/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: Robert Collins
  • Date: 2010-05-06 11:08:10 UTC
  • mto: This revision was merged to the branch mainline in revision 5223.
  • Revision ID: robertc@robertcollins.net-20100506110810-h3j07fh5gmw54s25
Cleaner matcher matching revised unlocking protocol.

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) 2005-2010 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 string
22
21
import sys
23
22
 
24
 
from .lazy_import import lazy_import
 
23
from bzrlib.lazy_import import lazy_import
25
24
lazy_import(globals(), """
26
25
import errno
27
 
import patiencediff
28
26
import subprocess
29
27
import tempfile
 
28
import time
30
29
 
31
 
from breezy import (
 
30
from bzrlib import (
 
31
    branch as _mod_branch,
 
32
    bzrdir,
 
33
    cmdline,
32
34
    cleanup,
33
 
    controldir,
34
35
    errors,
35
36
    osutils,
 
37
    patiencediff,
36
38
    textfile,
37
39
    timestamp,
38
40
    views,
39
41
    )
40
42
 
41
 
from breezy.workingtree import WorkingTree
42
 
from breezy.i18n import gettext
 
43
from bzrlib.workingtree import WorkingTree
43
44
""")
44
45
 
45
 
from .registry import (
 
46
from bzrlib.registry import (
46
47
    Registry,
47
48
    )
48
 
from .sixish import text_type
49
 
from .trace import mutter, note, warning
50
 
from .tree import FileTimestampUnavailable
51
 
 
52
 
 
53
 
DEFAULT_CONTEXT_AMOUNT = 3
 
49
from bzrlib.symbol_versioning import (
 
50
    deprecated_function,
 
51
    deprecated_in,
 
52
    )
 
53
from bzrlib.trace import mutter, note, warning
 
54
 
 
55
 
 
56
class AtTemplate(string.Template):
 
57
    """Templating class that uses @ instead of $."""
 
58
 
 
59
    delimiter = '@'
54
60
 
55
61
 
56
62
# TODO: Rather than building a changeset object, we should probably
67
73
        self.opcodes = None
68
74
 
69
75
 
70
 
def internal_diff(old_label, oldlines, new_label, newlines, to_file,
 
76
def internal_diff(old_filename, oldlines, new_filename, newlines, to_file,
71
77
                  allow_binary=False, sequence_matcher=None,
72
 
                  path_encoding='utf8', context_lines=DEFAULT_CONTEXT_AMOUNT):
 
78
                  path_encoding='utf8'):
73
79
    # FIXME: difflib is wrong if there is no trailing newline.
74
80
    # The syntax used by patch seems to be "\ No newline at
75
81
    # end of file" following the last diff line from that
80
86
    # In the meantime we at least make sure the patch isn't
81
87
    # mangled.
82
88
 
 
89
 
 
90
    # Special workaround for Python2.3, where difflib fails if
 
91
    # both sequences are empty.
 
92
    if not oldlines and not newlines:
 
93
        return
 
94
 
83
95
    if allow_binary is False:
84
96
        textfile.check_text_lines(oldlines)
85
97
        textfile.check_text_lines(newlines)
86
98
 
87
99
    if sequence_matcher is None:
88
100
        sequence_matcher = patiencediff.PatienceSequenceMatcher
89
 
    ud = unified_diff_bytes(
90
 
        oldlines, newlines,
91
 
        fromfile=old_label.encode(path_encoding, 'replace'),
92
 
        tofile=new_label.encode(path_encoding, 'replace'),
93
 
        n=context_lines, sequencematcher=sequence_matcher)
 
101
    ud = patiencediff.unified_diff(oldlines, newlines,
 
102
                      fromfile=old_filename.encode(path_encoding),
 
103
                      tofile=new_filename.encode(path_encoding),
 
104
                      sequencematcher=sequence_matcher)
94
105
 
95
106
    ud = list(ud)
96
 
    if len(ud) == 0:  # Identical contents, nothing to do
 
107
    if len(ud) == 0: # Identical contents, nothing to do
97
108
        return
98
109
    # work-around for difflib being too smart for its own good
99
110
    # if /dev/null is "1,0", patch won't recognize it as /dev/null
100
111
    if not oldlines:
101
 
        ud[2] = ud[2].replace(b'-1,0', b'-0,0')
 
112
        ud[2] = ud[2].replace('-1,0', '-0,0')
102
113
    elif not newlines:
103
 
        ud[2] = ud[2].replace(b'+1,0', b'+0,0')
 
114
        ud[2] = ud[2].replace('+1,0', '+0,0')
104
115
 
105
116
    for line in ud:
106
117
        to_file.write(line)
107
 
        if not line.endswith(b'\n'):
108
 
            to_file.write(b"\n\\ No newline at end of file\n")
109
 
    to_file.write(b'\n')
110
 
 
111
 
 
112
 
def unified_diff_bytes(a, b, fromfile=b'', tofile=b'', fromfiledate=b'',
113
 
                       tofiledate=b'', n=3, lineterm=b'\n', sequencematcher=None):
114
 
    r"""
115
 
    Compare two sequences of lines; generate the delta as a unified diff.
116
 
 
117
 
    Unified diffs are a compact way of showing line changes and a few
118
 
    lines of context.  The number of context lines is set by 'n' which
119
 
    defaults to three.
120
 
 
121
 
    By default, the diff control lines (those with ---, +++, or @@) are
122
 
    created with a trailing newline.  This is helpful so that inputs
123
 
    created from file.readlines() result in diffs that are suitable for
124
 
    file.writelines() since both the inputs and outputs have trailing
125
 
    newlines.
126
 
 
127
 
    For inputs that do not have trailing newlines, set the lineterm
128
 
    argument to "" so that the output will be uniformly newline free.
129
 
 
130
 
    The unidiff format normally has a header for filenames and modification
131
 
    times.  Any or all of these may be specified using strings for
132
 
    'fromfile', 'tofile', 'fromfiledate', and 'tofiledate'.  The modification
133
 
    times are normally expressed in the format returned by time.ctime().
134
 
 
135
 
    Example:
136
 
 
137
 
    >>> for line in bytes_unified_diff(b'one two three four'.split(),
138
 
    ...             b'zero one tree four'.split(), b'Original', b'Current',
139
 
    ...             b'Sat Jan 26 23:30:50 1991', b'Fri Jun 06 10:20:52 2003',
140
 
    ...             lineterm=b''):
141
 
    ...     print line
142
 
    --- Original Sat Jan 26 23:30:50 1991
143
 
    +++ Current Fri Jun 06 10:20:52 2003
144
 
    @@ -1,4 +1,4 @@
145
 
    +zero
146
 
     one
147
 
    -two
148
 
    -three
149
 
    +tree
150
 
     four
151
 
    """
152
 
    if sequencematcher is None:
153
 
        sequencematcher = difflib.SequenceMatcher
154
 
 
155
 
    if fromfiledate:
156
 
        fromfiledate = b'\t' + bytes(fromfiledate)
157
 
    if tofiledate:
158
 
        tofiledate = b'\t' + bytes(tofiledate)
159
 
 
160
 
    started = False
161
 
    for group in sequencematcher(None, a, b).get_grouped_opcodes(n):
162
 
        if not started:
163
 
            yield b'--- %s%s%s' % (fromfile, fromfiledate, lineterm)
164
 
            yield b'+++ %s%s%s' % (tofile, tofiledate, lineterm)
165
 
            started = True
166
 
        i1, i2, j1, j2 = group[0][1], group[-1][2], group[0][3], group[-1][4]
167
 
        yield b"@@ -%d,%d +%d,%d @@%s" % (i1 + 1, i2 - i1, j1 + 1, j2 - j1, lineterm)
168
 
        for tag, i1, i2, j1, j2 in group:
169
 
            if tag == 'equal':
170
 
                for line in a[i1:i2]:
171
 
                    yield b' ' + line
172
 
                continue
173
 
            if tag == 'replace' or tag == 'delete':
174
 
                for line in a[i1:i2]:
175
 
                    yield b'-' + line
176
 
            if tag == 'replace' or tag == 'insert':
177
 
                for line in b[j1:j2]:
178
 
                    yield b'+' + line
 
118
        if not line.endswith('\n'):
 
119
            to_file.write("\n\\ No newline at end of file\n")
 
120
    to_file.write('\n')
179
121
 
180
122
 
181
123
def _spawn_external_diff(diffcmd, capture_errors=True):
182
 
    """Spawn the external diff process, and return the child handle.
 
124
    """Spawn the externall diff process, and return the child handle.
183
125
 
184
126
    :param diffcmd: The command list to spawn
185
127
    :param capture_errors: Capture stderr as well as setting LANG=C
207
149
                                stdout=subprocess.PIPE,
208
150
                                stderr=stderr,
209
151
                                env=env)
210
 
    except OSError as e:
 
152
    except OSError, e:
211
153
        if e.errno == errno.ENOENT:
212
154
            raise errors.NoDiff(str(e))
213
155
        raise
215
157
    return pipe
216
158
 
217
159
 
218
 
# diff style options as of GNU diff v3.2
219
 
style_option_list = ['-c', '-C', '--context',
220
 
                     '-e', '--ed',
221
 
                     '-f', '--forward-ed',
222
 
                     '-q', '--brief',
223
 
                     '--normal',
224
 
                     '-n', '--rcs',
225
 
                     '-u', '-U', '--unified',
226
 
                     '-y', '--side-by-side',
227
 
                     '-D', '--ifdef']
228
 
 
229
 
 
230
 
def default_style_unified(diff_opts):
231
 
    """Default to unified diff style if alternative not specified in diff_opts.
232
 
 
233
 
        diff only allows one style to be specified; they don't override.
234
 
        Note that some of these take optargs, and the optargs can be
235
 
        directly appended to the options.
236
 
        This is only an approximate parser; it doesn't properly understand
237
 
        the grammar.
238
 
 
239
 
    :param diff_opts: List of options for external (GNU) diff.
240
 
    :return: List of options with default style=='unified'.
241
 
    """
242
 
    for s in style_option_list:
243
 
        for j in diff_opts:
244
 
            if j.startswith(s):
245
 
                break
246
 
        else:
247
 
            continue
248
 
        break
249
 
    else:
250
 
        diff_opts.append('-u')
251
 
    return diff_opts
252
 
 
253
 
 
254
 
def external_diff(old_label, oldlines, new_label, newlines, to_file,
 
160
def external_diff(old_filename, oldlines, new_filename, newlines, to_file,
255
161
                  diff_opts):
256
162
    """Display a diff by calling out to the external diff program."""
257
163
    # make sure our own output is properly ordered before the diff
258
164
    to_file.flush()
259
165
 
260
 
    oldtmp_fd, old_abspath = tempfile.mkstemp(prefix='brz-diff-old-')
261
 
    newtmp_fd, new_abspath = tempfile.mkstemp(prefix='brz-diff-new-')
 
166
    oldtmp_fd, old_abspath = tempfile.mkstemp(prefix='bzr-diff-old-')
 
167
    newtmp_fd, new_abspath = tempfile.mkstemp(prefix='bzr-diff-new-')
262
168
    oldtmpf = os.fdopen(oldtmp_fd, 'wb')
263
169
    newtmpf = os.fdopen(newtmp_fd, 'wb')
264
170
 
281
187
        if sys.platform == 'win32':
282
188
            # Popen doesn't do the proper encoding for external commands
283
189
            # Since we are dealing with an ANSI api, use mbcs encoding
284
 
            old_label = old_label.encode('mbcs')
285
 
            new_label = new_label.encode('mbcs')
 
190
            old_filename = old_filename.encode('mbcs')
 
191
            new_filename = new_filename.encode('mbcs')
286
192
        diffcmd = ['diff',
287
 
                   '--label', old_label,
 
193
                   '--label', old_filename,
288
194
                   old_abspath,
289
 
                   '--label', new_label,
 
195
                   '--label', new_filename,
290
196
                   new_abspath,
291
197
                   '--binary',
292
 
                   ]
 
198
                  ]
293
199
 
294
 
        diff_opts = default_style_unified(diff_opts)
 
200
        # diff only allows one style to be specified; they don't override.
 
201
        # note that some of these take optargs, and the optargs can be
 
202
        # directly appended to the options.
 
203
        # this is only an approximate parser; it doesn't properly understand
 
204
        # the grammar.
 
205
        for s in ['-c', '-u', '-C', '-U',
 
206
                  '-e', '--ed',
 
207
                  '-q', '--brief',
 
208
                  '--normal',
 
209
                  '-n', '--rcs',
 
210
                  '-y', '--side-by-side',
 
211
                  '-D', '--ifdef']:
 
212
            for j in diff_opts:
 
213
                if j.startswith(s):
 
214
                    break
 
215
            else:
 
216
                continue
 
217
            break
 
218
        else:
 
219
            diffcmd.append('-u')
295
220
 
296
221
        if diff_opts:
297
222
            diffcmd.extend(diff_opts)
298
223
 
299
224
        pipe = _spawn_external_diff(diffcmd, capture_errors=True)
300
 
        out, err = pipe.communicate()
 
225
        out,err = pipe.communicate()
301
226
        rc = pipe.returncode
302
227
 
303
228
        # internal_diff() adds a trailing newline, add one here for consistency
304
 
        out += b'\n'
 
229
        out += '\n'
305
230
        if rc == 2:
306
231
            # 'diff' gives retcode == 2 for all sorts of errors
307
232
            # one of those is 'Binary files differ'.
314
239
            out, err = pipe.communicate()
315
240
 
316
241
            # Write out the new i18n diff response
317
 
            to_file.write(out + b'\n')
 
242
            to_file.write(out+'\n')
318
243
            if pipe.returncode != 2:
319
244
                raise errors.BzrError(
320
 
                    'external diff failed with exit code 2'
321
 
                    ' when run with LANG=C and LC_ALL=C,'
322
 
                    ' but not when run natively: %r' % (diffcmd,))
 
245
                               'external diff failed with exit code 2'
 
246
                               ' when run with LANG=C and LC_ALL=C,'
 
247
                               ' but not when run natively: %r' % (diffcmd,))
323
248
 
324
 
            first_line = lang_c_out.split(b'\n', 1)[0]
 
249
            first_line = lang_c_out.split('\n', 1)[0]
325
250
            # Starting with diffutils 2.8.4 the word "binary" was dropped.
326
 
            m = re.match(b'^(binary )?files.*differ$', first_line, re.I)
 
251
            m = re.match('^(binary )?files.*differ$', first_line, re.I)
327
252
            if m is None:
328
253
                raise errors.BzrError('external diff failed with exit code 2;'
329
254
                                      ' command: %r' % (diffcmd,))
342
267
                msg = 'exit code %d' % rc
343
268
 
344
269
            raise errors.BzrError('external diff failed with %s; command: %r'
345
 
                                  % (msg, diffcmd))
 
270
                                  % (rc, diffcmd))
 
271
 
346
272
 
347
273
    finally:
348
274
        oldtmpf.close()                 # and delete
349
275
        newtmpf.close()
350
 
 
351
 
        def cleanup(path):
352
 
            # Warn in case the file couldn't be deleted (in case windows still
353
 
            # holds the file open, but not if the files have already been
354
 
            # deleted)
355
 
            try:
356
 
                os.remove(path)
357
 
            except OSError as e:
358
 
                if e.errno not in (errno.ENOENT,):
359
 
                    warning('Failed to delete temporary file: %s %s', path, e)
360
 
 
361
 
        cleanup(old_abspath)
362
 
        cleanup(new_abspath)
363
 
 
 
276
        # Clean up. Warn in case the files couldn't be deleted
 
277
        # (in case windows still holds the file open, but not
 
278
        # if the files have already been deleted)
 
279
        try:
 
280
            os.remove(old_abspath)
 
281
        except OSError, e:
 
282
            if e.errno not in (errno.ENOENT,):
 
283
                warning('Failed to delete temporary file: %s %s',
 
284
                        old_abspath, e)
 
285
        try:
 
286
            os.remove(new_abspath)
 
287
        except OSError:
 
288
            if e.errno not in (errno.ENOENT,):
 
289
                warning('Failed to delete temporary file: %s %s',
 
290
                        new_abspath, e)
 
291
 
 
292
 
 
293
@deprecated_function(deprecated_in((2, 2, 0)))
 
294
def get_trees_and_branches_to_diff(path_list, revision_specs, old_url, new_url,
 
295
                                   apply_view=True):
 
296
    """Get the trees and specific files to diff given a list of paths.
 
297
 
 
298
    This method works out the trees to be diff'ed and the files of
 
299
    interest within those trees.
 
300
 
 
301
    :param path_list:
 
302
        the list of arguments passed to the diff command
 
303
    :param revision_specs:
 
304
        Zero, one or two RevisionSpecs from the diff command line,
 
305
        saying what revisions to compare.
 
306
    :param old_url:
 
307
        The url of the old branch or tree. If None, the tree to use is
 
308
        taken from the first path, if any, or the current working tree.
 
309
    :param new_url:
 
310
        The url of the new branch or tree. If None, the tree to use is
 
311
        taken from the first path, if any, or the current working tree.
 
312
    :param apply_view:
 
313
        if True and a view is set, apply the view or check that the paths
 
314
        are within it
 
315
    :returns:
 
316
        a tuple of (old_tree, new_tree, old_branch, new_branch,
 
317
        specific_files, extra_trees) where extra_trees is a sequence of
 
318
        additional trees to search in for file-ids.  The trees and branches
 
319
        are not locked.
 
320
    """
 
321
    op = cleanup.OperationWithCleanups(get_trees_and_branches_to_diff_locked)
 
322
    return op.run_simple(path_list, revision_specs, old_url, new_url,
 
323
            op.add_cleanup, apply_view=apply_view)
 
324
    
364
325
 
365
326
def get_trees_and_branches_to_diff_locked(
366
 
        path_list, revision_specs, old_url, new_url, exit_stack, apply_view=True):
 
327
    path_list, revision_specs, old_url, new_url, add_cleanup, apply_view=True):
367
328
    """Get the trees and specific files to diff given a list of paths.
368
329
 
369
330
    This method works out the trees to be diff'ed and the files of
380
341
    :param new_url:
381
342
        The url of the new branch or tree. If None, the tree to use is
382
343
        taken from the first path, if any, or the current working tree.
383
 
    :param exit_stack:
384
 
        an ExitStack object. get_trees_and_branches_to_diff
 
344
    :param add_cleanup:
 
345
        a callable like Command.add_cleanup.  get_trees_and_branches_to_diff
385
346
        will register cleanups that must be run to unlock the trees, etc.
386
347
    :param apply_view:
387
348
        if True and a view is set, apply the view or check that the paths
390
351
        a tuple of (old_tree, new_tree, old_branch, new_branch,
391
352
        specific_files, extra_trees) where extra_trees is a sequence of
392
353
        additional trees to search in for file-ids.  The trees and branches
393
 
        will be read-locked until the cleanups registered via the exit_stack
 
354
        will be read-locked until the cleanups registered via the add_cleanup
394
355
        param are run.
395
356
    """
396
357
    # Get the old and new revision specs
422
383
 
423
384
    def lock_tree_or_branch(wt, br):
424
385
        if wt is not None:
425
 
            exit_stack.enter_context(wt.lock_read())
 
386
            wt.lock_read()
 
387
            add_cleanup(wt.unlock)
426
388
        elif br is not None:
427
 
            exit_stack.enter_context(br.lock_read())
 
389
            br.lock_read()
 
390
            add_cleanup(br.unlock)
428
391
 
429
392
    # Get the old location
430
393
    specific_files = []
431
394
    if old_url is None:
432
395
        old_url = default_location
433
396
    working_tree, branch, relpath = \
434
 
        controldir.ControlDir.open_containing_tree_or_branch(old_url)
 
397
        bzrdir.BzrDir.open_containing_tree_or_branch(old_url)
435
398
    lock_tree_or_branch(working_tree, branch)
436
399
    if consider_relpath and relpath != '':
437
400
        if working_tree is not None and apply_view:
445
408
        new_url = default_location
446
409
    if new_url != old_url:
447
410
        working_tree, branch, relpath = \
448
 
            controldir.ControlDir.open_containing_tree_or_branch(new_url)
 
411
            bzrdir.BzrDir.open_containing_tree_or_branch(new_url)
449
412
        lock_tree_or_branch(working_tree, branch)
450
413
        if consider_relpath and relpath != '':
451
414
            if working_tree is not None and apply_view:
452
415
                views.check_path_in_view(working_tree, relpath)
453
416
            specific_files.append(relpath)
454
417
    new_tree = _get_tree_to_diff(new_revision_spec, working_tree, branch,
455
 
                                 basis_is_default=working_tree is None)
 
418
        basis_is_default=working_tree is None)
456
419
    new_branch = branch
457
420
 
458
421
    # Get the specific files (all files is None, no files is [])
459
422
    if make_paths_wt_relative and working_tree is not None:
460
 
        other_paths = working_tree.safe_relpath_files(
461
 
            other_paths,
 
423
        try:
 
424
            from bzrlib.builtins import safe_relpath_files
 
425
            other_paths = safe_relpath_files(working_tree, other_paths,
462
426
            apply_view=apply_view)
 
427
        except errors.FileInWrongBranch:
 
428
            raise errors.BzrCommandError("Files are in different branches")
463
429
    specific_files.extend(other_paths)
464
430
    if len(specific_files) == 0:
465
431
        specific_files = None
466
 
        if (working_tree is not None and working_tree.supports_views() and
467
 
                apply_view):
 
432
        if (working_tree is not None and working_tree.supports_views()
 
433
            and apply_view):
468
434
            view_files = working_tree.views.lookup_view()
469
435
            if view_files:
470
436
                specific_files = view_files
471
437
                view_str = views.view_display_str(view_files)
472
 
                note(gettext("*** Ignoring files outside view. View is %s") % view_str)
 
438
                note("*** Ignoring files outside view. View is %s" % view_str)
473
439
 
474
440
    # Get extra trees that ought to be searched for file-ids
475
441
    extra_trees = None
476
442
    if working_tree is not None and working_tree not in (old_tree, new_tree):
477
443
        extra_trees = (working_tree,)
478
 
    return (old_tree, new_tree, old_branch, new_branch,
479
 
            specific_files, extra_trees)
 
444
    return old_tree, new_tree, old_branch, new_branch, specific_files, extra_trees
480
445
 
481
446
 
482
447
def _get_tree_to_diff(spec, tree=None, branch=None, basis_is_default=True):
499
464
                    extra_trees=None,
500
465
                    path_encoding='utf8',
501
466
                    using=None,
502
 
                    format_cls=None,
503
 
                    context=DEFAULT_CONTEXT_AMOUNT):
 
467
                    format_cls=None):
504
468
    """Show in text form the changes from one tree to another.
505
469
 
506
470
    :param to_file: The output stream.
507
 
    :param specific_files: Include only changes to these files - None for all
 
471
    :param specific_files:Include only changes to these files - None for all
508
472
        changes.
509
 
    :param external_diff_options: If set, use an external GNU diff and pass
 
473
    :param external_diff_options: If set, use an external GNU diff and pass 
510
474
        these options.
511
475
    :param extra_trees: If set, more Trees to use for looking up file ids
512
 
    :param path_encoding: If set, the path will be encoded as specified,
 
476
    :param path_encoding: If set, the path will be encoded as specified, 
513
477
        otherwise is supposed to be utf8
514
478
    :param format_cls: Formatter class (DiffTree subclass)
515
479
    """
516
 
    if context is None:
517
 
        context = DEFAULT_CONTEXT_AMOUNT
518
480
    if format_cls is None:
519
481
        format_cls = DiffTree
520
 
    with cleanup.ExitStack() as exit_stack:
521
 
        exit_stack.enter_context(old_tree.lock_read())
 
482
    old_tree.lock_read()
 
483
    try:
522
484
        if extra_trees is not None:
523
485
            for tree in extra_trees:
524
 
                exit_stack.enter_context(tree.lock_read())
525
 
        exit_stack.enter_context(new_tree.lock_read())
526
 
        differ = format_cls.from_trees_options(old_tree, new_tree, to_file,
527
 
                                               path_encoding,
528
 
                                               external_diff_options,
529
 
                                               old_label, new_label, using,
530
 
                                               context_lines=context)
531
 
        return differ.show_diff(specific_files, extra_trees)
532
 
 
533
 
 
534
 
def _patch_header_date(tree, path):
 
486
                tree.lock_read()
 
487
        new_tree.lock_read()
 
488
        try:
 
489
            differ = format_cls.from_trees_options(old_tree, new_tree, to_file,
 
490
                                                   path_encoding,
 
491
                                                   external_diff_options,
 
492
                                                   old_label, new_label, using)
 
493
            return differ.show_diff(specific_files, extra_trees)
 
494
        finally:
 
495
            new_tree.unlock()
 
496
            if extra_trees is not None:
 
497
                for tree in extra_trees:
 
498
                    tree.unlock()
 
499
    finally:
 
500
        old_tree.unlock()
 
501
 
 
502
 
 
503
def _patch_header_date(tree, file_id, path):
535
504
    """Returns a timestamp suitable for use in a patch header."""
536
505
    try:
537
 
        mtime = tree.get_file_mtime(path)
538
 
    except FileTimestampUnavailable:
 
506
        mtime = tree.get_file_mtime(file_id, path)
 
507
    except errors.FileTimestampUnavailable:
539
508
        mtime = 0
540
509
    return timestamp.format_patch_date(mtime)
541
510
 
542
511
 
543
512
def get_executable_change(old_is_x, new_is_x):
544
 
    descr = {True: b"+x", False: b"-x", None: b"??"}
 
513
    descr = { True:"+x", False:"-x", None:"??" }
545
514
    if old_is_x != new_is_x:
546
 
        return [b"%s to %s" % (descr[old_is_x], descr[new_is_x],)]
 
515
        return ["%s to %s" % (descr[old_is_x], descr[new_is_x],)]
547
516
    else:
548
517
        return []
549
518
 
580
549
                     diff_tree.to_file, diff_tree.path_encoding)
581
550
 
582
551
    @staticmethod
583
 
    def _diff_many(differs, old_path, new_path, old_kind, new_kind):
 
552
    def _diff_many(differs, file_id, old_path, new_path, old_kind, new_kind):
584
553
        for file_differ in differs:
585
 
            result = file_differ.diff(old_path, new_path, old_kind, new_kind)
 
554
            result = file_differ.diff(file_id, old_path, new_path, old_kind,
 
555
                                      new_kind)
586
556
            if result is not DiffPath.CANNOT_DIFF:
587
557
                return result
588
558
        else:
595
565
    Represents kind change as deletion + creation.  Uses the other differs
596
566
    to do this.
597
567
    """
598
 
 
599
568
    def __init__(self, differs):
600
569
        self.differs = differs
601
570
 
606
575
    def from_diff_tree(klass, diff_tree):
607
576
        return klass(diff_tree.differs)
608
577
 
609
 
    def diff(self, old_path, new_path, old_kind, new_kind):
 
578
    def diff(self, file_id, old_path, new_path, old_kind, new_kind):
610
579
        """Perform comparison
611
580
 
 
581
        :param file_id: The file_id of the file to compare
612
582
        :param old_path: Path of the file in the old tree
613
583
        :param new_path: Path of the file in the new tree
614
584
        :param old_kind: Old file-kind of the file
616
586
        """
617
587
        if None in (old_kind, new_kind):
618
588
            return DiffPath.CANNOT_DIFF
619
 
        result = DiffPath._diff_many(
620
 
            self.differs, old_path, new_path, old_kind, None)
 
589
        result = DiffPath._diff_many(self.differs, file_id, old_path,
 
590
                                       new_path, old_kind, None)
621
591
        if result is DiffPath.CANNOT_DIFF:
622
592
            return result
623
 
        return DiffPath._diff_many(
624
 
            self.differs, old_path, new_path, None, new_kind)
625
 
 
626
 
 
627
 
class DiffTreeReference(DiffPath):
628
 
 
629
 
    def diff(self, old_path, new_path, old_kind, new_kind):
630
 
        """Perform comparison between two tree references.  (dummy)
631
 
 
632
 
        """
633
 
        if 'tree-reference' not in (old_kind, new_kind):
634
 
            return self.CANNOT_DIFF
635
 
        if old_kind not in ('tree-reference', None):
636
 
            return self.CANNOT_DIFF
637
 
        if new_kind not in ('tree-reference', None):
638
 
            return self.CANNOT_DIFF
639
 
        return self.CHANGED
 
593
        return DiffPath._diff_many(self.differs, file_id, old_path, new_path,
 
594
                                     None, new_kind)
640
595
 
641
596
 
642
597
class DiffDirectory(DiffPath):
643
598
 
644
 
    def diff(self, old_path, new_path, old_kind, new_kind):
 
599
    def diff(self, file_id, old_path, new_path, old_kind, new_kind):
645
600
        """Perform comparison between two directories.  (dummy)
646
601
 
647
602
        """
656
611
 
657
612
class DiffSymlink(DiffPath):
658
613
 
659
 
    def diff(self, old_path, new_path, old_kind, new_kind):
 
614
    def diff(self, file_id, old_path, new_path, old_kind, new_kind):
660
615
        """Perform comparison between two symlinks
661
616
 
 
617
        :param file_id: The file_id of the file to compare
662
618
        :param old_path: Path of the file in the old tree
663
619
        :param new_path: Path of the file in the new tree
664
620
        :param old_kind: Old file-kind of the file
667
623
        if 'symlink' not in (old_kind, new_kind):
668
624
            return self.CANNOT_DIFF
669
625
        if old_kind == 'symlink':
670
 
            old_target = self.old_tree.get_symlink_target(old_path)
 
626
            old_target = self.old_tree.get_symlink_target(file_id)
671
627
        elif old_kind is None:
672
628
            old_target = None
673
629
        else:
674
630
            return self.CANNOT_DIFF
675
631
        if new_kind == 'symlink':
676
 
            new_target = self.new_tree.get_symlink_target(new_path)
 
632
            new_target = self.new_tree.get_symlink_target(file_id)
677
633
        elif new_kind is None:
678
634
            new_target = None
679
635
        else:
682
638
 
683
639
    def diff_symlink(self, old_target, new_target):
684
640
        if old_target is None:
685
 
            self.to_file.write(b'=== target is \'%s\'\n' %
686
 
                               new_target.encode(self.path_encoding, 'replace'))
 
641
            self.to_file.write('=== target is %r\n' % new_target)
687
642
        elif new_target is None:
688
 
            self.to_file.write(b'=== target was \'%s\'\n' %
689
 
                               old_target.encode(self.path_encoding, 'replace'))
 
643
            self.to_file.write('=== target was %r\n' % old_target)
690
644
        else:
691
 
            self.to_file.write(b'=== target changed \'%s\' => \'%s\'\n' %
692
 
                               (old_target.encode(self.path_encoding, 'replace'),
693
 
                                new_target.encode(self.path_encoding, 'replace')))
 
645
            self.to_file.write('=== target changed %r => %r\n' %
 
646
                              (old_target, new_target))
694
647
        return self.CHANGED
695
648
 
696
649
 
701
654
    EPOCH_DATE = '1970-01-01 00:00:00 +0000'
702
655
 
703
656
    def __init__(self, old_tree, new_tree, to_file, path_encoding='utf-8',
704
 
                 old_label='', new_label='', text_differ=internal_diff,
705
 
                 context_lines=DEFAULT_CONTEXT_AMOUNT):
 
657
                 old_label='', new_label='', text_differ=internal_diff):
706
658
        DiffPath.__init__(self, old_tree, new_tree, to_file, path_encoding)
707
659
        self.text_differ = text_differ
708
660
        self.old_label = old_label
709
661
        self.new_label = new_label
710
662
        self.path_encoding = path_encoding
711
 
        self.context_lines = context_lines
712
663
 
713
 
    def diff(self, old_path, new_path, old_kind, new_kind):
 
664
    def diff(self, file_id, old_path, new_path, old_kind, new_kind):
714
665
        """Compare two files in unified diff format
715
666
 
 
667
        :param file_id: The file_id of the file to compare
716
668
        :param old_path: Path of the file in the old tree
717
669
        :param new_path: Path of the file in the new tree
718
670
        :param old_kind: Old file-kind of the file
720
672
        """
721
673
        if 'file' not in (old_kind, new_kind):
722
674
            return self.CANNOT_DIFF
 
675
        from_file_id = to_file_id = file_id
723
676
        if old_kind == 'file':
724
 
            old_date = _patch_header_date(self.old_tree, old_path)
 
677
            old_date = _patch_header_date(self.old_tree, file_id, old_path)
725
678
        elif old_kind is None:
726
679
            old_date = self.EPOCH_DATE
 
680
            from_file_id = None
727
681
        else:
728
682
            return self.CANNOT_DIFF
729
683
        if new_kind == 'file':
730
 
            new_date = _patch_header_date(self.new_tree, new_path)
 
684
            new_date = _patch_header_date(self.new_tree, file_id, new_path)
731
685
        elif new_kind is None:
732
686
            new_date = self.EPOCH_DATE
 
687
            to_file_id = None
733
688
        else:
734
689
            return self.CANNOT_DIFF
735
 
        from_label = '%s%s\t%s' % (self.old_label, old_path,
736
 
                                   old_date)
737
 
        to_label = '%s%s\t%s' % (self.new_label, new_path,
738
 
                                 new_date)
739
 
        return self.diff_text(old_path, new_path, from_label, to_label)
 
690
        from_label = '%s%s\t%s' % (self.old_label, old_path, old_date)
 
691
        to_label = '%s%s\t%s' % (self.new_label, new_path, new_date)
 
692
        return self.diff_text(from_file_id, to_file_id, from_label, to_label,
 
693
            old_path, new_path)
740
694
 
741
 
    def diff_text(self, from_path, to_path, from_label, to_label):
 
695
    def diff_text(self, from_file_id, to_file_id, from_label, to_label,
 
696
        from_path=None, to_path=None):
742
697
        """Diff the content of given files in two trees
743
698
 
744
 
        :param from_path: The path in the from tree. If None,
 
699
        :param from_file_id: The id of the file in the from tree.  If None,
745
700
            the file is not present in the from tree.
746
 
        :param to_path: The path in the to tree. This may refer
747
 
            to a different file from from_path.  If None,
 
701
        :param to_file_id: The id of the file in the to tree.  This may refer
 
702
            to a different file from from_file_id.  If None,
748
703
            the file is not present in the to tree.
 
704
        :param from_path: The path in the from tree or None if unknown.
 
705
        :param to_path: The path in the to tree or None if unknown.
749
706
        """
750
 
        def _get_text(tree, path):
751
 
            if path is None:
752
 
                return []
753
 
            try:
754
 
                return tree.get_file_lines(path)
755
 
            except errors.NoSuchFile:
 
707
        def _get_text(tree, file_id, path):
 
708
            if file_id is not None:
 
709
                return tree.get_file(file_id, path).readlines()
 
710
            else:
756
711
                return []
757
712
        try:
758
 
            from_text = _get_text(self.old_tree, from_path)
759
 
            to_text = _get_text(self.new_tree, to_path)
 
713
            from_text = _get_text(self.old_tree, from_file_id, from_path)
 
714
            to_text = _get_text(self.new_tree, to_file_id, to_path)
760
715
            self.text_differ(from_label, from_text, to_label, to_text,
761
 
                             self.to_file, path_encoding=self.path_encoding,
762
 
                             context_lines=self.context_lines)
 
716
                             self.to_file)
763
717
        except errors.BinaryFile:
764
718
            self.to_file.write(
765
 
                ("Binary files %s%s and %s%s differ\n" %
766
 
                 (self.old_label, from_path, self.new_label, to_path)).encode(self.path_encoding, 'replace'))
 
719
                  ("Binary files %s and %s differ\n" %
 
720
                  (from_label, to_label)).encode(self.path_encoding))
767
721
        return self.CHANGED
768
722
 
769
723
 
773
727
                 path_encoding='utf-8'):
774
728
        DiffPath.__init__(self, old_tree, new_tree, to_file, path_encoding)
775
729
        self.command_template = command_template
776
 
        self._root = osutils.mkdtemp(prefix='brz-diff-')
 
730
        self._root = osutils.mkdtemp(prefix='bzr-diff-')
777
731
 
778
732
    @classmethod
779
 
    def from_string(klass, command_template, old_tree, new_tree, to_file,
 
733
    def from_string(klass, command_string, old_tree, new_tree, to_file,
780
734
                    path_encoding='utf-8'):
 
735
        command_template = cmdline.split(command_string)
 
736
        if '@' not in command_string:
 
737
            command_template.extend(['@old_path', '@new_path'])
781
738
        return klass(command_template, old_tree, new_tree, to_file,
782
739
                     path_encoding)
783
740
 
784
741
    @classmethod
785
 
    def make_from_diff_tree(klass, command_string, external_diff_options=None):
 
742
    def make_from_diff_tree(klass, command_string):
786
743
        def from_diff_tree(diff_tree):
787
 
            full_command_string = [command_string]
788
 
            if external_diff_options is not None:
789
 
                full_command_string += ' ' + external_diff_options
790
 
            return klass.from_string(full_command_string, diff_tree.old_tree,
 
744
            return klass.from_string(command_string, diff_tree.old_tree,
791
745
                                     diff_tree.new_tree, diff_tree.to_file)
792
746
        return from_diff_tree
793
747
 
794
748
    def _get_command(self, old_path, new_path):
795
749
        my_map = {'old_path': old_path, 'new_path': new_path}
796
 
        command = [t.format(**my_map) for t in
797
 
                   self.command_template]
798
 
        if sys.platform == 'win32':  # Popen doesn't accept unicode on win32
799
 
            command_encoded = []
800
 
            for c in command:
801
 
                if isinstance(c, text_type):
802
 
                    command_encoded.append(c.encode('mbcs'))
803
 
                else:
804
 
                    command_encoded.append(c)
805
 
            return command_encoded
806
 
        else:
807
 
            return command
 
750
        return [AtTemplate(t).substitute(my_map) for t in
 
751
                self.command_template]
808
752
 
809
753
    def _execute(self, old_path, new_path):
810
754
        command = self._get_command(old_path, new_path)
811
755
        try:
812
756
            proc = subprocess.Popen(command, stdout=subprocess.PIPE,
813
757
                                    cwd=self._root)
814
 
        except OSError as e:
 
758
        except OSError, e:
815
759
            if e.errno == errno.ENOENT:
816
760
                raise errors.ExecutableMissing(command[0])
817
761
            else:
818
762
                raise
819
763
        self.to_file.write(proc.stdout.read())
820
 
        proc.stdout.close()
821
764
        return proc.wait()
822
765
 
823
766
    def _try_symlink_root(self, tree, prefix):
824
 
        if (getattr(tree, 'abspath', None) is None or
825
 
                not osutils.host_os_dereferences_symlinks()):
 
767
        if (getattr(tree, 'abspath', None) is None
 
768
            or not osutils.host_os_dereferences_symlinks()):
826
769
            return False
827
770
        try:
828
771
            os.symlink(tree.abspath(''), osutils.pathjoin(self._root, prefix))
829
 
        except OSError as e:
 
772
        except OSError, e:
830
773
            if e.errno != errno.EEXIST:
831
774
                raise
832
775
        return True
833
776
 
834
 
    @staticmethod
835
 
    def _fenc():
836
 
        """Returns safe encoding for passing file path to diff tool"""
837
 
        if sys.platform == 'win32':
838
 
            return 'mbcs'
839
 
        else:
840
 
            # Don't fallback to 'utf-8' because subprocess may not be able to
841
 
            # handle utf-8 correctly when locale is not utf-8.
842
 
            return sys.getfilesystemencoding() or 'ascii'
843
 
 
844
 
    def _is_safepath(self, path):
845
 
        """Return true if `path` may be able to pass to subprocess."""
846
 
        fenc = self._fenc()
847
 
        try:
848
 
            return path == path.encode(fenc).decode(fenc)
849
 
        except UnicodeError:
850
 
            return False
851
 
 
852
 
    def _safe_filename(self, prefix, relpath):
853
 
        """Replace unsafe character in `relpath` then join `self._root`,
854
 
        `prefix` and `relpath`."""
855
 
        fenc = self._fenc()
856
 
        # encoded_str.replace('?', '_') may break multibyte char.
857
 
        # So we should encode, decode, then replace(u'?', u'_')
858
 
        relpath_tmp = relpath.encode(fenc, 'replace').decode(fenc, 'replace')
859
 
        relpath_tmp = relpath_tmp.replace(u'?', u'_')
860
 
        return osutils.pathjoin(self._root, prefix, relpath_tmp)
861
 
 
862
 
    def _write_file(self, relpath, tree, prefix, force_temp=False,
 
777
    def _write_file(self, file_id, tree, prefix, relpath, force_temp=False,
863
778
                    allow_write=False):
864
779
        if not force_temp and isinstance(tree, WorkingTree):
865
 
            full_path = tree.abspath(relpath)
866
 
            if self._is_safepath(full_path):
867
 
                return full_path
868
 
 
869
 
        full_path = self._safe_filename(prefix, relpath)
 
780
            return tree.abspath(tree.id2path(file_id))
 
781
        
 
782
        full_path = osutils.pathjoin(self._root, prefix, relpath)
870
783
        if not force_temp and self._try_symlink_root(tree, prefix):
871
784
            return full_path
872
785
        parent_dir = osutils.dirname(full_path)
873
786
        try:
874
787
            os.makedirs(parent_dir)
875
 
        except OSError as e:
 
788
        except OSError, e:
876
789
            if e.errno != errno.EEXIST:
877
790
                raise
878
 
        with tree.get_file(relpath) as source, \
879
 
                open(full_path, 'wb') as target:
880
 
            osutils.pumpfile(source, target)
881
 
        try:
882
 
            mtime = tree.get_file_mtime(relpath)
883
 
        except FileTimestampUnavailable:
 
791
        source = tree.get_file(file_id, relpath)
 
792
        try:
 
793
            target = open(full_path, 'wb')
 
794
            try:
 
795
                osutils.pumpfile(source, target)
 
796
            finally:
 
797
                target.close()
 
798
        finally:
 
799
            source.close()
 
800
        try:
 
801
            mtime = tree.get_file_mtime(file_id)
 
802
        except errors.FileTimestampUnavailable:
884
803
            pass
885
804
        else:
886
805
            os.utime(full_path, (mtime, mtime))
888
807
            osutils.make_readonly(full_path)
889
808
        return full_path
890
809
 
891
 
    def _prepare_files(self, old_path, new_path, force_temp=False,
 
810
    def _prepare_files(self, file_id, old_path, new_path, force_temp=False,
892
811
                       allow_write_new=False):
893
 
        old_disk_path = self._write_file(
894
 
            old_path, self.old_tree, 'old', force_temp)
895
 
        new_disk_path = self._write_file(
896
 
            new_path, self.new_tree, 'new', force_temp,
897
 
            allow_write=allow_write_new)
 
812
        old_disk_path = self._write_file(file_id, self.old_tree, 'old',
 
813
                                         old_path, force_temp)
 
814
        new_disk_path = self._write_file(file_id, self.new_tree, 'new',
 
815
                                         new_path, force_temp,
 
816
                                         allow_write=allow_write_new)
898
817
        return old_disk_path, new_disk_path
899
818
 
900
819
    def finish(self):
901
820
        try:
902
821
            osutils.rmtree(self._root)
903
 
        except OSError as e:
 
822
        except OSError, e:
904
823
            if e.errno != errno.ENOENT:
905
824
                mutter("The temporary directory \"%s\" was not "
906
 
                       "cleanly removed: %s." % (self._root, e))
 
825
                        "cleanly removed: %s." % (self._root, e))
907
826
 
908
 
    def diff(self, old_path, new_path, old_kind, new_kind):
 
827
    def diff(self, file_id, old_path, new_path, old_kind, new_kind):
909
828
        if (old_kind, new_kind) != ('file', 'file'):
910
829
            return DiffPath.CANNOT_DIFF
911
830
        (old_disk_path, new_disk_path) = self._prepare_files(
912
 
            old_path, new_path)
 
831
                                                file_id, old_path, new_path)
913
832
        self._execute(old_disk_path, new_disk_path)
914
833
 
915
 
    def edit_file(self, old_path, new_path):
 
834
    def edit_file(self, file_id):
916
835
        """Use this tool to edit a file.
917
836
 
918
837
        A temporary copy will be edited, and the new contents will be
919
838
        returned.
920
839
 
 
840
        :param file_id: The id of the file to edit.
921
841
        :return: The new contents of the file.
922
842
        """
923
 
        old_abs_path, new_abs_path = self._prepare_files(
924
 
            old_path, new_path, allow_write_new=True, force_temp=True)
925
 
        command = self._get_command(old_abs_path, new_abs_path)
 
843
        old_path = self.old_tree.id2path(file_id)
 
844
        new_path = self.new_tree.id2path(file_id)
 
845
        new_abs_path = self._prepare_files(file_id, old_path, new_path,
 
846
                                           allow_write_new=True,
 
847
                                           force_temp=True)[1]
 
848
        command = self._get_command(osutils.pathjoin('old', old_path),
 
849
                                    osutils.pathjoin('new', new_path))
926
850
        subprocess.call(command, cwd=self._root)
927
 
        with open(new_abs_path, 'rb') as new_file:
 
851
        new_file = open(new_abs_path, 'r')
 
852
        try:
928
853
            return new_file.read()
 
854
        finally:
 
855
            new_file.close()
929
856
 
930
857
 
931
858
class DiffTree(object):
943
870
    # list of factories that can provide instances of DiffPath objects
944
871
    # may be extended by plugins.
945
872
    diff_factories = [DiffSymlink.from_diff_tree,
946
 
                      DiffDirectory.from_diff_tree,
947
 
                      DiffTreeReference.from_diff_tree]
 
873
                      DiffDirectory.from_diff_tree]
948
874
 
949
875
    def __init__(self, old_tree, new_tree, to_file, path_encoding='utf-8',
950
876
                 diff_text=None, extra_factories=None):
960
886
            DiffPaths"""
961
887
        if diff_text is None:
962
888
            diff_text = DiffText(old_tree, new_tree, to_file, path_encoding,
963
 
                                 '', '', internal_diff)
 
889
                                 '', '',  internal_diff)
964
890
        self.old_tree = old_tree
965
891
        self.new_tree = new_tree
966
892
        self.to_file = to_file
974
900
    @classmethod
975
901
    def from_trees_options(klass, old_tree, new_tree, to_file,
976
902
                           path_encoding, external_diff_options, old_label,
977
 
                           new_label, using, context_lines):
 
903
                           new_label, using):
978
904
        """Factory for producing a DiffTree.
979
905
 
980
906
        Designed to accept options used by show_diff_trees.
981
 
 
982
907
        :param old_tree: The tree to show as old in the comparison
983
908
        :param new_tree: The tree to show as new in the comparison
984
909
        :param to_file: File to write comparisons to
990
915
        :param using: Commandline to use to invoke an external diff tool
991
916
        """
992
917
        if using is not None:
993
 
            extra_factories = [DiffFromTool.make_from_diff_tree(
994
 
                using, external_diff_options)]
 
918
            extra_factories = [DiffFromTool.make_from_diff_tree(using)]
995
919
        else:
996
920
            extra_factories = []
997
921
        if external_diff_options:
998
922
            opts = external_diff_options.split()
999
 
 
1000
 
            def diff_file(olab, olines, nlab, nlines, to_file, path_encoding=None, context_lines=None):
1001
 
                """:param path_encoding: not used but required
1002
 
                        to match the signature of internal_diff.
1003
 
                """
 
923
            def diff_file(olab, olines, nlab, nlines, to_file):
1004
924
                external_diff(olab, olines, nlab, nlines, to_file, opts)
1005
925
        else:
1006
926
            diff_file = internal_diff
1007
927
        diff_text = DiffText(old_tree, new_tree, to_file, path_encoding,
1008
 
                             old_label, new_label, diff_file, context_lines=context_lines)
 
928
                             old_label, new_label, diff_file)
1009
929
        return klass(old_tree, new_tree, to_file, path_encoding, diff_text,
1010
930
                     extra_factories)
1011
931
 
1025
945
        # TODO: Generation of pseudo-diffs for added/deleted files could
1026
946
        # be usefully made into a much faster special case.
1027
947
        iterator = self.new_tree.iter_changes(self.old_tree,
1028
 
                                              specific_files=specific_files,
1029
 
                                              extra_trees=extra_trees,
1030
 
                                              require_versioned=True)
 
948
                                               specific_files=specific_files,
 
949
                                               extra_trees=extra_trees,
 
950
                                               require_versioned=True)
1031
951
        has_changes = 0
1032
 
 
1033
952
        def changes_key(change):
1034
 
            old_path, new_path = change.path
 
953
            old_path, new_path = change[1]
1035
954
            path = new_path
1036
955
            if path is None:
1037
956
                path = old_path
1038
957
            return path
1039
 
 
1040
958
        def get_encoded_path(path):
1041
959
            if path is not None:
1042
960
                return path.encode(self.path_encoding, "replace")
1043
 
        for change in sorted(iterator, key=changes_key):
 
961
        for (file_id, paths, changed_content, versioned, parent, name, kind,
 
962
             executable) in sorted(iterator, key=changes_key):
1044
963
            # The root does not get diffed, and items with no known kind (that
1045
964
            # is, missing) in both trees are skipped as well.
1046
 
            if change.parent_id == (None, None) or change.kind == (None, None):
1047
 
                continue
1048
 
            if change.kind[0] == 'symlink' and not self.new_tree.supports_symlinks():
1049
 
                warning(
1050
 
                    'Ignoring "%s" as symlinks are not '
1051
 
                    'supported on this filesystem.' % (change.path[0],))
1052
 
                continue
1053
 
            oldpath, newpath = change.path
1054
 
            oldpath_encoded = get_encoded_path(change.path[0])
1055
 
            newpath_encoded = get_encoded_path(change.path[1])
1056
 
            old_present = (change.kind[0] is not None and change.versioned[0])
1057
 
            new_present = (change.kind[1] is not None and change.versioned[1])
1058
 
            executable = change.executable
1059
 
            kind = change.kind
1060
 
            renamed = (change.parent_id[0], change.name[0]) != (change.parent_id[1], change.name[1])
 
965
            if parent == (None, None) or kind == (None, None):
 
966
                continue
 
967
            oldpath, newpath = paths
 
968
            oldpath_encoded = get_encoded_path(paths[0])
 
969
            newpath_encoded = get_encoded_path(paths[1])
 
970
            old_present = (kind[0] is not None and versioned[0])
 
971
            new_present = (kind[1] is not None and versioned[1])
 
972
            renamed = (parent[0], name[0]) != (parent[1], name[1])
1061
973
 
1062
974
            properties_changed = []
1063
 
            properties_changed.extend(
1064
 
                get_executable_change(executable[0], executable[1]))
 
975
            properties_changed.extend(get_executable_change(executable[0], executable[1]))
1065
976
 
1066
977
            if properties_changed:
1067
 
                prop_str = b" (properties changed: %s)" % (
1068
 
                    b", ".join(properties_changed),)
 
978
                prop_str = " (properties changed: %s)" % (", ".join(properties_changed),)
1069
979
            else:
1070
 
                prop_str = b""
 
980
                prop_str = ""
1071
981
 
1072
982
            if (old_present, new_present) == (True, False):
1073
 
                self.to_file.write(b"=== removed %s '%s'\n" %
1074
 
                                   (kind[0].encode('ascii'), oldpath_encoded))
 
983
                self.to_file.write("=== removed %s '%s'\n" %
 
984
                                   (kind[0], oldpath_encoded))
1075
985
                newpath = oldpath
1076
986
            elif (old_present, new_present) == (False, True):
1077
 
                self.to_file.write(b"=== added %s '%s'\n" %
1078
 
                                   (kind[1].encode('ascii'), newpath_encoded))
 
987
                self.to_file.write("=== added %s '%s'\n" %
 
988
                                   (kind[1], newpath_encoded))
1079
989
                oldpath = newpath
1080
990
            elif renamed:
1081
 
                self.to_file.write(b"=== renamed %s '%s' => '%s'%s\n" %
1082
 
                                   (kind[0].encode('ascii'), oldpath_encoded, newpath_encoded, prop_str))
 
991
                self.to_file.write("=== renamed %s '%s' => '%s'%s\n" %
 
992
                    (kind[0], oldpath_encoded, newpath_encoded, prop_str))
1083
993
            else:
1084
994
                # if it was produced by iter_changes, it must be
1085
995
                # modified *somehow*, either content or execute bit.
1086
 
                self.to_file.write(b"=== modified %s '%s'%s\n" % (kind[0].encode('ascii'),
1087
 
                                                                  newpath_encoded, prop_str))
1088
 
            if change.changed_content:
1089
 
                self._diff(oldpath, newpath, kind[0], kind[1])
 
996
                self.to_file.write("=== modified %s '%s'%s\n" % (kind[0],
 
997
                                   newpath_encoded, prop_str))
 
998
            if changed_content:
 
999
                self._diff(file_id, oldpath, newpath, kind[0], kind[1])
1090
1000
                has_changes = 1
1091
1001
            if renamed:
1092
1002
                has_changes = 1
1093
1003
        return has_changes
1094
1004
 
1095
 
    def diff(self, old_path, new_path):
 
1005
    def diff(self, file_id, old_path, new_path):
1096
1006
        """Perform a diff of a single file
1097
1007
 
 
1008
        :param file_id: file-id of the file
1098
1009
        :param old_path: The path of the file in the old tree
1099
1010
        :param new_path: The path of the file in the new tree
1100
1011
        """
1101
 
        if old_path is None:
 
1012
        try:
 
1013
            old_kind = self.old_tree.kind(file_id)
 
1014
        except (errors.NoSuchId, errors.NoSuchFile):
1102
1015
            old_kind = None
1103
 
        else:
1104
 
            old_kind = self.old_tree.kind(old_path)
1105
 
        if new_path is None:
 
1016
        try:
 
1017
            new_kind = self.new_tree.kind(file_id)
 
1018
        except (errors.NoSuchId, errors.NoSuchFile):
1106
1019
            new_kind = None
1107
 
        else:
1108
 
            new_kind = self.new_tree.kind(new_path)
1109
 
        self._diff(old_path, new_path, old_kind, new_kind)
1110
 
 
1111
 
    def _diff(self, old_path, new_path, old_kind, new_kind):
1112
 
        result = DiffPath._diff_many(
1113
 
            self.differs, old_path, new_path, old_kind, new_kind)
 
1020
        self._diff(file_id, old_path, new_path, old_kind, new_kind)
 
1021
 
 
1022
 
 
1023
    def _diff(self, file_id, old_path, new_path, old_kind, new_kind):
 
1024
        result = DiffPath._diff_many(self.differs, file_id, old_path,
 
1025
                                       new_path, old_kind, new_kind)
1114
1026
        if result is DiffPath.CANNOT_DIFF:
1115
1027
            error_path = new_path
1116
1028
            if error_path is None: