/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: Martin Pool
  • Date: 2010-04-01 04:38:26 UTC
  • mto: This revision was merged to the branch mainline in revision 5128.
  • Revision ID: mbp@sourcefrog.net-20100401043826-zl8mv33nrmd1hcxp
Lazy-load sign-my-commits

Show diffs side-by-side

added added

removed removed

Lines of Context:
bzrlib/diff.py
1
 
# Copyright (C) 2005-2014 Canonical Ltd.
 
1
# Copyright (C) 2005-2010 Canonical Ltd.
2
2
#
3
3
# This program is free software; you can redistribute it and/or modify
4
4
# it under the terms of the GNU General Public License as published by
14
14
# along with this program; if not, write to the Free Software
15
15
# Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
16
16
 
17
 
from __future__ import absolute_import
18
 
 
19
17
import difflib
20
18
import os
21
19
import re
 
20
import shutil
22
21
import string
23
22
import sys
24
23
 
25
 
from .lazy_import import lazy_import
 
24
from bzrlib.lazy_import import lazy_import
26
25
lazy_import(globals(), """
27
26
import errno
28
27
import subprocess
29
28
import tempfile
 
29
import time
30
30
 
31
 
from breezy import (
 
31
from bzrlib import (
 
32
    branch as _mod_branch,
 
33
    bzrdir,
32
34
    cmdline,
33
 
    controldir,
34
35
    errors,
35
36
    osutils,
36
37
    patiencediff,
39
40
    views,
40
41
    )
41
42
 
42
 
from breezy.workingtree import WorkingTree
43
 
from breezy.i18n import gettext
 
43
from bzrlib.workingtree import WorkingTree
44
44
""")
45
45
 
46
 
from .registry import (
47
 
    Registry,
 
46
from bzrlib.symbol_versioning import (
 
47
    deprecated_function,
48
48
    )
49
 
from .sixish import text_type
50
 
from .trace import mutter, note, warning
51
 
from .tree import FileTimestampUnavailable
52
 
 
53
 
 
54
 
DEFAULT_CONTEXT_AMOUNT = 3
 
49
from bzrlib.trace import mutter, note, warning
55
50
 
56
51
 
57
52
class AtTemplate(string.Template):
74
69
        self.opcodes = None
75
70
 
76
71
 
77
 
def internal_diff(old_label, oldlines, new_label, newlines, to_file,
 
72
def internal_diff(old_filename, oldlines, new_filename, newlines, to_file,
78
73
                  allow_binary=False, sequence_matcher=None,
79
 
                  path_encoding='utf8', context_lines=DEFAULT_CONTEXT_AMOUNT):
 
74
                  path_encoding='utf8'):
80
75
    # FIXME: difflib is wrong if there is no trailing newline.
81
76
    # The syntax used by patch seems to be "\ No newline at
82
77
    # end of file" following the last diff line from that
87
82
    # In the meantime we at least make sure the patch isn't
88
83
    # mangled.
89
84
 
 
85
 
 
86
    # Special workaround for Python2.3, where difflib fails if
 
87
    # both sequences are empty.
 
88
    if not oldlines and not newlines:
 
89
        return
 
90
 
90
91
    if allow_binary is False:
91
92
        textfile.check_text_lines(oldlines)
92
93
        textfile.check_text_lines(newlines)
93
94
 
94
95
    if sequence_matcher is None:
95
96
        sequence_matcher = patiencediff.PatienceSequenceMatcher
96
 
    ud = patiencediff.unified_diff_bytes(oldlines, newlines,
97
 
                                         fromfile=old_label.encode(
98
 
                                             path_encoding, 'replace'),
99
 
                                         tofile=new_label.encode(
100
 
                                             path_encoding, 'replace'),
101
 
                                         n=context_lines, sequencematcher=sequence_matcher)
 
97
    ud = patiencediff.unified_diff(oldlines, newlines,
 
98
                      fromfile=old_filename.encode(path_encoding),
 
99
                      tofile=new_filename.encode(path_encoding),
 
100
                      sequencematcher=sequence_matcher)
102
101
 
103
102
    ud = list(ud)
104
 
    if len(ud) == 0:  # Identical contents, nothing to do
 
103
    if len(ud) == 0: # Identical contents, nothing to do
105
104
        return
106
105
    # work-around for difflib being too smart for its own good
107
106
    # if /dev/null is "1,0", patch won't recognize it as /dev/null
108
107
    if not oldlines:
109
 
        ud[2] = ud[2].replace(b'-1,0', b'-0,0')
 
108
        ud[2] = ud[2].replace('-1,0', '-0,0')
110
109
    elif not newlines:
111
 
        ud[2] = ud[2].replace(b'+1,0', b'+0,0')
 
110
        ud[2] = ud[2].replace('+1,0', '+0,0')
112
111
 
113
112
    for line in ud:
114
113
        to_file.write(line)
115
 
        if not line.endswith(b'\n'):
116
 
            to_file.write(b"\n\\ No newline at end of file\n")
117
 
    to_file.write(b'\n')
 
114
        if not line.endswith('\n'):
 
115
            to_file.write("\n\\ No newline at end of file\n")
 
116
    to_file.write('\n')
118
117
 
119
118
 
120
119
def _spawn_external_diff(diffcmd, capture_errors=True):
121
 
    """Spawn the external diff process, and return the child handle.
 
120
    """Spawn the externall diff process, and return the child handle.
122
121
 
123
122
    :param diffcmd: The command list to spawn
124
123
    :param capture_errors: Capture stderr as well as setting LANG=C
146
145
                                stdout=subprocess.PIPE,
147
146
                                stderr=stderr,
148
147
                                env=env)
149
 
    except OSError as e:
 
148
    except OSError, e:
150
149
        if e.errno == errno.ENOENT:
151
150
            raise errors.NoDiff(str(e))
152
151
        raise
154
153
    return pipe
155
154
 
156
155
 
157
 
# diff style options as of GNU diff v3.2
158
 
style_option_list = ['-c', '-C', '--context',
159
 
                     '-e', '--ed',
160
 
                     '-f', '--forward-ed',
161
 
                     '-q', '--brief',
162
 
                     '--normal',
163
 
                     '-n', '--rcs',
164
 
                     '-u', '-U', '--unified',
165
 
                     '-y', '--side-by-side',
166
 
                     '-D', '--ifdef']
167
 
 
168
 
 
169
 
def default_style_unified(diff_opts):
170
 
    """Default to unified diff style if alternative not specified in diff_opts.
171
 
 
172
 
        diff only allows one style to be specified; they don't override.
173
 
        Note that some of these take optargs, and the optargs can be
174
 
        directly appended to the options.
175
 
        This is only an approximate parser; it doesn't properly understand
176
 
        the grammar.
177
 
 
178
 
    :param diff_opts: List of options for external (GNU) diff.
179
 
    :return: List of options with default style=='unified'.
180
 
    """
181
 
    for s in style_option_list:
182
 
        for j in diff_opts:
183
 
            if j.startswith(s):
184
 
                break
185
 
        else:
186
 
            continue
187
 
        break
188
 
    else:
189
 
        diff_opts.append('-u')
190
 
    return diff_opts
191
 
 
192
 
 
193
 
def external_diff(old_label, oldlines, new_label, newlines, to_file,
 
156
def external_diff(old_filename, oldlines, new_filename, newlines, to_file,
194
157
                  diff_opts):
195
158
    """Display a diff by calling out to the external diff program."""
196
159
    # make sure our own output is properly ordered before the diff
197
160
    to_file.flush()
198
161
 
199
 
    oldtmp_fd, old_abspath = tempfile.mkstemp(prefix='brz-diff-old-')
200
 
    newtmp_fd, new_abspath = tempfile.mkstemp(prefix='brz-diff-new-')
 
162
    oldtmp_fd, old_abspath = tempfile.mkstemp(prefix='bzr-diff-old-')
 
163
    newtmp_fd, new_abspath = tempfile.mkstemp(prefix='bzr-diff-new-')
201
164
    oldtmpf = os.fdopen(oldtmp_fd, 'wb')
202
165
    newtmpf = os.fdopen(newtmp_fd, 'wb')
203
166
 
220
183
        if sys.platform == 'win32':
221
184
            # Popen doesn't do the proper encoding for external commands
222
185
            # Since we are dealing with an ANSI api, use mbcs encoding
223
 
            old_label = old_label.encode('mbcs')
224
 
            new_label = new_label.encode('mbcs')
 
186
            old_filename = old_filename.encode('mbcs')
 
187
            new_filename = new_filename.encode('mbcs')
225
188
        diffcmd = ['diff',
226
 
                   '--label', old_label,
 
189
                   '--label', old_filename,
227
190
                   old_abspath,
228
 
                   '--label', new_label,
 
191
                   '--label', new_filename,
229
192
                   new_abspath,
230
193
                   '--binary',
231
 
                   ]
 
194
                  ]
232
195
 
233
 
        diff_opts = default_style_unified(diff_opts)
 
196
        # diff only allows one style to be specified; they don't override.
 
197
        # note that some of these take optargs, and the optargs can be
 
198
        # directly appended to the options.
 
199
        # this is only an approximate parser; it doesn't properly understand
 
200
        # the grammar.
 
201
        for s in ['-c', '-u', '-C', '-U',
 
202
                  '-e', '--ed',
 
203
                  '-q', '--brief',
 
204
                  '--normal',
 
205
                  '-n', '--rcs',
 
206
                  '-y', '--side-by-side',
 
207
                  '-D', '--ifdef']:
 
208
            for j in diff_opts:
 
209
                if j.startswith(s):
 
210
                    break
 
211
            else:
 
212
                continue
 
213
            break
 
214
        else:
 
215
            diffcmd.append('-u')
234
216
 
235
217
        if diff_opts:
236
218
            diffcmd.extend(diff_opts)
237
219
 
238
220
        pipe = _spawn_external_diff(diffcmd, capture_errors=True)
239
 
        out, err = pipe.communicate()
 
221
        out,err = pipe.communicate()
240
222
        rc = pipe.returncode
241
223
 
242
224
        # internal_diff() adds a trailing newline, add one here for consistency
243
 
        out += b'\n'
 
225
        out += '\n'
244
226
        if rc == 2:
245
227
            # 'diff' gives retcode == 2 for all sorts of errors
246
228
            # one of those is 'Binary files differ'.
253
235
            out, err = pipe.communicate()
254
236
 
255
237
            # Write out the new i18n diff response
256
 
            to_file.write(out + b'\n')
 
238
            to_file.write(out+'\n')
257
239
            if pipe.returncode != 2:
258
240
                raise errors.BzrError(
259
 
                    'external diff failed with exit code 2'
260
 
                    ' when run with LANG=C and LC_ALL=C,'
261
 
                    ' but not when run natively: %r' % (diffcmd,))
 
241
                               'external diff failed with exit code 2'
 
242
                               ' when run with LANG=C and LC_ALL=C,'
 
243
                               ' but not when run natively: %r' % (diffcmd,))
262
244
 
263
 
            first_line = lang_c_out.split(b'\n', 1)[0]
 
245
            first_line = lang_c_out.split('\n', 1)[0]
264
246
            # Starting with diffutils 2.8.4 the word "binary" was dropped.
265
 
            m = re.match(b'^(binary )?files.*differ$', first_line, re.I)
 
247
            m = re.match('^(binary )?files.*differ$', first_line, re.I)
266
248
            if m is None:
267
249
                raise errors.BzrError('external diff failed with exit code 2;'
268
250
                                      ' command: %r' % (diffcmd,))
281
263
                msg = 'exit code %d' % rc
282
264
 
283
265
            raise errors.BzrError('external diff failed with %s; command: %r'
284
 
                                  % (msg, diffcmd))
 
266
                                  % (rc, diffcmd))
 
267
 
285
268
 
286
269
    finally:
287
270
        oldtmpf.close()                 # and delete
288
271
        newtmpf.close()
289
 
 
290
 
        def cleanup(path):
291
 
            # Warn in case the file couldn't be deleted (in case windows still
292
 
            # holds the file open, but not if the files have already been
293
 
            # deleted)
294
 
            try:
295
 
                os.remove(path)
296
 
            except OSError as e:
297
 
                if e.errno not in (errno.ENOENT,):
298
 
                    warning('Failed to delete temporary file: %s %s', path, e)
299
 
 
300
 
        cleanup(old_abspath)
301
 
        cleanup(new_abspath)
302
 
 
303
 
 
304
 
def get_trees_and_branches_to_diff_locked(
305
 
        path_list, revision_specs, old_url, new_url, add_cleanup, apply_view=True):
 
272
        # Clean up. Warn in case the files couldn't be deleted
 
273
        # (in case windows still holds the file open, but not
 
274
        # if the files have already been deleted)
 
275
        try:
 
276
            os.remove(old_abspath)
 
277
        except OSError, e:
 
278
            if e.errno not in (errno.ENOENT,):
 
279
                warning('Failed to delete temporary file: %s %s',
 
280
                        old_abspath, e)
 
281
        try:
 
282
            os.remove(new_abspath)
 
283
        except OSError:
 
284
            if e.errno not in (errno.ENOENT,):
 
285
                warning('Failed to delete temporary file: %s %s',
 
286
                        new_abspath, e)
 
287
 
 
288
 
 
289
def get_trees_and_branches_to_diff(path_list, revision_specs, old_url, new_url,
 
290
                                   apply_view=True):
306
291
    """Get the trees and specific files to diff given a list of paths.
307
292
 
308
293
    This method works out the trees to be diff'ed and the files of
319
304
    :param new_url:
320
305
        The url of the new branch or tree. If None, the tree to use is
321
306
        taken from the first path, if any, or the current working tree.
322
 
    :param add_cleanup:
323
 
        a callable like Command.add_cleanup.  get_trees_and_branches_to_diff
324
 
        will register cleanups that must be run to unlock the trees, etc.
325
307
    :param apply_view:
326
308
        if True and a view is set, apply the view or check that the paths
327
309
        are within it
328
310
    :returns:
329
311
        a tuple of (old_tree, new_tree, old_branch, new_branch,
330
312
        specific_files, extra_trees) where extra_trees is a sequence of
331
 
        additional trees to search in for file-ids.  The trees and branches
332
 
        will be read-locked until the cleanups registered via the add_cleanup
333
 
        param are run.
 
313
        additional trees to search in for file-ids.
334
314
    """
335
315
    # Get the old and new revision specs
336
316
    old_revision_spec = None
359
339
        default_location = path_list[0]
360
340
        other_paths = path_list[1:]
361
341
 
362
 
    def lock_tree_or_branch(wt, br):
363
 
        if wt is not None:
364
 
            wt.lock_read()
365
 
            add_cleanup(wt.unlock)
366
 
        elif br is not None:
367
 
            br.lock_read()
368
 
            add_cleanup(br.unlock)
369
 
 
370
342
    # Get the old location
371
343
    specific_files = []
372
344
    if old_url is None:
373
345
        old_url = default_location
374
346
    working_tree, branch, relpath = \
375
 
        controldir.ControlDir.open_containing_tree_or_branch(old_url)
376
 
    lock_tree_or_branch(working_tree, branch)
 
347
        bzrdir.BzrDir.open_containing_tree_or_branch(old_url)
377
348
    if consider_relpath and relpath != '':
378
349
        if working_tree is not None and apply_view:
379
350
            views.check_path_in_view(working_tree, relpath)
386
357
        new_url = default_location
387
358
    if new_url != old_url:
388
359
        working_tree, branch, relpath = \
389
 
            controldir.ControlDir.open_containing_tree_or_branch(new_url)
390
 
        lock_tree_or_branch(working_tree, branch)
 
360
            bzrdir.BzrDir.open_containing_tree_or_branch(new_url)
391
361
        if consider_relpath and relpath != '':
392
362
            if working_tree is not None and apply_view:
393
363
                views.check_path_in_view(working_tree, relpath)
394
364
            specific_files.append(relpath)
395
365
    new_tree = _get_tree_to_diff(new_revision_spec, working_tree, branch,
396
 
                                 basis_is_default=working_tree is None)
 
366
        basis_is_default=working_tree is None)
397
367
    new_branch = branch
398
368
 
399
369
    # Get the specific files (all files is None, no files is [])
400
370
    if make_paths_wt_relative and working_tree is not None:
401
 
        other_paths = working_tree.safe_relpath_files(
402
 
            other_paths,
 
371
        try:
 
372
            from bzrlib.builtins import safe_relpath_files
 
373
            other_paths = safe_relpath_files(working_tree, other_paths,
403
374
            apply_view=apply_view)
 
375
        except errors.FileInWrongBranch:
 
376
            raise errors.BzrCommandError("Files are in different branches")
404
377
    specific_files.extend(other_paths)
405
378
    if len(specific_files) == 0:
406
379
        specific_files = None
407
 
        if (working_tree is not None and working_tree.supports_views() and
408
 
                apply_view):
 
380
        if (working_tree is not None and working_tree.supports_views()
 
381
            and apply_view):
409
382
            view_files = working_tree.views.lookup_view()
410
383
            if view_files:
411
384
                specific_files = view_files
412
385
                view_str = views.view_display_str(view_files)
413
 
                note(gettext("*** Ignoring files outside view. View is %s") % view_str)
 
386
                note("*** Ignoring files outside view. View is %s" % view_str)
414
387
 
415
388
    # Get extra trees that ought to be searched for file-ids
416
389
    extra_trees = None
417
390
    if working_tree is not None and working_tree not in (old_tree, new_tree):
418
391
        extra_trees = (working_tree,)
419
 
    return (old_tree, new_tree, old_branch, new_branch,
420
 
            specific_files, extra_trees)
 
392
    return old_tree, new_tree, old_branch, new_branch, specific_files, extra_trees
421
393
 
422
394
 
423
395
def _get_tree_to_diff(spec, tree=None, branch=None, basis_is_default=True):
439
411
                    old_label='a/', new_label='b/',
440
412
                    extra_trees=None,
441
413
                    path_encoding='utf8',
442
 
                    using=None,
443
 
                    format_cls=None,
444
 
                    context=DEFAULT_CONTEXT_AMOUNT):
 
414
                    using=None):
445
415
    """Show in text form the changes from one tree to another.
446
416
 
447
 
    :param to_file: The output stream.
448
 
    :param specific_files: Include only changes to these files - None for all
449
 
        changes.
450
 
    :param external_diff_options: If set, use an external GNU diff and pass
451
 
        these options.
452
 
    :param extra_trees: If set, more Trees to use for looking up file ids
453
 
    :param path_encoding: If set, the path will be encoded as specified,
454
 
        otherwise is supposed to be utf8
455
 
    :param format_cls: Formatter class (DiffTree subclass)
 
417
    to_file
 
418
        The output stream.
 
419
 
 
420
    specific_files
 
421
        Include only changes to these files - None for all changes.
 
422
 
 
423
    external_diff_options
 
424
        If set, use an external GNU diff and pass these options.
 
425
 
 
426
    extra_trees
 
427
        If set, more Trees to use for looking up file ids
 
428
 
 
429
    path_encoding
 
430
        If set, the path will be encoded as specified, otherwise is supposed
 
431
        to be utf8
456
432
    """
457
 
    if context is None:
458
 
        context = DEFAULT_CONTEXT_AMOUNT
459
 
    if format_cls is None:
460
 
        format_cls = DiffTree
461
 
    with old_tree.lock_read():
 
433
    old_tree.lock_read()
 
434
    try:
462
435
        if extra_trees is not None:
463
436
            for tree in extra_trees:
464
437
                tree.lock_read()
465
438
        new_tree.lock_read()
466
439
        try:
467
 
            differ = format_cls.from_trees_options(old_tree, new_tree, to_file,
468
 
                                                   path_encoding,
469
 
                                                   external_diff_options,
470
 
                                                   old_label, new_label, using,
471
 
                                                   context_lines=context)
 
440
            differ = DiffTree.from_trees_options(old_tree, new_tree, to_file,
 
441
                                                 path_encoding,
 
442
                                                 external_diff_options,
 
443
                                                 old_label, new_label, using)
472
444
            return differ.show_diff(specific_files, extra_trees)
473
445
        finally:
474
446
            new_tree.unlock()
475
447
            if extra_trees is not None:
476
448
                for tree in extra_trees:
477
449
                    tree.unlock()
478
 
 
479
 
 
480
 
def _patch_header_date(tree, path):
 
450
    finally:
 
451
        old_tree.unlock()
 
452
 
 
453
 
 
454
def _patch_header_date(tree, file_id, path):
481
455
    """Returns a timestamp suitable for use in a patch header."""
482
456
    try:
483
 
        mtime = tree.get_file_mtime(path)
484
 
    except FileTimestampUnavailable:
 
457
        mtime = tree.get_file_mtime(file_id, path)
 
458
    except errors.FileTimestampUnavailable:
485
459
        mtime = 0
486
460
    return timestamp.format_patch_date(mtime)
487
461
 
488
462
 
489
463
def get_executable_change(old_is_x, new_is_x):
490
 
    descr = {True: b"+x", False: b"-x", None: b"??"}
 
464
    descr = { True:"+x", False:"-x", None:"??" }
491
465
    if old_is_x != new_is_x:
492
 
        return [b"%s to %s" % (descr[old_is_x], descr[new_is_x],)]
 
466
        return ["%s to %s" % (descr[old_is_x], descr[new_is_x],)]
493
467
    else:
494
468
        return []
495
469
 
526
500
                     diff_tree.to_file, diff_tree.path_encoding)
527
501
 
528
502
    @staticmethod
529
 
    def _diff_many(differs, old_path, new_path, old_kind, new_kind):
 
503
    def _diff_many(differs, file_id, old_path, new_path, old_kind, new_kind):
530
504
        for file_differ in differs:
531
 
            result = file_differ.diff(old_path, new_path, old_kind, new_kind)
 
505
            result = file_differ.diff(file_id, old_path, new_path, old_kind,
 
506
                                      new_kind)
532
507
            if result is not DiffPath.CANNOT_DIFF:
533
508
                return result
534
509
        else:
541
516
    Represents kind change as deletion + creation.  Uses the other differs
542
517
    to do this.
543
518
    """
544
 
 
545
519
    def __init__(self, differs):
546
520
        self.differs = differs
547
521
 
552
526
    def from_diff_tree(klass, diff_tree):
553
527
        return klass(diff_tree.differs)
554
528
 
555
 
    def diff(self, old_path, new_path, old_kind, new_kind):
 
529
    def diff(self, file_id, old_path, new_path, old_kind, new_kind):
556
530
        """Perform comparison
557
531
 
 
532
        :param file_id: The file_id of the file to compare
558
533
        :param old_path: Path of the file in the old tree
559
534
        :param new_path: Path of the file in the new tree
560
535
        :param old_kind: Old file-kind of the file
562
537
        """
563
538
        if None in (old_kind, new_kind):
564
539
            return DiffPath.CANNOT_DIFF
565
 
        result = DiffPath._diff_many(
566
 
            self.differs, old_path, new_path, old_kind, None)
 
540
        result = DiffPath._diff_many(self.differs, file_id, old_path,
 
541
                                       new_path, old_kind, None)
567
542
        if result is DiffPath.CANNOT_DIFF:
568
543
            return result
569
 
        return DiffPath._diff_many(
570
 
            self.differs, old_path, new_path, None, new_kind)
 
544
        return DiffPath._diff_many(self.differs, file_id, old_path, new_path,
 
545
                                     None, new_kind)
571
546
 
572
547
 
573
548
class DiffDirectory(DiffPath):
574
549
 
575
 
    def diff(self, old_path, new_path, old_kind, new_kind):
 
550
    def diff(self, file_id, old_path, new_path, old_kind, new_kind):
576
551
        """Perform comparison between two directories.  (dummy)
577
552
 
578
553
        """
587
562
 
588
563
class DiffSymlink(DiffPath):
589
564
 
590
 
    def diff(self, old_path, new_path, old_kind, new_kind):
 
565
    def diff(self, file_id, old_path, new_path, old_kind, new_kind):
591
566
        """Perform comparison between two symlinks
592
567
 
 
568
        :param file_id: The file_id of the file to compare
593
569
        :param old_path: Path of the file in the old tree
594
570
        :param new_path: Path of the file in the new tree
595
571
        :param old_kind: Old file-kind of the file
598
574
        if 'symlink' not in (old_kind, new_kind):
599
575
            return self.CANNOT_DIFF
600
576
        if old_kind == 'symlink':
601
 
            old_target = self.old_tree.get_symlink_target(old_path)
 
577
            old_target = self.old_tree.get_symlink_target(file_id)
602
578
        elif old_kind is None:
603
579
            old_target = None
604
580
        else:
605
581
            return self.CANNOT_DIFF
606
582
        if new_kind == 'symlink':
607
 
            new_target = self.new_tree.get_symlink_target(new_path)
 
583
            new_target = self.new_tree.get_symlink_target(file_id)
608
584
        elif new_kind is None:
609
585
            new_target = None
610
586
        else:
613
589
 
614
590
    def diff_symlink(self, old_target, new_target):
615
591
        if old_target is None:
616
 
            self.to_file.write(b'=== target is \'%s\'\n' %
617
 
                               new_target.encode(self.path_encoding, 'replace'))
 
592
            self.to_file.write('=== target is %r\n' % new_target)
618
593
        elif new_target is None:
619
 
            self.to_file.write(b'=== target was \'%s\'\n' %
620
 
                               old_target.encode(self.path_encoding, 'replace'))
 
594
            self.to_file.write('=== target was %r\n' % old_target)
621
595
        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')))
 
596
            self.to_file.write('=== target changed %r => %r\n' %
 
597
                              (old_target, new_target))
625
598
        return self.CHANGED
626
599
 
627
600
 
632
605
    EPOCH_DATE = '1970-01-01 00:00:00 +0000'
633
606
 
634
607
    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):
 
608
                 old_label='', new_label='', text_differ=internal_diff):
637
609
        DiffPath.__init__(self, old_tree, new_tree, to_file, path_encoding)
638
610
        self.text_differ = text_differ
639
611
        self.old_label = old_label
640
612
        self.new_label = new_label
641
613
        self.path_encoding = path_encoding
642
 
        self.context_lines = context_lines
643
614
 
644
 
    def diff(self, old_path, new_path, old_kind, new_kind):
 
615
    def diff(self, file_id, old_path, new_path, old_kind, new_kind):
645
616
        """Compare two files in unified diff format
646
617
 
 
618
        :param file_id: The file_id of the file to compare
647
619
        :param old_path: Path of the file in the old tree
648
620
        :param new_path: Path of the file in the new tree
649
621
        :param old_kind: Old file-kind of the file
651
623
        """
652
624
        if 'file' not in (old_kind, new_kind):
653
625
            return self.CANNOT_DIFF
 
626
        from_file_id = to_file_id = file_id
654
627
        if old_kind == 'file':
655
 
            old_date = _patch_header_date(self.old_tree, old_path)
 
628
            old_date = _patch_header_date(self.old_tree, file_id, old_path)
656
629
        elif old_kind is None:
657
630
            old_date = self.EPOCH_DATE
 
631
            from_file_id = None
658
632
        else:
659
633
            return self.CANNOT_DIFF
660
634
        if new_kind == 'file':
661
 
            new_date = _patch_header_date(self.new_tree, new_path)
 
635
            new_date = _patch_header_date(self.new_tree, file_id, new_path)
662
636
        elif new_kind is None:
663
637
            new_date = self.EPOCH_DATE
 
638
            to_file_id = None
664
639
        else:
665
640
            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)
 
641
        from_label = '%s%s\t%s' % (self.old_label, old_path, old_date)
 
642
        to_label = '%s%s\t%s' % (self.new_label, new_path, new_date)
 
643
        return self.diff_text(from_file_id, to_file_id, from_label, to_label,
 
644
            old_path, new_path)
671
645
 
672
 
    def diff_text(self, from_path, to_path, from_label, to_label):
 
646
    def diff_text(self, from_file_id, to_file_id, from_label, to_label,
 
647
        from_path=None, to_path=None):
673
648
        """Diff the content of given files in two trees
674
649
 
675
 
        :param from_path: The path in the from tree. If None,
 
650
        :param from_file_id: The id of the file in the from tree.  If None,
676
651
            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,
 
652
        :param to_file_id: The id of the file in the to tree.  This may refer
 
653
            to a different file from from_file_id.  If None,
679
654
            the file is not present in the to tree.
 
655
        :param from_path: The path in the from tree or None if unknown.
 
656
        :param to_path: The path in the to tree or None if unknown.
680
657
        """
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:
 
658
        def _get_text(tree, file_id, path):
 
659
            if file_id is not None:
 
660
                return tree.get_file(file_id, path).readlines()
 
661
            else:
687
662
                return []
688
663
        try:
689
 
            from_text = _get_text(self.old_tree, from_path)
690
 
            to_text = _get_text(self.new_tree, to_path)
 
664
            from_text = _get_text(self.old_tree, from_file_id, from_path)
 
665
            to_text = _get_text(self.new_tree, to_file_id, to_path)
691
666
            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)
 
667
                             self.to_file)
694
668
        except errors.BinaryFile:
695
669
            self.to_file.write(
696
 
                ("Binary files %s and %s differ\n" %
697
 
                 (from_label, to_label)).encode(self.path_encoding, 'replace'))
 
670
                  ("Binary files %s and %s differ\n" %
 
671
                  (from_label, to_label)).encode(self.path_encoding))
698
672
        return self.CHANGED
699
673
 
700
674
 
704
678
                 path_encoding='utf-8'):
705
679
        DiffPath.__init__(self, old_tree, new_tree, to_file, path_encoding)
706
680
        self.command_template = command_template
707
 
        self._root = osutils.mkdtemp(prefix='brz-diff-')
 
681
        self._root = osutils.mkdtemp(prefix='bzr-diff-')
708
682
 
709
683
    @classmethod
710
684
    def from_string(klass, command_string, old_tree, new_tree, to_file,
716
690
                     path_encoding)
717
691
 
718
692
    @classmethod
719
 
    def make_from_diff_tree(klass, command_string, external_diff_options=None):
 
693
    def make_from_diff_tree(klass, command_string):
720
694
        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,
 
695
            return klass.from_string(command_string, diff_tree.old_tree,
725
696
                                     diff_tree.new_tree, diff_tree.to_file)
726
697
        return from_diff_tree
727
698
 
728
699
    def _get_command(self, old_path, new_path):
729
700
        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
 
701
        return [AtTemplate(t).substitute(my_map) for t in
 
702
                self.command_template]
742
703
 
743
704
    def _execute(self, old_path, new_path):
744
705
        command = self._get_command(old_path, new_path)
745
706
        try:
746
707
            proc = subprocess.Popen(command, stdout=subprocess.PIPE,
747
708
                                    cwd=self._root)
748
 
        except OSError as e:
 
709
        except OSError, e:
749
710
            if e.errno == errno.ENOENT:
750
711
                raise errors.ExecutableMissing(command[0])
751
712
            else:
752
713
                raise
753
714
        self.to_file.write(proc.stdout.read())
754
 
        proc.stdout.close()
755
715
        return proc.wait()
756
716
 
757
717
    def _try_symlink_root(self, tree, prefix):
758
 
        if (getattr(tree, 'abspath', None) is None or
759
 
                not osutils.host_os_dereferences_symlinks()):
 
718
        if (getattr(tree, 'abspath', None) is None
 
719
            or not osutils.host_os_dereferences_symlinks()):
760
720
            return False
761
721
        try:
762
722
            os.symlink(tree.abspath(''), osutils.pathjoin(self._root, prefix))
763
 
        except OSError as e:
 
723
        except OSError, e:
764
724
            if e.errno != errno.EEXIST:
765
725
                raise
766
726
        return True
767
727
 
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,
 
728
    def _write_file(self, file_id, tree, prefix, relpath, force_temp=False,
797
729
                    allow_write=False):
798
730
        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)
 
731
            return tree.abspath(tree.id2path(file_id))
 
732
        
 
733
        full_path = osutils.pathjoin(self._root, prefix, relpath)
804
734
        if not force_temp and self._try_symlink_root(tree, prefix):
805
735
            return full_path
806
736
        parent_dir = osutils.dirname(full_path)
807
737
        try:
808
738
            os.makedirs(parent_dir)
809
 
        except OSError as e:
 
739
        except OSError, e:
810
740
            if e.errno != errno.EEXIST:
811
741
                raise
812
 
        source = tree.get_file(relpath)
 
742
        source = tree.get_file(file_id, relpath)
813
743
        try:
814
 
            with open(full_path, 'wb') as target:
 
744
            target = open(full_path, 'wb')
 
745
            try:
815
746
                osutils.pumpfile(source, target)
 
747
            finally:
 
748
                target.close()
816
749
        finally:
817
750
            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
751
        if not allow_write:
825
752
            osutils.make_readonly(full_path)
 
753
        try:
 
754
            mtime = tree.get_file_mtime(file_id)
 
755
        except errors.FileTimestampUnavailable:
 
756
            mtime = 0
 
757
        os.utime(full_path, (mtime, mtime))
826
758
        return full_path
827
759
 
828
 
    def _prepare_files(self, old_path, new_path, force_temp=False,
 
760
    def _prepare_files(self, file_id, old_path, new_path, force_temp=False,
829
761
                       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)
 
762
        old_disk_path = self._write_file(file_id, self.old_tree, 'old',
 
763
                                         old_path, force_temp)
 
764
        new_disk_path = self._write_file(file_id, self.new_tree, 'new',
 
765
                                         new_path, force_temp,
 
766
                                         allow_write=allow_write_new)
835
767
        return old_disk_path, new_disk_path
836
768
 
837
769
    def finish(self):
838
770
        try:
839
771
            osutils.rmtree(self._root)
840
 
        except OSError as e:
 
772
        except OSError, e:
841
773
            if e.errno != errno.ENOENT:
842
774
                mutter("The temporary directory \"%s\" was not "
843
 
                       "cleanly removed: %s." % (self._root, e))
 
775
                        "cleanly removed: %s." % (self._root, e))
844
776
 
845
 
    def diff(self, old_path, new_path, old_kind, new_kind):
 
777
    def diff(self, file_id, old_path, new_path, old_kind, new_kind):
846
778
        if (old_kind, new_kind) != ('file', 'file'):
847
779
            return DiffPath.CANNOT_DIFF
848
780
        (old_disk_path, new_disk_path) = self._prepare_files(
849
 
            old_path, new_path)
 
781
                                                file_id, old_path, new_path)
850
782
        self._execute(old_disk_path, new_disk_path)
851
783
 
852
 
    def edit_file(self, old_path, new_path):
 
784
    def edit_file(self, file_id):
853
785
        """Use this tool to edit a file.
854
786
 
855
787
        A temporary copy will be edited, and the new contents will be
856
788
        returned.
857
789
 
 
790
        :param file_id: The id of the file to edit.
858
791
        :return: The new contents of the file.
859
792
        """
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)
 
793
        old_path = self.old_tree.id2path(file_id)
 
794
        new_path = self.new_tree.id2path(file_id)
 
795
        new_abs_path = self._prepare_files(file_id, old_path, new_path,
 
796
                                           allow_write_new=True,
 
797
                                           force_temp=True)[1]
 
798
        command = self._get_command(osutils.pathjoin('old', old_path),
 
799
                                    osutils.pathjoin('new', new_path))
863
800
        subprocess.call(command, cwd=self._root)
864
 
        with open(new_abs_path, 'rb') as new_file:
 
801
        new_file = open(new_abs_path, 'r')
 
802
        try:
865
803
            return new_file.read()
 
804
        finally:
 
805
            new_file.close()
866
806
 
867
807
 
868
808
class DiffTree(object):
896
836
            DiffPaths"""
897
837
        if diff_text is None:
898
838
            diff_text = DiffText(old_tree, new_tree, to_file, path_encoding,
899
 
                                 '', '', internal_diff)
 
839
                                 '', '',  internal_diff)
900
840
        self.old_tree = old_tree
901
841
        self.new_tree = new_tree
902
842
        self.to_file = to_file
910
850
    @classmethod
911
851
    def from_trees_options(klass, old_tree, new_tree, to_file,
912
852
                           path_encoding, external_diff_options, old_label,
913
 
                           new_label, using, context_lines):
 
853
                           new_label, using):
914
854
        """Factory for producing a DiffTree.
915
855
 
916
856
        Designed to accept options used by show_diff_trees.
917
 
 
918
857
        :param old_tree: The tree to show as old in the comparison
919
858
        :param new_tree: The tree to show as new in the comparison
920
859
        :param to_file: File to write comparisons to
926
865
        :param using: Commandline to use to invoke an external diff tool
927
866
        """
928
867
        if using is not None:
929
 
            extra_factories = [DiffFromTool.make_from_diff_tree(
930
 
                using, external_diff_options)]
 
868
            extra_factories = [DiffFromTool.make_from_diff_tree(using)]
931
869
        else:
932
870
            extra_factories = []
933
871
        if external_diff_options:
934
872
            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
 
                """
 
873
            def diff_file(olab, olines, nlab, nlines, to_file):
940
874
                external_diff(olab, olines, nlab, nlines, to_file, opts)
941
875
        else:
942
876
            diff_file = internal_diff
943
877
        diff_text = DiffText(old_tree, new_tree, to_file, path_encoding,
944
 
                             old_label, new_label, diff_file, context_lines=context_lines)
 
878
                             old_label, new_label, diff_file)
945
879
        return klass(old_tree, new_tree, to_file, path_encoding, diff_text,
946
880
                     extra_factories)
947
881
 
948
882
    def show_diff(self, specific_files, extra_trees=None):
949
883
        """Write tree diff to self.to_file
950
884
 
951
 
        :param specific_files: the specific files to compare (recursive)
 
885
        :param sepecific_files: the specific files to compare (recursive)
952
886
        :param extra_trees: extra trees to use for mapping paths to file_ids
953
887
        """
954
888
        try:
961
895
        # TODO: Generation of pseudo-diffs for added/deleted files could
962
896
        # be usefully made into a much faster special case.
963
897
        iterator = self.new_tree.iter_changes(self.old_tree,
964
 
                                              specific_files=specific_files,
965
 
                                              extra_trees=extra_trees,
966
 
                                              require_versioned=True)
 
898
                                               specific_files=specific_files,
 
899
                                               extra_trees=extra_trees,
 
900
                                               require_versioned=True)
967
901
        has_changes = 0
968
 
 
969
902
        def changes_key(change):
970
903
            old_path, new_path = change[1]
971
904
            path = new_path
972
905
            if path is None:
973
906
                path = old_path
974
907
            return path
975
 
 
976
908
        def get_encoded_path(path):
977
909
            if path is not None:
978
910
                return path.encode(self.path_encoding, "replace")
990
922
            renamed = (parent[0], name[0]) != (parent[1], name[1])
991
923
 
992
924
            properties_changed = []
993
 
            properties_changed.extend(
994
 
                get_executable_change(executable[0], executable[1]))
 
925
            properties_changed.extend(get_executable_change(executable[0], executable[1]))
995
926
 
996
927
            if properties_changed:
997
 
                prop_str = b" (properties changed: %s)" % (
998
 
                    b", ".join(properties_changed),)
 
928
                prop_str = " (properties changed: %s)" % (", ".join(properties_changed),)
999
929
            else:
1000
 
                prop_str = b""
 
930
                prop_str = ""
1001
931
 
1002
932
            if (old_present, new_present) == (True, False):
1003
 
                self.to_file.write(b"=== removed %s '%s'\n" %
1004
 
                                   (kind[0].encode('ascii'), oldpath_encoded))
 
933
                self.to_file.write("=== removed %s '%s'\n" %
 
934
                                   (kind[0], oldpath_encoded))
1005
935
                newpath = oldpath
1006
936
            elif (old_present, new_present) == (False, True):
1007
 
                self.to_file.write(b"=== added %s '%s'\n" %
1008
 
                                   (kind[1].encode('ascii'), newpath_encoded))
 
937
                self.to_file.write("=== added %s '%s'\n" %
 
938
                                   (kind[1], newpath_encoded))
1009
939
                oldpath = newpath
1010
940
            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))
 
941
                self.to_file.write("=== renamed %s '%s' => '%s'%s\n" %
 
942
                    (kind[0], oldpath_encoded, newpath_encoded, prop_str))
1013
943
            else:
1014
944
                # if it was produced by iter_changes, it must be
1015
945
                # 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))
 
946
                self.to_file.write("=== modified %s '%s'%s\n" % (kind[0],
 
947
                                   newpath_encoded, prop_str))
1018
948
            if changed_content:
1019
 
                self._diff(oldpath, newpath, kind[0], kind[1])
 
949
                self._diff(file_id, oldpath, newpath, kind[0], kind[1])
1020
950
                has_changes = 1
1021
951
            if renamed:
1022
952
                has_changes = 1
1023
953
        return has_changes
1024
954
 
1025
 
    def diff(self, old_path, new_path):
 
955
    def diff(self, file_id, old_path, new_path):
1026
956
        """Perform a diff of a single file
1027
957
 
 
958
        :param file_id: file-id of the file
1028
959
        :param old_path: The path of the file in the old tree
1029
960
        :param new_path: The path of the file in the new tree
1030
961
        """
1031
 
        if old_path is None:
 
962
        try:
 
963
            old_kind = self.old_tree.kind(file_id)
 
964
        except (errors.NoSuchId, errors.NoSuchFile):
1032
965
            old_kind = None
1033
 
        else:
1034
 
            old_kind = self.old_tree.kind(old_path)
1035
 
        if new_path is None:
 
966
        try:
 
967
            new_kind = self.new_tree.kind(file_id)
 
968
        except (errors.NoSuchId, errors.NoSuchFile):
1036
969
            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
 
 
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)
 
970
        self._diff(file_id, old_path, new_path, old_kind, new_kind)
 
971
 
 
972
 
 
973
    def _diff(self, file_id, old_path, new_path, old_kind, new_kind):
 
974
        result = DiffPath._diff_many(self.differs, file_id, old_path,
 
975
                                       new_path, old_kind, new_kind)
1044
976
        if result is DiffPath.CANNOT_DIFF:
1045
977
            error_path = new_path
1046
978
            if error_path is None:
1047
979
                error_path = old_path
1048
980
            raise errors.NoDiffFound(error_path)
1049
 
 
1050
 
 
1051
 
format_registry = Registry()
1052
 
format_registry.register('default', DiffTree)