/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: Ian Clatworthy
  • Date: 2008-07-17 01:08:21 UTC
  • mto: This revision was merged to the branch mainline in revision 3546.
  • Revision ID: ian.clatworthy@canonical.com-20080717010821-qg05ah7iobvl8hin
workaround docutils dot-in-option name bug (Mark Hammond)

Show diffs side-by-side

added added

removed removed

Lines of Context:
bzrlib/diff.py
1
 
# Copyright (C) 2005-2014 Canonical Ltd.
 
1
# Copyright (C) 2004, 2005, 2006 Canonical Ltd.
2
2
#
3
3
# This program is free software; you can redistribute it and/or modify
4
4
# it under the terms of the GNU General Public License as published by
12
12
#
13
13
# You should have received a copy of the GNU General Public License
14
14
# along with this program; if not, write to the Free Software
15
 
# Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
16
 
 
17
 
from __future__ import absolute_import
 
15
# Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
18
16
 
19
17
import difflib
20
18
import os
21
19
import re
22
 
import string
 
20
import shutil
23
21
import sys
24
22
 
25
 
from .lazy_import import lazy_import
 
23
from bzrlib.lazy_import import lazy_import
26
24
lazy_import(globals(), """
27
25
import errno
28
26
import subprocess
29
27
import tempfile
 
28
import time
30
29
 
31
 
from breezy import (
32
 
    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,
38
37
    textfile,
39
38
    timestamp,
40
 
    views,
41
39
    )
42
 
 
43
 
from breezy.workingtree import WorkingTree
44
 
from breezy.i18n import gettext
45
40
""")
46
41
 
47
 
from .registry import (
48
 
    Registry,
49
 
    )
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 = '@'
 
42
from bzrlib.symbol_versioning import (
 
43
        deprecated_function,
 
44
        one_three
 
45
        )
 
46
from bzrlib.trace import mutter, 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')
 
102
    # work around for difflib emitting random spaces after the label
 
103
    ud[0] = ud[0][:-2] + '\n'
 
104
    ud[1] = ud[1][:-2] + '\n'
110
105
 
111
106
    for line in ud:
112
107
        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')
 
108
        if not line.endswith('\n'):
 
109
            to_file.write("\n\\ No newline at end of file\n")
 
110
    to_file.write('\n')
116
111
 
117
112
 
118
113
def _spawn_external_diff(diffcmd, capture_errors=True):
119
 
    """Spawn the external diff process, and return the child handle.
 
114
    """Spawn the externall diff process, and return the child handle.
120
115
 
121
116
    :param diffcmd: The command list to spawn
122
117
    :param capture_errors: Capture stderr as well as setting LANG=C
144
139
                                stdout=subprocess.PIPE,
145
140
                                stderr=stderr,
146
141
                                env=env)
147
 
    except OSError as e:
 
142
    except OSError, e:
148
143
        if e.errno == errno.ENOENT:
149
144
            raise errors.NoDiff(str(e))
150
145
        raise
151
146
 
152
147
    return pipe
153
148
 
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,
 
149
 
 
150
def external_diff(old_filename, oldlines, new_filename, newlines, to_file,
190
151
                  diff_opts):
191
152
    """Display a diff by calling out to the external diff program."""
192
153
    # make sure our own output is properly ordered before the diff
193
154
    to_file.flush()
194
155
 
195
 
    oldtmp_fd, old_abspath = tempfile.mkstemp(prefix='brz-diff-old-')
196
 
    newtmp_fd, new_abspath = tempfile.mkstemp(prefix='brz-diff-new-')
 
156
    oldtmp_fd, old_abspath = tempfile.mkstemp(prefix='bzr-diff-old-')
 
157
    newtmp_fd, new_abspath = tempfile.mkstemp(prefix='bzr-diff-new-')
197
158
    oldtmpf = os.fdopen(oldtmp_fd, 'wb')
198
159
    newtmpf = os.fdopen(newtmp_fd, 'wb')
199
160
 
213
174
 
214
175
        if not diff_opts:
215
176
            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
177
        diffcmd = ['diff',
222
 
                   '--label', old_label,
 
178
                   '--label', old_filename,
223
179
                   old_abspath,
224
 
                   '--label', new_label,
 
180
                   '--label', new_filename,
225
181
                   new_abspath,
226
182
                   '--binary',
227
183
                  ]
228
184
 
229
 
        diff_opts = default_style_unified(diff_opts)
230
 
 
 
185
        # diff only allows one style to be specified; they don't override.
 
186
        # note that some of these take optargs, and the optargs can be
 
187
        # directly appended to the options.
 
188
        # this is only an approximate parser; it doesn't properly understand
 
189
        # the grammar.
 
190
        for s in ['-c', '-u', '-C', '-U',
 
191
                  '-e', '--ed',
 
192
                  '-q', '--brief',
 
193
                  '--normal',
 
194
                  '-n', '--rcs',
 
195
                  '-y', '--side-by-side',
 
196
                  '-D', '--ifdef']:
 
197
            for j in diff_opts:
 
198
                if j.startswith(s):
 
199
                    break
 
200
            else:
 
201
                continue
 
202
            break
 
203
        else:
 
204
            diffcmd.append('-u')
 
205
                  
231
206
        if diff_opts:
232
207
            diffcmd.extend(diff_opts)
233
208
 
234
209
        pipe = _spawn_external_diff(diffcmd, capture_errors=True)
235
 
        out, err = pipe.communicate()
 
210
        out,err = pipe.communicate()
236
211
        rc = pipe.returncode
237
 
 
 
212
        
238
213
        # internal_diff() adds a trailing newline, add one here for consistency
239
 
        out += b'\n'
 
214
        out += '\n'
240
215
        if rc == 2:
241
216
            # 'diff' gives retcode == 2 for all sorts of errors
242
217
            # one of those is 'Binary files differ'.
249
224
            out, err = pipe.communicate()
250
225
 
251
226
            # Write out the new i18n diff response
252
 
            to_file.write(out+b'\n')
 
227
            to_file.write(out+'\n')
253
228
            if pipe.returncode != 2:
254
229
                raise errors.BzrError(
255
230
                               'external diff failed with exit code 2'
256
231
                               ' when run with LANG=C and LC_ALL=C,'
257
232
                               ' but not when run natively: %r' % (diffcmd,))
258
233
 
259
 
            first_line = lang_c_out.split(b'\n', 1)[0]
 
234
            first_line = lang_c_out.split('\n', 1)[0]
260
235
            # Starting with diffutils 2.8.4 the word "binary" was dropped.
261
 
            m = re.match(b'^(binary )?files.*differ$', first_line, re.I)
 
236
            m = re.match('^(binary )?files.*differ$', first_line, re.I)
262
237
            if m is None:
263
238
                raise errors.BzrError('external diff failed with exit code 2;'
264
239
                                      ' command: %r' % (diffcmd,))
275
250
                msg = 'signal %d' % (-rc)
276
251
            else:
277
252
                msg = 'exit code %d' % rc
278
 
 
279
 
            raise errors.BzrError('external diff failed with %s; command: %r'
280
 
                                  % (msg, diffcmd))
 
253
                
 
254
            raise errors.BzrError('external diff failed with %s; command: %r' 
 
255
                                  % (rc, diffcmd))
281
256
 
282
257
 
283
258
    finally:
284
259
        oldtmpf.close()                 # and delete
285
260
        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):
 
261
        # Clean up. Warn in case the files couldn't be deleted
 
262
        # (in case windows still holds the file open, but not
 
263
        # if the files have already been deleted)
 
264
        try:
 
265
            os.remove(old_abspath)
 
266
        except OSError, e:
 
267
            if e.errno not in (errno.ENOENT,):
 
268
                warning('Failed to delete temporary file: %s %s',
 
269
                        old_abspath, e)
 
270
        try:
 
271
            os.remove(new_abspath)
 
272
        except OSError:
 
273
            if e.errno not in (errno.ENOENT,):
 
274
                warning('Failed to delete temporary file: %s %s',
 
275
                        new_abspath, e)
 
276
 
 
277
 
 
278
def _get_trees_to_diff(path_list, revision_specs, old_url, new_url):
303
279
    """Get the trees and specific files to diff given a list of paths.
304
280
 
305
281
    This method works out the trees to be diff'ed and the files of
316
292
    :param new_url:
317
293
        The url of the new branch or tree. If None, the tree to use is
318
294
        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
 
    :param apply_view:
323
 
        if True and a view is set, apply the view or check that the paths
324
 
        are within it
325
295
    :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.
 
296
        a tuple of (old_tree, new_tree, specific_files, extra_trees) where
 
297
        extra_trees is a sequence of additional trees to search in for
 
298
        file-ids.
331
299
    """
332
300
    # Get the old and new revision specs
333
301
    old_revision_spec = None
356
324
        default_location = path_list[0]
357
325
        other_paths = path_list[1:]
358
326
 
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
327
    # Get the old location
368
328
    specific_files = []
369
329
    if old_url is None:
370
330
        old_url = default_location
371
331
    working_tree, branch, relpath = \
372
 
        controldir.ControlDir.open_containing_tree_or_branch(old_url)
373
 
    lock_tree_or_branch(working_tree, branch)
 
332
        bzrdir.BzrDir.open_containing_tree_or_branch(old_url)
374
333
    if consider_relpath and relpath != '':
375
 
        if working_tree is not None and apply_view:
376
 
            views.check_path_in_view(working_tree, relpath)
377
334
        specific_files.append(relpath)
378
335
    old_tree = _get_tree_to_diff(old_revision_spec, working_tree, branch)
379
 
    old_branch = branch
380
336
 
381
337
    # Get the new location
382
338
    if new_url is None:
383
339
        new_url = default_location
384
340
    if new_url != old_url:
385
341
        working_tree, branch, relpath = \
386
 
            controldir.ControlDir.open_containing_tree_or_branch(new_url)
387
 
        lock_tree_or_branch(working_tree, branch)
 
342
            bzrdir.BzrDir.open_containing_tree_or_branch(new_url)
388
343
        if consider_relpath and relpath != '':
389
 
            if working_tree is not None and apply_view:
390
 
                views.check_path_in_view(working_tree, relpath)
391
344
            specific_files.append(relpath)
392
345
    new_tree = _get_tree_to_diff(new_revision_spec, working_tree, branch,
393
346
        basis_is_default=working_tree is None)
394
 
    new_branch = branch
395
347
 
396
348
    # Get the specific files (all files is None, no files is [])
397
349
    if make_paths_wt_relative and working_tree is not None:
398
 
        other_paths = working_tree.safe_relpath_files(
399
 
            other_paths,
400
 
            apply_view=apply_view)
 
350
        other_paths = _relative_paths_in_tree(working_tree, other_paths)
401
351
    specific_files.extend(other_paths)
402
352
    if len(specific_files) == 0:
403
353
        specific_files = None
404
 
        if (working_tree is not None and working_tree.supports_views()
405
 
            and apply_view):
406
 
            view_files = working_tree.views.lookup_view()
407
 
            if view_files:
408
 
                specific_files = view_files
409
 
                view_str = views.view_display_str(view_files)
410
 
                note(gettext("*** Ignoring files outside view. View is %s") % view_str)
411
354
 
412
355
    # Get extra trees that ought to be searched for file-ids
413
356
    extra_trees = None
414
357
    if working_tree is not None and working_tree not in (old_tree, new_tree):
415
358
        extra_trees = (working_tree,)
416
 
    return (old_tree, new_tree, old_branch, new_branch,
417
 
            specific_files, extra_trees)
 
359
    return old_tree, new_tree, specific_files, extra_trees
418
360
 
419
361
 
420
362
def _get_tree_to_diff(spec, tree=None, branch=None, basis_is_default=True):
428
370
                return branch.basis_tree()
429
371
        else:
430
372
            return tree
431
 
    return spec.as_tree(branch)
 
373
    if not spec.needs_branch():
 
374
        branch = _mod_branch.Branch.open(spec.get_branch())
 
375
    revision_id = spec.as_revision_id(branch)
 
376
    return branch.repository.revision_tree(revision_id)
 
377
 
 
378
 
 
379
def _relative_paths_in_tree(tree, paths):
 
380
    """Get the relative paths within a working tree.
 
381
 
 
382
    Each path may be either an absolute path or a path relative to the
 
383
    current working directory.
 
384
    """
 
385
    result = []
 
386
    for filename in paths:
 
387
        try:
 
388
            result.append(tree.relpath(osutils.dereference_path(filename)))
 
389
        except errors.PathNotChild:
 
390
            raise errors.BzrCommandError("Files are in different branches")
 
391
    return result
432
392
 
433
393
 
434
394
def show_diff_trees(old_tree, new_tree, to_file, specific_files=None,
436
396
                    old_label='a/', new_label='b/',
437
397
                    extra_trees=None,
438
398
                    path_encoding='utf8',
439
 
                    using=None,
440
 
                    format_cls=None,
441
 
                    context=DEFAULT_CONTEXT_AMOUNT):
 
399
                    using=None):
442
400
    """Show in text form the changes from one tree to another.
443
401
 
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)
 
402
    to_file
 
403
        The output stream.
 
404
 
 
405
    specific_files
 
406
        Include only changes to these files - None for all changes.
 
407
 
 
408
    external_diff_options
 
409
        If set, use an external GNU diff and pass these options.
 
410
 
 
411
    extra_trees
 
412
        If set, more Trees to use for looking up file ids
 
413
 
 
414
    path_encoding
 
415
        If set, the path will be encoded as specified, otherwise is supposed
 
416
        to be utf8
453
417
    """
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():
 
418
    old_tree.lock_read()
 
419
    try:
459
420
        if extra_trees is not None:
460
421
            for tree in extra_trees:
461
422
                tree.lock_read()
462
423
        new_tree.lock_read()
463
424
        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)
 
425
            differ = DiffTree.from_trees_options(old_tree, new_tree, to_file,
 
426
                                                 path_encoding,
 
427
                                                 external_diff_options,
 
428
                                                 old_label, new_label, using)
469
429
            return differ.show_diff(specific_files, extra_trees)
470
430
        finally:
471
431
            new_tree.unlock()
472
432
            if extra_trees is not None:
473
433
                for tree in extra_trees:
474
434
                    tree.unlock()
 
435
    finally:
 
436
        old_tree.unlock()
475
437
 
476
438
 
477
439
def _patch_header_date(tree, file_id, path):
478
440
    """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
 
441
    mtime = tree.get_file_mtime(file_id, path)
483
442
    return timestamp.format_patch_date(mtime)
484
443
 
485
444
 
 
445
def _raise_if_nonexistent(paths, old_tree, new_tree):
 
446
    """Complain if paths are not in either inventory or tree.
 
447
 
 
448
    It's OK with the files exist in either tree's inventory, or 
 
449
    if they exist in the tree but are not versioned.
 
450
    
 
451
    This can be used by operations such as bzr status that can accept
 
452
    unknown or ignored files.
 
453
    """
 
454
    mutter("check paths: %r", paths)
 
455
    if not paths:
 
456
        return
 
457
    s = old_tree.filter_unversioned_files(paths)
 
458
    s = new_tree.filter_unversioned_files(s)
 
459
    s = [path for path in s if not new_tree.has_filename(path)]
 
460
    if s:
 
461
        raise errors.PathsDoNotExist(sorted(s))
 
462
 
 
463
 
 
464
@deprecated_function(one_three)
 
465
def get_prop_change(meta_modified):
 
466
    if meta_modified:
 
467
        return " (properties changed)"
 
468
    else:
 
469
        return  ""
 
470
 
486
471
def get_executable_change(old_is_x, new_is_x):
487
 
    descr = { True:b"+x", False:b"-x", None:b"??" }
 
472
    descr = { True:"+x", False:"-x", None:"??" }
488
473
    if old_is_x != new_is_x:
489
 
        return [b"%s to %s" % (descr[old_is_x], descr[new_is_x],)]
 
474
        return ["%s to %s" % (descr[old_is_x], descr[new_is_x],)]
490
475
    else:
491
476
        return []
492
477
 
597
582
        if 'symlink' not in (old_kind, new_kind):
598
583
            return self.CANNOT_DIFF
599
584
        if old_kind == 'symlink':
600
 
            old_target = self.old_tree.get_symlink_target(old_path, file_id)
 
585
            old_target = self.old_tree.get_symlink_target(file_id)
601
586
        elif old_kind is None:
602
587
            old_target = None
603
588
        else:
604
589
            return self.CANNOT_DIFF
605
590
        if new_kind == 'symlink':
606
 
            new_target = self.new_tree.get_symlink_target(new_path, file_id)
 
591
            new_target = self.new_tree.get_symlink_target(file_id)
607
592
        elif new_kind is None:
608
593
            new_target = None
609
594
        else:
612
597
 
613
598
    def diff_symlink(self, old_target, new_target):
614
599
        if old_target is None:
615
 
            self.to_file.write(b'=== target is \'%s\'\n' %
616
 
                new_target.encode(self.path_encoding, 'replace'))
 
600
            self.to_file.write('=== target is %r\n' % new_target)
617
601
        elif new_target is None:
618
 
            self.to_file.write(b'=== target was \'%s\'\n' %
619
 
                old_target.encode(self.path_encoding, 'replace'))
 
602
            self.to_file.write('=== target was %r\n' % old_target)
620
603
        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')))
 
604
            self.to_file.write('=== target changed %r => %r\n' %
 
605
                              (old_target, new_target))
624
606
        return self.CHANGED
625
607
 
626
608
 
630
612
    # or removed in a diff.
631
613
    EPOCH_DATE = '1970-01-01 00:00:00 +0000'
632
614
 
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):
 
615
    def __init__(self, old_tree, new_tree, to_file, path_encoding='utf-8',
 
616
                 old_label='', new_label='', text_differ=internal_diff):
636
617
        DiffPath.__init__(self, old_tree, new_tree, to_file, path_encoding)
637
618
        self.text_differ = text_differ
638
619
        self.old_label = old_label
639
620
        self.new_label = new_label
640
621
        self.path_encoding = path_encoding
641
 
        self.context_lines = context_lines
642
622
 
643
623
    def diff(self, file_id, old_path, new_path, old_kind, new_kind):
644
624
        """Compare two files in unified diff format
666
646
            to_file_id = None
667
647
        else:
668
648
            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)
 
649
        from_label = '%s%s\t%s' % (self.old_label, old_path, old_date)
 
650
        to_label = '%s%s\t%s' % (self.new_label, new_path, new_date)
 
651
        return self.diff_text(from_file_id, to_file_id, from_label, to_label)
675
652
 
676
 
    def diff_text(self, from_path, to_path, from_label, to_label,
677
 
        from_file_id=None, to_file_id=None):
 
653
    def diff_text(self, from_file_id, to_file_id, from_label, to_label):
678
654
        """Diff the content of given files in two trees
679
655
 
680
 
        :param from_path: The path in the from tree. If None,
 
656
        :param from_file_id: The id of the file in the from tree.  If None,
681
657
            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,
 
658
        :param to_file_id: The id of the file in the to tree.  This may refer
 
659
            to a different file from from_file_id.  If None,
684
660
            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.
689
661
        """
690
 
        def _get_text(tree, file_id, path):
691
 
            if file_id is None:
 
662
        def _get_text(tree, file_id):
 
663
            if file_id is not None:
 
664
                return tree.get_file(file_id).readlines()
 
665
            else:
692
666
                return []
693
 
            return tree.get_file_lines(path, file_id)
694
667
        try:
695
 
            from_text = _get_text(self.old_tree, from_file_id, from_path)
696
 
            to_text = _get_text(self.new_tree, to_file_id, to_path)
 
668
            from_text = _get_text(self.old_tree, from_file_id)
 
669
            to_text = _get_text(self.new_tree, to_file_id)
697
670
            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)
 
671
                             self.to_file)
700
672
        except errors.BinaryFile:
701
673
            self.to_file.write(
702
674
                  ("Binary files %s and %s differ\n" %
703
 
                  (from_label, to_label)).encode(self.path_encoding, 'replace'))
 
675
                  (from_label, to_label)).encode(self.path_encoding))
704
676
        return self.CHANGED
705
677
 
706
678
 
710
682
                 path_encoding='utf-8'):
711
683
        DiffPath.__init__(self, old_tree, new_tree, to_file, path_encoding)
712
684
        self.command_template = command_template
713
 
        self._root = osutils.mkdtemp(prefix='brz-diff-')
 
685
        self._root = tempfile.mkdtemp(prefix='bzr-diff-')
714
686
 
715
687
    @classmethod
716
688
    def from_string(klass, command_string, old_tree, new_tree, to_file,
717
689
                    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'])
 
690
        command_template = commands.shlex_split_unicode(command_string)
 
691
        command_template.extend(['%(old_path)s', '%(new_path)s'])
721
692
        return klass(command_template, old_tree, new_tree, to_file,
722
693
                     path_encoding)
723
694
 
724
695
    @classmethod
725
 
    def make_from_diff_tree(klass, command_string, external_diff_options=None):
 
696
    def make_from_diff_tree(klass, command_string):
726
697
        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,
 
698
            return klass.from_string(command_string, diff_tree.old_tree,
731
699
                                     diff_tree.new_tree, diff_tree.to_file)
732
700
        return from_diff_tree
733
701
 
734
702
    def _get_command(self, old_path, new_path):
735
703
        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
 
704
        return [t % my_map for t in self.command_template]
748
705
 
749
706
    def _execute(self, old_path, new_path):
750
707
        command = self._get_command(old_path, new_path)
751
708
        try:
752
709
            proc = subprocess.Popen(command, stdout=subprocess.PIPE,
753
710
                                    cwd=self._root)
754
 
        except OSError as e:
 
711
        except OSError, e:
755
712
            if e.errno == errno.ENOENT:
756
713
                raise errors.ExecutableMissing(command[0])
757
714
            else:
758
715
                raise
759
716
        self.to_file.write(proc.stdout.read())
760
 
        proc.stdout.close()
761
717
        return proc.wait()
762
718
 
763
719
    def _try_symlink_root(self, tree, prefix):
766
722
            return False
767
723
        try:
768
724
            os.symlink(tree.abspath(''), osutils.pathjoin(self._root, prefix))
769
 
        except OSError as e:
 
725
        except OSError, e:
770
726
            if e.errno != errno.EEXIST:
771
727
                raise
772
728
        return True
773
729
 
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):
 
730
    def _write_file(self, file_id, tree, prefix, relpath):
 
731
        full_path = osutils.pathjoin(self._root, prefix, relpath)
 
732
        if self._try_symlink_root(tree, prefix):
811
733
            return full_path
812
734
        parent_dir = osutils.dirname(full_path)
813
735
        try:
814
736
            os.makedirs(parent_dir)
815
 
        except OSError as e:
 
737
        except OSError, e:
816
738
            if e.errno != errno.EEXIST:
817
739
                raise
818
 
        source = tree.get_file(relpath, file_id)
 
740
        source = tree.get_file(file_id, relpath)
819
741
        try:
820
 
            with open(full_path, 'wb') as target:
 
742
            target = open(full_path, 'wb')
 
743
            try:
821
744
                osutils.pumpfile(source, target)
 
745
            finally:
 
746
                target.close()
822
747
        finally:
823
748
            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)
 
749
        osutils.make_readonly(full_path)
 
750
        mtime = tree.get_file_mtime(file_id)
 
751
        os.utime(full_path, (mtime, mtime))
832
752
        return full_path
833
753
 
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)
 
754
    def _prepare_files(self, file_id, old_path, new_path):
 
755
        old_disk_path = self._write_file(file_id, self.old_tree, 'old',
 
756
                                         old_path)
 
757
        new_disk_path = self._write_file(file_id, self.new_tree, 'new',
 
758
                                         new_path)
841
759
        return old_disk_path, new_disk_path
842
760
 
843
761
    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))
 
762
        osutils.rmtree(self._root)
850
763
 
851
764
    def diff(self, file_id, old_path, new_path, old_kind, new_kind):
852
765
        if (old_kind, new_kind) != ('file', 'file'):
853
766
            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()
 
767
        self._prepare_files(file_id, old_path, new_path)
 
768
        self._execute(osutils.pathjoin('old', old_path),
 
769
                      osutils.pathjoin('new', new_path))
874
770
 
875
771
 
876
772
class DiffTree(object):
918
814
    @classmethod
919
815
    def from_trees_options(klass, old_tree, new_tree, to_file,
920
816
                           path_encoding, external_diff_options, old_label,
921
 
                           new_label, using, context_lines):
 
817
                           new_label, using):
922
818
        """Factory for producing a DiffTree.
923
819
 
924
820
        Designed to accept options used by show_diff_trees.
925
 
 
926
821
        :param old_tree: The tree to show as old in the comparison
927
822
        :param new_tree: The tree to show as new in the comparison
928
823
        :param to_file: File to write comparisons to
934
829
        :param using: Commandline to use to invoke an external diff tool
935
830
        """
936
831
        if using is not None:
937
 
            extra_factories = [DiffFromTool.make_from_diff_tree(using, external_diff_options)]
 
832
            extra_factories = [DiffFromTool.make_from_diff_tree(using)]
938
833
        else:
939
834
            extra_factories = []
940
835
        if external_diff_options:
941
836
            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
 
                """
 
837
            def diff_file(olab, olines, nlab, nlines, to_file):
946
838
                external_diff(olab, olines, nlab, nlines, to_file, opts)
947
839
        else:
948
840
            diff_file = internal_diff
949
841
        diff_text = DiffText(old_tree, new_tree, to_file, path_encoding,
950
 
                             old_label, new_label, diff_file, context_lines=context_lines)
 
842
                             old_label, new_label, diff_file)
951
843
        return klass(old_tree, new_tree, to_file, path_encoding, diff_text,
952
844
                     extra_factories)
953
845
 
954
846
    def show_diff(self, specific_files, extra_trees=None):
955
847
        """Write tree diff to self.to_file
956
848
 
957
 
        :param specific_files: the specific files to compare (recursive)
 
849
        :param sepecific_files: the specific files to compare (recursive)
958
850
        :param extra_trees: extra trees to use for mapping paths to file_ids
959
851
        """
960
852
        try:
982
874
                return path.encode(self.path_encoding, "replace")
983
875
        for (file_id, paths, changed_content, versioned, parent, name, kind,
984
876
             executable) in sorted(iterator, key=changes_key):
985
 
            # The root does not get diffed, and items with no known kind (that
986
 
            # is, missing) in both trees are skipped as well.
987
 
            if parent == (None, None) or kind == (None, None):
 
877
            if parent == (None, None):
988
878
                continue
989
879
            oldpath, newpath = paths
990
880
            oldpath_encoded = get_encoded_path(paths[0])
997
887
            properties_changed.extend(get_executable_change(executable[0], executable[1]))
998
888
 
999
889
            if properties_changed:
1000
 
                prop_str = b" (properties changed: %s)" % (
1001
 
                        b", ".join(properties_changed),)
 
890
                prop_str = " (properties changed: %s)" % (", ".join(properties_changed),)
1002
891
            else:
1003
 
                prop_str = b""
 
892
                prop_str = ""
1004
893
 
1005
894
            if (old_present, new_present) == (True, False):
1006
 
                self.to_file.write(b"=== removed %s '%s'\n" %
1007
 
                                   (kind[0].encode('ascii'), oldpath_encoded))
 
895
                self.to_file.write("=== removed %s '%s'\n" %
 
896
                                   (kind[0], oldpath_encoded))
1008
897
                newpath = oldpath
1009
898
            elif (old_present, new_present) == (False, True):
1010
 
                self.to_file.write(b"=== added %s '%s'\n" %
1011
 
                                   (kind[1].encode('ascii'), newpath_encoded))
 
899
                self.to_file.write("=== added %s '%s'\n" %
 
900
                                   (kind[1], newpath_encoded))
1012
901
                oldpath = newpath
1013
902
            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))
 
903
                self.to_file.write("=== renamed %s '%s' => '%s'%s\n" %
 
904
                    (kind[0], oldpath_encoded, newpath_encoded, prop_str))
1016
905
            else:
1017
906
                # if it was produced by iter_changes, it must be
1018
907
                # modified *somehow*, either content or execute bit.
1019
 
                self.to_file.write(b"=== modified %s '%s'%s\n" % (kind[0].encode('ascii'),
 
908
                self.to_file.write("=== modified %s '%s'%s\n" % (kind[0],
1020
909
                                   newpath_encoded, prop_str))
1021
910
            if changed_content:
1022
 
                self._diff(oldpath, newpath, kind[0], kind[1], file_id=file_id)
 
911
                self.diff(file_id, oldpath, newpath)
1023
912
                has_changes = 1
1024
913
            if renamed:
1025
914
                has_changes = 1
1032
921
        :param old_path: The path of the file in the old tree
1033
922
        :param new_path: The path of the file in the new tree
1034
923
        """
1035
 
        if old_path is None:
 
924
        try:
 
925
            old_kind = self.old_tree.kind(file_id)
 
926
        except (errors.NoSuchId, errors.NoSuchFile):
1036
927
            old_kind = None
1037
 
        else:
1038
 
            old_kind = self.old_tree.kind(old_path, file_id)
1039
 
        if new_path is None:
 
928
        try:
 
929
            new_kind = self.new_tree.kind(file_id)
 
930
        except (errors.NoSuchId, errors.NoSuchFile):
1040
931
            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
932
 
1045
 
    def _diff(self, old_path, new_path, old_kind, new_kind, file_id):
1046
933
        result = DiffPath._diff_many(self.differs, file_id, old_path,
1047
 
                                     new_path, old_kind, new_kind)
 
934
                                       new_path, old_kind, new_kind)
1048
935
        if result is DiffPath.CANNOT_DIFF:
1049
936
            error_path = new_path
1050
937
            if error_path is None:
1051
938
                error_path = old_path
1052
939
            raise errors.NoDiffFound(error_path)
1053
 
 
1054
 
 
1055
 
format_registry = Registry()
1056
 
format_registry.register('default', DiffTree)