/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: Colin D Bennett
  • Date: 2009-02-12 16:57:39 UTC
  • mto: This revision was merged to the branch mainline in revision 4008.
  • Revision ID: colin@gibibit.com-20090212165739-02xv63odccfmxomw
Generate PDF version of the User Guide.

Show diffs side-by-side

added added

removed removed

Lines of Context:
bzrlib/diff.py
1
 
# Copyright (C) 2005-2014 Canonical Ltd.
 
1
# Copyright (C) 2004, 2005, 2006 Canonical Ltd.
2
2
#
3
3
# This program is free software; you can redistribute it and/or modify
4
4
# it under the terms of the GNU General Public License as published by
12
12
#
13
13
# You should have received a copy of the GNU General Public License
14
14
# along with this program; if not, write to the Free Software
15
 
# Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
16
 
 
17
 
from __future__ import absolute_import
 
15
# Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
18
16
 
19
17
import difflib
20
18
import os
21
19
import re
22
 
import string
 
20
import shutil
23
21
import sys
24
22
 
25
 
from .lazy_import import lazy_import
 
23
from bzrlib.lazy_import import lazy_import
26
24
lazy_import(globals(), """
27
25
import errno
28
26
import subprocess
29
27
import tempfile
 
28
import time
30
29
 
31
 
from breezy import (
32
 
    cmdline,
33
 
    controldir,
 
30
from bzrlib import (
 
31
    branch as _mod_branch,
 
32
    bzrdir,
 
33
    commands,
34
34
    errors,
35
35
    osutils,
36
36
    patiencediff,
37
37
    textfile,
38
38
    timestamp,
39
 
    views,
40
39
    )
41
 
 
42
 
from breezy.workingtree import WorkingTree
43
 
from breezy.i18n import gettext
44
40
""")
45
41
 
46
 
from .registry import (
47
 
    Registry,
48
 
    )
49
 
from .sixish import text_type
50
 
from .trace import mutter, note, warning
51
 
from .tree import FileTimestampUnavailable
52
 
 
53
 
 
54
 
DEFAULT_CONTEXT_AMOUNT = 3
55
 
 
56
 
 
57
 
class AtTemplate(string.Template):
58
 
    """Templating class that uses @ instead of $."""
59
 
 
60
 
    delimiter = '@'
 
42
from bzrlib.symbol_versioning import (
 
43
        deprecated_function,
 
44
        one_three
 
45
        )
 
46
from bzrlib.trace import warning
61
47
 
62
48
 
63
49
# TODO: Rather than building a changeset object, we should probably
74
60
        self.opcodes = None
75
61
 
76
62
 
77
 
def internal_diff(old_label, oldlines, new_label, newlines, to_file,
 
63
def internal_diff(old_filename, oldlines, new_filename, newlines, to_file,
78
64
                  allow_binary=False, sequence_matcher=None,
79
 
                  path_encoding='utf8', context_lines=DEFAULT_CONTEXT_AMOUNT):
 
65
                  path_encoding='utf8'):
80
66
    # FIXME: difflib is wrong if there is no trailing newline.
81
67
    # The syntax used by patch seems to be "\ No newline at
82
68
    # end of file" following the last diff line from that
87
73
    # In the meantime we at least make sure the patch isn't
88
74
    # mangled.
89
75
 
 
76
 
 
77
    # Special workaround for Python2.3, where difflib fails if
 
78
    # both sequences are empty.
 
79
    if not oldlines and not newlines:
 
80
        return
 
81
    
90
82
    if allow_binary is False:
91
83
        textfile.check_text_lines(oldlines)
92
84
        textfile.check_text_lines(newlines)
93
85
 
94
86
    if sequence_matcher is None:
95
87
        sequence_matcher = patiencediff.PatienceSequenceMatcher
96
 
    ud = patiencediff.unified_diff_bytes(oldlines, newlines,
97
 
                                         fromfile=old_label.encode(
98
 
                                             path_encoding, 'replace'),
99
 
                                         tofile=new_label.encode(
100
 
                                             path_encoding, 'replace'),
101
 
                                         n=context_lines, sequencematcher=sequence_matcher)
 
88
    ud = patiencediff.unified_diff(oldlines, newlines,
 
89
                      fromfile=old_filename.encode(path_encoding),
 
90
                      tofile=new_filename.encode(path_encoding),
 
91
                      sequencematcher=sequence_matcher)
102
92
 
103
93
    ud = list(ud)
104
 
    if len(ud) == 0:  # Identical contents, nothing to do
 
94
    if len(ud) == 0: # Identical contents, nothing to do
105
95
        return
106
96
    # work-around for difflib being too smart for its own good
107
97
    # if /dev/null is "1,0", patch won't recognize it as /dev/null
108
98
    if not oldlines:
109
 
        ud[2] = ud[2].replace(b'-1,0', b'-0,0')
 
99
        ud[2] = ud[2].replace('-1,0', '-0,0')
110
100
    elif not newlines:
111
 
        ud[2] = ud[2].replace(b'+1,0', b'+0,0')
 
101
        ud[2] = ud[2].replace('+1,0', '+0,0')
112
102
 
113
103
    for line in ud:
114
104
        to_file.write(line)
115
 
        if not line.endswith(b'\n'):
116
 
            to_file.write(b"\n\\ No newline at end of file\n")
117
 
    to_file.write(b'\n')
 
105
        if not line.endswith('\n'):
 
106
            to_file.write("\n\\ No newline at end of file\n")
 
107
    to_file.write('\n')
118
108
 
119
109
 
120
110
def _spawn_external_diff(diffcmd, capture_errors=True):
121
 
    """Spawn the external diff process, and return the child handle.
 
111
    """Spawn the externall diff process, and return the child handle.
122
112
 
123
113
    :param diffcmd: The command list to spawn
124
114
    :param capture_errors: Capture stderr as well as setting LANG=C
146
136
                                stdout=subprocess.PIPE,
147
137
                                stderr=stderr,
148
138
                                env=env)
149
 
    except OSError as e:
 
139
    except OSError, e:
150
140
        if e.errno == errno.ENOENT:
151
141
            raise errors.NoDiff(str(e))
152
142
        raise
154
144
    return pipe
155
145
 
156
146
 
157
 
# diff style options as of GNU diff v3.2
158
 
style_option_list = ['-c', '-C', '--context',
159
 
                     '-e', '--ed',
160
 
                     '-f', '--forward-ed',
161
 
                     '-q', '--brief',
162
 
                     '--normal',
163
 
                     '-n', '--rcs',
164
 
                     '-u', '-U', '--unified',
165
 
                     '-y', '--side-by-side',
166
 
                     '-D', '--ifdef']
167
 
 
168
 
 
169
 
def default_style_unified(diff_opts):
170
 
    """Default to unified diff style if alternative not specified in diff_opts.
171
 
 
172
 
        diff only allows one style to be specified; they don't override.
173
 
        Note that some of these take optargs, and the optargs can be
174
 
        directly appended to the options.
175
 
        This is only an approximate parser; it doesn't properly understand
176
 
        the grammar.
177
 
 
178
 
    :param diff_opts: List of options for external (GNU) diff.
179
 
    :return: List of options with default style=='unified'.
180
 
    """
181
 
    for s in style_option_list:
182
 
        for j in diff_opts:
183
 
            if j.startswith(s):
184
 
                break
185
 
        else:
186
 
            continue
187
 
        break
188
 
    else:
189
 
        diff_opts.append('-u')
190
 
    return diff_opts
191
 
 
192
 
 
193
 
def external_diff(old_label, oldlines, new_label, newlines, to_file,
 
147
def external_diff(old_filename, oldlines, new_filename, newlines, to_file,
194
148
                  diff_opts):
195
149
    """Display a diff by calling out to the external diff program."""
196
150
    # make sure our own output is properly ordered before the diff
197
151
    to_file.flush()
198
152
 
199
 
    oldtmp_fd, old_abspath = tempfile.mkstemp(prefix='brz-diff-old-')
200
 
    newtmp_fd, new_abspath = tempfile.mkstemp(prefix='brz-diff-new-')
 
153
    oldtmp_fd, old_abspath = tempfile.mkstemp(prefix='bzr-diff-old-')
 
154
    newtmp_fd, new_abspath = tempfile.mkstemp(prefix='bzr-diff-new-')
201
155
    oldtmpf = os.fdopen(oldtmp_fd, 'wb')
202
156
    newtmpf = os.fdopen(newtmp_fd, 'wb')
203
157
 
217
171
 
218
172
        if not diff_opts:
219
173
            diff_opts = []
220
 
        if sys.platform == 'win32':
221
 
            # Popen doesn't do the proper encoding for external commands
222
 
            # Since we are dealing with an ANSI api, use mbcs encoding
223
 
            old_label = old_label.encode('mbcs')
224
 
            new_label = new_label.encode('mbcs')
225
174
        diffcmd = ['diff',
226
 
                   '--label', old_label,
 
175
                   '--label', old_filename,
227
176
                   old_abspath,
228
 
                   '--label', new_label,
 
177
                   '--label', new_filename,
229
178
                   new_abspath,
230
179
                   '--binary',
231
 
                   ]
232
 
 
233
 
        diff_opts = default_style_unified(diff_opts)
234
 
 
 
180
                  ]
 
181
 
 
182
        # diff only allows one style to be specified; they don't override.
 
183
        # note that some of these take optargs, and the optargs can be
 
184
        # directly appended to the options.
 
185
        # this is only an approximate parser; it doesn't properly understand
 
186
        # the grammar.
 
187
        for s in ['-c', '-u', '-C', '-U',
 
188
                  '-e', '--ed',
 
189
                  '-q', '--brief',
 
190
                  '--normal',
 
191
                  '-n', '--rcs',
 
192
                  '-y', '--side-by-side',
 
193
                  '-D', '--ifdef']:
 
194
            for j in diff_opts:
 
195
                if j.startswith(s):
 
196
                    break
 
197
            else:
 
198
                continue
 
199
            break
 
200
        else:
 
201
            diffcmd.append('-u')
 
202
                  
235
203
        if diff_opts:
236
204
            diffcmd.extend(diff_opts)
237
205
 
238
206
        pipe = _spawn_external_diff(diffcmd, capture_errors=True)
239
 
        out, err = pipe.communicate()
 
207
        out,err = pipe.communicate()
240
208
        rc = pipe.returncode
241
 
 
 
209
        
242
210
        # internal_diff() adds a trailing newline, add one here for consistency
243
 
        out += b'\n'
 
211
        out += '\n'
244
212
        if rc == 2:
245
213
            # 'diff' gives retcode == 2 for all sorts of errors
246
214
            # one of those is 'Binary files differ'.
253
221
            out, err = pipe.communicate()
254
222
 
255
223
            # Write out the new i18n diff response
256
 
            to_file.write(out + b'\n')
 
224
            to_file.write(out+'\n')
257
225
            if pipe.returncode != 2:
258
226
                raise errors.BzrError(
259
 
                    'external diff failed with exit code 2'
260
 
                    ' when run with LANG=C and LC_ALL=C,'
261
 
                    ' but not when run natively: %r' % (diffcmd,))
 
227
                               'external diff failed with exit code 2'
 
228
                               ' when run with LANG=C and LC_ALL=C,'
 
229
                               ' but not when run natively: %r' % (diffcmd,))
262
230
 
263
 
            first_line = lang_c_out.split(b'\n', 1)[0]
 
231
            first_line = lang_c_out.split('\n', 1)[0]
264
232
            # Starting with diffutils 2.8.4 the word "binary" was dropped.
265
 
            m = re.match(b'^(binary )?files.*differ$', first_line, re.I)
 
233
            m = re.match('^(binary )?files.*differ$', first_line, re.I)
266
234
            if m is None:
267
235
                raise errors.BzrError('external diff failed with exit code 2;'
268
236
                                      ' command: %r' % (diffcmd,))
279
247
                msg = 'signal %d' % (-rc)
280
248
            else:
281
249
                msg = 'exit code %d' % rc
 
250
                
 
251
            raise errors.BzrError('external diff failed with %s; command: %r' 
 
252
                                  % (rc, diffcmd))
282
253
 
283
 
            raise errors.BzrError('external diff failed with %s; command: %r'
284
 
                                  % (msg, diffcmd))
285
254
 
286
255
    finally:
287
256
        oldtmpf.close()                 # and delete
288
257
        newtmpf.close()
289
 
 
290
 
        def cleanup(path):
291
 
            # Warn in case the file couldn't be deleted (in case windows still
292
 
            # holds the file open, but not if the files have already been
293
 
            # deleted)
294
 
            try:
295
 
                os.remove(path)
296
 
            except OSError as e:
297
 
                if e.errno not in (errno.ENOENT,):
298
 
                    warning('Failed to delete temporary file: %s %s', path, e)
299
 
 
300
 
        cleanup(old_abspath)
301
 
        cleanup(new_abspath)
302
 
 
303
 
 
304
 
def get_trees_and_branches_to_diff_locked(
305
 
        path_list, revision_specs, old_url, new_url, add_cleanup, apply_view=True):
 
258
        # Clean up. Warn in case the files couldn't be deleted
 
259
        # (in case windows still holds the file open, but not
 
260
        # if the files have already been deleted)
 
261
        try:
 
262
            os.remove(old_abspath)
 
263
        except OSError, e:
 
264
            if e.errno not in (errno.ENOENT,):
 
265
                warning('Failed to delete temporary file: %s %s',
 
266
                        old_abspath, e)
 
267
        try:
 
268
            os.remove(new_abspath)
 
269
        except OSError:
 
270
            if e.errno not in (errno.ENOENT,):
 
271
                warning('Failed to delete temporary file: %s %s',
 
272
                        new_abspath, e)
 
273
 
 
274
 
 
275
def _get_trees_to_diff(path_list, revision_specs, old_url, new_url):
306
276
    """Get the trees and specific files to diff given a list of paths.
307
277
 
308
278
    This method works out the trees to be diff'ed and the files of
319
289
    :param new_url:
320
290
        The url of the new branch or tree. If None, the tree to use is
321
291
        taken from the first path, if any, or the current working tree.
322
 
    :param add_cleanup:
323
 
        a callable like Command.add_cleanup.  get_trees_and_branches_to_diff
324
 
        will register cleanups that must be run to unlock the trees, etc.
325
 
    :param apply_view:
326
 
        if True and a view is set, apply the view or check that the paths
327
 
        are within it
328
292
    :returns:
329
 
        a tuple of (old_tree, new_tree, old_branch, new_branch,
330
 
        specific_files, extra_trees) where extra_trees is a sequence of
331
 
        additional trees to search in for file-ids.  The trees and branches
332
 
        will be read-locked until the cleanups registered via the add_cleanup
333
 
        param are run.
 
293
        a tuple of (old_tree, new_tree, specific_files, extra_trees) where
 
294
        extra_trees is a sequence of additional trees to search in for
 
295
        file-ids.
334
296
    """
335
297
    # Get the old and new revision specs
336
298
    old_revision_spec = None
359
321
        default_location = path_list[0]
360
322
        other_paths = path_list[1:]
361
323
 
362
 
    def lock_tree_or_branch(wt, br):
363
 
        if wt is not None:
364
 
            wt.lock_read()
365
 
            add_cleanup(wt.unlock)
366
 
        elif br is not None:
367
 
            br.lock_read()
368
 
            add_cleanup(br.unlock)
369
 
 
370
324
    # Get the old location
371
325
    specific_files = []
372
326
    if old_url is None:
373
327
        old_url = default_location
374
328
    working_tree, branch, relpath = \
375
 
        controldir.ControlDir.open_containing_tree_or_branch(old_url)
376
 
    lock_tree_or_branch(working_tree, branch)
 
329
        bzrdir.BzrDir.open_containing_tree_or_branch(old_url)
377
330
    if consider_relpath and relpath != '':
378
 
        if working_tree is not None and apply_view:
379
 
            views.check_path_in_view(working_tree, relpath)
380
331
        specific_files.append(relpath)
381
332
    old_tree = _get_tree_to_diff(old_revision_spec, working_tree, branch)
382
 
    old_branch = branch
383
333
 
384
334
    # Get the new location
385
335
    if new_url is None:
386
336
        new_url = default_location
387
337
    if new_url != old_url:
388
338
        working_tree, branch, relpath = \
389
 
            controldir.ControlDir.open_containing_tree_or_branch(new_url)
390
 
        lock_tree_or_branch(working_tree, branch)
 
339
            bzrdir.BzrDir.open_containing_tree_or_branch(new_url)
391
340
        if consider_relpath and relpath != '':
392
 
            if working_tree is not None and apply_view:
393
 
                views.check_path_in_view(working_tree, relpath)
394
341
            specific_files.append(relpath)
395
342
    new_tree = _get_tree_to_diff(new_revision_spec, working_tree, branch,
396
 
                                 basis_is_default=working_tree is None)
397
 
    new_branch = branch
 
343
        basis_is_default=working_tree is None)
398
344
 
399
345
    # Get the specific files (all files is None, no files is [])
400
346
    if make_paths_wt_relative and working_tree is not None:
401
 
        other_paths = working_tree.safe_relpath_files(
402
 
            other_paths,
403
 
            apply_view=apply_view)
 
347
        other_paths = _relative_paths_in_tree(working_tree, other_paths)
404
348
    specific_files.extend(other_paths)
405
349
    if len(specific_files) == 0:
406
350
        specific_files = None
407
 
        if (working_tree is not None and working_tree.supports_views() and
408
 
                apply_view):
409
 
            view_files = working_tree.views.lookup_view()
410
 
            if view_files:
411
 
                specific_files = view_files
412
 
                view_str = views.view_display_str(view_files)
413
 
                note(gettext("*** Ignoring files outside view. View is %s") % view_str)
414
351
 
415
352
    # Get extra trees that ought to be searched for file-ids
416
353
    extra_trees = None
417
354
    if working_tree is not None and working_tree not in (old_tree, new_tree):
418
355
        extra_trees = (working_tree,)
419
 
    return (old_tree, new_tree, old_branch, new_branch,
420
 
            specific_files, extra_trees)
 
356
    return old_tree, new_tree, specific_files, extra_trees
421
357
 
422
358
 
423
359
def _get_tree_to_diff(spec, tree=None, branch=None, basis_is_default=True):
434
370
    return spec.as_tree(branch)
435
371
 
436
372
 
 
373
def _relative_paths_in_tree(tree, paths):
 
374
    """Get the relative paths within a working tree.
 
375
 
 
376
    Each path may be either an absolute path or a path relative to the
 
377
    current working directory.
 
378
    """
 
379
    result = []
 
380
    for filename in paths:
 
381
        try:
 
382
            result.append(tree.relpath(osutils.dereference_path(filename)))
 
383
        except errors.PathNotChild:
 
384
            raise errors.BzrCommandError("Files are in different branches")
 
385
    return result
 
386
 
 
387
 
437
388
def show_diff_trees(old_tree, new_tree, to_file, specific_files=None,
438
389
                    external_diff_options=None,
439
390
                    old_label='a/', new_label='b/',
440
391
                    extra_trees=None,
441
392
                    path_encoding='utf8',
442
 
                    using=None,
443
 
                    format_cls=None,
444
 
                    context=DEFAULT_CONTEXT_AMOUNT):
 
393
                    using=None):
445
394
    """Show in text form the changes from one tree to another.
446
395
 
447
 
    :param to_file: The output stream.
448
 
    :param specific_files: Include only changes to these files - None for all
449
 
        changes.
450
 
    :param external_diff_options: If set, use an external GNU diff and pass
451
 
        these options.
452
 
    :param extra_trees: If set, more Trees to use for looking up file ids
453
 
    :param path_encoding: If set, the path will be encoded as specified,
454
 
        otherwise is supposed to be utf8
455
 
    :param format_cls: Formatter class (DiffTree subclass)
 
396
    to_file
 
397
        The output stream.
 
398
 
 
399
    specific_files
 
400
        Include only changes to these files - None for all changes.
 
401
 
 
402
    external_diff_options
 
403
        If set, use an external GNU diff and pass these options.
 
404
 
 
405
    extra_trees
 
406
        If set, more Trees to use for looking up file ids
 
407
 
 
408
    path_encoding
 
409
        If set, the path will be encoded as specified, otherwise is supposed
 
410
        to be utf8
456
411
    """
457
 
    if context is None:
458
 
        context = DEFAULT_CONTEXT_AMOUNT
459
 
    if format_cls is None:
460
 
        format_cls = DiffTree
461
 
    with old_tree.lock_read():
 
412
    old_tree.lock_read()
 
413
    try:
462
414
        if extra_trees is not None:
463
415
            for tree in extra_trees:
464
416
                tree.lock_read()
465
417
        new_tree.lock_read()
466
418
        try:
467
 
            differ = format_cls.from_trees_options(old_tree, new_tree, to_file,
468
 
                                                   path_encoding,
469
 
                                                   external_diff_options,
470
 
                                                   old_label, new_label, using,
471
 
                                                   context_lines=context)
 
419
            differ = DiffTree.from_trees_options(old_tree, new_tree, to_file,
 
420
                                                 path_encoding,
 
421
                                                 external_diff_options,
 
422
                                                 old_label, new_label, using)
472
423
            return differ.show_diff(specific_files, extra_trees)
473
424
        finally:
474
425
            new_tree.unlock()
475
426
            if extra_trees is not None:
476
427
                for tree in extra_trees:
477
428
                    tree.unlock()
478
 
 
479
 
 
480
 
def _patch_header_date(tree, path):
 
429
    finally:
 
430
        old_tree.unlock()
 
431
 
 
432
 
 
433
def _patch_header_date(tree, file_id, path):
481
434
    """Returns a timestamp suitable for use in a patch header."""
482
 
    try:
483
 
        mtime = tree.get_file_mtime(path)
484
 
    except FileTimestampUnavailable:
485
 
        mtime = 0
 
435
    mtime = tree.get_file_mtime(file_id, path)
486
436
    return timestamp.format_patch_date(mtime)
487
437
 
488
438
 
 
439
@deprecated_function(one_three)
 
440
def get_prop_change(meta_modified):
 
441
    if meta_modified:
 
442
        return " (properties changed)"
 
443
    else:
 
444
        return  ""
 
445
 
489
446
def get_executable_change(old_is_x, new_is_x):
490
 
    descr = {True: b"+x", False: b"-x", None: b"??"}
 
447
    descr = { True:"+x", False:"-x", None:"??" }
491
448
    if old_is_x != new_is_x:
492
 
        return [b"%s to %s" % (descr[old_is_x], descr[new_is_x],)]
 
449
        return ["%s to %s" % (descr[old_is_x], descr[new_is_x],)]
493
450
    else:
494
451
        return []
495
452
 
526
483
                     diff_tree.to_file, diff_tree.path_encoding)
527
484
 
528
485
    @staticmethod
529
 
    def _diff_many(differs, old_path, new_path, old_kind, new_kind):
 
486
    def _diff_many(differs, file_id, old_path, new_path, old_kind, new_kind):
530
487
        for file_differ in differs:
531
 
            result = file_differ.diff(old_path, new_path, old_kind, new_kind)
 
488
            result = file_differ.diff(file_id, old_path, new_path, old_kind,
 
489
                                      new_kind)
532
490
            if result is not DiffPath.CANNOT_DIFF:
533
491
                return result
534
492
        else:
541
499
    Represents kind change as deletion + creation.  Uses the other differs
542
500
    to do this.
543
501
    """
544
 
 
545
502
    def __init__(self, differs):
546
503
        self.differs = differs
547
504
 
552
509
    def from_diff_tree(klass, diff_tree):
553
510
        return klass(diff_tree.differs)
554
511
 
555
 
    def diff(self, old_path, new_path, old_kind, new_kind):
 
512
    def diff(self, file_id, old_path, new_path, old_kind, new_kind):
556
513
        """Perform comparison
557
514
 
 
515
        :param file_id: The file_id of the file to compare
558
516
        :param old_path: Path of the file in the old tree
559
517
        :param new_path: Path of the file in the new tree
560
518
        :param old_kind: Old file-kind of the file
562
520
        """
563
521
        if None in (old_kind, new_kind):
564
522
            return DiffPath.CANNOT_DIFF
565
 
        result = DiffPath._diff_many(
566
 
            self.differs, old_path, new_path, old_kind, None)
 
523
        result = DiffPath._diff_many(self.differs, file_id, old_path,
 
524
                                       new_path, old_kind, None)
567
525
        if result is DiffPath.CANNOT_DIFF:
568
526
            return result
569
 
        return DiffPath._diff_many(
570
 
            self.differs, old_path, new_path, None, new_kind)
 
527
        return DiffPath._diff_many(self.differs, file_id, old_path, new_path,
 
528
                                     None, new_kind)
571
529
 
572
530
 
573
531
class DiffDirectory(DiffPath):
574
532
 
575
 
    def diff(self, old_path, new_path, old_kind, new_kind):
 
533
    def diff(self, file_id, old_path, new_path, old_kind, new_kind):
576
534
        """Perform comparison between two directories.  (dummy)
577
535
 
578
536
        """
587
545
 
588
546
class DiffSymlink(DiffPath):
589
547
 
590
 
    def diff(self, old_path, new_path, old_kind, new_kind):
 
548
    def diff(self, file_id, old_path, new_path, old_kind, new_kind):
591
549
        """Perform comparison between two symlinks
592
550
 
 
551
        :param file_id: The file_id of the file to compare
593
552
        :param old_path: Path of the file in the old tree
594
553
        :param new_path: Path of the file in the new tree
595
554
        :param old_kind: Old file-kind of the file
598
557
        if 'symlink' not in (old_kind, new_kind):
599
558
            return self.CANNOT_DIFF
600
559
        if old_kind == 'symlink':
601
 
            old_target = self.old_tree.get_symlink_target(old_path)
 
560
            old_target = self.old_tree.get_symlink_target(file_id)
602
561
        elif old_kind is None:
603
562
            old_target = None
604
563
        else:
605
564
            return self.CANNOT_DIFF
606
565
        if new_kind == 'symlink':
607
 
            new_target = self.new_tree.get_symlink_target(new_path)
 
566
            new_target = self.new_tree.get_symlink_target(file_id)
608
567
        elif new_kind is None:
609
568
            new_target = None
610
569
        else:
613
572
 
614
573
    def diff_symlink(self, old_target, new_target):
615
574
        if old_target is None:
616
 
            self.to_file.write(b'=== target is \'%s\'\n' %
617
 
                               new_target.encode(self.path_encoding, 'replace'))
 
575
            self.to_file.write('=== target is %r\n' % new_target)
618
576
        elif new_target is None:
619
 
            self.to_file.write(b'=== target was \'%s\'\n' %
620
 
                               old_target.encode(self.path_encoding, 'replace'))
 
577
            self.to_file.write('=== target was %r\n' % old_target)
621
578
        else:
622
 
            self.to_file.write(b'=== target changed \'%s\' => \'%s\'\n' %
623
 
                               (old_target.encode(self.path_encoding, 'replace'),
624
 
                                new_target.encode(self.path_encoding, 'replace')))
 
579
            self.to_file.write('=== target changed %r => %r\n' %
 
580
                              (old_target, new_target))
625
581
        return self.CHANGED
626
582
 
627
583
 
632
588
    EPOCH_DATE = '1970-01-01 00:00:00 +0000'
633
589
 
634
590
    def __init__(self, old_tree, new_tree, to_file, path_encoding='utf-8',
635
 
                 old_label='', new_label='', text_differ=internal_diff,
636
 
                 context_lines=DEFAULT_CONTEXT_AMOUNT):
 
591
                 old_label='', new_label='', text_differ=internal_diff):
637
592
        DiffPath.__init__(self, old_tree, new_tree, to_file, path_encoding)
638
593
        self.text_differ = text_differ
639
594
        self.old_label = old_label
640
595
        self.new_label = new_label
641
596
        self.path_encoding = path_encoding
642
 
        self.context_lines = context_lines
643
597
 
644
 
    def diff(self, old_path, new_path, old_kind, new_kind):
 
598
    def diff(self, file_id, old_path, new_path, old_kind, new_kind):
645
599
        """Compare two files in unified diff format
646
600
 
 
601
        :param file_id: The file_id of the file to compare
647
602
        :param old_path: Path of the file in the old tree
648
603
        :param new_path: Path of the file in the new tree
649
604
        :param old_kind: Old file-kind of the file
651
606
        """
652
607
        if 'file' not in (old_kind, new_kind):
653
608
            return self.CANNOT_DIFF
 
609
        from_file_id = to_file_id = file_id
654
610
        if old_kind == 'file':
655
 
            old_date = _patch_header_date(self.old_tree, old_path)
 
611
            old_date = _patch_header_date(self.old_tree, file_id, old_path)
656
612
        elif old_kind is None:
657
613
            old_date = self.EPOCH_DATE
 
614
            from_file_id = None
658
615
        else:
659
616
            return self.CANNOT_DIFF
660
617
        if new_kind == 'file':
661
 
            new_date = _patch_header_date(self.new_tree, new_path)
 
618
            new_date = _patch_header_date(self.new_tree, file_id, new_path)
662
619
        elif new_kind is None:
663
620
            new_date = self.EPOCH_DATE
 
621
            to_file_id = None
664
622
        else:
665
623
            return self.CANNOT_DIFF
666
 
        from_label = '%s%s\t%s' % (self.old_label, old_path,
667
 
                                   old_date)
668
 
        to_label = '%s%s\t%s' % (self.new_label, new_path,
669
 
                                 new_date)
670
 
        return self.diff_text(old_path, new_path, from_label, to_label)
 
624
        from_label = '%s%s\t%s' % (self.old_label, old_path, old_date)
 
625
        to_label = '%s%s\t%s' % (self.new_label, new_path, new_date)
 
626
        return self.diff_text(from_file_id, to_file_id, from_label, to_label)
671
627
 
672
 
    def diff_text(self, from_path, to_path, from_label, to_label):
 
628
    def diff_text(self, from_file_id, to_file_id, from_label, to_label):
673
629
        """Diff the content of given files in two trees
674
630
 
675
 
        :param from_path: The path in the from tree. If None,
 
631
        :param from_file_id: The id of the file in the from tree.  If None,
676
632
            the file is not present in the from tree.
677
 
        :param to_path: The path in the to tree. This may refer
678
 
            to a different file from from_path.  If None,
 
633
        :param to_file_id: The id of the file in the to tree.  This may refer
 
634
            to a different file from from_file_id.  If None,
679
635
            the file is not present in the to tree.
680
636
        """
681
 
        def _get_text(tree, path):
682
 
            if path is None:
683
 
                return []
684
 
            try:
685
 
                return tree.get_file_lines(path)
686
 
            except errors.NoSuchFile:
 
637
        def _get_text(tree, file_id):
 
638
            if file_id is not None:
 
639
                return tree.get_file(file_id).readlines()
 
640
            else:
687
641
                return []
688
642
        try:
689
 
            from_text = _get_text(self.old_tree, from_path)
690
 
            to_text = _get_text(self.new_tree, to_path)
 
643
            from_text = _get_text(self.old_tree, from_file_id)
 
644
            to_text = _get_text(self.new_tree, to_file_id)
691
645
            self.text_differ(from_label, from_text, to_label, to_text,
692
 
                             self.to_file, path_encoding=self.path_encoding,
693
 
                             context_lines=self.context_lines)
 
646
                             self.to_file)
694
647
        except errors.BinaryFile:
695
648
            self.to_file.write(
696
 
                ("Binary files %s and %s differ\n" %
697
 
                 (from_label, to_label)).encode(self.path_encoding, 'replace'))
 
649
                  ("Binary files %s and %s differ\n" %
 
650
                  (from_label, to_label)).encode(self.path_encoding))
698
651
        return self.CHANGED
699
652
 
700
653
 
704
657
                 path_encoding='utf-8'):
705
658
        DiffPath.__init__(self, old_tree, new_tree, to_file, path_encoding)
706
659
        self.command_template = command_template
707
 
        self._root = osutils.mkdtemp(prefix='brz-diff-')
 
660
        self._root = osutils.mkdtemp(prefix='bzr-diff-')
708
661
 
709
662
    @classmethod
710
663
    def from_string(klass, command_string, old_tree, new_tree, to_file,
711
664
                    path_encoding='utf-8'):
712
 
        command_template = cmdline.split(command_string)
713
 
        if '@' not in command_string:
714
 
            command_template.extend(['@old_path', '@new_path'])
 
665
        command_template = commands.shlex_split_unicode(command_string)
 
666
        command_template.extend(['%(old_path)s', '%(new_path)s'])
715
667
        return klass(command_template, old_tree, new_tree, to_file,
716
668
                     path_encoding)
717
669
 
718
670
    @classmethod
719
 
    def make_from_diff_tree(klass, command_string, external_diff_options=None):
 
671
    def make_from_diff_tree(klass, command_string):
720
672
        def from_diff_tree(diff_tree):
721
 
            full_command_string = [command_string]
722
 
            if external_diff_options is not None:
723
 
                full_command_string += ' ' + external_diff_options
724
 
            return klass.from_string(full_command_string, diff_tree.old_tree,
 
673
            return klass.from_string(command_string, diff_tree.old_tree,
725
674
                                     diff_tree.new_tree, diff_tree.to_file)
726
675
        return from_diff_tree
727
676
 
728
677
    def _get_command(self, old_path, new_path):
729
678
        my_map = {'old_path': old_path, 'new_path': new_path}
730
 
        command = [AtTemplate(t).substitute(my_map) for t in
731
 
                   self.command_template]
732
 
        if sys.platform == 'win32':  # Popen doesn't accept unicode on win32
733
 
            command_encoded = []
734
 
            for c in command:
735
 
                if isinstance(c, text_type):
736
 
                    command_encoded.append(c.encode('mbcs'))
737
 
                else:
738
 
                    command_encoded.append(c)
739
 
            return command_encoded
740
 
        else:
741
 
            return command
 
679
        return [t % my_map for t in self.command_template]
742
680
 
743
681
    def _execute(self, old_path, new_path):
744
682
        command = self._get_command(old_path, new_path)
745
683
        try:
746
684
            proc = subprocess.Popen(command, stdout=subprocess.PIPE,
747
685
                                    cwd=self._root)
748
 
        except OSError as e:
 
686
        except OSError, e:
749
687
            if e.errno == errno.ENOENT:
750
688
                raise errors.ExecutableMissing(command[0])
751
689
            else:
752
690
                raise
753
691
        self.to_file.write(proc.stdout.read())
754
 
        proc.stdout.close()
755
692
        return proc.wait()
756
693
 
757
694
    def _try_symlink_root(self, tree, prefix):
758
 
        if (getattr(tree, 'abspath', None) is None or
759
 
                not osutils.host_os_dereferences_symlinks()):
 
695
        if (getattr(tree, 'abspath', None) is None
 
696
            or not osutils.host_os_dereferences_symlinks()):
760
697
            return False
761
698
        try:
762
699
            os.symlink(tree.abspath(''), osutils.pathjoin(self._root, prefix))
763
 
        except OSError as e:
 
700
        except OSError, e:
764
701
            if e.errno != errno.EEXIST:
765
702
                raise
766
703
        return True
767
704
 
768
 
    @staticmethod
769
 
    def _fenc():
770
 
        """Returns safe encoding for passing file path to diff tool"""
771
 
        if sys.platform == 'win32':
772
 
            return 'mbcs'
773
 
        else:
774
 
            # Don't fallback to 'utf-8' because subprocess may not be able to
775
 
            # handle utf-8 correctly when locale is not utf-8.
776
 
            return sys.getfilesystemencoding() or 'ascii'
777
 
 
778
 
    def _is_safepath(self, path):
779
 
        """Return true if `path` may be able to pass to subprocess."""
780
 
        fenc = self._fenc()
781
 
        try:
782
 
            return path == path.encode(fenc).decode(fenc)
783
 
        except UnicodeError:
784
 
            return False
785
 
 
786
 
    def _safe_filename(self, prefix, relpath):
787
 
        """Replace unsafe character in `relpath` then join `self._root`,
788
 
        `prefix` and `relpath`."""
789
 
        fenc = self._fenc()
790
 
        # encoded_str.replace('?', '_') may break multibyte char.
791
 
        # So we should encode, decode, then replace(u'?', u'_')
792
 
        relpath_tmp = relpath.encode(fenc, 'replace').decode(fenc, 'replace')
793
 
        relpath_tmp = relpath_tmp.replace(u'?', u'_')
794
 
        return osutils.pathjoin(self._root, prefix, relpath_tmp)
795
 
 
796
 
    def _write_file(self, relpath, tree, prefix, force_temp=False,
797
 
                    allow_write=False):
798
 
        if not force_temp and isinstance(tree, WorkingTree):
799
 
            full_path = tree.abspath(relpath)
800
 
            if self._is_safepath(full_path):
801
 
                return full_path
802
 
 
803
 
        full_path = self._safe_filename(prefix, relpath)
804
 
        if not force_temp and self._try_symlink_root(tree, prefix):
 
705
    def _write_file(self, file_id, tree, prefix, relpath):
 
706
        full_path = osutils.pathjoin(self._root, prefix, relpath)
 
707
        if self._try_symlink_root(tree, prefix):
805
708
            return full_path
806
709
        parent_dir = osutils.dirname(full_path)
807
710
        try:
808
711
            os.makedirs(parent_dir)
809
 
        except OSError as e:
 
712
        except OSError, e:
810
713
            if e.errno != errno.EEXIST:
811
714
                raise
812
 
        source = tree.get_file(relpath)
 
715
        source = tree.get_file(file_id, relpath)
813
716
        try:
814
 
            with open(full_path, 'wb') as target:
 
717
            target = open(full_path, 'wb')
 
718
            try:
815
719
                osutils.pumpfile(source, target)
 
720
            finally:
 
721
                target.close()
816
722
        finally:
817
723
            source.close()
818
 
        try:
819
 
            mtime = tree.get_file_mtime(relpath)
820
 
        except FileTimestampUnavailable:
821
 
            pass
822
 
        else:
823
 
            os.utime(full_path, (mtime, mtime))
824
 
        if not allow_write:
825
 
            osutils.make_readonly(full_path)
 
724
        osutils.make_readonly(full_path)
 
725
        mtime = tree.get_file_mtime(file_id)
 
726
        os.utime(full_path, (mtime, mtime))
826
727
        return full_path
827
728
 
828
 
    def _prepare_files(self, old_path, new_path, force_temp=False,
829
 
                       allow_write_new=False):
830
 
        old_disk_path = self._write_file(
831
 
            old_path, self.old_tree, 'old', force_temp)
832
 
        new_disk_path = self._write_file(
833
 
            new_path, self.new_tree, 'new', force_temp,
834
 
            allow_write=allow_write_new)
 
729
    def _prepare_files(self, file_id, old_path, new_path):
 
730
        old_disk_path = self._write_file(file_id, self.old_tree, 'old',
 
731
                                         old_path)
 
732
        new_disk_path = self._write_file(file_id, self.new_tree, 'new',
 
733
                                         new_path)
835
734
        return old_disk_path, new_disk_path
836
735
 
837
736
    def finish(self):
838
 
        try:
839
 
            osutils.rmtree(self._root)
840
 
        except OSError as e:
841
 
            if e.errno != errno.ENOENT:
842
 
                mutter("The temporary directory \"%s\" was not "
843
 
                       "cleanly removed: %s." % (self._root, e))
 
737
        osutils.rmtree(self._root)
844
738
 
845
 
    def diff(self, old_path, new_path, old_kind, new_kind):
 
739
    def diff(self, file_id, old_path, new_path, old_kind, new_kind):
846
740
        if (old_kind, new_kind) != ('file', 'file'):
847
741
            return DiffPath.CANNOT_DIFF
848
 
        (old_disk_path, new_disk_path) = self._prepare_files(
849
 
            old_path, new_path)
850
 
        self._execute(old_disk_path, new_disk_path)
851
 
 
852
 
    def edit_file(self, old_path, new_path):
853
 
        """Use this tool to edit a file.
854
 
 
855
 
        A temporary copy will be edited, and the new contents will be
856
 
        returned.
857
 
 
858
 
        :return: The new contents of the file.
859
 
        """
860
 
        old_abs_path, new_abs_path = self._prepare_files(
861
 
            old_path, new_path, allow_write_new=True, force_temp=True)
862
 
        command = self._get_command(old_abs_path, new_abs_path)
863
 
        subprocess.call(command, cwd=self._root)
864
 
        with open(new_abs_path, 'rb') as new_file:
865
 
            return new_file.read()
 
742
        self._prepare_files(file_id, old_path, new_path)
 
743
        self._execute(osutils.pathjoin('old', old_path),
 
744
                      osutils.pathjoin('new', new_path))
866
745
 
867
746
 
868
747
class DiffTree(object):
896
775
            DiffPaths"""
897
776
        if diff_text is None:
898
777
            diff_text = DiffText(old_tree, new_tree, to_file, path_encoding,
899
 
                                 '', '', internal_diff)
 
778
                                 '', '',  internal_diff)
900
779
        self.old_tree = old_tree
901
780
        self.new_tree = new_tree
902
781
        self.to_file = to_file
910
789
    @classmethod
911
790
    def from_trees_options(klass, old_tree, new_tree, to_file,
912
791
                           path_encoding, external_diff_options, old_label,
913
 
                           new_label, using, context_lines):
 
792
                           new_label, using):
914
793
        """Factory for producing a DiffTree.
915
794
 
916
795
        Designed to accept options used by show_diff_trees.
917
 
 
918
796
        :param old_tree: The tree to show as old in the comparison
919
797
        :param new_tree: The tree to show as new in the comparison
920
798
        :param to_file: File to write comparisons to
926
804
        :param using: Commandline to use to invoke an external diff tool
927
805
        """
928
806
        if using is not None:
929
 
            extra_factories = [DiffFromTool.make_from_diff_tree(
930
 
                using, external_diff_options)]
 
807
            extra_factories = [DiffFromTool.make_from_diff_tree(using)]
931
808
        else:
932
809
            extra_factories = []
933
810
        if external_diff_options:
934
811
            opts = external_diff_options.split()
935
 
 
936
 
            def diff_file(olab, olines, nlab, nlines, to_file, path_encoding=None, context_lines=None):
937
 
                """:param path_encoding: not used but required
938
 
                        to match the signature of internal_diff.
939
 
                """
 
812
            def diff_file(olab, olines, nlab, nlines, to_file):
940
813
                external_diff(olab, olines, nlab, nlines, to_file, opts)
941
814
        else:
942
815
            diff_file = internal_diff
943
816
        diff_text = DiffText(old_tree, new_tree, to_file, path_encoding,
944
 
                             old_label, new_label, diff_file, context_lines=context_lines)
 
817
                             old_label, new_label, diff_file)
945
818
        return klass(old_tree, new_tree, to_file, path_encoding, diff_text,
946
819
                     extra_factories)
947
820
 
948
821
    def show_diff(self, specific_files, extra_trees=None):
949
822
        """Write tree diff to self.to_file
950
823
 
951
 
        :param specific_files: the specific files to compare (recursive)
 
824
        :param sepecific_files: the specific files to compare (recursive)
952
825
        :param extra_trees: extra trees to use for mapping paths to file_ids
953
826
        """
954
827
        try:
961
834
        # TODO: Generation of pseudo-diffs for added/deleted files could
962
835
        # be usefully made into a much faster special case.
963
836
        iterator = self.new_tree.iter_changes(self.old_tree,
964
 
                                              specific_files=specific_files,
965
 
                                              extra_trees=extra_trees,
966
 
                                              require_versioned=True)
 
837
                                               specific_files=specific_files,
 
838
                                               extra_trees=extra_trees,
 
839
                                               require_versioned=True)
967
840
        has_changes = 0
968
 
 
969
841
        def changes_key(change):
970
842
            old_path, new_path = change[1]
971
843
            path = new_path
972
844
            if path is None:
973
845
                path = old_path
974
846
            return path
975
 
 
976
847
        def get_encoded_path(path):
977
848
            if path is not None:
978
849
                return path.encode(self.path_encoding, "replace")
990
861
            renamed = (parent[0], name[0]) != (parent[1], name[1])
991
862
 
992
863
            properties_changed = []
993
 
            properties_changed.extend(
994
 
                get_executable_change(executable[0], executable[1]))
 
864
            properties_changed.extend(get_executable_change(executable[0], executable[1]))
995
865
 
996
866
            if properties_changed:
997
 
                prop_str = b" (properties changed: %s)" % (
998
 
                    b", ".join(properties_changed),)
 
867
                prop_str = " (properties changed: %s)" % (", ".join(properties_changed),)
999
868
            else:
1000
 
                prop_str = b""
 
869
                prop_str = ""
1001
870
 
1002
871
            if (old_present, new_present) == (True, False):
1003
 
                self.to_file.write(b"=== removed %s '%s'\n" %
1004
 
                                   (kind[0].encode('ascii'), oldpath_encoded))
 
872
                self.to_file.write("=== removed %s '%s'\n" %
 
873
                                   (kind[0], oldpath_encoded))
1005
874
                newpath = oldpath
1006
875
            elif (old_present, new_present) == (False, True):
1007
 
                self.to_file.write(b"=== added %s '%s'\n" %
1008
 
                                   (kind[1].encode('ascii'), newpath_encoded))
 
876
                self.to_file.write("=== added %s '%s'\n" %
 
877
                                   (kind[1], newpath_encoded))
1009
878
                oldpath = newpath
1010
879
            elif renamed:
1011
 
                self.to_file.write(b"=== renamed %s '%s' => '%s'%s\n" %
1012
 
                                   (kind[0].encode('ascii'), oldpath_encoded, newpath_encoded, prop_str))
 
880
                self.to_file.write("=== renamed %s '%s' => '%s'%s\n" %
 
881
                    (kind[0], oldpath_encoded, newpath_encoded, prop_str))
1013
882
            else:
1014
883
                # if it was produced by iter_changes, it must be
1015
884
                # modified *somehow*, either content or execute bit.
1016
 
                self.to_file.write(b"=== modified %s '%s'%s\n" % (kind[0].encode('ascii'),
1017
 
                                                                  newpath_encoded, prop_str))
 
885
                self.to_file.write("=== modified %s '%s'%s\n" % (kind[0],
 
886
                                   newpath_encoded, prop_str))
1018
887
            if changed_content:
1019
 
                self._diff(oldpath, newpath, kind[0], kind[1])
 
888
                self.diff(file_id, oldpath, newpath)
1020
889
                has_changes = 1
1021
890
            if renamed:
1022
891
                has_changes = 1
1023
892
        return has_changes
1024
893
 
1025
 
    def diff(self, old_path, new_path):
 
894
    def diff(self, file_id, old_path, new_path):
1026
895
        """Perform a diff of a single file
1027
896
 
 
897
        :param file_id: file-id of the file
1028
898
        :param old_path: The path of the file in the old tree
1029
899
        :param new_path: The path of the file in the new tree
1030
900
        """
1031
 
        if old_path is None:
 
901
        try:
 
902
            old_kind = self.old_tree.kind(file_id)
 
903
        except (errors.NoSuchId, errors.NoSuchFile):
1032
904
            old_kind = None
1033
 
        else:
1034
 
            old_kind = self.old_tree.kind(old_path)
1035
 
        if new_path is None:
 
905
        try:
 
906
            new_kind = self.new_tree.kind(file_id)
 
907
        except (errors.NoSuchId, errors.NoSuchFile):
1036
908
            new_kind = None
1037
 
        else:
1038
 
            new_kind = self.new_tree.kind(new_path)
1039
 
        self._diff(old_path, new_path, old_kind, new_kind)
1040
909
 
1041
 
    def _diff(self, old_path, new_path, old_kind, new_kind):
1042
 
        result = DiffPath._diff_many(
1043
 
            self.differs, old_path, new_path, old_kind, new_kind)
 
910
        result = DiffPath._diff_many(self.differs, file_id, old_path,
 
911
                                       new_path, old_kind, new_kind)
1044
912
        if result is DiffPath.CANNOT_DIFF:
1045
913
            error_path = new_path
1046
914
            if error_path is None:
1047
915
                error_path = old_path
1048
916
            raise errors.NoDiffFound(error_path)
1049
 
 
1050
 
 
1051
 
format_registry = Registry()
1052
 
format_registry.register('default', DiffTree)