/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 breezy/diff.py

  • Committer: Jelmer Vernooij
  • Date: 2019-12-28 14:10:36 UTC
  • mto: This revision was merged to the branch mainline in revision 7431.
  • Revision ID: jelmer@jelmer.uk-20191228141036-hsqitjor9m5fsri1
Python3 compatibility.

Show diffs side-by-side

added added

removed removed

Lines of Context:
breezy/diff.py
1
 
# Copyright (C) 2005-2010 Canonical Ltd.
 
1
# Copyright (C) 2005-2014 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
 
17
19
import difflib
18
20
import os
19
21
import re
20
 
import string
21
22
import sys
22
23
 
23
 
from bzrlib.lazy_import import lazy_import
 
24
from .lazy_import import lazy_import
24
25
lazy_import(globals(), """
25
26
import errno
 
27
import patiencediff
26
28
import subprocess
27
29
import tempfile
28
 
import time
29
30
 
30
 
from bzrlib import (
31
 
    branch as _mod_branch,
32
 
    bzrdir,
33
 
    cmdline,
 
31
from breezy import (
34
32
    cleanup,
 
33
    controldir,
35
34
    errors,
36
35
    osutils,
37
 
    patiencediff,
38
36
    textfile,
39
37
    timestamp,
40
38
    views,
41
39
    )
42
40
 
43
 
from bzrlib.workingtree import WorkingTree
 
41
from breezy.workingtree import WorkingTree
 
42
from breezy.i18n import gettext
44
43
""")
45
44
 
46
 
from bzrlib.registry import (
 
45
from .registry import (
47
46
    Registry,
48
47
    )
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 = '@'
 
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
60
54
 
61
55
 
62
56
# TODO: Rather than building a changeset object, we should probably
73
67
        self.opcodes = None
74
68
 
75
69
 
76
 
def internal_diff(old_filename, oldlines, new_filename, newlines, to_file,
 
70
def internal_diff(old_label, oldlines, new_label, newlines, to_file,
77
71
                  allow_binary=False, sequence_matcher=None,
78
 
                  path_encoding='utf8'):
 
72
                  path_encoding='utf8', context_lines=DEFAULT_CONTEXT_AMOUNT):
79
73
    # FIXME: difflib is wrong if there is no trailing newline.
80
74
    # The syntax used by patch seems to be "\ No newline at
81
75
    # end of file" following the last diff line from that
86
80
    # In the meantime we at least make sure the patch isn't
87
81
    # mangled.
88
82
 
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
 
 
95
83
    if allow_binary is False:
96
84
        textfile.check_text_lines(oldlines)
97
85
        textfile.check_text_lines(newlines)
98
86
 
99
87
    if sequence_matcher is None:
100
88
        sequence_matcher = patiencediff.PatienceSequenceMatcher
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)
 
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)
105
94
 
106
95
    ud = list(ud)
107
 
    if len(ud) == 0: # Identical contents, nothing to do
 
96
    if len(ud) == 0:  # Identical contents, nothing to do
108
97
        return
109
98
    # work-around for difflib being too smart for its own good
110
99
    # if /dev/null is "1,0", patch won't recognize it as /dev/null
111
100
    if not oldlines:
112
 
        ud[2] = ud[2].replace('-1,0', '-0,0')
 
101
        ud[2] = ud[2].replace(b'-1,0', b'-0,0')
113
102
    elif not newlines:
114
 
        ud[2] = ud[2].replace('+1,0', '+0,0')
 
103
        ud[2] = ud[2].replace(b'+1,0', b'+0,0')
115
104
 
116
105
    for line in ud:
117
106
        to_file.write(line)
118
 
        if not line.endswith('\n'):
119
 
            to_file.write("\n\\ No newline at end of file\n")
120
 
    to_file.write('\n')
 
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
121
179
 
122
180
 
123
181
def _spawn_external_diff(diffcmd, capture_errors=True):
124
 
    """Spawn the externall diff process, and return the child handle.
 
182
    """Spawn the external diff process, and return the child handle.
125
183
 
126
184
    :param diffcmd: The command list to spawn
127
185
    :param capture_errors: Capture stderr as well as setting LANG=C
149
207
                                stdout=subprocess.PIPE,
150
208
                                stderr=stderr,
151
209
                                env=env)
152
 
    except OSError, e:
 
210
    except OSError as e:
153
211
        if e.errno == errno.ENOENT:
154
212
            raise errors.NoDiff(str(e))
155
213
        raise
157
215
    return pipe
158
216
 
159
217
 
160
 
def external_diff(old_filename, oldlines, new_filename, newlines, to_file,
 
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,
161
255
                  diff_opts):
162
256
    """Display a diff by calling out to the external diff program."""
163
257
    # make sure our own output is properly ordered before the diff
164
258
    to_file.flush()
165
259
 
166
 
    oldtmp_fd, old_abspath = tempfile.mkstemp(prefix='bzr-diff-old-')
167
 
    newtmp_fd, new_abspath = tempfile.mkstemp(prefix='bzr-diff-new-')
 
260
    oldtmp_fd, old_abspath = tempfile.mkstemp(prefix='brz-diff-old-')
 
261
    newtmp_fd, new_abspath = tempfile.mkstemp(prefix='brz-diff-new-')
168
262
    oldtmpf = os.fdopen(oldtmp_fd, 'wb')
169
263
    newtmpf = os.fdopen(newtmp_fd, 'wb')
170
264
 
187
281
        if sys.platform == 'win32':
188
282
            # Popen doesn't do the proper encoding for external commands
189
283
            # Since we are dealing with an ANSI api, use mbcs encoding
190
 
            old_filename = old_filename.encode('mbcs')
191
 
            new_filename = new_filename.encode('mbcs')
 
284
            old_label = old_label.encode('mbcs')
 
285
            new_label = new_label.encode('mbcs')
192
286
        diffcmd = ['diff',
193
 
                   '--label', old_filename,
 
287
                   '--label', old_label,
194
288
                   old_abspath,
195
 
                   '--label', new_filename,
 
289
                   '--label', new_label,
196
290
                   new_abspath,
197
291
                   '--binary',
198
 
                  ]
 
292
                   ]
199
293
 
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')
 
294
        diff_opts = default_style_unified(diff_opts)
220
295
 
221
296
        if diff_opts:
222
297
            diffcmd.extend(diff_opts)
223
298
 
224
299
        pipe = _spawn_external_diff(diffcmd, capture_errors=True)
225
 
        out,err = pipe.communicate()
 
300
        out, err = pipe.communicate()
226
301
        rc = pipe.returncode
227
302
 
228
303
        # internal_diff() adds a trailing newline, add one here for consistency
229
 
        out += '\n'
 
304
        out += b'\n'
230
305
        if rc == 2:
231
306
            # 'diff' gives retcode == 2 for all sorts of errors
232
307
            # one of those is 'Binary files differ'.
239
314
            out, err = pipe.communicate()
240
315
 
241
316
            # Write out the new i18n diff response
242
 
            to_file.write(out+'\n')
 
317
            to_file.write(out + b'\n')
243
318
            if pipe.returncode != 2:
244
319
                raise errors.BzrError(
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,))
 
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,))
248
323
 
249
 
            first_line = lang_c_out.split('\n', 1)[0]
 
324
            first_line = lang_c_out.split(b'\n', 1)[0]
250
325
            # Starting with diffutils 2.8.4 the word "binary" was dropped.
251
 
            m = re.match('^(binary )?files.*differ$', first_line, re.I)
 
326
            m = re.match(b'^(binary )?files.*differ$', first_line, re.I)
252
327
            if m is None:
253
328
                raise errors.BzrError('external diff failed with exit code 2;'
254
329
                                      ' command: %r' % (diffcmd,))
267
342
                msg = 'exit code %d' % rc
268
343
 
269
344
            raise errors.BzrError('external diff failed with %s; command: %r'
270
 
                                  % (rc, diffcmd))
271
 
 
 
345
                                  % (msg, diffcmd))
272
346
 
273
347
    finally:
274
348
        oldtmpf.close()                 # and delete
275
349
        newtmpf.close()
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
 
    
 
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
 
325
364
 
326
365
def get_trees_and_branches_to_diff_locked(
327
 
    path_list, revision_specs, old_url, new_url, add_cleanup, apply_view=True):
 
366
        path_list, revision_specs, old_url, new_url, exit_stack, apply_view=True):
328
367
    """Get the trees and specific files to diff given a list of paths.
329
368
 
330
369
    This method works out the trees to be diff'ed and the files of
341
380
    :param new_url:
342
381
        The url of the new branch or tree. If None, the tree to use is
343
382
        taken from the first path, if any, or the current working tree.
344
 
    :param add_cleanup:
345
 
        a callable like Command.add_cleanup.  get_trees_and_branches_to_diff
 
383
    :param exit_stack:
 
384
        an ExitStack object. get_trees_and_branches_to_diff
346
385
        will register cleanups that must be run to unlock the trees, etc.
347
386
    :param apply_view:
348
387
        if True and a view is set, apply the view or check that the paths
351
390
        a tuple of (old_tree, new_tree, old_branch, new_branch,
352
391
        specific_files, extra_trees) where extra_trees is a sequence of
353
392
        additional trees to search in for file-ids.  The trees and branches
354
 
        will be read-locked until the cleanups registered via the add_cleanup
 
393
        will be read-locked until the cleanups registered via the exit_stack
355
394
        param are run.
356
395
    """
357
396
    # Get the old and new revision specs
383
422
 
384
423
    def lock_tree_or_branch(wt, br):
385
424
        if wt is not None:
386
 
            wt.lock_read()
387
 
            add_cleanup(wt.unlock)
 
425
            exit_stack.enter_context(wt.lock_read())
388
426
        elif br is not None:
389
 
            br.lock_read()
390
 
            add_cleanup(br.unlock)
 
427
            exit_stack.enter_context(br.lock_read())
391
428
 
392
429
    # Get the old location
393
430
    specific_files = []
394
431
    if old_url is None:
395
432
        old_url = default_location
396
433
    working_tree, branch, relpath = \
397
 
        bzrdir.BzrDir.open_containing_tree_or_branch(old_url)
 
434
        controldir.ControlDir.open_containing_tree_or_branch(old_url)
398
435
    lock_tree_or_branch(working_tree, branch)
399
436
    if consider_relpath and relpath != '':
400
437
        if working_tree is not None and apply_view:
408
445
        new_url = default_location
409
446
    if new_url != old_url:
410
447
        working_tree, branch, relpath = \
411
 
            bzrdir.BzrDir.open_containing_tree_or_branch(new_url)
 
448
            controldir.ControlDir.open_containing_tree_or_branch(new_url)
412
449
        lock_tree_or_branch(working_tree, branch)
413
450
        if consider_relpath and relpath != '':
414
451
            if working_tree is not None and apply_view:
415
452
                views.check_path_in_view(working_tree, relpath)
416
453
            specific_files.append(relpath)
417
454
    new_tree = _get_tree_to_diff(new_revision_spec, working_tree, branch,
418
 
        basis_is_default=working_tree is None)
 
455
                                 basis_is_default=working_tree is None)
419
456
    new_branch = branch
420
457
 
421
458
    # Get the specific files (all files is None, no files is [])
422
459
    if make_paths_wt_relative and working_tree is not None:
423
 
        try:
424
 
            from bzrlib.builtins import safe_relpath_files
425
 
            other_paths = safe_relpath_files(working_tree, other_paths,
 
460
        other_paths = working_tree.safe_relpath_files(
 
461
            other_paths,
426
462
            apply_view=apply_view)
427
 
        except errors.FileInWrongBranch:
428
 
            raise errors.BzrCommandError("Files are in different branches")
429
463
    specific_files.extend(other_paths)
430
464
    if len(specific_files) == 0:
431
465
        specific_files = None
432
 
        if (working_tree is not None and working_tree.supports_views()
433
 
            and apply_view):
 
466
        if (working_tree is not None and working_tree.supports_views() and
 
467
                apply_view):
434
468
            view_files = working_tree.views.lookup_view()
435
469
            if view_files:
436
470
                specific_files = view_files
437
471
                view_str = views.view_display_str(view_files)
438
 
                note("*** Ignoring files outside view. View is %s" % view_str)
 
472
                note(gettext("*** Ignoring files outside view. View is %s") % view_str)
439
473
 
440
474
    # Get extra trees that ought to be searched for file-ids
441
475
    extra_trees = None
442
476
    if working_tree is not None and working_tree not in (old_tree, new_tree):
443
477
        extra_trees = (working_tree,)
444
 
    return old_tree, new_tree, old_branch, new_branch, specific_files, extra_trees
 
478
    return (old_tree, new_tree, old_branch, new_branch,
 
479
            specific_files, extra_trees)
445
480
 
446
481
 
447
482
def _get_tree_to_diff(spec, tree=None, branch=None, basis_is_default=True):
464
499
                    extra_trees=None,
465
500
                    path_encoding='utf8',
466
501
                    using=None,
467
 
                    format_cls=None):
 
502
                    format_cls=None,
 
503
                    context=DEFAULT_CONTEXT_AMOUNT):
468
504
    """Show in text form the changes from one tree to another.
469
505
 
470
506
    :param to_file: The output stream.
471
 
    :param specific_files:Include only changes to these files - None for all
 
507
    :param specific_files: Include only changes to these files - None for all
472
508
        changes.
473
 
    :param external_diff_options: If set, use an external GNU diff and pass 
 
509
    :param external_diff_options: If set, use an external GNU diff and pass
474
510
        these options.
475
511
    :param extra_trees: If set, more Trees to use for looking up file ids
476
 
    :param path_encoding: If set, the path will be encoded as specified, 
 
512
    :param path_encoding: If set, the path will be encoded as specified,
477
513
        otherwise is supposed to be utf8
478
514
    :param format_cls: Formatter class (DiffTree subclass)
479
515
    """
 
516
    if context is None:
 
517
        context = DEFAULT_CONTEXT_AMOUNT
480
518
    if format_cls is None:
481
519
        format_cls = DiffTree
482
 
    old_tree.lock_read()
483
 
    try:
 
520
    with cleanup.ExitStack() as exit_stack:
 
521
        exit_stack.enter_context(old_tree.lock_read())
484
522
        if extra_trees is not None:
485
523
            for tree in extra_trees:
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):
 
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):
504
535
    """Returns a timestamp suitable for use in a patch header."""
505
536
    try:
506
 
        mtime = tree.get_file_mtime(file_id, path)
507
 
    except errors.FileTimestampUnavailable:
 
537
        mtime = tree.get_file_mtime(path)
 
538
    except FileTimestampUnavailable:
508
539
        mtime = 0
509
540
    return timestamp.format_patch_date(mtime)
510
541
 
511
542
 
512
543
def get_executable_change(old_is_x, new_is_x):
513
 
    descr = { True:"+x", False:"-x", None:"??" }
 
544
    descr = {True: b"+x", False: b"-x", None: b"??"}
514
545
    if old_is_x != new_is_x:
515
 
        return ["%s to %s" % (descr[old_is_x], descr[new_is_x],)]
 
546
        return [b"%s to %s" % (descr[old_is_x], descr[new_is_x],)]
516
547
    else:
517
548
        return []
518
549
 
549
580
                     diff_tree.to_file, diff_tree.path_encoding)
550
581
 
551
582
    @staticmethod
552
 
    def _diff_many(differs, file_id, old_path, new_path, old_kind, new_kind):
 
583
    def _diff_many(differs, old_path, new_path, old_kind, new_kind):
553
584
        for file_differ in differs:
554
 
            result = file_differ.diff(file_id, old_path, new_path, old_kind,
555
 
                                      new_kind)
 
585
            result = file_differ.diff(old_path, new_path, old_kind, new_kind)
556
586
            if result is not DiffPath.CANNOT_DIFF:
557
587
                return result
558
588
        else:
565
595
    Represents kind change as deletion + creation.  Uses the other differs
566
596
    to do this.
567
597
    """
 
598
 
568
599
    def __init__(self, differs):
569
600
        self.differs = differs
570
601
 
575
606
    def from_diff_tree(klass, diff_tree):
576
607
        return klass(diff_tree.differs)
577
608
 
578
 
    def diff(self, file_id, old_path, new_path, old_kind, new_kind):
 
609
    def diff(self, old_path, new_path, old_kind, new_kind):
579
610
        """Perform comparison
580
611
 
581
 
        :param file_id: The file_id of the file to compare
582
612
        :param old_path: Path of the file in the old tree
583
613
        :param new_path: Path of the file in the new tree
584
614
        :param old_kind: Old file-kind of the file
586
616
        """
587
617
        if None in (old_kind, new_kind):
588
618
            return DiffPath.CANNOT_DIFF
589
 
        result = DiffPath._diff_many(self.differs, file_id, old_path,
590
 
                                       new_path, old_kind, None)
 
619
        result = DiffPath._diff_many(
 
620
            self.differs, old_path, new_path, old_kind, None)
591
621
        if result is DiffPath.CANNOT_DIFF:
592
622
            return result
593
 
        return DiffPath._diff_many(self.differs, file_id, old_path, new_path,
594
 
                                     None, new_kind)
 
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
595
640
 
596
641
 
597
642
class DiffDirectory(DiffPath):
598
643
 
599
 
    def diff(self, file_id, old_path, new_path, old_kind, new_kind):
 
644
    def diff(self, old_path, new_path, old_kind, new_kind):
600
645
        """Perform comparison between two directories.  (dummy)
601
646
 
602
647
        """
611
656
 
612
657
class DiffSymlink(DiffPath):
613
658
 
614
 
    def diff(self, file_id, old_path, new_path, old_kind, new_kind):
 
659
    def diff(self, old_path, new_path, old_kind, new_kind):
615
660
        """Perform comparison between two symlinks
616
661
 
617
 
        :param file_id: The file_id of the file to compare
618
662
        :param old_path: Path of the file in the old tree
619
663
        :param new_path: Path of the file in the new tree
620
664
        :param old_kind: Old file-kind of the file
623
667
        if 'symlink' not in (old_kind, new_kind):
624
668
            return self.CANNOT_DIFF
625
669
        if old_kind == 'symlink':
626
 
            old_target = self.old_tree.get_symlink_target(file_id)
 
670
            old_target = self.old_tree.get_symlink_target(old_path)
627
671
        elif old_kind is None:
628
672
            old_target = None
629
673
        else:
630
674
            return self.CANNOT_DIFF
631
675
        if new_kind == 'symlink':
632
 
            new_target = self.new_tree.get_symlink_target(file_id)
 
676
            new_target = self.new_tree.get_symlink_target(new_path)
633
677
        elif new_kind is None:
634
678
            new_target = None
635
679
        else:
638
682
 
639
683
    def diff_symlink(self, old_target, new_target):
640
684
        if old_target is None:
641
 
            self.to_file.write('=== target is %r\n' % new_target)
 
685
            self.to_file.write(b'=== target is \'%s\'\n' %
 
686
                               new_target.encode(self.path_encoding, 'replace'))
642
687
        elif new_target is None:
643
 
            self.to_file.write('=== target was %r\n' % old_target)
 
688
            self.to_file.write(b'=== target was \'%s\'\n' %
 
689
                               old_target.encode(self.path_encoding, 'replace'))
644
690
        else:
645
 
            self.to_file.write('=== target changed %r => %r\n' %
646
 
                              (old_target, new_target))
 
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')))
647
694
        return self.CHANGED
648
695
 
649
696
 
654
701
    EPOCH_DATE = '1970-01-01 00:00:00 +0000'
655
702
 
656
703
    def __init__(self, old_tree, new_tree, to_file, path_encoding='utf-8',
657
 
                 old_label='', new_label='', text_differ=internal_diff):
 
704
                 old_label='', new_label='', text_differ=internal_diff,
 
705
                 context_lines=DEFAULT_CONTEXT_AMOUNT):
658
706
        DiffPath.__init__(self, old_tree, new_tree, to_file, path_encoding)
659
707
        self.text_differ = text_differ
660
708
        self.old_label = old_label
661
709
        self.new_label = new_label
662
710
        self.path_encoding = path_encoding
 
711
        self.context_lines = context_lines
663
712
 
664
 
    def diff(self, file_id, old_path, new_path, old_kind, new_kind):
 
713
    def diff(self, old_path, new_path, old_kind, new_kind):
665
714
        """Compare two files in unified diff format
666
715
 
667
 
        :param file_id: The file_id of the file to compare
668
716
        :param old_path: Path of the file in the old tree
669
717
        :param new_path: Path of the file in the new tree
670
718
        :param old_kind: Old file-kind of the file
672
720
        """
673
721
        if 'file' not in (old_kind, new_kind):
674
722
            return self.CANNOT_DIFF
675
 
        from_file_id = to_file_id = file_id
676
723
        if old_kind == 'file':
677
 
            old_date = _patch_header_date(self.old_tree, file_id, old_path)
 
724
            old_date = _patch_header_date(self.old_tree, old_path)
678
725
        elif old_kind is None:
679
726
            old_date = self.EPOCH_DATE
680
 
            from_file_id = None
681
727
        else:
682
728
            return self.CANNOT_DIFF
683
729
        if new_kind == 'file':
684
 
            new_date = _patch_header_date(self.new_tree, file_id, new_path)
 
730
            new_date = _patch_header_date(self.new_tree, new_path)
685
731
        elif new_kind is None:
686
732
            new_date = self.EPOCH_DATE
687
 
            to_file_id = None
688
733
        else:
689
734
            return self.CANNOT_DIFF
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)
 
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)
694
740
 
695
 
    def diff_text(self, from_file_id, to_file_id, from_label, to_label,
696
 
        from_path=None, to_path=None):
 
741
    def diff_text(self, from_path, to_path, from_label, to_label):
697
742
        """Diff the content of given files in two trees
698
743
 
699
 
        :param from_file_id: The id of the file in the from tree.  If None,
 
744
        :param from_path: The path in the from tree. If None,
700
745
            the file is not present in the from tree.
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,
 
746
        :param to_path: The path in the to tree. This may refer
 
747
            to a different file from from_path.  If None,
703
748
            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.
706
749
        """
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:
 
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:
711
756
                return []
712
757
        try:
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)
 
758
            from_text = _get_text(self.old_tree, from_path)
 
759
            to_text = _get_text(self.new_tree, to_path)
715
760
            self.text_differ(from_label, from_text, to_label, to_text,
716
 
                             self.to_file)
 
761
                             self.to_file, path_encoding=self.path_encoding,
 
762
                             context_lines=self.context_lines)
717
763
        except errors.BinaryFile:
718
764
            self.to_file.write(
719
 
                  ("Binary files %s and %s differ\n" %
720
 
                  (from_label, to_label)).encode(self.path_encoding))
 
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'))
721
767
        return self.CHANGED
722
768
 
723
769
 
727
773
                 path_encoding='utf-8'):
728
774
        DiffPath.__init__(self, old_tree, new_tree, to_file, path_encoding)
729
775
        self.command_template = command_template
730
 
        self._root = osutils.mkdtemp(prefix='bzr-diff-')
 
776
        self._root = osutils.mkdtemp(prefix='brz-diff-')
731
777
 
732
778
    @classmethod
733
 
    def from_string(klass, command_string, old_tree, new_tree, to_file,
 
779
    def from_string(klass, command_template, old_tree, new_tree, to_file,
734
780
                    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'])
738
781
        return klass(command_template, old_tree, new_tree, to_file,
739
782
                     path_encoding)
740
783
 
741
784
    @classmethod
742
 
    def make_from_diff_tree(klass, command_string):
 
785
    def make_from_diff_tree(klass, command_string, external_diff_options=None):
743
786
        def from_diff_tree(diff_tree):
744
 
            return klass.from_string(command_string, diff_tree.old_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,
745
791
                                     diff_tree.new_tree, diff_tree.to_file)
746
792
        return from_diff_tree
747
793
 
748
794
    def _get_command(self, old_path, new_path):
749
795
        my_map = {'old_path': old_path, 'new_path': new_path}
750
 
        return [AtTemplate(t).substitute(my_map) for t in
751
 
                self.command_template]
 
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
752
808
 
753
809
    def _execute(self, old_path, new_path):
754
810
        command = self._get_command(old_path, new_path)
755
811
        try:
756
812
            proc = subprocess.Popen(command, stdout=subprocess.PIPE,
757
813
                                    cwd=self._root)
758
 
        except OSError, e:
 
814
        except OSError as e:
759
815
            if e.errno == errno.ENOENT:
760
816
                raise errors.ExecutableMissing(command[0])
761
817
            else:
762
818
                raise
763
819
        self.to_file.write(proc.stdout.read())
 
820
        proc.stdout.close()
764
821
        return proc.wait()
765
822
 
766
823
    def _try_symlink_root(self, tree, prefix):
767
 
        if (getattr(tree, 'abspath', None) is None
768
 
            or not osutils.host_os_dereferences_symlinks()):
 
824
        if (getattr(tree, 'abspath', None) is None or
 
825
                not osutils.host_os_dereferences_symlinks()):
769
826
            return False
770
827
        try:
771
828
            os.symlink(tree.abspath(''), osutils.pathjoin(self._root, prefix))
772
 
        except OSError, e:
 
829
        except OSError as e:
773
830
            if e.errno != errno.EEXIST:
774
831
                raise
775
832
        return True
776
833
 
777
 
    def _write_file(self, file_id, tree, prefix, relpath, force_temp=False,
 
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,
778
863
                    allow_write=False):
779
864
        if not force_temp and isinstance(tree, WorkingTree):
780
 
            return tree.abspath(tree.id2path(file_id))
781
 
        
782
 
        full_path = osutils.pathjoin(self._root, prefix, relpath)
 
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)
783
870
        if not force_temp and self._try_symlink_root(tree, prefix):
784
871
            return full_path
785
872
        parent_dir = osutils.dirname(full_path)
786
873
        try:
787
874
            os.makedirs(parent_dir)
788
 
        except OSError, e:
 
875
        except OSError as e:
789
876
            if e.errno != errno.EEXIST:
790
877
                raise
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:
 
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:
803
884
            pass
804
885
        else:
805
886
            os.utime(full_path, (mtime, mtime))
807
888
            osutils.make_readonly(full_path)
808
889
        return full_path
809
890
 
810
 
    def _prepare_files(self, file_id, old_path, new_path, force_temp=False,
 
891
    def _prepare_files(self, old_path, new_path, force_temp=False,
811
892
                       allow_write_new=False):
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)
 
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)
817
898
        return old_disk_path, new_disk_path
818
899
 
819
900
    def finish(self):
820
901
        try:
821
902
            osutils.rmtree(self._root)
822
 
        except OSError, e:
 
903
        except OSError as e:
823
904
            if e.errno != errno.ENOENT:
824
905
                mutter("The temporary directory \"%s\" was not "
825
 
                        "cleanly removed: %s." % (self._root, e))
 
906
                       "cleanly removed: %s." % (self._root, e))
826
907
 
827
 
    def diff(self, file_id, old_path, new_path, old_kind, new_kind):
 
908
    def diff(self, old_path, new_path, old_kind, new_kind):
828
909
        if (old_kind, new_kind) != ('file', 'file'):
829
910
            return DiffPath.CANNOT_DIFF
830
911
        (old_disk_path, new_disk_path) = self._prepare_files(
831
 
                                                file_id, old_path, new_path)
 
912
            old_path, new_path)
832
913
        self._execute(old_disk_path, new_disk_path)
833
914
 
834
 
    def edit_file(self, file_id):
 
915
    def edit_file(self, old_path, new_path):
835
916
        """Use this tool to edit a file.
836
917
 
837
918
        A temporary copy will be edited, and the new contents will be
838
919
        returned.
839
920
 
840
 
        :param file_id: The id of the file to edit.
841
921
        :return: The new contents of the file.
842
922
        """
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))
 
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)
850
926
        subprocess.call(command, cwd=self._root)
851
 
        new_file = open(new_abs_path, 'r')
852
 
        try:
 
927
        with open(new_abs_path, 'rb') as new_file:
853
928
            return new_file.read()
854
 
        finally:
855
 
            new_file.close()
856
929
 
857
930
 
858
931
class DiffTree(object):
870
943
    # list of factories that can provide instances of DiffPath objects
871
944
    # may be extended by plugins.
872
945
    diff_factories = [DiffSymlink.from_diff_tree,
873
 
                      DiffDirectory.from_diff_tree]
 
946
                      DiffDirectory.from_diff_tree,
 
947
                      DiffTreeReference.from_diff_tree]
874
948
 
875
949
    def __init__(self, old_tree, new_tree, to_file, path_encoding='utf-8',
876
950
                 diff_text=None, extra_factories=None):
886
960
            DiffPaths"""
887
961
        if diff_text is None:
888
962
            diff_text = DiffText(old_tree, new_tree, to_file, path_encoding,
889
 
                                 '', '',  internal_diff)
 
963
                                 '', '', internal_diff)
890
964
        self.old_tree = old_tree
891
965
        self.new_tree = new_tree
892
966
        self.to_file = to_file
900
974
    @classmethod
901
975
    def from_trees_options(klass, old_tree, new_tree, to_file,
902
976
                           path_encoding, external_diff_options, old_label,
903
 
                           new_label, using):
 
977
                           new_label, using, context_lines):
904
978
        """Factory for producing a DiffTree.
905
979
 
906
980
        Designed to accept options used by show_diff_trees.
 
981
 
907
982
        :param old_tree: The tree to show as old in the comparison
908
983
        :param new_tree: The tree to show as new in the comparison
909
984
        :param to_file: File to write comparisons to
915
990
        :param using: Commandline to use to invoke an external diff tool
916
991
        """
917
992
        if using is not None:
918
 
            extra_factories = [DiffFromTool.make_from_diff_tree(using)]
 
993
            extra_factories = [DiffFromTool.make_from_diff_tree(
 
994
                using, external_diff_options)]
919
995
        else:
920
996
            extra_factories = []
921
997
        if external_diff_options:
922
998
            opts = external_diff_options.split()
923
 
            def diff_file(olab, olines, nlab, nlines, to_file):
 
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
                """
924
1004
                external_diff(olab, olines, nlab, nlines, to_file, opts)
925
1005
        else:
926
1006
            diff_file = internal_diff
927
1007
        diff_text = DiffText(old_tree, new_tree, to_file, path_encoding,
928
 
                             old_label, new_label, diff_file)
 
1008
                             old_label, new_label, diff_file, context_lines=context_lines)
929
1009
        return klass(old_tree, new_tree, to_file, path_encoding, diff_text,
930
1010
                     extra_factories)
931
1011
 
945
1025
        # TODO: Generation of pseudo-diffs for added/deleted files could
946
1026
        # be usefully made into a much faster special case.
947
1027
        iterator = self.new_tree.iter_changes(self.old_tree,
948
 
                                               specific_files=specific_files,
949
 
                                               extra_trees=extra_trees,
950
 
                                               require_versioned=True)
 
1028
                                              specific_files=specific_files,
 
1029
                                              extra_trees=extra_trees,
 
1030
                                              require_versioned=True)
951
1031
        has_changes = 0
 
1032
 
952
1033
        def changes_key(change):
953
 
            old_path, new_path = change[1]
 
1034
            old_path, new_path = change.path
954
1035
            path = new_path
955
1036
            if path is None:
956
1037
                path = old_path
957
1038
            return path
 
1039
 
958
1040
        def get_encoded_path(path):
959
1041
            if path is not None:
960
1042
                return path.encode(self.path_encoding, "replace")
961
 
        for (file_id, paths, changed_content, versioned, parent, name, kind,
962
 
             executable) in sorted(iterator, key=changes_key):
 
1043
        for change in sorted(iterator, key=changes_key):
963
1044
            # The root does not get diffed, and items with no known kind (that
964
1045
            # is, missing) in both trees are skipped as well.
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])
 
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])
973
1061
 
974
1062
            properties_changed = []
975
 
            properties_changed.extend(get_executable_change(executable[0], executable[1]))
 
1063
            properties_changed.extend(
 
1064
                get_executable_change(executable[0], executable[1]))
976
1065
 
977
1066
            if properties_changed:
978
 
                prop_str = " (properties changed: %s)" % (", ".join(properties_changed),)
 
1067
                prop_str = b" (properties changed: %s)" % (
 
1068
                    b", ".join(properties_changed),)
979
1069
            else:
980
 
                prop_str = ""
 
1070
                prop_str = b""
981
1071
 
982
1072
            if (old_present, new_present) == (True, False):
983
 
                self.to_file.write("=== removed %s '%s'\n" %
984
 
                                   (kind[0], oldpath_encoded))
 
1073
                self.to_file.write(b"=== removed %s '%s'\n" %
 
1074
                                   (kind[0].encode('ascii'), oldpath_encoded))
985
1075
                newpath = oldpath
986
1076
            elif (old_present, new_present) == (False, True):
987
 
                self.to_file.write("=== added %s '%s'\n" %
988
 
                                   (kind[1], newpath_encoded))
 
1077
                self.to_file.write(b"=== added %s '%s'\n" %
 
1078
                                   (kind[1].encode('ascii'), newpath_encoded))
989
1079
                oldpath = newpath
990
1080
            elif renamed:
991
 
                self.to_file.write("=== renamed %s '%s' => '%s'%s\n" %
992
 
                    (kind[0], oldpath_encoded, newpath_encoded, prop_str))
 
1081
                self.to_file.write(b"=== renamed %s '%s' => '%s'%s\n" %
 
1082
                                   (kind[0].encode('ascii'), oldpath_encoded, newpath_encoded, prop_str))
993
1083
            else:
994
1084
                # if it was produced by iter_changes, it must be
995
1085
                # modified *somehow*, either content or execute bit.
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])
 
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])
1000
1090
                has_changes = 1
1001
1091
            if renamed:
1002
1092
                has_changes = 1
1003
1093
        return has_changes
1004
1094
 
1005
 
    def diff(self, file_id, old_path, new_path):
 
1095
    def diff(self, old_path, new_path):
1006
1096
        """Perform a diff of a single file
1007
1097
 
1008
 
        :param file_id: file-id of the file
1009
1098
        :param old_path: The path of the file in the old tree
1010
1099
        :param new_path: The path of the file in the new tree
1011
1100
        """
1012
 
        try:
1013
 
            old_kind = self.old_tree.kind(file_id)
1014
 
        except (errors.NoSuchId, errors.NoSuchFile):
 
1101
        if old_path is None:
1015
1102
            old_kind = None
1016
 
        try:
1017
 
            new_kind = self.new_tree.kind(file_id)
1018
 
        except (errors.NoSuchId, errors.NoSuchFile):
 
1103
        else:
 
1104
            old_kind = self.old_tree.kind(old_path)
 
1105
        if new_path is None:
1019
1106
            new_kind = None
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)
 
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)
1026
1114
        if result is DiffPath.CANNOT_DIFF:
1027
1115
            error_path = new_path
1028
1116
            if error_path is None: