3 # Copyright (c) 2012 Felipe Contreras
7 # Just copy to your ~/bin, or anywhere in your $PATH.
8 # Then you can clone with:
9 # % git clone bzr::/path/to/bzr/repo/or/url
12 # % git clone bzr::$HOME/myrepo
14 # % git clone bzr::lp:myrepo
20 if hasattr(bzrlib
, "initialize"):
24 bzrlib
.plugin
.load_plugins()
26 import bzrlib
.generate_ids
27 import bzrlib
.transport
36 import atexit
, shutil
, hashlib
, urlparse
, subprocess
38 NAME_RE
= re
.compile('^([^<>]+)')
39 AUTHOR_RE
= re
.compile('^([^<>]+?)? ?<([^<>]*)>$')
40 RAW_AUTHOR_RE
= re
.compile('^(\w+) (.+)? <(.*)> (\d+) ([+-]\d+)')
43 sys
.stderr
.write('ERROR: %s\n' % (msg
% args
))
47 sys
.stderr
.write('WARNING: %s\n' % (msg
% args
))
50 return '%+03d%02d' % (tz
/ 3600, tz
% 3600 / 60)
54 def __init__(self
, path
):
63 if not os
.path
.exists(self
.path
):
66 tmp
= json
.load(open(self
.path
))
67 self
.tips
= tmp
['tips']
68 self
.marks
= tmp
['marks']
69 self
.last_mark
= tmp
['last-mark']
71 for rev
, mark
in self
.marks
.iteritems():
72 self
.rev_marks
[mark
] = rev
75 return { 'tips': self
.tips
, 'marks': self
.marks
, 'last-mark' : self
.last_mark
}
78 json
.dump(self
.dict(), open(self
.path
, 'w'))
81 return str(self
.dict())
83 def from_rev(self
, rev
):
84 return self
.marks
[rev
]
86 def to_rev(self
, mark
):
87 return self
.rev_marks
[mark
]
93 def get_mark(self
, rev
):
95 self
.marks
[rev
] = self
.last_mark
98 def is_marked(self
, rev
):
99 return rev
in self
.marks
101 def new_mark(self
, rev
, mark
):
102 self
.marks
[rev
] = mark
103 self
.rev_marks
[mark
] = rev
104 self
.last_mark
= mark
106 def get_tip(self
, branch
):
107 return self
.tips
.get(branch
, None)
109 def set_tip(self
, branch
, tip
):
110 self
.tips
[branch
] = tip
114 def __init__(self
, repo
):
116 self
.line
= self
.get_line()
119 return sys
.stdin
.readline().strip()
121 def __getitem__(self
, i
):
122 return self
.line
.split()[i
]
124 def check(self
, word
):
125 return self
.line
.startswith(word
)
127 def each_block(self
, separator
):
128 while self
.line
!= separator
:
130 self
.line
= self
.get_line()
133 return self
.each_block('')
136 self
.line
= self
.get_line()
137 if self
.line
== 'done':
141 i
= self
.line
.index(':') + 1
142 return int(self
.line
[i
:])
145 if not self
.check('data'):
147 i
= self
.line
.index(' ') + 1
148 size
= int(self
.line
[i
:])
149 return sys
.stdin
.read(size
)
151 def get_author(self
):
152 m
= RAW_AUTHOR_RE
.match(self
.line
)
155 _
, name
, email
, date
, tz
= m
.groups()
156 committer
= '%s <%s>' % (name
, email
)
158 tz
= ((tz
/ 100) * 3600) + ((tz
% 100) * 60)
159 return (committer
, int(date
), tz
)
161 def rev_to_mark(rev
):
163 return marks
.from_rev(rev
)
165 def mark_to_rev(mark
):
167 return marks
.to_rev(mark
)
169 def fixup_user(user
):
171 user
= user
.replace('"', '')
172 m
= AUTHOR_RE
.match(user
)
175 mail
= m
.group(2).strip()
177 m
= NAME_RE
.match(user
)
179 name
= m
.group(1).strip()
181 return '%s <%s>' % (name
, mail
)
183 def get_filechanges(cur
, prev
):
187 changes
= cur
.changes_from(prev
)
190 return s
.encode('utf-8')
192 for path
, fid
, kind
in changes
.added
:
193 modified
[u(path
)] = fid
194 for path
, fid
, kind
in changes
.removed
:
195 removed
[u(path
)] = None
196 for path
, fid
, kind
, mod
, _
in changes
.modified
:
197 modified
[u(path
)] = fid
198 for oldpath
, newpath
, fid
, kind
, mod
, _
in changes
.renamed
:
199 removed
[u(oldpath
)] = None
200 if kind
== 'directory':
201 lst
= cur
.list_files(from_dir
=newpath
, recursive
=True)
202 for path
, file_class
, kind
, fid
, entry
in lst
:
203 if kind
!= 'directory':
204 modified
[u(newpath
+ '/' + path
)] = fid
206 modified
[u(newpath
)] = fid
208 return modified
, removed
210 def export_files(tree
, files
):
211 global marks
, filenodes
214 for path
, fid
in files
.iteritems():
215 kind
= tree
.kind(fid
)
217 h
= tree
.get_file_sha1(fid
)
219 if kind
== 'symlink':
220 d
= tree
.get_symlink_target(fid
)
224 if tree
.is_executable(fid
):
229 # is the blob already exported?
232 final
.append((mode
, mark
, path
))
235 d
= tree
.get_file_text(fid
)
236 elif kind
== 'directory':
239 die("Unhandled kind '%s' for path '%s'" % (kind
, path
))
241 mark
= marks
.next_mark()
245 print "mark :%u" % mark
246 print "data %d" % len(d
)
249 final
.append((mode
, mark
, path
))
253 def export_branch(branch
, name
):
256 ref
= '%s/heads/%s' % (prefix
, name
)
257 tip
= marks
.get_tip(name
)
259 repo
= branch
.repository
262 revs
= branch
.iter_merge_sorted_revisions(None, tip
, 'exclude', 'forward')
265 revs
= [revid
for revid
, _
, _
, _
in revs
if not marks
.is_marked(revid
)]
269 rev
= repo
.get_revision(revid
)
271 parents
= rev
.parent_ids
274 committer
= rev
.committer
.encode('utf-8')
275 committer
= "%s %u %s" % (fixup_user(committer
), time
, gittz(tz
))
276 authors
= rev
.get_apparent_authors()
278 author
= authors
[0].encode('utf-8')
279 author
= "%s %u %s" % (fixup_user(author
), time
, gittz(tz
))
282 msg
= rev
.message
.encode('utf-8')
286 if len(parents
) == 0:
287 parent
= bzrlib
.revision
.NULL_REVISION
291 cur_tree
= repo
.revision_tree(revid
)
292 prev
= repo
.revision_tree(parent
)
293 modified
, removed
= get_filechanges(cur_tree
, prev
)
295 modified_final
= export_files(cur_tree
, modified
)
297 if len(parents
) == 0:
298 print 'reset %s' % ref
300 print "commit %s" % ref
301 print "mark :%d" % (marks
.get_mark(revid
))
302 print "author %s" % (author
)
303 print "committer %s" % (committer
)
304 print "data %d" % (len(msg
))
307 for i
, p
in enumerate(parents
):
316 print "merge :%s" % m
320 for f
in modified_final
:
321 print "M %s :%u %s" % f
325 if (count
% 100 == 0):
326 print "progress revision %s (%d/%d)" % (revid
, count
, len(revs
))
327 print "#############################################################"
331 revid
= branch
.last_revision()
333 # make sure the ref is updated
334 print "reset %s" % ref
335 print "from :%u" % rev_to_mark(revid
)
338 marks
.set_tip(name
, revid
)
340 def export_tag(repo
, name
):
343 ref
= '%s/tags/%s' % (prefix
, name
)
344 print "reset %s" % ref
345 print "from :%u" % rev_to_mark(tags
[name
])
348 def do_import(parser
):
352 path
= os
.path
.join(dirname
, 'marks-git')
355 if os
.path
.exists(path
):
356 print "feature import-marks=%s" % path
357 print "feature export-marks=%s" % path
358 print "feature force"
361 while parser
.check('import'):
363 if ref
.startswith('refs/heads/'):
364 name
= ref
[len('refs/heads/'):]
365 export_branch(branch
, name
)
366 if ref
.startswith('refs/tags/'):
367 name
= ref
[len('refs/tags/'):]
368 export_tag(branch
, name
)
375 def parse_blob(parser
):
379 mark
= parser
.get_mark()
381 data
= parser
.get_data()
382 blob_marks
[mark
] = data
387 def __init__(self
, branch
, revid
, parents
, files
):
393 def copy_tree(revid
):
394 files
= files_cache
[revid
] = {}
396 tree
= branch
.repository
.revision_tree(revid
)
398 for path
, entry
in tree
.iter_entries_by_dir():
399 files
[path
] = [entry
.file_id
, None]
404 if len(parents
) == 0:
405 self
.base_id
= bzrlib
.revision
.NULL_REVISION
408 self
.base_id
= parents
[0]
409 self
.base_files
= files_cache
.get(self
.base_id
, None)
410 if not self
.base_files
:
411 self
.base_files
= copy_tree(self
.base_id
)
413 self
.files
= files_cache
[revid
] = self
.base_files
.copy()
416 for path
, data
in self
.files
.iteritems():
418 self
.rev_files
[fid
] = [path
, mark
]
420 for path
, f
in files
.iteritems():
421 fid
, mark
= self
.files
.get(path
, [None, None])
423 fid
= bzrlib
.generate_ids
.gen_file_id(path
)
425 self
.rev_files
[fid
] = [path
, mark
]
426 self
.updates
[fid
] = f
428 def last_revision(self
):
431 def iter_changes(self
):
434 def get_parent(dirname
, basename
):
435 parent_fid
, mark
= self
.base_files
.get(dirname
, [None, None])
438 parent_fid
, mark
= self
.files
.get(dirname
, [None, None])
443 fid
= bzrlib
.generate_ids
.gen_file_id(path
)
444 add_entry(fid
, dirname
, 'directory')
447 def add_entry(fid
, path
, kind
, mode
= None):
448 dirname
, basename
= os
.path
.split(path
)
449 parent_fid
= get_parent(dirname
, basename
)
454 elif mode
== '120000':
465 self
.files
[path
] = [change
[0], None]
466 changes
.append(change
)
468 def update_entry(fid
, path
, kind
, mode
= None):
469 dirname
, basename
= os
.path
.split(path
)
470 parent_fid
= get_parent(dirname
, basename
)
475 elif mode
== '120000':
486 self
.files
[path
] = [change
[0], None]
487 changes
.append(change
)
489 def remove_entry(fid
, path
, kind
):
490 dirname
, basename
= os
.path
.split(path
)
491 parent_fid
= get_parent(dirname
, basename
)
501 changes
.append(change
)
503 for fid
, f
in self
.updates
.iteritems():
507 remove_entry(fid
, path
, 'file')
510 if path
in self
.base_files
:
511 update_entry(fid
, path
, 'file', f
['mode'])
513 add_entry(fid
, path
, 'file', f
['mode'])
515 self
.files
[path
][1] = f
['mark']
516 self
.rev_files
[fid
][1] = f
['mark']
520 def get_content(self
, file_id
):
521 path
, mark
= self
.rev_files
[file_id
]
523 return blob_marks
[mark
]
526 tree
= self
.branch
.repository
.revision_tree(self
.base_id
)
527 return tree
.get_file_text(file_id
)
529 def get_file_with_stat(self
, file_id
, path
=None):
530 content
= self
.get_content(file_id
)
531 return (StringIO
.StringIO(content
), None)
533 def get_symlink_target(self
, file_id
):
534 return self
.get_content(file_id
)
536 def id2path(self
, file_id
):
537 path
, mark
= self
.rev_files
[file_id
]
540 def c_style_unescape(string
):
541 if string
[0] == string
[-1] == '"':
542 return string
.decode('string-escape')[1:-1]
545 def parse_commit(parser
):
546 global marks
, blob_marks
, parsed_refs
554 if ref
!= 'refs/heads/master':
555 die("bzr doesn't support multiple branches; use 'master'")
557 commit_mark
= parser
.get_mark()
559 author
= parser
.get_author()
561 committer
= parser
.get_author()
563 data
= parser
.get_data()
565 if parser
.check('from'):
566 parents
.append(parser
.get_mark())
568 while parser
.check('merge'):
569 parents
.append(parser
.get_mark())
572 # fast-export adds an extra newline
579 if parser
.check('M'):
580 t
, m
, mark_ref
, path
= line
.split(' ', 3)
581 mark
= int(mark_ref
[1:])
582 f
= { 'mode' : m
, 'mark' : mark
}
583 elif parser
.check('D'):
584 t
, path
= line
.split(' ')
585 f
= { 'deleted' : True }
587 die('Unknown file command: %s' % line
)
588 path
= c_style_unescape(path
).decode('utf-8')
593 committer
, date
, tz
= committer
594 parents
= [str(mark_to_rev(p
)) for p
in parents
]
595 revid
= bzrlib
.generate_ids
.gen_revision_id(committer
, date
)
597 props
['branch-nick'] = branch
.nick
599 mtree
= CustomTree(branch
, revid
, parents
, files
)
600 changes
= mtree
.iter_changes()
604 builder
= branch
.get_commit_builder(parents
, None, date
, tz
, committer
, props
, revid
)
606 list(builder
.record_iter_changes(mtree
, mtree
.last_revision(), changes
))
607 builder
.finish_inventory()
608 builder
.commit(data
.decode('utf-8', 'replace'))
615 parsed_refs
[ref
] = revid
616 marks
.new_mark(revid
, commit_mark
)
618 def parse_reset(parser
):
624 if ref
!= 'refs/heads/master':
625 die("bzr doesn't support multiple branches; use 'master'")
628 if parser
.check('commit'):
631 if not parser
.check('from'):
633 from_mark
= parser
.get_mark()
636 parsed_refs
[ref
] = mark_to_rev(from_mark
)
638 def do_export(parser
):
639 global parsed_refs
, dirname
, peer
643 for line
in parser
.each_block('done'):
644 if parser
.check('blob'):
646 elif parser
.check('commit'):
648 elif parser
.check('reset'):
650 elif parser
.check('tag'):
652 elif parser
.check('feature'):
655 die('unhandled export command: %s' % line
)
659 for ref
, revid
in parsed_refs
.iteritems():
660 if ref
== 'refs/heads/master':
661 branch
.generate_revision_history(revid
, marks
.get_tip('master'))
664 branch
.push(peer
, stop_revision
=revid
)
665 except bzrlib
.errors
.DivergedBranches
:
666 print "error %s non-fast forward" % ref
670 wt
= branch
.bzrdir
.open_workingtree()
672 except bzrlib
.errors
.NoWorkingTree
:
679 def do_capabilities(parser
):
684 print "refspec refs/heads/*:%s/heads/*" % prefix
685 print "refspec refs/tags/*:%s/tags/*" % prefix
687 path
= os
.path
.join(dirname
, 'marks-git')
689 if os
.path
.exists(path
):
690 print "*import-marks %s" % path
691 print "*export-marks %s" % path
695 def ref_is_valid(name
):
696 return not True in [c
in name
for c
in '~^: \\']
700 print "? refs/heads/%s" % 'master'
704 for tag
, revid
in branch
.tags
.get_tag_dict().items():
706 branch
.revision_id_to_dotted_revno(revid
)
707 except bzrlib
.errors
.NoSuchRevision
:
709 if not ref_is_valid(tag
):
711 print "? refs/tags/%s" % tag
714 print "@refs/heads/%s HEAD" % 'master'
717 def get_repo(url
, alias
):
720 origin
= bzrlib
.bzrdir
.BzrDir
.open(url
)
721 branch
= origin
.open_branch()
723 if not isinstance(origin
.transport
, bzrlib
.transport
.local
.LocalTransport
):
724 clone_path
= os
.path
.join(dirname
, 'clone')
725 remote_branch
= branch
726 if os
.path
.exists(clone_path
):
728 d
= bzrlib
.bzrdir
.BzrDir
.open(clone_path
)
729 branch
= d
.open_branch()
731 result
= branch
.pull(remote_branch
, [], None, False)
732 except bzrlib
.errors
.DivergedBranches
:
733 # use remote branch for now
738 d
= origin
.sprout(clone_path
, None,
739 hardlink
=True, create_tree_if_local
=False,
740 source_branch
=remote_branch
)
741 branch
= d
.open_branch()
742 branch
.bind(remote_branch
)
750 def fix_path(alias
, orig_url
):
751 url
= urlparse
.urlparse(orig_url
, 'file')
752 if url
.scheme
!= 'file' or os
.path
.isabs(url
.path
):
754 abs_url
= urlparse
.urljoin("%s/" % os
.getcwd(), orig_url
)
755 cmd
= ['git', 'config', 'remote.%s.url' % alias
, "bzr::%s" % abs_url
]
759 global marks
, prefix
, dirname
760 global tags
, filenodes
778 alias
= hashlib
.sha1(alias
).hexdigest()
782 prefix
= 'refs/bzr/%s' % alias
783 gitdir
= os
.environ
['GIT_DIR']
784 dirname
= os
.path
.join(gitdir
, 'bzr', alias
)
789 if not os
.path
.exists(dirname
):
792 bzrlib
.ui
.ui_factory
.be_quiet(True)
794 repo
= get_repo(url
, alias
)
796 marks_path
= os
.path
.join(dirname
, 'marks-int')
797 marks
= Marks(marks_path
)
799 parser
= Parser(repo
)
801 if parser
.check('capabilities'):
802 do_capabilities(parser
)
803 elif parser
.check('list'):
805 elif parser
.check('import'):
807 elif parser
.check('export'):
810 die('unhandled command: %s' % line
)
819 shutil
.rmtree(dirname
)
822 sys
.exit(main(sys
.argv
))