/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 git_repository.py

GitBranchBuilder now handles file names with newlines correctly.

Show diffs side-by-side

added added

removed removed

Lines of Context:
16
16
 
17
17
"""An adapter between a Git Repository and a Bazaar Branch"""
18
18
 
19
 
import git
20
 
import os
21
 
import time
22
 
 
23
 
import bzrlib
24
19
from bzrlib import (
25
20
    deprecated_graph,
26
 
    errors,
27
 
    inventory,
28
 
    osutils,
29
21
    repository,
30
22
    revision,
31
 
    revisiontree,
32
23
    urlutils,
33
 
    versionedfile,
34
 
    )
35
 
from bzrlib.foreign import (
36
 
        ForeignRevision,
37
 
        )
38
 
from bzrlib.transport import get_transport
39
 
 
40
 
from bzrlib.plugins.git.foreign import (
41
 
    ForeignRepository,
42
 
    versionedfiles,
43
 
    )
44
 
from bzrlib.plugins.git.mapping import default_mapping
45
 
 
46
 
 
47
 
class GitRepository(ForeignRepository):
 
24
    )
 
25
 
 
26
from bzrlib.plugins.git import (
 
27
    ids,
 
28
    model,
 
29
    )
 
30
 
 
31
 
 
32
class GitRepository(repository.Repository):
48
33
    """An adapter to git repositories for bzr."""
49
34
 
50
 
    _serializer = None
51
 
 
52
35
    def __init__(self, gitdir, lockfiles):
53
 
        self.base = gitdir.root_transport.base
54
36
        self.bzrdir = gitdir
55
37
        self.control_files = lockfiles
56
 
        self._git = gitdir._git
57
 
        self.texts = None
58
 
        self.signatures = versionedfiles.VirtualSignatureTexts(self)
59
 
        self.revisions = versionedfiles.VirtualRevisionTexts(self)
60
 
        self._format = GitFormat()
61
 
        self._fallback_repositories = []
 
38
        gitdirectory = gitdir.transport.local_abspath('.')
 
39
        self._git = model.GitModel(gitdirectory)
 
40
        self._revision_cache = {}
62
41
 
63
 
    def _all_revision_ids(self):
64
 
        if self._git.heads == []:
65
 
            return set()
66
 
        ret = set()
67
 
        skip = 0
68
 
        max_count = 1000
69
 
        cms = None
70
 
        while cms != []:
71
 
            cms = self._git.commits("--all", max_count=max_count, skip=skip)
72
 
            skip += max_count
73
 
            ret.update([default_mapping.revision_id_foreign_to_bzr(cm.id) for cm in cms])
74
 
        return ret
 
42
    def _ancestor_revisions(self, revision_ids):
 
43
        if revision_ids is not None:
 
44
            git_revisions = [gitrevid_from_bzr(r) for r in revision_ids]
 
45
        else:
 
46
            git_revisions = None
 
47
        for lines in self._git.ancestor_lines(git_revisions):
 
48
            yield self._parse_rev(lines)
75
49
 
76
50
    def is_shared(self):
77
51
        return True
78
52
 
79
 
    def supports_rich_root(self):
80
 
        return False
81
 
 
82
 
    #def get_revision_delta(self, revision_id):
83
 
    #    parent_revid = self.get_revision(revision_id).parent_ids[0]
84
 
    #    diff = self._git.diff(ids.convert_revision_id_bzr_to_git(parent_revid),
85
 
    #                   ids.convert_revision_id_bzr_to_git(revision_id))
86
 
 
87
 
    def get_ancestry(self, revision_id):
88
 
        revision_id = revision.ensure_null(revision_id)
89
 
        ret = []
90
 
        if revision_id != revision.NULL_REVISION:
91
 
            skip = 0
92
 
            max_count = 1000
93
 
            cms = None
94
 
            while cms != []:
95
 
                cms = self._git.commits(self.lookup_git_revid(revision_id, default_mapping), max_count=max_count, skip=skip)
96
 
                skip += max_count
97
 
                ret += [default_mapping.revision_id_foreign_to_bzr(cm.id) for cm in cms]
98
 
        return [None] + ret
99
 
 
100
 
    def get_signature_text(self, revision_id):
101
 
        raise errors.NoSuchRevision(self, revision_id)
102
 
 
103
 
    def has_signature_for_revision_id(self, revision_id):
104
 
        return False
105
 
 
106
 
    def get_parent_map(self, revision_ids):
107
 
        ret = {}
108
 
        for revid in revision_ids:
109
 
            if revid == revision.NULL_REVISION:
110
 
                ret[revid] = ()
111
 
            else:
112
 
                commit = self._git.commit(self.lookup_git_revid(revid, default_mapping))
113
 
                ret[revid] = tuple([default_mapping.revision_id_foreign_to_bzr(p.id) for p in commit.parents])
114
 
        return ret
115
 
 
116
 
    def lookup_git_revid(self, bzr_revid, mapping):
117
 
        try:
118
 
            return mapping.revision_id_bzr_to_foreign(bzr_revid)
119
 
        except errors.InvalidRevisionId:
120
 
            raise errors.NoSuchRevision(bzr_revid, self)
 
53
    def get_revision_graph(self, revision_id=None):
 
54
        result = {}
 
55
        if revision_id is not None:
 
56
            param = [ids.convert_revision_id_bzr_to_git(revision_id)]
 
57
        else:
 
58
            param = None
 
59
        for node, parents in self._git.ancestry(param).iteritems():
 
60
            bzr_node = ids.convert_revision_id_git_to_bzr(node)
 
61
            bzr_parents = [ids.convert_revision_id_git_to_bzr(n)
 
62
                           for n in parents]
 
63
            result[bzr_node] = bzr_parents
 
64
        return result
 
65
 
 
66
    def get_revision_graph_with_ghosts(self, revision_ids=None):
 
67
        graph = deprecated_graph.Graph()
 
68
        if revision_ids is not None:
 
69
            revision_ids = [ids.convert_revision_id_bzr_to_git(r)
 
70
                            for r in revision_ids]
 
71
        for node, parents in self._git.ancestry(revision_ids).iteritems():
 
72
            bzr_node = ids.convert_revision_id_git_to_bzr(node)
 
73
            bzr_parents = [ids.convert_revision_id_git_to_bzr(n)
 
74
                           for n in parents]
 
75
 
 
76
            graph.add_node(bzr_node, bzr_parents)
 
77
        return graph
121
78
 
122
79
    def get_revision(self, revision_id):
123
 
        git_commit_id = self.lookup_git_revid(revision_id, default_mapping)
124
 
        commit = self._git.commit(git_commit_id)
125
 
        # print "fetched revision:", git_commit_id
126
 
        revision = self._parse_rev(commit, default_mapping)
127
 
        return revision
 
80
        if revision_id in self._revision_cache:
 
81
            return self._revision_cache[revision_id]
 
82
        raw = self._git.rev_list(
 
83
            [ids.convert_revision_id_bzr_to_git(revision_id)],
 
84
            max_count=1, header=True)
 
85
        return self._parse_rev(raw)
128
86
 
129
87
    def has_revision(self, revision_id):
130
88
        try:
138
96
        return [self.get_revision(r) for r in revisions]
139
97
 
140
98
    @classmethod
141
 
    def _parse_rev(klass, commit, mapping):
142
 
        """Convert a git commit to a bzr revision.
143
 
 
 
99
    def _parse_rev(klass, raw):
 
100
        """Parse a single git revision.
 
101
 
 
102
        * The first line is the git commit id.
 
103
        * Following lines conform to the 'name value' structure, until the
 
104
          first blank line.
 
105
        * All lines after the first blank line and until the NULL line have 4
 
106
          leading spaces and constitute the commit message.
 
107
 
 
108
        :param raw: sequence of newline-terminated strings, its last item is a
 
109
            single NULL character.
144
110
        :return: a `bzrlib.revision.Revision` object.
145
111
        """
146
 
        rev = ForeignRevision(commit.id, mapping, mapping.revision_id_foreign_to_bzr(commit.id))
147
 
        rev.parent_ids = tuple([mapping.revision_id_foreign_to_bzr(p.id) for p in commit.parents])
148
 
        rev.message = commit.message.decode("utf-8", "replace")
149
 
        rev.committer = str(commit.committer).decode("utf-8", "replace")
150
 
        rev.properties['author'] = str(commit.author).decode("utf-8", "replace")
151
 
        rev.timestamp = time.mktime(commit.committed_date)
152
 
        rev.timezone = 0
 
112
        parents = []
 
113
        message_lines = []
 
114
        in_message = False
 
115
        committer_was_set = False
 
116
        revision_id = ids.convert_revision_id_git_to_bzr(raw[0][:-1])
 
117
        rev = revision.Revision(revision_id)
 
118
        rev.inventory_sha1 = ""
 
119
        assert raw[-1] == '\x00', (
 
120
            "Last item of raw was not a single NULL character.")
 
121
        for line in raw[1:-1]:
 
122
            if in_message:
 
123
                assert line[:4] == '    ', (
 
124
                    "Unexpected line format in commit message: %r" % line)
 
125
                message_lines.append(line[4:])
 
126
                continue
 
127
            if line == '\n':
 
128
                in_message = True
 
129
                continue
 
130
            name, value = line[:-1].split(' ', 1)
 
131
            if name == 'parent':
 
132
                rev.parent_ids.append(
 
133
                    ids.convert_revision_id_git_to_bzr(value))
 
134
                continue
 
135
            if name == 'author':
 
136
                author, timestamp, timezone = value.rsplit(' ', 2)
 
137
                rev.properties['author'] = author
 
138
                rev.properties['git-author-timestamp'] = timestamp
 
139
                rev.properties['git-author-timezone'] = timezone
 
140
                if not committer_was_set:
 
141
                    rev.committer = author
 
142
                    rev.timestamp = float(timestamp)
 
143
                    rev.timezone = klass._parse_tz(timezone)
 
144
                continue
 
145
            if name == 'committer':
 
146
                committer_was_set = True
 
147
                committer, timestamp, timezone = value.rsplit(' ', 2)
 
148
                rev.committer = committer
 
149
                rev.timestamp = float(timestamp)
 
150
                rev.timezone = klass._parse_tz(timezone)
 
151
                continue
 
152
            if name == 'tree':
 
153
                rev.properties['git-tree-id'] = value
 
154
                continue
 
155
 
 
156
        rev.message = ''.join(message_lines)
153
157
        return rev
154
158
 
 
159
    @classmethod
 
160
    def _parse_tz(klass, tz):
 
161
        """Parse a timezone specification in the [+|-]HHMM format.
 
162
 
 
163
        :return: the timezone offset in seconds.
 
164
        """
 
165
        assert len(tz) == 5
 
166
        sign = {'+': +1, '-': -1}[tz[0]]
 
167
        hours = int(tz[1:3])
 
168
        minutes = int(tz[3:])
 
169
        return float(sign * 60 * (60 * hours + minutes))
 
170
 
155
171
    def revision_trees(self, revids):
156
172
        for revid in revids:
157
173
            yield self.revision_tree(revid)
158
174
 
159
175
    def revision_tree(self, revision_id):
160
 
        revision_id = revision.ensure_null(revision_id)
161
 
 
162
 
        if revision_id == revision.NULL_REVISION:
163
 
            inv = inventory.Inventory(root_id=None)
164
 
            inv.revision_id = revision_id
165
 
            return revisiontree.RevisionTree(self, inv, revision_id)
166
 
 
167
176
        return GitRevisionTree(self, revision_id)
168
177
 
169
178
    def get_inventory(self, revision_id):
170
 
        assert revision_id != None
171
 
        return self.revision_tree(revision_id).inventory
172
 
 
173
 
    def set_make_working_trees(self, trees):
174
 
        pass
175
 
 
176
 
 
177
 
def escape_file_id(file_id):
178
 
    return file_id.replace('_', '__').replace(' ', '_s')
179
 
 
180
 
 
181
 
def unescape_file_id(file_id):
182
 
    return file_id.replace("_s", " ").replace("__", "_")
183
 
 
184
 
 
185
 
class GitRevisionTree(revisiontree.RevisionTree):
 
179
        revision = self.get_revision(revision_id)
 
180
        inventory = GitInventory(revision_id)
 
181
        tree_id = revision.properties['git-tree-id']
 
182
        type_map = {'blob': 'file', 'tree': 'directory' }
 
183
        def get_inventory(tree_id, prefix):
 
184
            for perms, type, obj_id, name in self._git.get_inventory(tree_id):
 
185
                full_path = prefix + name
 
186
                if type == 'blob':
 
187
                    text_sha1 = obj_id
 
188
                else:
 
189
                    text_sha1 = None
 
190
                executable = (perms[-3] in ('1', '3', '5', '7'))
 
191
                entry = GitEntry(full_path, type_map[type], revision_id,
 
192
                                 text_sha1, executable)
 
193
                inventory.entries[full_path] = entry
 
194
                if type == 'tree':
 
195
                    get_inventory(obj_id, full_path+'/')
 
196
        get_inventory(tree_id, '')
 
197
        return inventory
 
198
 
 
199
 
 
200
class GitRevisionTree(object):
186
201
 
187
202
    def __init__(self, repository, revision_id):
188
 
        self._repository = repository
 
203
        self.repository = repository
189
204
        self.revision_id = revision_id
190
 
        git_id = repository.lookup_git_revid(revision_id, default_mapping)
191
 
        self.tree = repository._git.commit(git_id).tree
192
 
        self._inventory = inventory.Inventory(revision_id=revision_id)
193
 
        self._inventory.root.revision = revision_id
194
 
        self._build_inventory(self.tree, self._inventory.root, "")
195
 
 
196
 
    def get_revision_id(self):
197
 
        return self.revision_id
198
 
 
199
 
    def get_file_text(self, file_id):
200
 
        entry = self._inventory[file_id]
201
 
        if entry.kind == 'directory': return ""
202
 
        return self._repository._git.blob(entry.text_id).data
203
 
 
204
 
    def _build_inventory(self, tree, ie, path):
205
 
        assert isinstance(path, str)
206
 
        for b in tree.contents:
207
 
            basename = b.name.decode("utf-8")
208
 
            if path == "":
209
 
                child_path = b.name
210
 
            else:
211
 
                child_path = urlutils.join(path, b.name)
212
 
            file_id = escape_file_id(child_path.encode('utf-8'))
213
 
            if b.mode[0] == '0':
214
 
                child_ie = inventory.InventoryDirectory(file_id, basename, ie.file_id)
215
 
            elif b.mode[0] == '1':
216
 
                if b.mode[1] == '0':
217
 
                    child_ie = inventory.InventoryFile(file_id, basename, ie.file_id)
218
 
                    child_ie.text_sha1 = osutils.sha_string(b.data)
219
 
                elif b.mode[1] == '2':
220
 
                    child_ie = inventory.InventoryLink(file_id, basename, ie.file_id)
221
 
                    child_ie.text_sha1 = osutils.sha_string("")
222
 
                else:
223
 
                    raise AssertionError(
224
 
                        "Unknown file kind, perms=%r." % (b.mode,))
225
 
                child_ie.text_id = b.id
226
 
                child_ie.text_size = b.size
227
 
            else:
228
 
                raise AssertionError(
229
 
                    "Unknown blob kind, perms=%r." % (b.mode,))
230
 
            child_ie.executable = bool(int(b.mode[3:], 8) & 0111)
231
 
            child_ie.revision = self.revision_id
232
 
            self._inventory.add(child_ie)
233
 
            if b.mode[0] == '0':
234
 
                self._build_inventory(b, child_ie, child_path)
235
 
 
236
 
 
237
 
class GitFormat(object):
238
 
 
239
 
    supports_tree_reference = False
240
 
 
241
 
    def get_format_description(self):
242
 
        return "Git Repository"
 
205
        self.inventory = repository.get_inventory(revision_id)
 
206
 
 
207
    def get_file(self, file_id):
 
208
        return iterablefile.IterableFile(self.get_file_lines(file_id))
 
209
 
 
210
    def get_file_lines(self, file_id):
 
211
        obj_id = self.inventory[file_id].text_sha1
 
212
        return self.repository._git.cat_file('blob', obj_id)
 
213
 
 
214
    def is_executable(self, file_id):
 
215
        return self.inventory[file_id].executable
 
216
 
 
217
 
 
218
class GitInventory(object):
 
219
 
 
220
    def __init__(self, revision_id):
 
221
        self.entries = {}
 
222
        self.root = GitEntry('', 'directory', revision_id)
 
223
        self.entries[''] = self.root
 
224
 
 
225
    def __getitem__(self, key):
 
226
        return self.entries[key]
 
227
 
 
228
    def iter_entries(self):
 
229
        return iter(sorted(self.entries.items()))
 
230
 
 
231
    def iter_entries_by_dir(self):
 
232
        return self.iter_entries()
 
233
 
 
234
    def __len__(self):
 
235
        return len(self.entries)
 
236
 
 
237
 
 
238
class GitEntry(object):
 
239
 
 
240
    def __init__(self, path, kind, revision, text_sha1=None, executable=False,
 
241
                 text_size=None):
 
242
        self.path = path
 
243
        self.file_id = path
 
244
        self.kind = kind
 
245
        self.executable = executable
 
246
        self.name = osutils.basename(path)
 
247
        if path == '':
 
248
            self.parent_id = None
 
249
        else:
 
250
            self.parent_id = osutils.dirname(path)
 
251
        self.revision = revision
 
252
        self.symlink_target = None
 
253
        self.text_sha1 = text_sha1
 
254
        self.text_size = None
 
255
 
 
256
    def __repr__(self):
 
257
        return "GitEntry(%r, %r, %r, %r)" % (self.path, self.kind,
 
258
                                             self.revision, self.parent_id)
 
259
 
 
260