/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 mapping.py

Default to using global user id rather than making one up.

Show diffs side-by-side

added added

removed removed

Lines of Context:
mapping.py
1
 
# Copyright (C) 2007-2008 Canonical Ltd
 
1
# Copyright (C) 2007 Canonical Ltd
 
2
# Copyright (C) 2008-2009 Jelmer Vernooij <jelmer@samba.org>
 
3
# Copyright (C) 2008 John Carr
2
4
#
3
5
# This program is free software; you can redistribute it and/or modify
4
6
# it under the terms of the GNU General Public License as published by
16
18
 
17
19
"""Converters, etc for going between Bazaar and Git ids."""
18
20
 
19
 
from bzrlib import errors, foreign, urlutils
20
 
from bzrlib.inventory import ROOT_ID
 
21
import stat
 
22
 
 
23
from bzrlib import (
 
24
    errors,
 
25
    foreign,
 
26
    osutils,
 
27
    trace,
 
28
    urlutils,
 
29
    )
 
30
from bzrlib.inventory import (
 
31
    ROOT_ID,
 
32
    )
21
33
from bzrlib.foreign import (
22
 
        ForeignVcs, 
23
 
        VcsMappingRegistry, 
24
 
        ForeignRevision,
25
 
        )
 
34
    ForeignVcs, 
 
35
    VcsMappingRegistry, 
 
36
    ForeignRevision,
 
37
    )
 
38
from bzrlib.xml_serializer import (
 
39
    escape_invalid_chars,
 
40
    )
 
41
 
 
42
DEFAULT_FILE_MODE = stat.S_IFREG | 0644
 
43
 
26
44
 
27
45
def escape_file_id(file_id):
28
46
    return file_id.replace('_', '__').replace(' ', '_s')
29
47
 
30
48
 
31
49
def unescape_file_id(file_id):
32
 
    return file_id.replace("_s", " ").replace("__", "_")
 
50
    ret = []
 
51
    i = 0
 
52
    while i < len(file_id):
 
53
        if file_id[i] != '_':
 
54
            ret.append(file_id[i])
 
55
        else:
 
56
            if file_id[i+1] == '_':
 
57
                ret.append("_")
 
58
            elif file_id[i+1] == 's':
 
59
                ret.append(" ")
 
60
            else:
 
61
                raise AssertionError("unknown escape character %s" % file_id[i+1])
 
62
            i += 1
 
63
        i += 1
 
64
    return "".join(ret)
 
65
 
 
66
 
 
67
def fix_person_identifier(text):
 
68
    if "<" in text and ">" in text:
 
69
        return text
 
70
    return "%s <%s>" % (text, text)
 
71
 
 
72
 
 
73
def warn_escaped(commit, num_escaped):
 
74
    trace.warning("Escaped %d XML-invalid characters in %s. Will be unable "
 
75
                  "to regenerate the SHA map.", num_escaped, commit)
 
76
 
 
77
 
 
78
def warn_unusual_mode(commit, path, mode):
 
79
    trace.warning("Unusual file mode %o for %s in %s. Will be unable to "
 
80
                  "regenerate the SHA map.", mode, path, commit)
33
81
 
34
82
 
35
83
class BzrGitMapping(foreign.VcsMapping):
55
103
        return bzr_rev_id[len(cls.revid_prefix)+1:], cls()
56
104
 
57
105
    def generate_file_id(self, path):
 
106
        # Git paths are just bytestrings
 
107
        # We must just hope they are valid UTF-8..
58
108
        if path == "":
59
109
            return ROOT_ID
60
 
        return escape_file_id(path.encode('utf-8'))
 
110
        return escape_file_id(path)
 
111
 
 
112
    def parse_file_id(self, file_id):
 
113
        if file_id == ROOT_ID:
 
114
            return ""
 
115
        return unescape_file_id(file_id)
61
116
 
62
117
    def import_commit(self, commit):
63
118
        """Convert a git commit to a bzr revision.
68
123
            raise AssertionError("Commit object can't be None")
69
124
        rev = ForeignRevision(commit.id, self, self.revision_id_foreign_to_bzr(commit.id))
70
125
        rev.parent_ids = tuple([self.revision_id_foreign_to_bzr(p) for p in commit.parents])
71
 
        rev.message = commit.message.decode("utf-8", "replace")
72
 
        rev.committer = str(commit.committer).decode("utf-8", "replace")
 
126
        rev.message, num_escaped = escape_invalid_chars(commit.message.decode("utf-8", "replace"))
 
127
        if num_escaped:
 
128
            warn_escaped(commit.id, num_escaped)
 
129
        rev.committer, num_escaped = escape_invalid_chars(str(commit.committer).decode("utf-8", "replace"))
 
130
        if num_escaped:
 
131
            warn_escaped(commit.id, num_escaped)
73
132
        if commit.committer != commit.author:
74
 
            rev.properties['author'] = str(commit.author).decode("utf-8", "replace")
 
133
            rev.properties['author'], num_escaped = escape_invalid_chars(str(commit.author).decode("utf-8", "replace"))
 
134
            if num_escaped:
 
135
                warn_escaped(commit.id, num_escaped)
 
136
 
 
137
        if commit.commit_time != commit.author_time:
 
138
            rev.properties['author-timestamp'] = str(commit.author_time)
 
139
        if commit.commit_timezone != commit.author_timezone:
 
140
            rev.properties['author-timezone'] = "%d" % (commit.author_timezone, )
75
141
        rev.timestamp = commit.commit_time
76
 
        rev.timezone = 0
 
142
        rev.timezone = commit.commit_timezone
77
143
        return rev
78
144
 
79
145
 
81
147
    revid_prefix = 'git-v1'
82
148
    experimental = False
83
149
 
 
150
    def __str__(self):
 
151
        return self.revid_prefix
 
152
 
84
153
 
85
154
class BzrGitMappingExperimental(BzrGitMappingv1):
86
155
    revid_prefix = 'git-experimental'
107
176
 
108
177
 
109
178
class ForeignGit(ForeignVcs):
110
 
    """Foreign Git."""
 
179
    """The Git Stupid Content Tracker"""
111
180
 
112
181
    def __init__(self):
113
182
        super(ForeignGit, self).__init__(mapping_registry)
121
190
default_mapping = BzrGitMappingv1()
122
191
 
123
192
 
124
 
def inventory_to_tree_and_blobs(repo, mapping, revision_id):
125
 
    from dulwich.objects import Tree, Blob
126
 
    from bzrlib.inventory import InventoryDirectory, InventoryFile
 
193
def text_to_blob(texts, entry):
 
194
    from dulwich.objects import Blob
 
195
    text = texts.get_record_stream([(entry.file_id, entry.revision)], 'unordered', True).next().get_bytes_as('fulltext')
 
196
    blob = Blob()
 
197
    blob._text = text
 
198
    return blob
 
199
 
 
200
 
 
201
def symlink_to_blob(entry):
 
202
    from dulwich.objects import Blob
 
203
    blob = Blob()
 
204
    blob._text = entry.symlink_target
 
205
    return blob
 
206
 
 
207
def mode_is_executable(mode):
 
208
    return bool(mode & 0111)
 
209
 
 
210
def mode_kind(mode):
 
211
    entry_kind = (mode & 0700000) / 0100000
 
212
    if entry_kind == 0:
 
213
        return 'directory'
 
214
    elif entry_kind == 1:
 
215
        file_kind = (mode & 070000) / 010000
 
216
        if file_kind == 0:
 
217
            return 'file'
 
218
        elif file_kind == 2:
 
219
            return 'symlink'
 
220
        elif file_kind == 6:
 
221
            return 'tree-reference'
 
222
        else:
 
223
            raise AssertionError(
 
224
                "Unknown file kind %d, perms=%o." % (file_kind, mode,))
 
225
    else:
 
226
        raise AssertionError(
 
227
            "Unknown kind, perms=%r." % (mode,))
 
228
 
 
229
 
 
230
def entry_mode(entry):
 
231
    if entry.kind == 'directory':
 
232
        return stat.S_IFDIR
 
233
    elif entry.kind == 'symlink':
 
234
        return stat.S_IFLNK
 
235
    elif entry.kind == 'file':
 
236
        mode = stat.S_IFREG | 0644
 
237
        if entry.executable:
 
238
            mode |= 0111
 
239
        return mode
 
240
    else:
 
241
        raise AssertionError
 
242
 
 
243
 
 
244
def directory_to_tree(entry, lookup_ie_sha1):
 
245
    from dulwich.objects import Tree
 
246
    tree = Tree()
 
247
    for name in sorted(entry.children.keys()):
 
248
        ie = entry.children[name]
 
249
        tree.add(entry_mode(ie), name.encode("utf-8"), lookup_ie_sha1(ie))
 
250
    tree.serialize()
 
251
    return tree
 
252
 
 
253
 
 
254
def inventory_to_tree_and_blobs(inventory, texts, mapping, cur=None):
 
255
    """Convert a Bazaar tree to a Git tree.
 
256
 
 
257
    :return: Yields tuples with object sha1, object and path
 
258
    """
 
259
    from dulwich.objects import Tree
127
260
    import stat
128
261
    stack = []
129
 
    cur = ""
 
262
    if cur is None:
 
263
        cur = ""
130
264
    tree = Tree()
131
265
 
132
 
    inv = repo.get_inventory(revision_id)
133
 
 
134
266
    # stack contains the set of trees that we haven't 
135
267
    # finished constructing
136
 
 
137
 
    for path, entry in inv.iter_entries():
138
 
        while stack and not path.startswith(cur):
 
268
    for path, entry in inventory.iter_entries():
 
269
        while stack and not path.startswith(osutils.pathjoin(cur, "")):
 
270
            # We've hit a file that's not a child of the previous path
139
271
            tree.serialize()
140
 
            sha = tree.sha().hexdigest()
141
 
            yield sha, tree, cur
 
272
            sha = tree.id
 
273
            yield sha, tree, cur.encode("utf-8")
142
274
            t = (stat.S_IFDIR, urlutils.basename(cur).encode('UTF-8'), sha)
143
275
            cur, tree = stack.pop()
144
276
            tree.add(*t)
145
277
 
146
 
        if type(entry) == InventoryDirectory:
 
278
        if entry.kind == "directory":
147
279
            stack.append((cur, tree))
148
280
            cur = path
149
281
            tree = Tree()
150
 
 
151
 
        if type(entry) == InventoryFile:
152
 
            #FIXME: We can make potentially make this Lazy to avoid shaing lots of stuff
153
 
            # and having all these objects in memory at once
154
 
            blob = Blob()
155
 
            _, blob._text = repo.iter_files_bytes([(entry.file_id, entry.revision, path)]).next()
156
 
            sha = blob.sha().hexdigest()
157
 
            yield sha, blob, path
158
 
 
 
282
        else:
 
283
            if entry.kind == "file":
 
284
                blob = text_to_blob(texts, entry)
 
285
            elif entry.kind == "symlink":
 
286
                blob = symlink_to_blob(entry)
 
287
            else:
 
288
                raise AssertionError("Unknown kind %s" % entry.kind)
 
289
            sha = blob.id
 
290
            yield sha, blob, path.encode("utf-8")
159
291
            name = urlutils.basename(path).encode("utf-8")
160
 
            mode = stat.S_IFREG | 0644
161
 
            if entry.executable:
162
 
                mode |= 0111
163
 
            tree.add(mode, name, sha)
 
292
            tree.add(entry_mode(entry), name, sha)
164
293
 
165
294
    while len(stack) > 1:
166
295
        tree.serialize()
167
 
        sha = tree.sha().hexdigest()
168
 
        yield sha, tree, cur
 
296
        sha = tree.id
 
297
        yield sha, tree, cur.encode("utf-8")
169
298
        t = (stat.S_IFDIR, urlutils.basename(cur).encode('UTF-8'), sha)
170
299
        cur, tree = stack.pop()
171
300
        tree.add(*t)
172
301
 
173
302
    tree.serialize()
174
 
    yield tree.sha().hexdigest(), tree, cur
 
303
    yield tree.id, tree, cur.encode("utf-8")
175
304
 
176
305
 
177
306
def revision_to_commit(rev, tree_sha, parent_lookup):
183
312
    """
184
313
    from dulwich.objects import Commit
185
314
    commit = Commit()
186
 
    commit._tree = tree_sha
 
315
    commit.tree = tree_sha
187
316
    for p in rev.parent_ids:
188
 
        commit._parents.append(parent_lookup(p))
189
 
    commit._message = rev.message
190
 
    commit._committer = rev.committer
191
 
    if 'author' in rev.properties:
192
 
        commit._author = rev.properties['author']
193
 
    else:
194
 
        commit._author = rev.committer
195
 
    commit._commit_time = long(rev.timestamp)
196
 
    commit.serialize()
 
317
        git_p = parent_lookup(p)
 
318
        if git_p is not None:
 
319
            assert len(git_p) == 40, "unexpected length for %r" % git_p
 
320
            commit.parents.append(git_p)
 
321
    commit.message = rev.message.encode("utf-8")
 
322
    commit.committer = fix_person_identifier(rev.committer.encode("utf-8"))
 
323
    commit.author = fix_person_identifier(rev.get_apparent_authors()[0].encode("utf-8"))
 
324
    commit.commit_time = long(rev.timestamp)
 
325
    if 'author-timestamp' in rev.properties:
 
326
        commit.author_time = long(rev.properties['author-timestamp'])
 
327
    else:
 
328
        commit.author_time = commit.commit_time
 
329
    commit.commit_timezone = rev.timezone
 
330
    if 'author-timezone' in rev.properties:
 
331
        commit.author_timezone = int(rev.properties['author-timezone'])
 
332
    else:
 
333
        commit.author_timezone = commit.commit_timezone 
197
334
    return commit