/brz/remove-bazaar

To get this branch, use:
bzr branch http://gegoxaren.bato24.eu/bzr/brz/remove-bazaar

« back to all changes in this revision

Viewing changes to breezy/diff.py

  • Committer: Breezy landing bot
  • Author(s): Jelmer Vernooij
  • Date: 2019-01-01 23:55:19 UTC
  • mfrom: (7211.12.2 fast-export-default-out)
  • Revision ID: breezy.the.bot@gmail.com-20190101235519-s3qiot9brigweze2
Properly unwrap sys.stdout when writing a fast-export stream.

Merged from https://code.launchpad.net/~jelmer/brz/fast-export-default-out/+merge/359558

Show diffs side-by-side

added added

removed removed

Lines of Context:
breezy/diff.py
1
 
# Copyright (C) 2004, 2005, 2006 Canonical Ltd.
2
 
 
 
1
# Copyright (C) 2005-2014 Canonical Ltd.
 
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
5
5
# the Free Software Foundation; either version 2 of the License, or
6
6
# (at your option) any later version.
7
 
 
 
7
#
8
8
# This program is distributed in the hope that it will be useful,
9
9
# but WITHOUT ANY WARRANTY; without even the implied warranty of
10
10
# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
11
11
# GNU General Public License for more details.
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., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
16
 
 
17
 
from bzrlib.delta import compare_trees
18
 
from bzrlib.errors import BzrError
19
 
import bzrlib.errors as errors
20
 
from bzrlib.symbol_versioning import *
21
 
from bzrlib.trace import mutter
 
15
# Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
 
16
 
 
17
from __future__ import absolute_import
 
18
 
 
19
import difflib
 
20
import os
 
21
import re
 
22
import string
 
23
import sys
 
24
 
 
25
from .lazy_import import lazy_import
 
26
lazy_import(globals(), """
 
27
import errno
 
28
import subprocess
 
29
import tempfile
 
30
 
 
31
from breezy import (
 
32
    cmdline,
 
33
    controldir,
 
34
    errors,
 
35
    osutils,
 
36
    patiencediff,
 
37
    textfile,
 
38
    timestamp,
 
39
    views,
 
40
    )
 
41
 
 
42
from breezy.workingtree import WorkingTree
 
43
from breezy.i18n import gettext
 
44
""")
 
45
 
 
46
from .registry import (
 
47
    Registry,
 
48
    )
 
49
from .sixish import text_type
 
50
from .trace import mutter, note, warning
 
51
from .tree import FileTimestampUnavailable
 
52
 
 
53
 
 
54
DEFAULT_CONTEXT_AMOUNT = 3
 
55
 
 
56
 
 
57
class AtTemplate(string.Template):
 
58
    """Templating class that uses @ instead of $."""
 
59
 
 
60
    delimiter = '@'
 
61
 
22
62
 
23
63
# TODO: Rather than building a changeset object, we should probably
24
64
# invoke callbacks on an object.  That object can either accumulate a
25
65
# list, write them out directly, etc etc.
26
66
 
27
 
def internal_diff(old_filename, oldlines, new_filename, newlines, to_file):
28
 
    import difflib
29
 
    
 
67
 
 
68
class _PrematchedMatcher(difflib.SequenceMatcher):
 
69
    """Allow SequenceMatcher operations to use predetermined blocks"""
 
70
 
 
71
    def __init__(self, matching_blocks):
 
72
        difflib.SequenceMatcher(self, None, None)
 
73
        self.matching_blocks = matching_blocks
 
74
        self.opcodes = None
 
75
 
 
76
 
 
77
def internal_diff(old_label, oldlines, new_label, newlines, to_file,
 
78
                  allow_binary=False, sequence_matcher=None,
 
79
                  path_encoding='utf8', context_lines=DEFAULT_CONTEXT_AMOUNT):
30
80
    # FIXME: difflib is wrong if there is no trailing newline.
31
81
    # The syntax used by patch seems to be "\ No newline at
32
82
    # end of file" following the last diff line from that
37
87
    # In the meantime we at least make sure the patch isn't
38
88
    # mangled.
39
89
 
40
 
 
41
 
    # Special workaround for Python2.3, where difflib fails if
42
 
    # both sequences are empty.
43
 
    if not oldlines and not newlines:
44
 
        return
45
 
 
46
 
    ud = difflib.unified_diff(oldlines, newlines,
47
 
                              fromfile=old_filename+'\t', 
48
 
                              tofile=new_filename+'\t')
 
90
    if allow_binary is False:
 
91
        textfile.check_text_lines(oldlines)
 
92
        textfile.check_text_lines(newlines)
 
93
 
 
94
    if sequence_matcher is None:
 
95
        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)
49
102
 
50
103
    ud = list(ud)
 
104
    if len(ud) == 0:  # Identical contents, nothing to do
 
105
        return
51
106
    # work-around for difflib being too smart for its own good
52
107
    # if /dev/null is "1,0", patch won't recognize it as /dev/null
53
108
    if not oldlines:
54
 
        ud[2] = ud[2].replace('-1,0', '-0,0')
 
109
        ud[2] = ud[2].replace(b'-1,0', b'-0,0')
55
110
    elif not newlines:
56
 
        ud[2] = ud[2].replace('+1,0', '+0,0')
57
 
    # work around for difflib emitting random spaces after the label
58
 
    ud[0] = ud[0][:-2] + '\n'
59
 
    ud[1] = ud[1][:-2] + '\n'
 
111
        ud[2] = ud[2].replace(b'+1,0', b'+0,0')
60
112
 
61
113
    for line in ud:
62
114
        to_file.write(line)
63
 
        if not line.endswith('\n'):
64
 
            to_file.write("\n\\ No newline at end of file\n")
65
 
    print >>to_file
66
 
 
67
 
 
68
 
def external_diff(old_filename, oldlines, new_filename, newlines, to_file,
 
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')
 
118
 
 
119
 
 
120
def _spawn_external_diff(diffcmd, capture_errors=True):
 
121
    """Spawn the external diff process, and return the child handle.
 
122
 
 
123
    :param diffcmd: The command list to spawn
 
124
    :param capture_errors: Capture stderr as well as setting LANG=C
 
125
        and LC_ALL=C. This lets us read and understand the output of diff,
 
126
        and respond to any errors.
 
127
    :return: A Popen object.
 
128
    """
 
129
    if capture_errors:
 
130
        # construct minimal environment
 
131
        env = {}
 
132
        path = os.environ.get('PATH')
 
133
        if path is not None:
 
134
            env['PATH'] = path
 
135
        env['LANGUAGE'] = 'C'   # on win32 only LANGUAGE has effect
 
136
        env['LANG'] = 'C'
 
137
        env['LC_ALL'] = 'C'
 
138
        stderr = subprocess.PIPE
 
139
    else:
 
140
        env = None
 
141
        stderr = None
 
142
 
 
143
    try:
 
144
        pipe = subprocess.Popen(diffcmd,
 
145
                                stdin=subprocess.PIPE,
 
146
                                stdout=subprocess.PIPE,
 
147
                                stderr=stderr,
 
148
                                env=env)
 
149
    except OSError as e:
 
150
        if e.errno == errno.ENOENT:
 
151
            raise errors.NoDiff(str(e))
 
152
        raise
 
153
 
 
154
    return pipe
 
155
 
 
156
 
 
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,
69
194
                  diff_opts):
70
195
    """Display a diff by calling out to the external diff program."""
71
 
    import sys
72
 
    
73
 
    if to_file != sys.stdout:
74
 
        raise NotImplementedError("sorry, can't send external diff other than to stdout yet",
75
 
                                  to_file)
76
 
 
77
196
    # make sure our own output is properly ordered before the diff
78
197
    to_file.flush()
79
198
 
80
 
    from tempfile import NamedTemporaryFile
81
 
    import os
82
 
 
83
 
    oldtmpf = NamedTemporaryFile()
84
 
    newtmpf = NamedTemporaryFile()
 
199
    oldtmp_fd, old_abspath = tempfile.mkstemp(prefix='brz-diff-old-')
 
200
    newtmp_fd, new_abspath = tempfile.mkstemp(prefix='brz-diff-new-')
 
201
    oldtmpf = os.fdopen(oldtmp_fd, 'wb')
 
202
    newtmpf = os.fdopen(newtmp_fd, 'wb')
85
203
 
86
204
    try:
87
205
        # TODO: perhaps a special case for comparing to or from the empty
94
212
        oldtmpf.writelines(oldlines)
95
213
        newtmpf.writelines(newlines)
96
214
 
97
 
        oldtmpf.flush()
98
 
        newtmpf.flush()
 
215
        oldtmpf.close()
 
216
        newtmpf.close()
99
217
 
100
218
        if not diff_opts:
101
219
            diff_opts = []
 
220
        if sys.platform == 'win32':
 
221
            # Popen doesn't do the proper encoding for external commands
 
222
            # Since we are dealing with an ANSI api, use mbcs encoding
 
223
            old_label = old_label.encode('mbcs')
 
224
            new_label = new_label.encode('mbcs')
102
225
        diffcmd = ['diff',
103
 
                   '--label', old_filename+'\t',
104
 
                   oldtmpf.name,
105
 
                   '--label', new_filename+'\t',
106
 
                   newtmpf.name]
107
 
 
108
 
        # diff only allows one style to be specified; they don't override.
109
 
        # note that some of these take optargs, and the optargs can be
110
 
        # directly appended to the options.
111
 
        # this is only an approximate parser; it doesn't properly understand
112
 
        # the grammar.
113
 
        for s in ['-c', '-u', '-C', '-U',
114
 
                  '-e', '--ed',
115
 
                  '-q', '--brief',
116
 
                  '--normal',
117
 
                  '-n', '--rcs',
118
 
                  '-y', '--side-by-side',
119
 
                  '-D', '--ifdef']:
120
 
            for j in diff_opts:
121
 
                if j.startswith(s):
122
 
                    break
123
 
            else:
124
 
                continue
125
 
            break
126
 
        else:
127
 
            diffcmd.append('-u')
128
 
                  
 
226
                   '--label', old_label,
 
227
                   old_abspath,
 
228
                   '--label', new_label,
 
229
                   new_abspath,
 
230
                   '--binary',
 
231
                   ]
 
232
 
 
233
        diff_opts = default_style_unified(diff_opts)
 
234
 
129
235
        if diff_opts:
130
236
            diffcmd.extend(diff_opts)
131
237
 
132
 
        rc = os.spawnvp(os.P_WAIT, 'diff', diffcmd)
133
 
        
134
 
        if rc != 0 and rc != 1:
 
238
        pipe = _spawn_external_diff(diffcmd, capture_errors=True)
 
239
        out, err = pipe.communicate()
 
240
        rc = pipe.returncode
 
241
 
 
242
        # internal_diff() adds a trailing newline, add one here for consistency
 
243
        out += b'\n'
 
244
        if rc == 2:
 
245
            # 'diff' gives retcode == 2 for all sorts of errors
 
246
            # one of those is 'Binary files differ'.
 
247
            # Bad options could also be the problem.
 
248
            # 'Binary files' is not a real error, so we suppress that error.
 
249
            lang_c_out = out
 
250
 
 
251
            # Since we got here, we want to make sure to give an i18n error
 
252
            pipe = _spawn_external_diff(diffcmd, capture_errors=False)
 
253
            out, err = pipe.communicate()
 
254
 
 
255
            # Write out the new i18n diff response
 
256
            to_file.write(out + b'\n')
 
257
            if pipe.returncode != 2:
 
258
                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,))
 
262
 
 
263
            first_line = lang_c_out.split(b'\n', 1)[0]
 
264
            # Starting with diffutils 2.8.4 the word "binary" was dropped.
 
265
            m = re.match(b'^(binary )?files.*differ$', first_line, re.I)
 
266
            if m is None:
 
267
                raise errors.BzrError('external diff failed with exit code 2;'
 
268
                                      ' command: %r' % (diffcmd,))
 
269
            else:
 
270
                # Binary files differ, just return
 
271
                return
 
272
 
 
273
        # If we got to here, we haven't written out the output of diff
 
274
        # do so now
 
275
        to_file.write(out)
 
276
        if rc not in (0, 1):
135
277
            # returns 1 if files differ; that's OK
136
278
            if rc < 0:
137
279
                msg = 'signal %d' % (-rc)
138
280
            else:
139
281
                msg = 'exit code %d' % rc
140
 
                
141
 
            raise BzrError('external diff failed with %s; command: %r' % (rc, diffcmd))
 
282
 
 
283
            raise errors.BzrError('external diff failed with %s; command: %r'
 
284
                                  % (msg, diffcmd))
 
285
 
142
286
    finally:
143
287
        oldtmpf.close()                 # and delete
144
288
        newtmpf.close()
145
289
 
146
 
 
147
 
@deprecated_function(zero_eight)
148
 
def show_diff(b, from_spec, specific_files, external_diff_options=None,
149
 
              revision2=None, output=None, b2=None):
150
 
    """Shortcut for showing the diff to the working tree.
151
 
 
152
 
    Please use show_diff_trees instead.
153
 
 
154
 
    b
155
 
        Branch.
156
 
 
157
 
    revision
158
 
        None for 'basis tree', or otherwise the old revision to compare against.
159
 
    
160
 
    The more general form is show_diff_trees(), where the caller
161
 
    supplies any two trees.
 
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):
 
306
    """Get the trees and specific files to diff given a list of paths.
 
307
 
 
308
    This method works out the trees to be diff'ed and the files of
 
309
    interest within those trees.
 
310
 
 
311
    :param path_list:
 
312
        the list of arguments passed to the diff command
 
313
    :param revision_specs:
 
314
        Zero, one or two RevisionSpecs from the diff command line,
 
315
        saying what revisions to compare.
 
316
    :param old_url:
 
317
        The url of the old branch or tree. If None, the tree to use is
 
318
        taken from the first path, if any, or the current working tree.
 
319
    :param new_url:
 
320
        The url of the new branch or tree. If None, the tree to use is
 
321
        taken from the first path, if any, or the current working tree.
 
322
    :param add_cleanup:
 
323
        a callable like Command.add_cleanup.  get_trees_and_branches_to_diff
 
324
        will register cleanups that must be run to unlock the trees, etc.
 
325
    :param apply_view:
 
326
        if True and a view is set, apply the view or check that the paths
 
327
        are within it
 
328
    :returns:
 
329
        a tuple of (old_tree, new_tree, old_branch, new_branch,
 
330
        specific_files, extra_trees) where extra_trees is a sequence of
 
331
        additional trees to search in for file-ids.  The trees and branches
 
332
        will be read-locked until the cleanups registered via the add_cleanup
 
333
        param are run.
162
334
    """
163
 
    if output is None:
164
 
        import sys
165
 
        output = sys.stdout
 
335
    # Get the old and new revision specs
 
336
    old_revision_spec = None
 
337
    new_revision_spec = None
 
338
    if revision_specs is not None:
 
339
        if len(revision_specs) > 0:
 
340
            old_revision_spec = revision_specs[0]
 
341
            if old_url is None:
 
342
                old_url = old_revision_spec.get_branch()
 
343
        if len(revision_specs) > 1:
 
344
            new_revision_spec = revision_specs[1]
 
345
            if new_url is None:
 
346
                new_url = new_revision_spec.get_branch()
166
347
 
167
 
    if from_spec is None:
168
 
        old_tree = b.bzrdir.open_workingtree()
169
 
        if b2 is None:
170
 
            old_tree = old_tree = old_tree.basis_tree()
 
348
    other_paths = []
 
349
    make_paths_wt_relative = True
 
350
    consider_relpath = True
 
351
    if path_list is None or len(path_list) == 0:
 
352
        # If no path is given, the current working tree is used
 
353
        default_location = u'.'
 
354
        consider_relpath = False
 
355
    elif old_url is not None and new_url is not None:
 
356
        other_paths = path_list
 
357
        make_paths_wt_relative = False
171
358
    else:
172
 
        old_tree = b.repository.revision_tree(from_spec.in_history(b).rev_id)
173
 
 
174
 
    if revision2 is None:
175
 
        if b2 is None:
176
 
            new_tree = b.bzrdir.open_workingtree()
 
359
        default_location = path_list[0]
 
360
        other_paths = path_list[1:]
 
361
 
 
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
    # Get the old location
 
371
    specific_files = []
 
372
    if old_url is None:
 
373
        old_url = default_location
 
374
    working_tree, branch, relpath = \
 
375
        controldir.ControlDir.open_containing_tree_or_branch(old_url)
 
376
    lock_tree_or_branch(working_tree, branch)
 
377
    if consider_relpath and relpath != '':
 
378
        if working_tree is not None and apply_view:
 
379
            views.check_path_in_view(working_tree, relpath)
 
380
        specific_files.append(relpath)
 
381
    old_tree = _get_tree_to_diff(old_revision_spec, working_tree, branch)
 
382
    old_branch = branch
 
383
 
 
384
    # Get the new location
 
385
    if new_url is None:
 
386
        new_url = default_location
 
387
    if new_url != old_url:
 
388
        working_tree, branch, relpath = \
 
389
            controldir.ControlDir.open_containing_tree_or_branch(new_url)
 
390
        lock_tree_or_branch(working_tree, branch)
 
391
        if consider_relpath and relpath != '':
 
392
            if working_tree is not None and apply_view:
 
393
                views.check_path_in_view(working_tree, relpath)
 
394
            specific_files.append(relpath)
 
395
    new_tree = _get_tree_to_diff(new_revision_spec, working_tree, branch,
 
396
                                 basis_is_default=working_tree is None)
 
397
    new_branch = branch
 
398
 
 
399
    # Get the specific files (all files is None, no files is [])
 
400
    if make_paths_wt_relative and working_tree is not None:
 
401
        other_paths = working_tree.safe_relpath_files(
 
402
            other_paths,
 
403
            apply_view=apply_view)
 
404
    specific_files.extend(other_paths)
 
405
    if len(specific_files) == 0:
 
406
        specific_files = None
 
407
        if (working_tree is not None and working_tree.supports_views() and
 
408
                apply_view):
 
409
            view_files = working_tree.views.lookup_view()
 
410
            if view_files:
 
411
                specific_files = view_files
 
412
                view_str = views.view_display_str(view_files)
 
413
                note(gettext("*** Ignoring files outside view. View is %s") % view_str)
 
414
 
 
415
    # Get extra trees that ought to be searched for file-ids
 
416
    extra_trees = None
 
417
    if working_tree is not None and working_tree not in (old_tree, new_tree):
 
418
        extra_trees = (working_tree,)
 
419
    return (old_tree, new_tree, old_branch, new_branch,
 
420
            specific_files, extra_trees)
 
421
 
 
422
 
 
423
def _get_tree_to_diff(spec, tree=None, branch=None, basis_is_default=True):
 
424
    if branch is None and tree is not None:
 
425
        branch = tree.branch
 
426
    if spec is None or spec.spec is None:
 
427
        if basis_is_default:
 
428
            if tree is not None:
 
429
                return tree.basis_tree()
 
430
            else:
 
431
                return branch.basis_tree()
177
432
        else:
178
 
            new_tree = b2.bzrdir.open_workingtree()
179
 
    else:
180
 
        new_tree = b.repository.revision_tree(revision2.in_history(b).rev_id)
181
 
 
182
 
    return show_diff_trees(old_tree, new_tree, output, specific_files,
183
 
                           external_diff_options)
184
 
 
185
 
 
186
 
def diff_cmd_helper(tree, specific_files, external_diff_options, 
187
 
                    old_revision_spec=None, new_revision_spec=None):
188
 
    """Helper for cmd_diff.
189
 
 
190
 
   tree 
191
 
        A WorkingTree
192
 
 
193
 
    specific_files
194
 
        The specific files to compare, or None
195
 
 
196
 
    external_diff_options
197
 
        If non-None, run an external diff, and pass it these options
198
 
 
199
 
    old_revision_spec
200
 
        If None, use basis tree as old revision, otherwise use the tree for
201
 
        the specified revision. 
202
 
 
203
 
    new_revision_spec
204
 
        If None, use working tree as new revision, otherwise use the tree for
205
 
        the specified revision.
206
 
    
207
 
    The more general form is show_diff_trees(), where the caller
208
 
    supplies any two trees.
209
 
    """
210
 
    import sys
211
 
    output = sys.stdout
212
 
    def spec_tree(spec):
213
 
        revision_id = spec.in_store(tree.branch).rev_id
214
 
        return tree.branch.repository.revision_tree(revision_id)
215
 
    if old_revision_spec is None:
216
 
        old_tree = tree.basis_tree()
217
 
    else:
218
 
        old_tree = spec_tree(old_revision_spec)
219
 
 
220
 
    if new_revision_spec is None:
221
 
        new_tree = tree
222
 
    else:
223
 
        new_tree = spec_tree(new_revision_spec)
224
 
 
225
 
    return show_diff_trees(old_tree, new_tree, sys.stdout, specific_files,
226
 
                           external_diff_options)
 
433
            return tree
 
434
    return spec.as_tree(branch)
227
435
 
228
436
 
229
437
def show_diff_trees(old_tree, new_tree, to_file, specific_files=None,
230
 
                    external_diff_options=None):
 
438
                    external_diff_options=None,
 
439
                    old_label='a/', new_label='b/',
 
440
                    extra_trees=None,
 
441
                    path_encoding='utf8',
 
442
                    using=None,
 
443
                    format_cls=None,
 
444
                    context=DEFAULT_CONTEXT_AMOUNT):
231
445
    """Show in text form the changes from one tree to another.
232
446
 
233
 
    to_files
234
 
        If set, include only changes to these files.
235
 
 
236
 
    external_diff_options
237
 
        If set, use an external GNU diff and pass these options.
 
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)
238
456
    """
239
 
    old_tree.lock_read()
240
 
    try:
 
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():
 
462
        if extra_trees is not None:
 
463
            for tree in extra_trees:
 
464
                tree.lock_read()
241
465
        new_tree.lock_read()
242
466
        try:
243
 
            return _show_diff_trees(old_tree, new_tree, to_file,
244
 
                                    specific_files, external_diff_options)
 
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)
 
472
            return differ.show_diff(specific_files, extra_trees)
245
473
        finally:
246
474
            new_tree.unlock()
247
 
    finally:
248
 
        old_tree.unlock()
249
 
 
250
 
 
251
 
def _show_diff_trees(old_tree, new_tree, to_file,
252
 
                     specific_files, external_diff_options):
253
 
 
254
 
    # TODO: Options to control putting on a prefix or suffix, perhaps
255
 
    # as a format string?
256
 
    old_label = 'a/'
257
 
    new_label = 'b/'
258
 
 
259
 
    DEVNULL = '/dev/null'
260
 
    # Windows users, don't panic about this filename -- it is a
261
 
    # special signal to GNU patch that the file should be created or
262
 
    # deleted respectively.
263
 
 
264
 
    # TODO: Generation of pseudo-diffs for added/deleted files could
265
 
    # be usefully made into a much faster special case.
266
 
 
267
 
    _raise_if_doubly_unversioned(specific_files, old_tree, new_tree)
268
 
 
269
 
    if external_diff_options:
270
 
        assert isinstance(external_diff_options, basestring)
271
 
        opts = external_diff_options.split()
272
 
        def diff_file(olab, olines, nlab, nlines, to_file):
273
 
            external_diff(olab, olines, nlab, nlines, to_file, opts)
274
 
    else:
275
 
        diff_file = internal_diff
276
 
    
277
 
    delta = compare_trees(old_tree, new_tree, want_unchanged=False,
278
 
                          specific_files=specific_files)
279
 
 
280
 
    has_changes = 0
281
 
    for path, file_id, kind in delta.removed:
282
 
        has_changes = 1
283
 
        print >>to_file, '=== removed %s %r' % (kind, old_label + path)
284
 
        old_tree.inventory[file_id].diff(diff_file, old_label + path, old_tree,
285
 
                                         DEVNULL, None, None, to_file)
286
 
    for path, file_id, kind in delta.added:
287
 
        has_changes = 1
288
 
        print >>to_file, '=== added %s %r' % (kind, new_label + path)
289
 
        new_tree.inventory[file_id].diff(diff_file, new_label + path, new_tree,
290
 
                                         DEVNULL, None, None, to_file, 
291
 
                                         reverse=True)
292
 
    for (old_path, new_path, file_id, kind,
293
 
         text_modified, meta_modified) in delta.renamed:
294
 
        has_changes = 1
295
 
        prop_str = get_prop_change(meta_modified)
296
 
        print >>to_file, '=== renamed %s %r => %r%s' % (
297
 
                    kind, old_label + old_path, new_label + new_path, prop_str)
298
 
        _maybe_diff_file_or_symlink(old_label, old_path, old_tree, file_id,
299
 
                                    new_label, new_path, new_tree,
300
 
                                    text_modified, kind, to_file, diff_file)
301
 
    for path, file_id, kind, text_modified, meta_modified in delta.modified:
302
 
        has_changes = 1
303
 
        prop_str = get_prop_change(meta_modified)
304
 
        print >>to_file, '=== modified %s %r%s' % (kind, old_label + path,
305
 
                    prop_str)
306
 
        if text_modified:
307
 
            _maybe_diff_file_or_symlink(old_label, path, old_tree, file_id,
308
 
                                        new_label, path, new_tree,
309
 
                                        True, kind, to_file, diff_file)
310
 
 
311
 
    return has_changes
312
 
 
313
 
 
314
 
def _raise_if_doubly_unversioned(specific_files, old_tree, new_tree):
315
 
    """Complain if paths are not versioned in either tree."""
316
 
    if not specific_files:
317
 
        return
318
 
    old_unversioned = old_tree.filter_unversioned_files(specific_files)
319
 
    new_unversioned = new_tree.filter_unversioned_files(specific_files)
320
 
    unversioned = old_unversioned.intersection(new_unversioned)
321
 
    if unversioned:
322
 
        raise errors.PathsNotVersionedError(sorted(unversioned))
323
 
    
324
 
 
325
 
def get_prop_change(meta_modified):
326
 
    if meta_modified:
327
 
        return " (properties changed)"
328
 
    else:
329
 
        return  ""
330
 
 
331
 
 
332
 
def _maybe_diff_file_or_symlink(old_label, old_path, old_tree, file_id,
333
 
                                new_label, new_path, new_tree, text_modified,
334
 
                                kind, to_file, diff_file):
335
 
    if text_modified:
336
 
        new_entry = new_tree.inventory[file_id]
337
 
        old_tree.inventory[file_id].diff(diff_file,
338
 
                                         old_label + old_path, old_tree,
339
 
                                         new_label + new_path, new_entry, 
340
 
                                         new_tree, to_file)
 
475
            if extra_trees is not None:
 
476
                for tree in extra_trees:
 
477
                    tree.unlock()
 
478
 
 
479
 
 
480
def _patch_header_date(tree, path):
 
481
    """Returns a timestamp suitable for use in a patch header."""
 
482
    try:
 
483
        mtime = tree.get_file_mtime(path)
 
484
    except FileTimestampUnavailable:
 
485
        mtime = 0
 
486
    return timestamp.format_patch_date(mtime)
 
487
 
 
488
 
 
489
def get_executable_change(old_is_x, new_is_x):
 
490
    descr = {True: b"+x", False: b"-x", None: b"??"}
 
491
    if old_is_x != new_is_x:
 
492
        return [b"%s to %s" % (descr[old_is_x], descr[new_is_x],)]
 
493
    else:
 
494
        return []
 
495
 
 
496
 
 
497
class DiffPath(object):
 
498
    """Base type for command object that compare files"""
 
499
 
 
500
    # The type or contents of the file were unsuitable for diffing
 
501
    CANNOT_DIFF = 'CANNOT_DIFF'
 
502
    # The file has changed in a semantic way
 
503
    CHANGED = 'CHANGED'
 
504
    # The file content may have changed, but there is no semantic change
 
505
    UNCHANGED = 'UNCHANGED'
 
506
 
 
507
    def __init__(self, old_tree, new_tree, to_file, path_encoding='utf-8'):
 
508
        """Constructor.
 
509
 
 
510
        :param old_tree: The tree to show as the old tree in the comparison
 
511
        :param new_tree: The tree to show as new in the comparison
 
512
        :param to_file: The file to write comparison data to
 
513
        :param path_encoding: The character encoding to write paths in
 
514
        """
 
515
        self.old_tree = old_tree
 
516
        self.new_tree = new_tree
 
517
        self.to_file = to_file
 
518
        self.path_encoding = path_encoding
 
519
 
 
520
    def finish(self):
 
521
        pass
 
522
 
 
523
    @classmethod
 
524
    def from_diff_tree(klass, diff_tree):
 
525
        return klass(diff_tree.old_tree, diff_tree.new_tree,
 
526
                     diff_tree.to_file, diff_tree.path_encoding)
 
527
 
 
528
    @staticmethod
 
529
    def _diff_many(differs, old_path, new_path, old_kind, new_kind):
 
530
        for file_differ in differs:
 
531
            result = file_differ.diff(old_path, new_path, old_kind, new_kind)
 
532
            if result is not DiffPath.CANNOT_DIFF:
 
533
                return result
 
534
        else:
 
535
            return DiffPath.CANNOT_DIFF
 
536
 
 
537
 
 
538
class DiffKindChange(object):
 
539
    """Special differ for file kind changes.
 
540
 
 
541
    Represents kind change as deletion + creation.  Uses the other differs
 
542
    to do this.
 
543
    """
 
544
 
 
545
    def __init__(self, differs):
 
546
        self.differs = differs
 
547
 
 
548
    def finish(self):
 
549
        pass
 
550
 
 
551
    @classmethod
 
552
    def from_diff_tree(klass, diff_tree):
 
553
        return klass(diff_tree.differs)
 
554
 
 
555
    def diff(self, old_path, new_path, old_kind, new_kind):
 
556
        """Perform comparison
 
557
 
 
558
        :param old_path: Path of the file in the old tree
 
559
        :param new_path: Path of the file in the new tree
 
560
        :param old_kind: Old file-kind of the file
 
561
        :param new_kind: New file-kind of the file
 
562
        """
 
563
        if None in (old_kind, new_kind):
 
564
            return DiffPath.CANNOT_DIFF
 
565
        result = DiffPath._diff_many(
 
566
            self.differs, old_path, new_path, old_kind, None)
 
567
        if result is DiffPath.CANNOT_DIFF:
 
568
            return result
 
569
        return DiffPath._diff_many(
 
570
            self.differs, old_path, new_path, None, new_kind)
 
571
 
 
572
 
 
573
class DiffDirectory(DiffPath):
 
574
 
 
575
    def diff(self, old_path, new_path, old_kind, new_kind):
 
576
        """Perform comparison between two directories.  (dummy)
 
577
 
 
578
        """
 
579
        if 'directory' not in (old_kind, new_kind):
 
580
            return self.CANNOT_DIFF
 
581
        if old_kind not in ('directory', None):
 
582
            return self.CANNOT_DIFF
 
583
        if new_kind not in ('directory', None):
 
584
            return self.CANNOT_DIFF
 
585
        return self.CHANGED
 
586
 
 
587
 
 
588
class DiffSymlink(DiffPath):
 
589
 
 
590
    def diff(self, old_path, new_path, old_kind, new_kind):
 
591
        """Perform comparison between two symlinks
 
592
 
 
593
        :param old_path: Path of the file in the old tree
 
594
        :param new_path: Path of the file in the new tree
 
595
        :param old_kind: Old file-kind of the file
 
596
        :param new_kind: New file-kind of the file
 
597
        """
 
598
        if 'symlink' not in (old_kind, new_kind):
 
599
            return self.CANNOT_DIFF
 
600
        if old_kind == 'symlink':
 
601
            old_target = self.old_tree.get_symlink_target(old_path)
 
602
        elif old_kind is None:
 
603
            old_target = None
 
604
        else:
 
605
            return self.CANNOT_DIFF
 
606
        if new_kind == 'symlink':
 
607
            new_target = self.new_tree.get_symlink_target(new_path)
 
608
        elif new_kind is None:
 
609
            new_target = None
 
610
        else:
 
611
            return self.CANNOT_DIFF
 
612
        return self.diff_symlink(old_target, new_target)
 
613
 
 
614
    def diff_symlink(self, old_target, new_target):
 
615
        if old_target is None:
 
616
            self.to_file.write(b'=== target is \'%s\'\n' %
 
617
                               new_target.encode(self.path_encoding, 'replace'))
 
618
        elif new_target is None:
 
619
            self.to_file.write(b'=== target was \'%s\'\n' %
 
620
                               old_target.encode(self.path_encoding, 'replace'))
 
621
        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')))
 
625
        return self.CHANGED
 
626
 
 
627
 
 
628
class DiffText(DiffPath):
 
629
 
 
630
    # GNU Patch uses the epoch date to detect files that are being added
 
631
    # or removed in a diff.
 
632
    EPOCH_DATE = '1970-01-01 00:00:00 +0000'
 
633
 
 
634
    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):
 
637
        DiffPath.__init__(self, old_tree, new_tree, to_file, path_encoding)
 
638
        self.text_differ = text_differ
 
639
        self.old_label = old_label
 
640
        self.new_label = new_label
 
641
        self.path_encoding = path_encoding
 
642
        self.context_lines = context_lines
 
643
 
 
644
    def diff(self, old_path, new_path, old_kind, new_kind):
 
645
        """Compare two files in unified diff format
 
646
 
 
647
        :param old_path: Path of the file in the old tree
 
648
        :param new_path: Path of the file in the new tree
 
649
        :param old_kind: Old file-kind of the file
 
650
        :param new_kind: New file-kind of the file
 
651
        """
 
652
        if 'file' not in (old_kind, new_kind):
 
653
            return self.CANNOT_DIFF
 
654
        if old_kind == 'file':
 
655
            old_date = _patch_header_date(self.old_tree, old_path)
 
656
        elif old_kind is None:
 
657
            old_date = self.EPOCH_DATE
 
658
        else:
 
659
            return self.CANNOT_DIFF
 
660
        if new_kind == 'file':
 
661
            new_date = _patch_header_date(self.new_tree, new_path)
 
662
        elif new_kind is None:
 
663
            new_date = self.EPOCH_DATE
 
664
        else:
 
665
            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)
 
671
 
 
672
    def diff_text(self, from_path, to_path, from_label, to_label):
 
673
        """Diff the content of given files in two trees
 
674
 
 
675
        :param from_path: The path in the from tree. If None,
 
676
            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,
 
679
            the file is not present in the to tree.
 
680
        """
 
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:
 
687
                return []
 
688
        try:
 
689
            from_text = _get_text(self.old_tree, from_path)
 
690
            to_text = _get_text(self.new_tree, to_path)
 
691
            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)
 
694
        except errors.BinaryFile:
 
695
            self.to_file.write(
 
696
                ("Binary files %s and %s differ\n" %
 
697
                 (from_label, to_label)).encode(self.path_encoding, 'replace'))
 
698
        return self.CHANGED
 
699
 
 
700
 
 
701
class DiffFromTool(DiffPath):
 
702
 
 
703
    def __init__(self, command_template, old_tree, new_tree, to_file,
 
704
                 path_encoding='utf-8'):
 
705
        DiffPath.__init__(self, old_tree, new_tree, to_file, path_encoding)
 
706
        self.command_template = command_template
 
707
        self._root = osutils.mkdtemp(prefix='brz-diff-')
 
708
 
 
709
    @classmethod
 
710
    def from_string(klass, command_string, old_tree, new_tree, to_file,
 
711
                    path_encoding='utf-8'):
 
712
        command_template = cmdline.split(command_string)
 
713
        if '@' not in command_string:
 
714
            command_template.extend(['@old_path', '@new_path'])
 
715
        return klass(command_template, old_tree, new_tree, to_file,
 
716
                     path_encoding)
 
717
 
 
718
    @classmethod
 
719
    def make_from_diff_tree(klass, command_string, external_diff_options=None):
 
720
        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,
 
725
                                     diff_tree.new_tree, diff_tree.to_file)
 
726
        return from_diff_tree
 
727
 
 
728
    def _get_command(self, old_path, new_path):
 
729
        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
 
742
 
 
743
    def _execute(self, old_path, new_path):
 
744
        command = self._get_command(old_path, new_path)
 
745
        try:
 
746
            proc = subprocess.Popen(command, stdout=subprocess.PIPE,
 
747
                                    cwd=self._root)
 
748
        except OSError as e:
 
749
            if e.errno == errno.ENOENT:
 
750
                raise errors.ExecutableMissing(command[0])
 
751
            else:
 
752
                raise
 
753
        self.to_file.write(proc.stdout.read())
 
754
        proc.stdout.close()
 
755
        return proc.wait()
 
756
 
 
757
    def _try_symlink_root(self, tree, prefix):
 
758
        if (getattr(tree, 'abspath', None) is None or
 
759
                not osutils.host_os_dereferences_symlinks()):
 
760
            return False
 
761
        try:
 
762
            os.symlink(tree.abspath(''), osutils.pathjoin(self._root, prefix))
 
763
        except OSError as e:
 
764
            if e.errno != errno.EEXIST:
 
765
                raise
 
766
        return True
 
767
 
 
768
    @staticmethod
 
769
    def _fenc():
 
770
        """Returns safe encoding for passing file path to diff tool"""
 
771
        if sys.platform == 'win32':
 
772
            return 'mbcs'
 
773
        else:
 
774
            # Don't fallback to 'utf-8' because subprocess may not be able to
 
775
            # handle utf-8 correctly when locale is not utf-8.
 
776
            return sys.getfilesystemencoding() or 'ascii'
 
777
 
 
778
    def _is_safepath(self, path):
 
779
        """Return true if `path` may be able to pass to subprocess."""
 
780
        fenc = self._fenc()
 
781
        try:
 
782
            return path == path.encode(fenc).decode(fenc)
 
783
        except UnicodeError:
 
784
            return False
 
785
 
 
786
    def _safe_filename(self, prefix, relpath):
 
787
        """Replace unsafe character in `relpath` then join `self._root`,
 
788
        `prefix` and `relpath`."""
 
789
        fenc = self._fenc()
 
790
        # encoded_str.replace('?', '_') may break multibyte char.
 
791
        # So we should encode, decode, then replace(u'?', u'_')
 
792
        relpath_tmp = relpath.encode(fenc, 'replace').decode(fenc, 'replace')
 
793
        relpath_tmp = relpath_tmp.replace(u'?', u'_')
 
794
        return osutils.pathjoin(self._root, prefix, relpath_tmp)
 
795
 
 
796
    def _write_file(self, relpath, tree, prefix, force_temp=False,
 
797
                    allow_write=False):
 
798
        if not force_temp and isinstance(tree, WorkingTree):
 
799
            full_path = tree.abspath(relpath)
 
800
            if self._is_safepath(full_path):
 
801
                return full_path
 
802
 
 
803
        full_path = self._safe_filename(prefix, relpath)
 
804
        if not force_temp and self._try_symlink_root(tree, prefix):
 
805
            return full_path
 
806
        parent_dir = osutils.dirname(full_path)
 
807
        try:
 
808
            os.makedirs(parent_dir)
 
809
        except OSError as e:
 
810
            if e.errno != errno.EEXIST:
 
811
                raise
 
812
        source = tree.get_file(relpath)
 
813
        try:
 
814
            with open(full_path, 'wb') as target:
 
815
                osutils.pumpfile(source, target)
 
816
        finally:
 
817
            source.close()
 
818
        try:
 
819
            mtime = tree.get_file_mtime(relpath)
 
820
        except FileTimestampUnavailable:
 
821
            pass
 
822
        else:
 
823
            os.utime(full_path, (mtime, mtime))
 
824
        if not allow_write:
 
825
            osutils.make_readonly(full_path)
 
826
        return full_path
 
827
 
 
828
    def _prepare_files(self, old_path, new_path, force_temp=False,
 
829
                       allow_write_new=False):
 
830
        old_disk_path = self._write_file(
 
831
            old_path, self.old_tree, 'old', force_temp)
 
832
        new_disk_path = self._write_file(
 
833
            new_path, self.new_tree, 'new', force_temp,
 
834
            allow_write=allow_write_new)
 
835
        return old_disk_path, new_disk_path
 
836
 
 
837
    def finish(self):
 
838
        try:
 
839
            osutils.rmtree(self._root)
 
840
        except OSError as e:
 
841
            if e.errno != errno.ENOENT:
 
842
                mutter("The temporary directory \"%s\" was not "
 
843
                       "cleanly removed: %s." % (self._root, e))
 
844
 
 
845
    def diff(self, old_path, new_path, old_kind, new_kind):
 
846
        if (old_kind, new_kind) != ('file', 'file'):
 
847
            return DiffPath.CANNOT_DIFF
 
848
        (old_disk_path, new_disk_path) = self._prepare_files(
 
849
            old_path, new_path)
 
850
        self._execute(old_disk_path, new_disk_path)
 
851
 
 
852
    def edit_file(self, old_path, new_path):
 
853
        """Use this tool to edit a file.
 
854
 
 
855
        A temporary copy will be edited, and the new contents will be
 
856
        returned.
 
857
 
 
858
        :return: The new contents of the file.
 
859
        """
 
860
        old_abs_path, new_abs_path = self._prepare_files(
 
861
            old_path, new_path, allow_write_new=True, force_temp=True)
 
862
        command = self._get_command(old_abs_path, new_abs_path)
 
863
        subprocess.call(command, cwd=self._root)
 
864
        with open(new_abs_path, 'rb') as new_file:
 
865
            return new_file.read()
 
866
 
 
867
 
 
868
class DiffTree(object):
 
869
    """Provides textual representations of the difference between two trees.
 
870
 
 
871
    A DiffTree examines two trees and where a file-id has altered
 
872
    between them, generates a textual representation of the difference.
 
873
    DiffTree uses a sequence of DiffPath objects which are each
 
874
    given the opportunity to handle a given altered fileid. The list
 
875
    of DiffPath objects can be extended globally by appending to
 
876
    DiffTree.diff_factories, or for a specific diff operation by
 
877
    supplying the extra_factories option to the appropriate method.
 
878
    """
 
879
 
 
880
    # list of factories that can provide instances of DiffPath objects
 
881
    # may be extended by plugins.
 
882
    diff_factories = [DiffSymlink.from_diff_tree,
 
883
                      DiffDirectory.from_diff_tree]
 
884
 
 
885
    def __init__(self, old_tree, new_tree, to_file, path_encoding='utf-8',
 
886
                 diff_text=None, extra_factories=None):
 
887
        """Constructor
 
888
 
 
889
        :param old_tree: Tree to show as old in the comparison
 
890
        :param new_tree: Tree to show as new in the comparison
 
891
        :param to_file: File to write comparision to
 
892
        :param path_encoding: Character encoding to write paths in
 
893
        :param diff_text: DiffPath-type object to use as a last resort for
 
894
            diffing text files.
 
895
        :param extra_factories: Factories of DiffPaths to try before any other
 
896
            DiffPaths"""
 
897
        if diff_text is None:
 
898
            diff_text = DiffText(old_tree, new_tree, to_file, path_encoding,
 
899
                                 '', '', internal_diff)
 
900
        self.old_tree = old_tree
 
901
        self.new_tree = new_tree
 
902
        self.to_file = to_file
 
903
        self.path_encoding = path_encoding
 
904
        self.differs = []
 
905
        if extra_factories is not None:
 
906
            self.differs.extend(f(self) for f in extra_factories)
 
907
        self.differs.extend(f(self) for f in self.diff_factories)
 
908
        self.differs.extend([diff_text, DiffKindChange.from_diff_tree(self)])
 
909
 
 
910
    @classmethod
 
911
    def from_trees_options(klass, old_tree, new_tree, to_file,
 
912
                           path_encoding, external_diff_options, old_label,
 
913
                           new_label, using, context_lines):
 
914
        """Factory for producing a DiffTree.
 
915
 
 
916
        Designed to accept options used by show_diff_trees.
 
917
 
 
918
        :param old_tree: The tree to show as old in the comparison
 
919
        :param new_tree: The tree to show as new in the comparison
 
920
        :param to_file: File to write comparisons to
 
921
        :param path_encoding: Character encoding to use for writing paths
 
922
        :param external_diff_options: If supplied, use the installed diff
 
923
            binary to perform file comparison, using supplied options.
 
924
        :param old_label: Prefix to use for old file labels
 
925
        :param new_label: Prefix to use for new file labels
 
926
        :param using: Commandline to use to invoke an external diff tool
 
927
        """
 
928
        if using is not None:
 
929
            extra_factories = [DiffFromTool.make_from_diff_tree(
 
930
                using, external_diff_options)]
 
931
        else:
 
932
            extra_factories = []
 
933
        if external_diff_options:
 
934
            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
                """
 
940
                external_diff(olab, olines, nlab, nlines, to_file, opts)
 
941
        else:
 
942
            diff_file = internal_diff
 
943
        diff_text = DiffText(old_tree, new_tree, to_file, path_encoding,
 
944
                             old_label, new_label, diff_file, context_lines=context_lines)
 
945
        return klass(old_tree, new_tree, to_file, path_encoding, diff_text,
 
946
                     extra_factories)
 
947
 
 
948
    def show_diff(self, specific_files, extra_trees=None):
 
949
        """Write tree diff to self.to_file
 
950
 
 
951
        :param specific_files: the specific files to compare (recursive)
 
952
        :param extra_trees: extra trees to use for mapping paths to file_ids
 
953
        """
 
954
        try:
 
955
            return self._show_diff(specific_files, extra_trees)
 
956
        finally:
 
957
            for differ in self.differs:
 
958
                differ.finish()
 
959
 
 
960
    def _show_diff(self, specific_files, extra_trees):
 
961
        # TODO: Generation of pseudo-diffs for added/deleted files could
 
962
        # be usefully made into a much faster special case.
 
963
        iterator = self.new_tree.iter_changes(self.old_tree,
 
964
                                              specific_files=specific_files,
 
965
                                              extra_trees=extra_trees,
 
966
                                              require_versioned=True)
 
967
        has_changes = 0
 
968
 
 
969
        def changes_key(change):
 
970
            old_path, new_path = change[1]
 
971
            path = new_path
 
972
            if path is None:
 
973
                path = old_path
 
974
            return path
 
975
 
 
976
        def get_encoded_path(path):
 
977
            if path is not None:
 
978
                return path.encode(self.path_encoding, "replace")
 
979
        for (file_id, paths, changed_content, versioned, parent, name, kind,
 
980
             executable) in sorted(iterator, key=changes_key):
 
981
            # The root does not get diffed, and items with no known kind (that
 
982
            # is, missing) in both trees are skipped as well.
 
983
            if parent == (None, None) or kind == (None, None):
 
984
                continue
 
985
            oldpath, newpath = paths
 
986
            oldpath_encoded = get_encoded_path(paths[0])
 
987
            newpath_encoded = get_encoded_path(paths[1])
 
988
            old_present = (kind[0] is not None and versioned[0])
 
989
            new_present = (kind[1] is not None and versioned[1])
 
990
            renamed = (parent[0], name[0]) != (parent[1], name[1])
 
991
 
 
992
            properties_changed = []
 
993
            properties_changed.extend(
 
994
                get_executable_change(executable[0], executable[1]))
 
995
 
 
996
            if properties_changed:
 
997
                prop_str = b" (properties changed: %s)" % (
 
998
                    b", ".join(properties_changed),)
 
999
            else:
 
1000
                prop_str = b""
 
1001
 
 
1002
            if (old_present, new_present) == (True, False):
 
1003
                self.to_file.write(b"=== removed %s '%s'\n" %
 
1004
                                   (kind[0].encode('ascii'), oldpath_encoded))
 
1005
                newpath = oldpath
 
1006
            elif (old_present, new_present) == (False, True):
 
1007
                self.to_file.write(b"=== added %s '%s'\n" %
 
1008
                                   (kind[1].encode('ascii'), newpath_encoded))
 
1009
                oldpath = newpath
 
1010
            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))
 
1013
            else:
 
1014
                # if it was produced by iter_changes, it must be
 
1015
                # 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))
 
1018
            if changed_content:
 
1019
                self._diff(oldpath, newpath, kind[0], kind[1])
 
1020
                has_changes = 1
 
1021
            if renamed:
 
1022
                has_changes = 1
 
1023
        return has_changes
 
1024
 
 
1025
    def diff(self, old_path, new_path):
 
1026
        """Perform a diff of a single file
 
1027
 
 
1028
        :param old_path: The path of the file in the old tree
 
1029
        :param new_path: The path of the file in the new tree
 
1030
        """
 
1031
        if old_path is None:
 
1032
            old_kind = None
 
1033
        else:
 
1034
            old_kind = self.old_tree.kind(old_path)
 
1035
        if new_path is None:
 
1036
            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)
 
1044
        if result is DiffPath.CANNOT_DIFF:
 
1045
            error_path = new_path
 
1046
            if error_path is None:
 
1047
                error_path = old_path
 
1048
            raise errors.NoDiffFound(error_path)
 
1049
 
 
1050
 
 
1051
format_registry = Registry()
 
1052
format_registry.register('default', DiffTree)