/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

  • Committer: Invité
  • Date: 2009-05-05 20:43:26 UTC
  • mto: (0.200.441 trunk)
  • mto: This revision was merged to the branch mainline in revision 6960.
  • Revision ID: guest@mirexpress-20090505204326-n0vcprylu2hyzq4v
Ensure git plugin is loaded in bzr-*-pack

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
    urlutils,
 
28
    )
 
29
from bzrlib.inventory import (
 
30
    ROOT_ID,
 
31
    )
21
32
from bzrlib.foreign import (
22
 
        ForeignVcs, 
23
 
        VcsMappingRegistry, 
24
 
        ForeignRevision,
25
 
        )
 
33
    ForeignVcs, 
 
34
    VcsMappingRegistry, 
 
35
    ForeignRevision,
 
36
    )
 
37
from bzrlib.xml_serializer import (
 
38
    escape_invalid_chars,
 
39
    )
 
40
 
 
41
DEFAULT_FILE_MODE = stat.S_IFREG | 0644
 
42
 
26
43
 
27
44
def escape_file_id(file_id):
28
45
    return file_id.replace('_', '__').replace(' ', '_s')
29
46
 
30
47
 
31
48
def unescape_file_id(file_id):
32
 
    return file_id.replace("_s", " ").replace("__", "_")
 
49
    ret = []
 
50
    i = 0
 
51
    while i < len(file_id):
 
52
        if file_id[i] != '_':
 
53
            ret.append(file_id[i])
 
54
        else:
 
55
            if file_id[i+1] == '_':
 
56
                ret.append("_")
 
57
            elif file_id[i+1] == 's':
 
58
                ret.append(" ")
 
59
            else:
 
60
                raise AssertionError("unknown escape character %s" % file_id[i+1])
 
61
            i += 1
 
62
        i += 1
 
63
    return "".join(ret)
 
64
 
 
65
 
 
66
def fix_person_identifier(text):
 
67
    if "<" in text and ">" in text:
 
68
        return text
 
69
    return "%s <%s>" % (text, text)
33
70
 
34
71
 
35
72
class BzrGitMapping(foreign.VcsMapping):
55
92
        return bzr_rev_id[len(cls.revid_prefix)+1:], cls()
56
93
 
57
94
    def generate_file_id(self, path):
 
95
        # Git paths are just bytestrings
 
96
        # We must just hope they are valid UTF-8..
58
97
        if path == "":
59
98
            return ROOT_ID
60
 
        return escape_file_id(path.encode('utf-8'))
 
99
        return escape_file_id(path)
 
100
 
 
101
    def parse_file_id(self, file_id):
 
102
        if file_id == ROOT_ID:
 
103
            return ""
 
104
        return unescape_file_id(file_id)
61
105
 
62
106
    def import_commit(self, commit):
63
107
        """Convert a git commit to a bzr revision.
68
112
            raise AssertionError("Commit object can't be None")
69
113
        rev = ForeignRevision(commit.id, self, self.revision_id_foreign_to_bzr(commit.id))
70
114
        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")
 
115
        rev.message = escape_invalid_chars(commit.message.decode("utf-8", "replace"))[0]
 
116
        rev.committer = escape_invalid_chars(str(commit.committer).decode("utf-8", "replace"))[0]
73
117
        if commit.committer != commit.author:
74
 
            rev.properties['author'] = str(commit.author).decode("utf-8", "replace")
 
118
            rev.properties['author'] = escape_invalid_chars(str(commit.author).decode("utf-8", "replace"))[0]
 
119
 
 
120
        if commit.commit_time != commit.author_time:
 
121
            rev.properties['author-timestamp'] = str(commit.author_time)
 
122
        if commit.commit_timezone != commit.author_timezone:
 
123
            rev.properties['author-timezone'] = "%f" % (commit.author_timezone * .6)
75
124
        rev.timestamp = commit.commit_time
76
 
        rev.timezone = 0
 
125
        rev.timezone = int(commit.commit_timezone * .6)
 
126
        if rev.timezone / .6 != commit.commit_timezone:
 
127
            rev.properties['commit-timezone'] = "%f" % (commit.commit_timezone * .6)
77
128
        return rev
78
129
 
79
130
 
81
132
    revid_prefix = 'git-v1'
82
133
    experimental = False
83
134
 
 
135
    def __str__(self):
 
136
        return self.revid_prefix
 
137
 
84
138
 
85
139
class BzrGitMappingExperimental(BzrGitMappingv1):
86
140
    revid_prefix = 'git-experimental'
107
161
 
108
162
 
109
163
class ForeignGit(ForeignVcs):
110
 
    """Foreign Git."""
 
164
    """The Git Stupid Content Tracker"""
111
165
 
112
166
    def __init__(self):
113
167
        super(ForeignGit, self).__init__(mapping_registry)
121
175
default_mapping = BzrGitMappingv1()
122
176
 
123
177
 
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
 
178
def text_to_blob(texts, entry):
 
179
    from dulwich.objects import Blob
 
180
    text = texts.get_record_stream([(entry.file_id, entry.revision)], 'unordered', True).next().get_bytes_as('fulltext')
 
181
    blob = Blob()
 
182
    blob._text = text
 
183
    return blob
 
184
 
 
185
 
 
186
def symlink_to_blob(entry):
 
187
    from dulwich.objects import Blob
 
188
    blob = Blob()
 
189
    blob._text = entry.symlink_target
 
190
    return blob
 
191
 
 
192
 
 
193
def entry_mode(entry):
 
194
    if entry.kind == 'directory':
 
195
        return stat.S_IFDIR
 
196
    elif entry.kind == 'symlink':
 
197
        return stat.S_IFLNK
 
198
    elif entry.kind == 'file':
 
199
        mode = stat.S_IFREG | 0644
 
200
        if entry.executable:
 
201
            mode |= 0111
 
202
        return mode
 
203
    else:
 
204
        raise AssertionError
 
205
 
 
206
 
 
207
def directory_to_tree(entry, lookup_ie_sha1):
 
208
    from dulwich.objects import Tree
 
209
    tree = Tree()
 
210
    for name in sorted(entry.children.keys()):
 
211
        ie = entry.children[name]
 
212
        tree.add(entry_mode(ie), name.encode("utf-8"), lookup_ie_sha1(ie))
 
213
    tree.serialize()
 
214
    return tree
 
215
 
 
216
 
 
217
def inventory_to_tree_and_blobs(inventory, texts, mapping, cur=None):
 
218
    """Convert a Bazaar tree to a Git tree.
 
219
 
 
220
    :return: Yields tuples with object sha1, object and path
 
221
    """
 
222
    from dulwich.objects import Tree
127
223
    import stat
128
224
    stack = []
129
 
    cur = ""
 
225
    if cur is None:
 
226
        cur = ""
130
227
    tree = Tree()
131
228
 
132
 
    inv = repo.get_inventory(revision_id)
133
 
 
134
229
    # stack contains the set of trees that we haven't 
135
230
    # finished constructing
136
 
 
137
 
    for path, entry in inv.iter_entries():
138
 
        while stack and not path.startswith(cur):
 
231
    for path, entry in inventory.iter_entries():
 
232
        while stack and not path.startswith(osutils.pathjoin(cur, "")):
 
233
            # We've hit a file that's not a child of the previous path
139
234
            tree.serialize()
140
 
            sha = tree.sha().hexdigest()
141
 
            yield sha, tree, cur
 
235
            sha = tree.id
 
236
            yield sha, tree, cur.encode("utf-8")
142
237
            t = (stat.S_IFDIR, urlutils.basename(cur).encode('UTF-8'), sha)
143
238
            cur, tree = stack.pop()
144
239
            tree.add(*t)
145
240
 
146
 
        if type(entry) == InventoryDirectory:
 
241
        if entry.kind == "directory":
147
242
            stack.append((cur, tree))
148
243
            cur = path
149
244
            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
 
 
 
245
        else:
 
246
            if entry.kind == "file":
 
247
                blob = text_to_blob(texts, entry)
 
248
            elif entry.kind == "symlink":
 
249
                blob = symlink_to_blob(entry)
 
250
            else:
 
251
                raise AssertionError("Unknown kind %s" % entry.kind)
 
252
            sha = blob.id
 
253
            yield sha, blob, path.encode("utf-8")
159
254
            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)
 
255
            tree.add(entry_mode(entry), name, sha)
164
256
 
165
257
    while len(stack) > 1:
166
258
        tree.serialize()
167
 
        sha = tree.sha().hexdigest()
168
 
        yield sha, tree, cur
 
259
        sha = tree.id
 
260
        yield sha, tree, cur.encode("utf-8")
169
261
        t = (stat.S_IFDIR, urlutils.basename(cur).encode('UTF-8'), sha)
170
262
        cur, tree = stack.pop()
171
263
        tree.add(*t)
172
264
 
173
265
    tree.serialize()
174
 
    yield tree.sha().hexdigest(), tree, cur
 
266
    yield tree.id, tree, cur.encode("utf-8")
175
267
 
176
268
 
177
269
def revision_to_commit(rev, tree_sha, parent_lookup):
183
275
    """
184
276
    from dulwich.objects import Commit
185
277
    commit = Commit()
186
 
    commit._tree = tree_sha
 
278
    commit.tree = tree_sha
187
279
    for p in rev.parent_ids:
188
280
        git_p = parent_lookup(p)
189
281
        if git_p is not None:
190
 
            commit._parents.append(git_p)
191
 
    commit._message = rev.message.encode("utf-8")
192
 
    commit._committer = rev.committer.encode("utf-8")
193
 
    commit._author = rev.get_apparent_author().encode("utf-8")
194
 
    commit._commit_time = long(rev.timestamp)
195
 
    commit.serialize()
 
282
            assert len(git_p) == 40, "unexpected length for %r" % git_p
 
283
            commit.parents.append(git_p)
 
284
    commit.message = rev.message.encode("utf-8")
 
285
    commit.committer = fix_person_identifier(rev.committer.encode("utf-8"))
 
286
    commit.author = fix_person_identifier(rev.get_apparent_authors()[0].encode("utf-8"))
 
287
    commit.commit_time = long(rev.timestamp)
 
288
    if 'author-timestamp' in rev.properties:
 
289
        commit.author_time = long(rev.properties['author-timestamp'])
 
290
    else:
 
291
        commit.author_time = commit.commit_time
 
292
    if 'committer-timezone' in rev.properties:
 
293
        commit.commit_timezone = int(float(rev.properties['commit-timezone']) / .6)
 
294
    else:
 
295
        commit.commit_timezone = int(rev.timezone / .6) 
 
296
    if 'author-timezone' in rev.properties:
 
297
        commit.author_timezone = int(float(rev.properties['author-timezone']) / .6)
 
298
    else:
 
299
        commit.author_timezone = commit.commit_timezone 
196
300
    return commit