/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

Cope with map for Tree objects becoming invalid.

Show diffs side-by-side

added added

removed removed

Lines of Context:
mapping.py
1
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
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
 
        ForeignRevision,
23
 
        )
 
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
24
43
 
25
44
 
26
45
def escape_file_id(file_id):
28
47
 
29
48
 
30
49
def unescape_file_id(file_id):
31
 
    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)
32
81
 
33
82
 
34
83
class BzrGitMapping(foreign.VcsMapping):
35
84
    """Class that maps between Git and Bazaar semantics."""
36
85
    experimental = False
37
86
 
38
 
    def revision_id_foreign_to_bzr(self, git_rev_id):
 
87
    def __init__(self):
 
88
        super(BzrGitMapping, self).__init__(foreign_git)
 
89
 
 
90
    def __eq__(self, other):
 
91
        return type(self) == type(other) and self.revid_prefix == other.revid_prefix
 
92
 
 
93
    @classmethod
 
94
    def revision_id_foreign_to_bzr(cls, git_rev_id):
39
95
        """Convert a git revision id handle to a Bazaar revision id."""
40
 
        return "%s:%s" % (self.revid_prefix, git_rev_id)
 
96
        return "%s:%s" % (cls.revid_prefix, git_rev_id)
41
97
 
42
 
    def revision_id_bzr_to_foreign(self, bzr_rev_id):
 
98
    @classmethod
 
99
    def revision_id_bzr_to_foreign(cls, bzr_rev_id):
43
100
        """Convert a Bazaar revision id to a git revision id handle."""
44
 
        if not bzr_rev_id.startswith("%s:" % self.revid_prefix):
45
 
            raise errors.InvalidRevisionId(bzr_rev_id, self)
46
 
        return bzr_rev_id[len(self.revid_prefix)+1:]
47
 
 
48
 
    def show_foreign_revid(self, foreign_revid):
49
 
        return { "git commit": foreign_revid }
 
101
        if not bzr_rev_id.startswith("%s:" % cls.revid_prefix):
 
102
            raise errors.InvalidRevisionId(bzr_rev_id, cls)
 
103
        return bzr_rev_id[len(cls.revid_prefix)+1:], cls()
50
104
 
51
105
    def generate_file_id(self, path):
 
106
        # Git paths are just bytestrings
 
107
        # We must just hope they are valid UTF-8..
52
108
        if path == "":
53
109
            return ROOT_ID
54
 
        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)
55
116
 
56
117
    def import_commit(self, commit):
57
118
        """Convert a git commit to a bzr revision.
62
123
            raise AssertionError("Commit object can't be None")
63
124
        rev = ForeignRevision(commit.id, self, self.revision_id_foreign_to_bzr(commit.id))
64
125
        rev.parent_ids = tuple([self.revision_id_foreign_to_bzr(p) for p in commit.parents])
65
 
        rev.message = commit.message.decode("utf-8", "replace")
66
 
        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)
67
132
        if commit.committer != commit.author:
68
 
            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, )
69
141
        rev.timestamp = commit.commit_time
70
 
        rev.timezone = 0
 
142
        rev.timezone = commit.commit_timezone
71
143
        return rev
72
144
 
73
145
 
74
 
class BzrGitMappingExperimental(BzrGitMapping):
 
146
class BzrGitMappingv1(BzrGitMapping):
 
147
    revid_prefix = 'git-v1'
 
148
    experimental = False
 
149
 
 
150
    def __str__(self):
 
151
        return self.revid_prefix
 
152
 
 
153
 
 
154
class BzrGitMappingExperimental(BzrGitMappingv1):
75
155
    revid_prefix = 'git-experimental'
76
156
    experimental = True
77
157
 
78
158
 
79
 
default_mapping = BzrGitMappingExperimental()
 
159
class GitMappingRegistry(VcsMappingRegistry):
 
160
 
 
161
    def revision_id_bzr_to_foreign(self, bzr_revid):
 
162
        if not bzr_revid.startswith("git-"):
 
163
            raise errors.InvalidRevisionId(bzr_revid, None)
 
164
        (mapping_version, git_sha) = bzr_revid.split(":", 1)
 
165
        mapping = self.get(mapping_version)
 
166
        return mapping.revision_id_bzr_to_foreign(bzr_revid)
 
167
 
 
168
    parse_revision_id = revision_id_bzr_to_foreign
 
169
 
 
170
 
 
171
mapping_registry = GitMappingRegistry()
 
172
mapping_registry.register_lazy('git-v1', "bzrlib.plugins.git.mapping",
 
173
                                   "BzrGitMappingv1")
 
174
mapping_registry.register_lazy('git-experimental', "bzrlib.plugins.git.mapping",
 
175
                                   "BzrGitMappingExperimental")
 
176
 
 
177
 
 
178
class ForeignGit(ForeignVcs):
 
179
    """The Git Stupid Content Tracker"""
 
180
 
 
181
    def __init__(self):
 
182
        super(ForeignGit, self).__init__(mapping_registry)
 
183
 
 
184
    @classmethod
 
185
    def show_foreign_revid(cls, foreign_revid):
 
186
        return { "git commit": foreign_revid }
 
187
 
 
188
 
 
189
foreign_git = ForeignGit()
 
190
default_mapping = BzrGitMappingv1()
 
191
 
 
192
 
 
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
 
 
208
def entry_mode(entry):
 
209
    if entry.kind == 'directory':
 
210
        return stat.S_IFDIR
 
211
    elif entry.kind == 'symlink':
 
212
        return stat.S_IFLNK
 
213
    elif entry.kind == 'file':
 
214
        mode = stat.S_IFREG | 0644
 
215
        if entry.executable:
 
216
            mode |= 0111
 
217
        return mode
 
218
    else:
 
219
        raise AssertionError
 
220
 
 
221
 
 
222
def directory_to_tree(entry, lookup_ie_sha1):
 
223
    from dulwich.objects import Tree
 
224
    tree = Tree()
 
225
    for name in sorted(entry.children.keys()):
 
226
        ie = entry.children[name]
 
227
        tree.add(entry_mode(ie), name.encode("utf-8"), lookup_ie_sha1(ie))
 
228
    tree.serialize()
 
229
    return tree
 
230
 
 
231
 
 
232
def inventory_to_tree_and_blobs(inventory, texts, mapping, cur=None):
 
233
    """Convert a Bazaar tree to a Git tree.
 
234
 
 
235
    :return: Yields tuples with object sha1, object and path
 
236
    """
 
237
    from dulwich.objects import Tree
 
238
    import stat
 
239
    stack = []
 
240
    if cur is None:
 
241
        cur = ""
 
242
    tree = Tree()
 
243
 
 
244
    # stack contains the set of trees that we haven't 
 
245
    # finished constructing
 
246
    for path, entry in inventory.iter_entries():
 
247
        while stack and not path.startswith(osutils.pathjoin(cur, "")):
 
248
            # We've hit a file that's not a child of the previous path
 
249
            tree.serialize()
 
250
            sha = tree.id
 
251
            yield sha, tree, cur.encode("utf-8")
 
252
            t = (stat.S_IFDIR, urlutils.basename(cur).encode('UTF-8'), sha)
 
253
            cur, tree = stack.pop()
 
254
            tree.add(*t)
 
255
 
 
256
        if entry.kind == "directory":
 
257
            stack.append((cur, tree))
 
258
            cur = path
 
259
            tree = Tree()
 
260
        else:
 
261
            if entry.kind == "file":
 
262
                blob = text_to_blob(texts, entry)
 
263
            elif entry.kind == "symlink":
 
264
                blob = symlink_to_blob(entry)
 
265
            else:
 
266
                raise AssertionError("Unknown kind %s" % entry.kind)
 
267
            sha = blob.id
 
268
            yield sha, blob, path.encode("utf-8")
 
269
            name = urlutils.basename(path).encode("utf-8")
 
270
            tree.add(entry_mode(entry), name, sha)
 
271
 
 
272
    while len(stack) > 1:
 
273
        tree.serialize()
 
274
        sha = tree.id
 
275
        yield sha, tree, cur.encode("utf-8")
 
276
        t = (stat.S_IFDIR, urlutils.basename(cur).encode('UTF-8'), sha)
 
277
        cur, tree = stack.pop()
 
278
        tree.add(*t)
 
279
 
 
280
    tree.serialize()
 
281
    yield tree.id, tree, cur.encode("utf-8")
 
282
 
 
283
 
 
284
def revision_to_commit(rev, tree_sha, parent_lookup):
 
285
    """Turn a Bazaar revision in to a Git commit
 
286
 
 
287
    :param tree_sha: Tree sha for the commit
 
288
    :param parent_lookup: Function for looking up the GIT sha equiv of a bzr revision
 
289
    :return dulwich.objects.Commit represent the revision:
 
290
    """
 
291
    from dulwich.objects import Commit
 
292
    commit = Commit()
 
293
    commit.tree = tree_sha
 
294
    for p in rev.parent_ids:
 
295
        git_p = parent_lookup(p)
 
296
        if git_p is not None:
 
297
            assert len(git_p) == 40, "unexpected length for %r" % git_p
 
298
            commit.parents.append(git_p)
 
299
    commit.message = rev.message.encode("utf-8")
 
300
    commit.committer = fix_person_identifier(rev.committer.encode("utf-8"))
 
301
    commit.author = fix_person_identifier(rev.get_apparent_authors()[0].encode("utf-8"))
 
302
    commit.commit_time = long(rev.timestamp)
 
303
    if 'author-timestamp' in rev.properties:
 
304
        commit.author_time = long(rev.properties['author-timestamp'])
 
305
    else:
 
306
        commit.author_time = commit.commit_time
 
307
    commit.commit_timezone = rev.timezone
 
308
    if 'author-timezone' in rev.properties:
 
309
        commit.author_timezone = int(rev.properties['author-timezone'])
 
310
    else:
 
311
        commit.author_timezone = commit.commit_timezone 
 
312
    return commit