:
 
202
            diffcmd.append('-u')
234
203
 
235
204
        if diff_opts:
236
205
            diffcmd.extend(diff_opts)
237
206
 
238
207
        pipe = _spawn_external_diff(diffcmd, capture_errors=True)
239
 
        out, err = pipe.communicate()
 
208
        out,err = pipe.communicate()
240
209
        rc = pipe.returncode
241
210
 
242
211
        # internal_diff() adds a trailing newline, add one here for consistency
243
 
        out += b'\n'
 
212
        out += '\n'
244
213
        if rc == 2:
245
214
            # 'diff' gives retcode == 2 for all sorts of errors
246
215
            # one of those is 'Binary files differ'.
253
222
            out, err = pipe.communicate()
254
223
 
255
224
            # Write out the new i18n diff response
256
 
            to_file.write(out + b'\n')
 
225
            to_file.write(out+'\n')
257
226
            if pipe.returncode != 2:
258
227
                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,))
 
228
                               'external diff failed with exit code 2'
 
229
                               ' when run with LANG=C and LC_ALL=C,'
 
230
                               ' but not when run natively: %r' % (diffcmd,))
262
231
 
263
 
            first_line = lang_c_out.split(b'\n', 1)[0]
 
232
            first_line = lang_c_out.split('\n', 1)[0]
264
233
            # Starting with diffutils 2.8.4 the word "binary" was dropped.
265
 
            m = re.match(b'^(binary )?files.*differ$', first_line, re.I)
 
234
            m = re.match('^(binary )?files.*differ$', first_line, re.I)
266
235
            if m is None:
267
236
                raise errors.BzrError('external diff failed with exit code 2;'
268
237
                                      ' command: %r' % (diffcmd,))
281
250
                msg = 'exit code %d' % rc
282
251
 
283
252
            raise errors.BzrError('external diff failed with %s; command: %r'
284
 
                                  % (msg, diffcmd))
 
253
                                  % (rc, diffcmd))
 
254
 
285
255
 
286
256
    finally:
287
257
        oldtmpf.close()                 # and delete
288
258
        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):
 
259
        # Clean up. Warn in case the files couldn't be deleted
 
260
        # (in case windows still holds the file open, but not
 
261
        # if the files have already been deleted)
 
262
        try:
 
263
            os.remove(old_abspath)
 
264
        except OSError, e:
 
265
            if e.errno not in (errno.ENOENT,):
 
266
                warning('Failed to delete temporary file: %s %s',
 
267
                        old_abspath, e)
 
268
        try:
 
269
            os.remove(new_abspath)
 
270
        except OSError:
 
271
            if e.errno not in (errno.ENOENT,):
 
272
                warning('Failed to delete temporary file: %s %s',
 
273
                        new_abspath, e)
 
274
 
 
275
 
 
276
def _get_trees_to_diff(path_list, revision_specs, old_url, new_url,
 
277
    apply_view=True):
306
278
    """Get the trees and specific files to diff given a list of paths.
307
279
 
308
280
    This method works out the trees to be diff'ed and the files of
319
291
    :param new_url:
320
292
        The url of the new branch or tree. If None, the tree to use is
321
293
        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
294
    :param apply_view:
326
295
        if True and a view is set, apply the view or check that the paths
327
296
        are within it
328
297
    :returns:
329
 
        a tuple of (old_tree, new_tree, old_branch, new_branch,
330
 
        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.
 
298
        a tuple of (old_tree, new_tree, specific_files, extra_trees) where
 
299
        extra_trees is a sequence of additional trees to search in for
 
300
        file-ids.
334
301
    """
335
302
    # Get the old and new revision specs
336
303
    old_revision_spec = None
359
326
        default_location = path_list[0]
360
327
        other_paths = path_list[1:]
361
328
 
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
329
    # Get the old location
371
330
    specific_files = []
372
331
    if old_url is None:
373
332
        old_url = default_location
374
333
    working_tree, branch, relpath = \
375
 
        controldir.ControlDir.open_containing_tree_or_branch(old_url)
376
 
    lock_tree_or_branch(working_tree, branch)
 
334
        bzrdir.BzrDir.open_containing_tree_or_branch(old_url)
377
335
    if consider_relpath and relpath != '':
378
336
        if working_tree is not None and apply_view:
379
337
            views.check_path_in_view(working_tree, relpath)
380
338
        specific_files.append(relpath)
381
339
    old_tree = _get_tree_to_diff(old_revision_spec, working_tree, branch)
382
 
    old_branch = branch
383
340
 
384
341
    # Get the new location
385
342
    if new_url is None:
386
343
        new_url = default_location
387
344
    if new_url != old_url:
388
345
        working_tree, branch, relpath = \
389
 
            controldir.ControlDir.open_containing_tree_or_branch(new_url)
390
 
        lock_tree_or_branch(working_tree, branch)
 
346
            bzrdir.BzrDir.open_containing_tree_or_branch(new_url)
391
347
        if consider_relpath and relpath != '':
392
348
            if working_tree is not None and apply_view:
393
349
                views.check_path_in_view(working_tree, relpath)
394
350
            specific_files.append(relpath)
395
351
    new_tree = _get_tree_to_diff(new_revision_spec, working_tree, branch,
396
 
                                 basis_is_default=working_tree is None)
397
 
    new_branch = branch
 
352
        basis_is_default=working_tree is None)
398
353
 
399
354
    # Get the specific files (all files is None, no files is [])
400
355
    if make_paths_wt_relative and working_tree is not None:
401
 
        other_paths = working_tree.safe_relpath_files(
402
 
            other_paths,
 
356
        try:
 
357
            from bzrlib.builtins import safe_relpath_files
 
358
            other_paths = safe_relpath_files(working_tree, other_paths,
403
359
            apply_view=apply_view)
 
360
        except errors.FileInWrongBranch:
 
361
            raise errors.BzrCommandError("Files are in different branches")
404
362
    specific_files.extend(other_paths)
405
363
    if len(specific_files) == 0:
406
364
        specific_files = None
407
 
        if (working_tree is not None and working_tree.supports_views() and
408
 
                apply_view):
 
365
        if (working_tree is not None and working_tree.supports_views()
 
366
            and apply_view):
409
367
            view_files = working_tree.views.lookup_view()
410
368
            if view_files:
411
369
                specific_files = view_files
412
370
                view_str = views.view_display_str(view_files)
413
 
                note(gettext("*** Ignoring files outside view. View is %s") % view_str)
 
371
                note("*** ignoring files outside view: %s" % view_str)
414
372
 
415
373
    # Get extra trees that ought to be searched for file-ids
416
374
    extra_trees = None
417
375
    if working_tree is not None and working_tree not in (old_tree, new_tree):
418
376
        extra_trees = (working_tree,)
419
 
    return (old_tree, new_tree, old_branch, new_branch,
420
 
            specific_files, extra_trees)
421
 
 
 
377
    return old_tree, new_tree, specific_files, extra_trees
422
378
 
423
379
def _get_tree_to_diff(spec, tree=None, branch=None, basis_is_default=True):
424
380
    if branch is None and tree is not None:
439
395
                    old_label='a/', new_label='b/',
440
396
                    extra_trees=None,
441
397
                    path_encoding='utf8',
442
 
                    using=None,
443
 
                    format_cls=None,
444
 
                    context=DEFAULT_CONTEXT_AMOUNT):
 
398
                    using=None):
445
399
    """Show in text form the changes from one tree to another.
446
400
 
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)
 
401
    to_file
 
402
        The output stream.
 
403
 
 
404
    specific_files
 
405
        Include only changes to these files - None for all changes.
 
406
 
 
407
    external_diff_options
 
408
        If set, use an external GNU diff and pass these options.
 
409
 
 
410
    extra_trees
 
411
        If set, more Trees to use for looking up file ids
 
412
 
 
413
    path_encoding
 
414
        If set, the path will be encoded as specified, otherwise is supposed
 
415
        to be utf8
456
416
    """
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():
 
417
    old_tree.lock_read()
 
418
    try:
462
419
        if extra_trees is not None:
463
420
            for tree in extra_trees:
464
421
                tree.lock_read()
465
422
        new_tree.lock_read()
466
423
        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)
 
424
            differ = DiffTree.from_trees_options(old_tree, new_tree, to_file,
 
425
                                                 path_encoding,
 
426
                                                 external_diff_options,
 
427
                                                 old_label, new_label, using)
472
428
            return differ.show_diff(specific_files, extra_trees)
473
429
        finally:
474
430
            new_tree.unlock()
475
431
            if extra_trees is not None:
476
432
                for tree in extra_trees:
477
433
                    tree.unlock()
 
434
    finally:
 
435
        old_tree.unlock()
478
436
 
479
437
 
480
438
def _patch_header_date(tree, file_id, path):
481
439
    """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
 
440
    mtime = tree.get_file_mtime(file_id, path)
486
441
    return timestamp.format_patch_date(mtime)
487
442
 
488
443
 
 
444
@deprecated_function(one_three)
 
445
def get_prop_change(meta_modified):
 
446
    if meta_modified:
 
447
        return " (properties changed)"
 
448
    else:
 
449
        return  ""
 
450
 
489
451
def get_executable_change(old_is_x, new_is_x):
490
 
    descr = {True: b"+x", False: b"-x", None: b"??"}
 
452
    descr = { True:"+x", False:"-x", None:"??" }
491
453
    if old_is_x != new_is_x:
492
 
        return [b"%s to %s" % (descr[old_is_x], descr[new_is_x],)]
 
454
        return ["%s to %s" % (descr[old_is_x], descr[new_is_x],)]
493
455
    else:
494
456
        return []
495
457
 
542
504
    Represents kind change as deletion + creation.  Uses the other differs
543
505
    to do this.
544
506
    """
545
 
 
546
507
    def __init__(self, differs):
547
508
        self.differs = differs
548
509
 
565
526
        if None in (old_kind, new_kind):
566
527
            return DiffPath.CANNOT_DIFF
567
528
        result = DiffPath._diff_many(self.differs, file_id, old_path,
568
 
                                     new_path, old_kind, None)
 
529
                                       new_path, old_kind, None)
569
530
        if result is DiffPath.CANNOT_DIFF:
570
531
            return result
571
532
        return DiffPath._diff_many(self.differs, file_id, old_path, new_path,
572
 
                                   None, new_kind)
 
533
                                     None, new_kind)
573
534
 
574
535
 
575
536
class DiffDirectory(DiffPath):
601
562
        if 'symlink' not in (old_kind, new_kind):
602
563
            return self.CANNOT_DIFF
603
564
        if old_kind == 'symlink':
604
 
            old_target = self.old_tree.get_symlink_target(old_path)
 
565
            old_target = self.old_tree.get_symlink_target(file_id)
605
566
        elif old_kind is None:
606
567
            old_target = None
607
568
        else:
608
569
            return self.CANNOT_DIFF
609
570
        if new_kind == 'symlink':
610
 
            new_target = self.new_tree.get_symlink_target(new_path)
 
571
            new_target = self.new_tree.get_symlink_target(file_id)
611
572
        elif new_kind is None:
612
573
            new_target = None
613
574
        else:
616
577
 
617
578
    def diff_symlink(self, old_target, new_target):
618
579
        if old_target is None:
619
 
            self.to_file.write(b'=== target is \'%s\'\n' %
620
 
                               new_target.encode(self.path_encoding, 'replace'))
 
580
            self.to_file.write('=== target is %r\n' % new_target)
621
581
        elif new_target is None:
622
 
            self.to_file.write(b'=== target was \'%s\'\n' %
623
 
                               old_target.encode(self.path_encoding, 'replace'))
 
582
            self.to_file.write('=== target was %r\n' % old_target)
624
583
        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')))
 
584
            self.to_file.write('=== target changed %r => %r\n' %
 
585
                              (old_target, new_target))
628
586
        return self.CHANGED
629
587
 
630
588
 
635
593
    EPOCH_DATE = '1970-01-01 00:00:00 +0000'
636
594
 
637
595
    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):
 
596
                 old_label='', new_label='', text_differ=internal_diff):
640
597
        DiffPath.__init__(self, old_tree, new_tree, to_file, path_encoding)
641
598
        self.text_differ = text_differ
642
599
        self.old_label = old_label
643
600
        self.new_label = new_label
644
601
        self.path_encoding = path_encoding
645
 
        self.context_lines = context_lines
646
602
 
647
603
    def diff(self, file_id, old_path, new_path, old_kind, new_kind):
648
604
        """Compare two files in unified diff format
670
626
            to_file_id = None
671
627
        else:
672
628
            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)
 
629
        from_label = '%s%s\t%s' % (self.old_label, old_path, old_date)
 
630
        to_label = '%s%s\t%s' % (self.new_label, new_path, new_date)
 
631
        return self.diff_text(from_file_id, to_file_id, from_label, to_label)
679
632
 
680
 
    def diff_text(self, from_path, to_path, from_label, to_label,
681
 
                  from_file_id=None, to_file_id=None):
 
633
    def diff_text(self, from_file_id, to_file_id, from_label, to_label):
682
634
        """Diff the content of given files in two trees
683
635
 
684
 
        :param from_path: The path in the from tree. If None,
 
636
        :param from_file_id: The id of the file in the from tree.  If None,
685
637
            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,
 
638
        :param to_file_id: The id of the file in the to tree.  This may refer
 
639
            to a different file from from_file_id.  If None,
688
640
            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.
693
641
        """
694
 
        def _get_text(tree, file_id, path):
695
 
            if file_id is None:
 
642
        def _get_text(tree, file_id):
 
643
            if file_id is not None:
 
644
                return tree.get_file(file_id).readlines()
 
645
            else:
696
646
                return []
697
 
            return tree.get_file_lines(path)
698
647
        try:
699
 
            from_text = _get_text(self.old_tree, from_file_id, from_path)
700
 
            to_text = _get_text(self.new_tree, to_file_id, to_path)
 
648
            from_text = _get_text(self.old_tree, from_file_id)
 
649
            to_text = _get_text(self.new_tree, to_file_id)
701
650
            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)
 
651
                             self.to_file)
704
652
        except errors.BinaryFile:
705
653
            self.to_file.write(
706
 
                ("Binary files %s and %s differ\n" %
707
 
                 (from_label, to_label)).encode(self.path_encoding, 'replace'))
 
654
                  ("Binary files %s and %s differ\n" %
 
655
                  (from_label, to_label)).encode(self.path_encoding))
708
656
        return self.CHANGED
709
657
 
710
658
 
714
662
                 path_encoding='utf-8'):
715
663
        DiffPath.__init__(self, old_tree, new_tree, to_file, path_encoding)
716
664
        self.command_template = command_template
717
 
        self._root = osutils.mkdtemp(prefix='brz-diff-')
 
665
        self._root = osutils.mkdtemp(prefix='bzr-diff-')
718
666
 
719
667
    @classmethod
720
668
    def from_string(klass, command_string, old_tree, new_tree, to_file,
721
669
                    path_encoding='utf-8'):
722
 
        command_template = cmdline.split(command_string)
723
 
        if '@' not in command_string:
724
 
            command_template.extend(['@old_path', '@new_path'])
 
670
        command_template = commands.shlex_split_unicode(command_string)
 
671
        command_template.extend(['%(old_path)s', '%(new_path)s'])
725
672
        return klass(command_template, old_tree, new_tree, to_file,
726
673
                     path_encoding)
727
674
 
728
675
    @classmethod
729
 
    def make_from_diff_tree(klass, command_string, external_diff_options=None):
 
676
    def make_from_diff_tree(klass, command_string):
730
677
        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,
 
678
            return klass.from_string(command_string, diff_tree.old_tree,
735
679
                                     diff_tree.new_tree, diff_tree.to_file)
736
680
        return from_diff_tree
737
681
 
738
682
    def _get_command(self, old_path, new_path):
739
683
        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
 
684
        return [t % my_map for t in self.command_template]
752
685
 
753
686
    def _execute(self, old_path, new_path):
754
687
        command = self._get_command(old_path, new_path)
755
688
        try:
756
689
            proc = subprocess.Popen(command, stdout=subprocess.PIPE,
757
690
                                    cwd=self._root)
758
 
        except OSError as e:
 
691
        except OSError, e:
759
692
            if e.errno == errno.ENOENT:
760
693
                raise errors.ExecutableMissing(command[0])
761
694
            else:
762
695
                raise
763
696
        self.to_file.write(proc.stdout.read())
764
 
        proc.stdout.close()
765
697
        return proc.wait()
766
698
 
767
699
    def _try_symlink_root(self, tree, prefix):
768
 
        if (getattr(tree, 'abspath', None) is None or
769
 
                not osutils.host_os_dereferences_symlinks()):
 
700
        if (getattr(tree, 'abspath', None) is None
 
701
            or not osutils.host_os_dereferences_symlinks()):
770
702
            return False
771
703
        try:
772
704
            os.symlink(tree.abspath(''), osutils.pathjoin(self._root, prefix))
773
 
        except OSError as e:
 
705
        except OSError, e:
774
706
            if e.errno != errno.EEXIST:
775
707
                raise
776
708
        return True
777
709
 
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):
808
 
        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)
814
 
        if not force_temp and self._try_symlink_root(tree, prefix):
 
710
    def _write_file(self, file_id, tree, prefix, relpath):
 
711
        full_path = osutils.pathjoin(self._root, prefix, relpath)
 
712
        if self._try_symlink_root(tree, prefix):
815
713
            return full_path
816
714
        parent_dir = osutils.dirname(full_path)
817
715
        try:
818
716
            os.makedirs(parent_dir)
819
 
        except OSError as e:
 
717
        except OSError, e:
820
718
            if e.errno != errno.EEXIST:
821
719
                raise
822
 
        source = tree.get_file(relpath)
 
720
        source = tree.get_file(file_id, relpath)
823
721
        try:
824
 
            with open(full_path, 'wb') as target:
 
722
            target = open(full_path, 'wb')
 
723
            try:
825
724
                osutils.pumpfile(source, target)
 
725
            finally:
 
726
                target.close()
826
727
        finally:
827
728
            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
 
        if not allow_write:
835
 
            osutils.make_readonly(full_path)
 
729
        osutils.make_readonly(full_path)
 
730
        mtime = tree.get_file_mtime(file_id)
 
731
        os.utime(full_path, (mtime, mtime))
836
732
        return full_path
837
733
 
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,
844
 
                                         allow_write=allow_write_new)
 
734
    def _prepare_files(self, file_id, old_path, new_path):
 
735
        old_disk_path = self._write_file(file_id, self.old_tree, 'old',
 
736
                                         old_path)
 
737
        new_disk_path = self._write_file(file_id, self.new_tree, 'new',
 
738
                                         new_path)
845
739
        return old_disk_path, new_disk_path
846
740
 
847
741
    def finish(self):
848
 
        try:
849
 
            osutils.rmtree(self._root)
850
 
        except OSError as e:
851
 
            if e.errno != errno.ENOENT:
852
 
                mutter("The temporary directory \"%s\" was not "
853
 
                       "cleanly removed: %s." % (self._root, e))
 
742
        osutils.rmtree(self._root)
854
743
 
855
744
    def diff(self, file_id, old_path, new_path, old_kind, new_kind):
856
745
        if (old_kind, new_kind) != ('file', 'file'):
857
746
            return DiffPath.CANNOT_DIFF
858
 
        (old_disk_path, new_disk_path) = self._prepare_files(
859
 
            old_path, new_path, file_id=file_id)
860
 
        self._execute(old_disk_path, new_disk_path)
861
 
 
862
 
    def edit_file(self, old_path, new_path, file_id=None):
863
 
        """Use this tool to edit a file.
864
 
 
865
 
        A temporary copy will be edited, and the new contents will be
866
 
        returned.
867
 
 
868
 
        :param file_id: The id of the file to edit.
869
 
        :return: The new contents of the file.
870
 
        """
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)
875
 
        subprocess.call(command, cwd=self._root)
876
 
        with open(new_abs_path, 'rb') as new_file:
877
 
            return new_file.read()
 
747
        self._prepare_files(file_id, old_path, new_path)
 
748
        self._execute(osutils.pathjoin('old', old_path),
 
749
                      osutils.pathjoin('new', new_path))
878
750
 
879
751
 
880
752
class DiffTree(object):
908
780
            DiffPaths"""
909
781
        if diff_text is None:
910
782
            diff_text = DiffText(old_tree, new_tree, to_file, path_encoding,
911
 
                                 '', '', internal_diff)
 
783
                                 '', '',  internal_diff)
912
784
        self.old_tree = old_tree
913
785
        self.new_tree = new_tree
914
786
        self.to_file = to_file
922
794
    @classmethod
923
795
    def from_trees_options(klass, old_tree, new_tree, to_file,
924
796
                           path_encoding, external_diff_options, old_label,
925
 
                           new_label, using, context_lines):
 
797
                           new_label, using):
926
798
        """Factory for producing a DiffTree.
927
799
 
928
800
        Designed to accept options used by show_diff_trees.
929
 
 
930
801
        :param old_tree: The tree to show as old in the comparison
931
802
        :param new_tree: The tree to show as new in the comparison
932
803
        :param to_file: File to write comparisons to
938
809
        :param using: Commandline to use to invoke an external diff tool
939
810
        """
940
811
        if using is not None:
941
 
            extra_factories = [DiffFromTool.make_from_diff_tree(
942
 
                using, external_diff_options)]
 
812
            extra_factories = [DiffFromTool.make_from_diff_tree(using)]
943
813
        else:
944
814
            extra_factories = []
945
815
        if external_diff_options:
946
816
            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
 
                """
 
817
            def diff_file(olab, olines, nlab, nlines, to_file):
952
818
                external_diff(olab, olines, nlab, nlines, to_file, opts)
953
819
        else:
954
820
            diff_file = internal_diff
955
821
        diff_text = DiffText(old_tree, new_tree, to_file, path_encoding,
956
 
                             old_label, new_label, diff_file, context_lines=context_lines)
 
822
                             old_label, new_label, diff_file)
957
823
        return klass(old_tree, new_tree, to_file, path_encoding, diff_text,
958
824
                     extra_factories)
959
825
 
960
826
    def show_diff(self, specific_files, extra_trees=None):
961
827
        """Write tree diff to self.to_file
962
828
 
963
 
        :param specific_files: the specific files to compare (recursive)
 
829
        :param sepecific_files: the specific files to compare (recursive)
964
830
        :param extra_trees: extra trees to use for mapping paths to file_ids
965
831
        """
966
832
        try:
973
839
        # TODO: Generation of pseudo-diffs for added/deleted files could
974
840
        # be usefully made into a much faster special case.
975
841
        iterator = self.new_tree.iter_changes(self.old_tree,
976
 
                                              specific_files=specific_files,
977
 
                                              extra_trees=extra_trees,
978
 
                                              require_versioned=True)
 
842
                                               specific_files=specific_files,
 
843
                                               extra_trees=extra_trees,
 
844
                                               require_versioned=True)
979
845
        has_changes = 0
980
 
 
981
846
        def changes_key(change):
982
847
            old_path, new_path = change[1]
983
848
            path = new_path
984
849
            if path is None:
985
850
                path = old_path
986
851
            return path
987
 
 
988
852
        def get_encoded_path(path):
989
853
            if path is not None:
990
854
                return path.encode(self.path_encoding, "replace")
1002
866
            renamed = (parent[0], name[0]) != (parent[1], name[1])
1003
867
 
1004
868
            properties_changed = []
1005
 
            properties_changed.extend(
1006
 
                get_executable_change(executable[0], executable[1]))
 
869
            properties_changed.extend(get_executable_change(executable[0], executable[1]))
1007
870
 
1008
871
            if properties_changed:
1009
 
                prop_str = b" (properties changed: %s)" % (
1010
 
                    b", ".join(properties_changed),)
 
872
                prop_str = " (properties changed: %s)" % (", ".join(properties_changed),)
1011
873
            else:
1012
 
                prop_str = b""
 
874
                prop_str = ""
1013
875
 
1014
876
            if (old_present, new_present) == (True, False):
1015
 
                self.to_file.write(b"=== removed %s '%s'\n" %
1016
 
                                   (kind[0].encode('ascii'), oldpath_encoded))
 
877
                self.to_file.write("=== removed %s '%s'\n" %
 
878
                                   (kind[0], oldpath_encoded))
1017
879
                newpath = oldpath
1018
880
            elif (old_present, new_present) == (False, True):
1019
 
                self.to_file.write(b"=== added %s '%s'\n" %
1020
 
                                   (kind[1].encode('ascii'), newpath_encoded))
 
881
                self.to_file.write("=== added %s '%s'\n" %
 
882
                                   (kind[1], newpath_encoded))
1021
883
                oldpath = newpath
1022
884
            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))
 
885
                self.to_file.write("=== renamed %s '%s' => '%s'%s\n" %
 
886
                    (kind[0], oldpath_encoded, newpath_encoded, prop_str))
1025
887
            else:
1026
888
                # if it was produced by iter_changes, it must be
1027
889
                # 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))
 
890
                self.to_file.write("=== modified %s '%s'%s\n" % (kind[0],
 
891
                                   newpath_encoded, prop_str))
1030
892
            if changed_content:
1031
 
                self._diff(oldpath, newpath, kind[0], kind[1], file_id=file_id)
 
893
                self.diff(file_id, oldpath, newpath)
1032
894
                has_changes = 1
1033
895
            if renamed:
1034
896
                has_changes = 1
1041
903
        :param old_path: The path of the file in the old tree
1042
904
        :param new_path: The path of the file in the new tree
1043
905
        """
1044
 
        if old_path is None:
 
906
        try:
 
907
            old_kind = self.old_tree.kind(file_id)
 
908
        except (errors.NoSuchId, errors.NoSuchFile):
1045
909
            old_kind = None
1046
 
        else:
1047
 
            old_kind = self.old_tree.kind(old_path)
1048
 
        if new_path is None:
 
910
        try:
 
911
            new_kind = self.new_tree.kind(file_id)
 
912
        except (errors.NoSuchId, errors.NoSuchFile):
1049
913
            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
914
 
1054
 
    def _diff(self, old_path, new_path, old_kind, new_kind, file_id):
1055
915
        result = DiffPath._diff_many(self.differs, file_id, old_path,
1056
 
                                     new_path, old_kind, new_kind)
 
916
                                       new_path, old_kind, new_kind)
1057
917
        if result is DiffPath.CANNOT_DIFF:
1058
918
            error_path = new_path
1059
919
            if error_path is None:
1060
920
                error_path = old_path
1061
921
            raise errors.NoDiffFound(error_path)
1062
 
 
1063
 
 
1064
 
format_registry = Registry()
1065
 
format_registry.register('default', DiffTree)