/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: John Arbash Meinel
  • Date: 2009-06-02 19:56:24 UTC
  • mto: This revision was merged to the branch mainline in revision 4469.
  • Revision ID: john@arbash-meinel.com-20090602195624-utljsyz0qgmq63lg
Add a chunks_to_gzip function.
This allows the _record_to_data code to build up a list of chunks,
rather than requiring a single string.
It should be ~ the same performance when using a single string, since
we are only adding a for() loop over the chunks and an if check.
We could possibly just remove the if check and not worry about adding
some empty strings in there.

Show diffs side-by-side

added added

removed removed

Lines of Context:
bzrlib/diff.py
1
 
# Copyright (C) 2005-2014 Canonical Ltd.
 
1
# Copyright (C) 2004, 2005, 2006 Canonical Ltd.
2
2
#
3
3
# This program is free software; you can redistribute it and/or modify
4
4
# it under the terms of the GNU General Public License as published by
14
14
# along with this program; if not, write to the Free Software
15
15
# Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
16
16
 
17
 
from __future__ import absolute_import
18
 
 
19
17
import difflib
20
18
import os
21
19
import re
22
 
import string
 
20
import shutil
23
21
import sys
24
22
 
25
 
from .lazy_import import lazy_import
 
23
from bzrlib.lazy_import import lazy_import
26
24
lazy_import(globals(), """
27
25
import errno
28
26
import subprocess
29
27
import tempfile
 
28
import time
30
29
 
31
 
from breezy import (
32
 
    cleanup,
33
 
    cmdline,
34
 
    controldir,
 
30
from bzrlib import (
 
31
    branch as _mod_branch,
 
32
    bzrdir,
 
33
    commands,
35
34
    errors,
36
35
    osutils,
37
36
    patiencediff,
39
38
    timestamp,
40
39
    views,
41
40
    )
42
 
 
43
 
from breezy.workingtree import WorkingTree
44
 
from breezy.i18n import gettext
45
41
""")
46
42
 
47
 
from .registry import (
48
 
    Registry,
 
43
from bzrlib.symbol_versioning import (
 
44
    deprecated_function,
49
45
    )
50
 
from .sixish import text_type
51
 
from .trace import mutter, note, warning
52
 
from .tree import FileTimestampUnavailable
53
 
 
54
 
 
55
 
DEFAULT_CONTEXT_AMOUNT = 3
56
 
 
57
 
class AtTemplate(string.Template):
58
 
    """Templating class that uses @ instead of $."""
59
 
 
60
 
    delimiter = '@'
 
46
from bzrlib.trace import mutter, note, 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(path_encoding, 'replace'),
98
 
                      tofile=new_label.encode(path_encoding, 'replace'),
99
 
                      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)
100
92
 
101
93
    ud = list(ud)
102
94
    if len(ud) == 0: # Identical contents, nothing to do
104
96
    # work-around for difflib being too smart for its own good
105
97
    # if /dev/null is "1,0", patch won't recognize it as /dev/null
106
98
    if not oldlines:
107
 
        ud[2] = ud[2].replace(b'-1,0', b'-0,0')
 
99
        ud[2] = ud[2].replace('-1,0', '-0,0')
108
100
    elif not newlines:
109
 
        ud[2] = ud[2].replace(b'+1,0', b'+0,0')
 
101
        ud[2] = ud[2].replace('+1,0', '+0,0')
110
102
 
111
103
    for line in ud:
112
104
        to_file.write(line)
113
 
        if not line.endswith(b'\n'):
114
 
            to_file.write(b"\n\\ No newline at end of file\n")
115
 
    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')
116
108
 
117
109
 
118
110
def _spawn_external_diff(diffcmd, capture_errors=True):
119
 
    """Spawn the external diff process, and return the child handle.
 
111
    """Spawn the externall diff process, and return the child handle.
120
112
 
121
113
    :param diffcmd: The command list to spawn
122
114
    :param capture_errors: Capture stderr as well as setting LANG=C
144
136
                                stdout=subprocess.PIPE,
145
137
                                stderr=stderr,
146
138
                                env=env)
147
 
    except OSError as e:
 
139
    except OSError, e:
148
140
        if e.errno == errno.ENOENT:
149
141
            raise errors.NoDiff(str(e))
150
142
        raise
151
143
 
152
144
    return pipe
153
145
 
154
 
# diff style options as of GNU diff v3.2
155
 
style_option_list = ['-c', '-C', '--context',
156
 
                     '-e', '--ed',
157
 
                     '-f', '--forward-ed',
158
 
                     '-q', '--brief',
159
 
                     '--normal',
160
 
                     '-n', '--rcs',
161
 
                     '-u', '-U', '--unified',
162
 
                     '-y', '--side-by-side',
163
 
                     '-D', '--ifdef']
164
 
 
165
 
def default_style_unified(diff_opts):
166
 
    """Default to unified diff style if alternative not specified in diff_opts.
167
 
 
168
 
        diff only allows one style to be specified; they don't override.
169
 
        Note that some of these take optargs, and the optargs can be
170
 
        directly appended to the options.
171
 
        This is only an approximate parser; it doesn't properly understand
172
 
        the grammar.
173
 
 
174
 
    :param diff_opts: List of options for external (GNU) diff.
175
 
    :return: List of options with default style=='unified'.
176
 
    """
177
 
    for s in style_option_list:
178
 
        for j in diff_opts:
179
 
            if j.startswith(s):
180
 
                break
181
 
        else:
182
 
            continue
183
 
        break
184
 
    else:
185
 
        diff_opts.append('-u')
186
 
    return diff_opts
187
 
 
188
 
 
189
 
def external_diff(old_label, oldlines, new_label, newlines, to_file,
 
146
 
 
147
def external_diff(old_filename, oldlines, new_filename, newlines, to_file,
190
148
                  diff_opts):
191
149
    """Display a diff by calling out to the external diff program."""
192
150
    # make sure our own output is properly ordered before the diff
193
151
    to_file.flush()
194
152
 
195
 
    oldtmp_fd, old_abspath = tempfile.mkstemp(prefix='brz-diff-old-')
196
 
    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-')
197
155
    oldtmpf = os.fdopen(oldtmp_fd, 'wb')
198
156
    newtmpf = os.fdopen(newtmp_fd, 'wb')
199
157
 
213
171
 
214
172
        if not diff_opts:
215
173
            diff_opts = []
216
 
        if sys.platform == 'win32':
217
 
            # Popen doesn't do the proper encoding for external commands
218
 
            # Since we are dealing with an ANSI api, use mbcs encoding
219
 
            old_label = old_label.encode('mbcs')
220
 
            new_label = new_label.encode('mbcs')
221
174
        diffcmd = ['diff',
222
 
                   '--label', old_label,
 
175
                   '--label', old_filename,
223
176
                   old_abspath,
224
 
                   '--label', new_label,
 
177
                   '--label', new_filename,
225
178
                   new_abspath,
226
179
                   '--binary',
227
180
                  ]
228
181
 
229
 
        diff_opts = default_style_unified(diff_opts)
 
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')
230
202
 
231
203
        if diff_opts:
232
204
            diffcmd.extend(diff_opts)
233
205
 
234
206
        pipe = _spawn_external_diff(diffcmd, capture_errors=True)
235
 
        out, err = pipe.communicate()
 
207
        out,err = pipe.communicate()
236
208
        rc = pipe.returncode
237
209
 
238
210
        # internal_diff() adds a trailing newline, add one here for consistency
239
 
        out += b'\n'
 
211
        out += '\n'
240
212
        if rc == 2:
241
213
            # 'diff' gives retcode == 2 for all sorts of errors
242
214
            # one of those is 'Binary files differ'.
249
221
            out, err = pipe.communicate()
250
222
 
251
223
            # Write out the new i18n diff response
252
 
            to_file.write(out+b'\n')
 
224
            to_file.write(out+'\n')
253
225
            if pipe.returncode != 2:
254
226
                raise errors.BzrError(
255
227
                               'external diff failed with exit code 2'
256
228
                               ' when run with LANG=C and LC_ALL=C,'
257
229
                               ' but not when run natively: %r' % (diffcmd,))
258
230
 
259
 
            first_line = lang_c_out.split(b'\n', 1)[0]
 
231
            first_line = lang_c_out.split('\n', 1)[0]
260
232
            # Starting with diffutils 2.8.4 the word "binary" was dropped.
261
 
            m = re.match(b'^(binary )?files.*differ$', first_line, re.I)
 
233
            m = re.match('^(binary )?files.*differ$', first_line, re.I)
262
234
            if m is None:
263
235
                raise errors.BzrError('external diff failed with exit code 2;'
264
236
                                      ' command: %r' % (diffcmd,))
277
249
                msg = 'exit code %d' % rc
278
250
 
279
251
            raise errors.BzrError('external diff failed with %s; command: %r'
280
 
                                  % (msg, diffcmd))
 
252
                                  % (rc, diffcmd))
281
253
 
282
254
 
283
255
    finally:
284
256
        oldtmpf.close()                 # and delete
285
257
        newtmpf.close()
286
 
 
287
 
        def cleanup(path):
288
 
            # Warn in case the file couldn't be deleted (in case windows still
289
 
            # holds the file open, but not if the files have already been
290
 
            # deleted)
291
 
            try:
292
 
                os.remove(path)
293
 
            except OSError as e:
294
 
                if e.errno not in (errno.ENOENT,):
295
 
                    warning('Failed to delete temporary file: %s %s', path, e)
296
 
 
297
 
        cleanup(old_abspath)
298
 
        cleanup(new_abspath)
299
 
 
300
 
 
301
 
def get_trees_and_branches_to_diff_locked(
302
 
    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,
 
276
    apply_view=True):
303
277
    """Get the trees and specific files to diff given a list of paths.
304
278
 
305
279
    This method works out the trees to be diff'ed and the files of
316
290
    :param new_url:
317
291
        The url of the new branch or tree. If None, the tree to use is
318
292
        taken from the first path, if any, or the current working tree.
319
 
    :param add_cleanup:
320
 
        a callable like Command.add_cleanup.  get_trees_and_branches_to_diff
321
 
        will register cleanups that must be run to unlock the trees, etc.
322
293
    :param apply_view:
323
294
        if True and a view is set, apply the view or check that the paths
324
295
        are within it
325
296
    :returns:
326
 
        a tuple of (old_tree, new_tree, old_branch, new_branch,
327
 
        specific_files, extra_trees) where extra_trees is a sequence of
328
 
        additional trees to search in for file-ids.  The trees and branches
329
 
        will be read-locked until the cleanups registered via the add_cleanup
330
 
        param are run.
 
297
        a tuple of (old_tree, new_tree, specific_files, extra_trees) where
 
298
        extra_trees is a sequence of additional trees to search in for
 
299
        file-ids.
331
300
    """
332
301
    # Get the old and new revision specs
333
302
    old_revision_spec = None
356
325
        default_location = path_list[0]
357
326
        other_paths = path_list[1:]
358
327
 
359
 
    def lock_tree_or_branch(wt, br):
360
 
        if wt is not None:
361
 
            wt.lock_read()
362
 
            add_cleanup(wt.unlock)
363
 
        elif br is not None:
364
 
            br.lock_read()
365
 
            add_cleanup(br.unlock)
366
 
 
367
328
    # Get the old location
368
329
    specific_files = []
369
330
    if old_url is None:
370
331
        old_url = default_location
371
332
    working_tree, branch, relpath = \
372
 
        controldir.ControlDir.open_containing_tree_or_branch(old_url)
373
 
    lock_tree_or_branch(working_tree, branch)
 
333
        bzrdir.BzrDir.open_containing_tree_or_branch(old_url)
374
334
    if consider_relpath and relpath != '':
375
335
        if working_tree is not None and apply_view:
376
336
            views.check_path_in_view(working_tree, relpath)
377
337
        specific_files.append(relpath)
378
338
    old_tree = _get_tree_to_diff(old_revision_spec, working_tree, branch)
379
 
    old_branch = branch
380
339
 
381
340
    # Get the new location
382
341
    if new_url is None:
383
342
        new_url = default_location
384
343
    if new_url != old_url:
385
344
        working_tree, branch, relpath = \
386
 
            controldir.ControlDir.open_containing_tree_or_branch(new_url)
387
 
        lock_tree_or_branch(working_tree, branch)
 
345
            bzrdir.BzrDir.open_containing_tree_or_branch(new_url)
388
346
        if consider_relpath and relpath != '':
389
347
            if working_tree is not None and apply_view:
390
348
                views.check_path_in_view(working_tree, relpath)
391
349
            specific_files.append(relpath)
392
350
    new_tree = _get_tree_to_diff(new_revision_spec, working_tree, branch,
393
351
        basis_is_default=working_tree is None)
394
 
    new_branch = branch
395
352
 
396
353
    # Get the specific files (all files is None, no files is [])
397
354
    if make_paths_wt_relative and working_tree is not None:
398
 
        other_paths = working_tree.safe_relpath_files(
399
 
            other_paths,
 
355
        try:
 
356
            from bzrlib.builtins import safe_relpath_files
 
357
            other_paths = safe_relpath_files(working_tree, other_paths,
400
358
            apply_view=apply_view)
 
359
        except errors.FileInWrongBranch:
 
360
            raise errors.BzrCommandError("Files are in different branches")
401
361
    specific_files.extend(other_paths)
402
362
    if len(specific_files) == 0:
403
363
        specific_files = None
407
367
            if view_files:
408
368
                specific_files = view_files
409
369
                view_str = views.view_display_str(view_files)
410
 
                note(gettext("*** Ignoring files outside view. View is %s") % view_str)
 
370
                note("*** Ignoring files outside view. View is %s" % view_str)
411
371
 
412
372
    # Get extra trees that ought to be searched for file-ids
413
373
    extra_trees = None
414
374
    if working_tree is not None and working_tree not in (old_tree, new_tree):
415
375
        extra_trees = (working_tree,)
416
 
    return (old_tree, new_tree, old_branch, new_branch,
417
 
            specific_files, extra_trees)
418
 
 
 
376
    return old_tree, new_tree, specific_files, extra_trees
419
377
 
420
378
def _get_tree_to_diff(spec, tree=None, branch=None, basis_is_default=True):
421
379
    if branch is None and tree is not None:
436
394
                    old_label='a/', new_label='b/',
437
395
                    extra_trees=None,
438
396
                    path_encoding='utf8',
439
 
                    using=None,
440
 
                    format_cls=None,
441
 
                    context=DEFAULT_CONTEXT_AMOUNT):
 
397
                    using=None):
442
398
    """Show in text form the changes from one tree to another.
443
399
 
444
 
    :param to_file: The output stream.
445
 
    :param specific_files: Include only changes to these files - None for all
446
 
        changes.
447
 
    :param external_diff_options: If set, use an external GNU diff and pass 
448
 
        these options.
449
 
    :param extra_trees: If set, more Trees to use for looking up file ids
450
 
    :param path_encoding: If set, the path will be encoded as specified, 
451
 
        otherwise is supposed to be utf8
452
 
    :param format_cls: Formatter class (DiffTree subclass)
 
400
    to_file
 
401
        The output stream.
 
402
 
 
403
    specific_files
 
404
        Include only changes to these files - None for all changes.
 
405
 
 
406
    external_diff_options
 
407
        If set, use an external GNU diff and pass these options.
 
408
 
 
409
    extra_trees
 
410
        If set, more Trees to use for looking up file ids
 
411
 
 
412
    path_encoding
 
413
        If set, the path will be encoded as specified, otherwise is supposed
 
414
        to be utf8
453
415
    """
454
 
    if context is None:
455
 
        context = DEFAULT_CONTEXT_AMOUNT
456
 
    if format_cls is None:
457
 
        format_cls = DiffTree
458
 
    with old_tree.lock_read():
 
416
    old_tree.lock_read()
 
417
    try:
459
418
        if extra_trees is not None:
460
419
            for tree in extra_trees:
461
420
                tree.lock_read()
462
421
        new_tree.lock_read()
463
422
        try:
464
 
            differ = format_cls.from_trees_options(old_tree, new_tree, to_file,
465
 
                                                   path_encoding,
466
 
                                                   external_diff_options,
467
 
                                                   old_label, new_label, using,
468
 
                                                   context_lines=context)
 
423
            differ = DiffTree.from_trees_options(old_tree, new_tree, to_file,
 
424
                                                 path_encoding,
 
425
                                                 external_diff_options,
 
426
                                                 old_label, new_label, using)
469
427
            return differ.show_diff(specific_files, extra_trees)
470
428
        finally:
471
429
            new_tree.unlock()
472
430
            if extra_trees is not None:
473
431
                for tree in extra_trees:
474
432
                    tree.unlock()
 
433
    finally:
 
434
        old_tree.unlock()
475
435
 
476
436
 
477
437
def _patch_header_date(tree, file_id, path):
478
438
    """Returns a timestamp suitable for use in a patch header."""
479
 
    try:
480
 
        mtime = tree.get_file_mtime(path, file_id)
481
 
    except FileTimestampUnavailable:
482
 
        mtime = 0
 
439
    mtime = tree.get_file_mtime(file_id, path)
483
440
    return timestamp.format_patch_date(mtime)
484
441
 
485
442
 
486
443
def get_executable_change(old_is_x, new_is_x):
487
 
    descr = { True:b"+x", False:b"-x", None:b"??" }
 
444
    descr = { True:"+x", False:"-x", None:"??" }
488
445
    if old_is_x != new_is_x:
489
 
        return [b"%s to %s" % (descr[old_is_x], descr[new_is_x],)]
 
446
        return ["%s to %s" % (descr[old_is_x], descr[new_is_x],)]
490
447
    else:
491
448
        return []
492
449
 
597
554
        if 'symlink' not in (old_kind, new_kind):
598
555
            return self.CANNOT_DIFF
599
556
        if old_kind == 'symlink':
600
 
            old_target = self.old_tree.get_symlink_target(old_path, file_id)
 
557
            old_target = self.old_tree.get_symlink_target(file_id)
601
558
        elif old_kind is None:
602
559
            old_target = None
603
560
        else:
604
561
            return self.CANNOT_DIFF
605
562
        if new_kind == 'symlink':
606
 
            new_target = self.new_tree.get_symlink_target(new_path, file_id)
 
563
            new_target = self.new_tree.get_symlink_target(file_id)
607
564
        elif new_kind is None:
608
565
            new_target = None
609
566
        else:
612
569
 
613
570
    def diff_symlink(self, old_target, new_target):
614
571
        if old_target is None:
615
 
            self.to_file.write(b'=== target is \'%s\'\n' %
616
 
                new_target.encode(self.path_encoding, 'replace'))
 
572
            self.to_file.write('=== target is %r\n' % new_target)
617
573
        elif new_target is None:
618
 
            self.to_file.write(b'=== target was \'%s\'\n' %
619
 
                old_target.encode(self.path_encoding, 'replace'))
 
574
            self.to_file.write('=== target was %r\n' % old_target)
620
575
        else:
621
 
            self.to_file.write(b'=== target changed \'%s\' => \'%s\'\n' %
622
 
                              (old_target.encode(self.path_encoding, 'replace'),
623
 
                               new_target.encode(self.path_encoding, 'replace')))
 
576
            self.to_file.write('=== target changed %r => %r\n' %
 
577
                              (old_target, new_target))
624
578
        return self.CHANGED
625
579
 
626
580
 
630
584
    # or removed in a diff.
631
585
    EPOCH_DATE = '1970-01-01 00:00:00 +0000'
632
586
 
633
 
    def __init__(self, old_tree, new_tree, to_file, path_encoding='utf-8', 
634
 
                 old_label='', new_label='', text_differ=internal_diff, 
635
 
                 context_lines=DEFAULT_CONTEXT_AMOUNT):
 
587
    def __init__(self, old_tree, new_tree, to_file, path_encoding='utf-8',
 
588
                 old_label='', new_label='', text_differ=internal_diff):
636
589
        DiffPath.__init__(self, old_tree, new_tree, to_file, path_encoding)
637
590
        self.text_differ = text_differ
638
591
        self.old_label = old_label
639
592
        self.new_label = new_label
640
593
        self.path_encoding = path_encoding
641
 
        self.context_lines = context_lines
642
594
 
643
595
    def diff(self, file_id, old_path, new_path, old_kind, new_kind):
644
596
        """Compare two files in unified diff format
666
618
            to_file_id = None
667
619
        else:
668
620
            return self.CANNOT_DIFF
669
 
        from_label = '%s%s\t%s' % (self.old_label, old_path,
670
 
                old_date)
671
 
        to_label = '%s%s\t%s' % (self.new_label, new_path,
672
 
                new_date)
673
 
        return self.diff_text(old_path, new_path, from_label, to_label,
674
 
            from_file_id, to_file_id)
 
621
        from_label = '%s%s\t%s' % (self.old_label, old_path, old_date)
 
622
        to_label = '%s%s\t%s' % (self.new_label, new_path, new_date)
 
623
        return self.diff_text(from_file_id, to_file_id, from_label, to_label,
 
624
            old_path, new_path)
675
625
 
676
 
    def diff_text(self, from_path, to_path, from_label, to_label,
677
 
        from_file_id=None, to_file_id=None):
 
626
    def diff_text(self, from_file_id, to_file_id, from_label, to_label,
 
627
        from_path=None, to_path=None):
678
628
        """Diff the content of given files in two trees
679
629
 
680
 
        :param from_path: The path in the from tree. If None,
 
630
        :param from_file_id: The id of the file in the from tree.  If None,
681
631
            the file is not present in the from tree.
682
 
        :param to_path: The path in the to tree. This may refer
683
 
            to a different file from from_path.  If None,
 
632
        :param to_file_id: The id of the file in the to tree.  This may refer
 
633
            to a different file from from_file_id.  If None,
684
634
            the file is not present in the to tree.
685
 
        :param from_file_id: The id of the file in the from tree or None if
686
 
            unknown.
687
 
        :param to_file_id: The id of the file in the to tree or None if
688
 
            unknown.
 
635
        :param from_path: The path in the from tree or None if unknown.
 
636
        :param to_path: The path in the to tree or None if unknown.
689
637
        """
690
638
        def _get_text(tree, file_id, path):
691
 
            if file_id is None:
 
639
            if file_id is not None:
 
640
                return tree.get_file(file_id, path).readlines()
 
641
            else:
692
642
                return []
693
 
            return tree.get_file_lines(path, file_id)
694
643
        try:
695
644
            from_text = _get_text(self.old_tree, from_file_id, from_path)
696
645
            to_text = _get_text(self.new_tree, to_file_id, to_path)
697
646
            self.text_differ(from_label, from_text, to_label, to_text,
698
 
                             self.to_file, path_encoding=self.path_encoding,
699
 
                             context_lines=self.context_lines)
 
647
                             self.to_file)
700
648
        except errors.BinaryFile:
701
649
            self.to_file.write(
702
650
                  ("Binary files %s and %s differ\n" %
703
 
                  (from_label, to_label)).encode(self.path_encoding, 'replace'))
 
651
                  (from_label, to_label)).encode(self.path_encoding))
704
652
        return self.CHANGED
705
653
 
706
654
 
710
658
                 path_encoding='utf-8'):
711
659
        DiffPath.__init__(self, old_tree, new_tree, to_file, path_encoding)
712
660
        self.command_template = command_template
713
 
        self._root = osutils.mkdtemp(prefix='brz-diff-')
 
661
        self._root = osutils.mkdtemp(prefix='bzr-diff-')
714
662
 
715
663
    @classmethod
716
664
    def from_string(klass, command_string, old_tree, new_tree, to_file,
717
665
                    path_encoding='utf-8'):
718
 
        command_template = cmdline.split(command_string)
719
 
        if '@' not in command_string:
720
 
            command_template.extend(['@old_path', '@new_path'])
 
666
        command_template = commands.shlex_split_unicode(command_string)
 
667
        command_template.extend(['%(old_path)s', '%(new_path)s'])
721
668
        return klass(command_template, old_tree, new_tree, to_file,
722
669
                     path_encoding)
723
670
 
724
671
    @classmethod
725
 
    def make_from_diff_tree(klass, command_string, external_diff_options=None):
 
672
    def make_from_diff_tree(klass, command_string):
726
673
        def from_diff_tree(diff_tree):
727
 
            full_command_string = [command_string]
728
 
            if external_diff_options is not None:
729
 
                full_command_string += ' ' + external_diff_options
730
 
            return klass.from_string(full_command_string, diff_tree.old_tree,
 
674
            return klass.from_string(command_string, diff_tree.old_tree,
731
675
                                     diff_tree.new_tree, diff_tree.to_file)
732
676
        return from_diff_tree
733
677
 
734
678
    def _get_command(self, old_path, new_path):
735
679
        my_map = {'old_path': old_path, 'new_path': new_path}
736
 
        command = [AtTemplate(t).substitute(my_map) for t in
737
 
                   self.command_template]
738
 
        if sys.platform == 'win32': # Popen doesn't accept unicode on win32
739
 
            command_encoded = []
740
 
            for c in command:
741
 
                if isinstance(c, text_type):
742
 
                    command_encoded.append(c.encode('mbcs'))
743
 
                else:
744
 
                    command_encoded.append(c)
745
 
            return command_encoded
746
 
        else:
747
 
            return command
 
680
        return [t % my_map for t in self.command_template]
748
681
 
749
682
    def _execute(self, old_path, new_path):
750
683
        command = self._get_command(old_path, new_path)
751
684
        try:
752
685
            proc = subprocess.Popen(command, stdout=subprocess.PIPE,
753
686
                                    cwd=self._root)
754
 
        except OSError as e:
 
687
        except OSError, e:
755
688
            if e.errno == errno.ENOENT:
756
689
                raise errors.ExecutableMissing(command[0])
757
690
            else:
758
691
                raise
759
692
        self.to_file.write(proc.stdout.read())
760
 
        proc.stdout.close()
761
693
        return proc.wait()
762
694
 
763
695
    def _try_symlink_root(self, tree, prefix):
766
698
            return False
767
699
        try:
768
700
            os.symlink(tree.abspath(''), osutils.pathjoin(self._root, prefix))
769
 
        except OSError as e:
 
701
        except OSError, e:
770
702
            if e.errno != errno.EEXIST:
771
703
                raise
772
704
        return True
773
705
 
774
 
    @staticmethod
775
 
    def _fenc():
776
 
        """Returns safe encoding for passing file path to diff tool"""
777
 
        if sys.platform == 'win32':
778
 
            return 'mbcs'
779
 
        else:
780
 
            # Don't fallback to 'utf-8' because subprocess may not be able to
781
 
            # handle utf-8 correctly when locale is not utf-8.
782
 
            return sys.getfilesystemencoding() or 'ascii'
783
 
 
784
 
    def _is_safepath(self, path):
785
 
        """Return true if `path` may be able to pass to subprocess."""
786
 
        fenc = self._fenc()
787
 
        try:
788
 
            return path == path.encode(fenc).decode(fenc)
789
 
        except UnicodeError:
790
 
            return False
791
 
 
792
 
    def _safe_filename(self, prefix, relpath):
793
 
        """Replace unsafe character in `relpath` then join `self._root`,
794
 
        `prefix` and `relpath`."""
795
 
        fenc = self._fenc()
796
 
        # encoded_str.replace('?', '_') may break multibyte char.
797
 
        # So we should encode, decode, then replace(u'?', u'_')
798
 
        relpath_tmp = relpath.encode(fenc, 'replace').decode(fenc, 'replace')
799
 
        relpath_tmp = relpath_tmp.replace(u'?', u'_')
800
 
        return osutils.pathjoin(self._root, prefix, relpath_tmp)
801
 
 
802
 
    def _write_file(self, relpath, tree, prefix, force_temp=False,
803
 
                    allow_write=False, file_id=None):
804
 
        if not force_temp and isinstance(tree, WorkingTree):
805
 
            full_path = tree.abspath(relpath)
806
 
            if self._is_safepath(full_path):
807
 
                return full_path
808
 
 
809
 
        full_path = self._safe_filename(prefix, relpath)
810
 
        if not force_temp and self._try_symlink_root(tree, prefix):
 
706
    def _write_file(self, file_id, tree, prefix, relpath):
 
707
        full_path = osutils.pathjoin(self._root, prefix, relpath)
 
708
        if self._try_symlink_root(tree, prefix):
811
709
            return full_path
812
710
        parent_dir = osutils.dirname(full_path)
813
711
        try:
814
712
            os.makedirs(parent_dir)
815
 
        except OSError as e:
 
713
        except OSError, e:
816
714
            if e.errno != errno.EEXIST:
817
715
                raise
818
 
        source = tree.get_file(relpath, file_id)
 
716
        source = tree.get_file(file_id, relpath)
819
717
        try:
820
 
            with open(full_path, 'wb') as target:
 
718
            target = open(full_path, 'wb')
 
719
            try:
821
720
                osutils.pumpfile(source, target)
 
721
            finally:
 
722
                target.close()
822
723
        finally:
823
724
            source.close()
824
 
        try:
825
 
            mtime = tree.get_file_mtime(relpath, file_id)
826
 
        except FileTimestampUnavailable:
827
 
            pass
828
 
        else:
829
 
            os.utime(full_path, (mtime, mtime))
830
 
        if not allow_write:
831
 
            osutils.make_readonly(full_path)
 
725
        osutils.make_readonly(full_path)
 
726
        mtime = tree.get_file_mtime(file_id)
 
727
        os.utime(full_path, (mtime, mtime))
832
728
        return full_path
833
729
 
834
 
    def _prepare_files(self, old_path, new_path, force_temp=False,
835
 
                       allow_write_new=False, file_id=None):
836
 
        old_disk_path = self._write_file(old_path, self.old_tree, 'old',
837
 
                                         force_temp, file_id=file_id)
838
 
        new_disk_path = self._write_file(new_path, self.new_tree, 'new',
839
 
                                         force_temp, file_id=file_id,
840
 
                                         allow_write=allow_write_new)
 
730
    def _prepare_files(self, file_id, old_path, new_path):
 
731
        old_disk_path = self._write_file(file_id, self.old_tree, 'old',
 
732
                                         old_path)
 
733
        new_disk_path = self._write_file(file_id, self.new_tree, 'new',
 
734
                                         new_path)
841
735
        return old_disk_path, new_disk_path
842
736
 
843
737
    def finish(self):
844
 
        try:
845
 
            osutils.rmtree(self._root)
846
 
        except OSError as e:
847
 
            if e.errno != errno.ENOENT:
848
 
                mutter("The temporary directory \"%s\" was not "
849
 
                        "cleanly removed: %s." % (self._root, e))
 
738
        osutils.rmtree(self._root)
850
739
 
851
740
    def diff(self, file_id, old_path, new_path, old_kind, new_kind):
852
741
        if (old_kind, new_kind) != ('file', 'file'):
853
742
            return DiffPath.CANNOT_DIFF
854
 
        (old_disk_path, new_disk_path) = self._prepare_files(
855
 
                old_path, new_path, file_id=file_id)
856
 
        self._execute(old_disk_path, new_disk_path)
857
 
 
858
 
    def edit_file(self, old_path, new_path, file_id=None):
859
 
        """Use this tool to edit a file.
860
 
 
861
 
        A temporary copy will be edited, and the new contents will be
862
 
        returned.
863
 
 
864
 
        :param file_id: The id of the file to edit.
865
 
        :return: The new contents of the file.
866
 
        """
867
 
        old_abs_path, new_abs_path = self._prepare_files(
868
 
                old_path, new_path, allow_write_new=True, force_temp=True,
869
 
                file_id=file_id)
870
 
        command = self._get_command(old_abs_path, new_abs_path)
871
 
        subprocess.call(command, cwd=self._root)
872
 
        with open(new_abs_path, 'rb') as new_file:
873
 
            return new_file.read()
 
743
        self._prepare_files(file_id, old_path, new_path)
 
744
        self._execute(osutils.pathjoin('old', old_path),
 
745
                      osutils.pathjoin('new', new_path))
874
746
 
875
747
 
876
748
class DiffTree(object):
918
790
    @classmethod
919
791
    def from_trees_options(klass, old_tree, new_tree, to_file,
920
792
                           path_encoding, external_diff_options, old_label,
921
 
                           new_label, using, context_lines):
 
793
                           new_label, using):
922
794
        """Factory for producing a DiffTree.
923
795
 
924
796
        Designed to accept options used by show_diff_trees.
925
 
 
926
797
        :param old_tree: The tree to show as old in the comparison
927
798
        :param new_tree: The tree to show as new in the comparison
928
799
        :param to_file: File to write comparisons to
934
805
        :param using: Commandline to use to invoke an external diff tool
935
806
        """
936
807
        if using is not None:
937
 
            extra_factories = [DiffFromTool.make_from_diff_tree(using, external_diff_options)]
 
808
            extra_factories = [DiffFromTool.make_from_diff_tree(using)]
938
809
        else:
939
810
            extra_factories = []
940
811
        if external_diff_options:
941
812
            opts = external_diff_options.split()
942
 
            def diff_file(olab, olines, nlab, nlines, to_file, path_encoding=None, context_lines=None):
943
 
                """:param path_encoding: not used but required
944
 
                        to match the signature of internal_diff.
945
 
                """
 
813
            def diff_file(olab, olines, nlab, nlines, to_file):
946
814
                external_diff(olab, olines, nlab, nlines, to_file, opts)
947
815
        else:
948
816
            diff_file = internal_diff
949
817
        diff_text = DiffText(old_tree, new_tree, to_file, path_encoding,
950
 
                             old_label, new_label, diff_file, context_lines=context_lines)
 
818
                             old_label, new_label, diff_file)
951
819
        return klass(old_tree, new_tree, to_file, path_encoding, diff_text,
952
820
                     extra_factories)
953
821
 
954
822
    def show_diff(self, specific_files, extra_trees=None):
955
823
        """Write tree diff to self.to_file
956
824
 
957
 
        :param specific_files: the specific files to compare (recursive)
 
825
        :param sepecific_files: the specific files to compare (recursive)
958
826
        :param extra_trees: extra trees to use for mapping paths to file_ids
959
827
        """
960
828
        try:
997
865
            properties_changed.extend(get_executable_change(executable[0], executable[1]))
998
866
 
999
867
            if properties_changed:
1000
 
                prop_str = b" (properties changed: %s)" % (
1001
 
                        b", ".join(properties_changed),)
 
868
                prop_str = " (properties changed: %s)" % (", ".join(properties_changed),)
1002
869
            else:
1003
 
                prop_str = b""
 
870
                prop_str = ""
1004
871
 
1005
872
            if (old_present, new_present) == (True, False):
1006
 
                self.to_file.write(b"=== removed %s '%s'\n" %
1007
 
                                   (kind[0].encode('ascii'), oldpath_encoded))
 
873
                self.to_file.write("=== removed %s '%s'\n" %
 
874
                                   (kind[0], oldpath_encoded))
1008
875
                newpath = oldpath
1009
876
            elif (old_present, new_present) == (False, True):
1010
 
                self.to_file.write(b"=== added %s '%s'\n" %
1011
 
                                   (kind[1].encode('ascii'), newpath_encoded))
 
877
                self.to_file.write("=== added %s '%s'\n" %
 
878
                                   (kind[1], newpath_encoded))
1012
879
                oldpath = newpath
1013
880
            elif renamed:
1014
 
                self.to_file.write(b"=== renamed %s '%s' => '%s'%s\n" %
1015
 
                    (kind[0].encode('ascii'), oldpath_encoded, newpath_encoded, prop_str))
 
881
                self.to_file.write("=== renamed %s '%s' => '%s'%s\n" %
 
882
                    (kind[0], oldpath_encoded, newpath_encoded, prop_str))
1016
883
            else:
1017
884
                # if it was produced by iter_changes, it must be
1018
885
                # modified *somehow*, either content or execute bit.
1019
 
                self.to_file.write(b"=== modified %s '%s'%s\n" % (kind[0].encode('ascii'),
 
886
                self.to_file.write("=== modified %s '%s'%s\n" % (kind[0],
1020
887
                                   newpath_encoded, prop_str))
1021
888
            if changed_content:
1022
 
                self._diff(oldpath, newpath, kind[0], kind[1], file_id=file_id)
 
889
                self._diff(file_id, oldpath, newpath, kind[0], kind[1])
1023
890
                has_changes = 1
1024
891
            if renamed:
1025
892
                has_changes = 1
1032
899
        :param old_path: The path of the file in the old tree
1033
900
        :param new_path: The path of the file in the new tree
1034
901
        """
1035
 
        if old_path is None:
 
902
        try:
 
903
            old_kind = self.old_tree.kind(file_id)
 
904
        except (errors.NoSuchId, errors.NoSuchFile):
1036
905
            old_kind = None
1037
 
        else:
1038
 
            old_kind = self.old_tree.kind(old_path, file_id)
1039
 
        if new_path is None:
 
906
        try:
 
907
            new_kind = self.new_tree.kind(file_id)
 
908
        except (errors.NoSuchId, errors.NoSuchFile):
1040
909
            new_kind = None
1041
 
        else:
1042
 
            new_kind = self.new_tree.kind(new_path, file_id)
1043
 
        self._diff(old_path, new_path, old_kind, new_kind, file_id=file_id)
1044
 
 
1045
 
    def _diff(self, old_path, new_path, old_kind, new_kind, file_id):
 
910
        self._diff(file_id, old_path, new_path, old_kind, new_kind)
 
911
 
 
912
 
 
913
    def _diff(self, file_id, old_path, new_path, old_kind, new_kind):
1046
914
        result = DiffPath._diff_many(self.differs, file_id, old_path,
1047
 
                                     new_path, old_kind, new_kind)
 
915
                                       new_path, old_kind, new_kind)
1048
916
        if result is DiffPath.CANNOT_DIFF:
1049
917
            error_path = new_path
1050
918
            if error_path is None:
1051
919
                error_path = old_path
1052
920
            raise errors.NoDiffFound(error_path)
1053
 
 
1054
 
 
1055
 
format_registry = Registry()
1056
 
format_registry.register('default', DiffTree)