1 """Python GIT interface
5 Copyright (C) 2005, Catalin Marinas <catalin.marinas@gmail.com>
7 This program is free software; you can redistribute it and/or modify
8 it under the terms of the GNU General Public License version 2 as
9 published by the Free Software Foundation.
11 This program is distributed in the hope that it will be useful,
12 but WITHOUT ANY WARRANTY; without even the implied warranty of
13 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
14 GNU General Public License for more details.
16 You should have received a copy of the GNU General Public License
17 along with this program; if not, write to the Free Software
18 Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
21 import sys, os, re, gitmergeonefile
22 from shutil import copyfile
24 from stgit.exception import *
25 from stgit import basedir
26 from stgit.utils import *
27 from stgit.out import *
28 from stgit.run import *
29 from stgit.config import config
32 class GitException(StgException):
35 # When a subprocess has a problem, we want the exception to be a
36 # subclass of GitException.
37 class GitRunException(GitException):
41 def __init__(self, *cmd):
42 """Initialise the Run object and insert the 'git' command name.
44 Run.__init__(self, 'git', *cmd)
51 """An author, committer, etc."""
52 def __init__(self, name = None, email = None, date = '',
54 self.name = self.email = self.date = None
55 if name or email or date:
61 assert not (name or email or date)
63 m = re.match(r'^(.+)<(.+)>(.*)$', s)
65 return [x.strip() or None for x in m.groups()]
66 self.name, self.email, self.date = parse_desc(desc)
67 def set_name(self, val):
70 def set_email(self, val):
73 def set_date(self, val):
77 if self.name and self.email:
78 return '%s <%s>' % (self.name, self.email)
80 raise GitException, 'not enough identity data'
83 """Handle the commit objects
85 def __init__(self, id_hash):
86 self.__id_hash = id_hash
88 lines = GRun('cat-file', 'commit', id_hash).output_lines()
89 for i in range(len(lines)):
92 break # we've seen all the header fields
93 key, val = line.split(' ', 1)
98 elif key == 'committer':
99 self.__committer = val
101 pass # ignore other headers
102 self.__log = '\n'.join(lines[i+1:])
104 def get_id_hash(self):
105 return self.__id_hash
110 def get_parent(self):
111 parents = self.get_parents()
117 def get_parents(self):
118 return GRun('rev-list', '--parents', '--max-count=1', self.__id_hash
119 ).output_one_line().split()[1:]
121 def get_author(self):
124 def get_committer(self):
125 return self.__committer
131 return self.get_id_hash()
133 # dictionary of Commit objects, used to avoid multiple calls to git
140 def get_commit(id_hash):
141 """Commit objects factory. Save/look-up them in the __commits
146 if id_hash in __commits:
147 return __commits[id_hash]
149 commit = Commit(id_hash)
150 __commits[id_hash] = commit
154 """Return the list of file conflicts
157 for line in GRun('ls-files', '-z', '--unmerged'
158 ).raw_output().split('\0')[:-1]:
159 stat, path = line.split('\t', 1)
164 files = [os.path.join(basedir.get(), 'info', 'exclude')]
165 user_exclude = config.get('core.excludesfile')
167 files.append(user_exclude)
170 def ls_files(files, tree = 'HEAD', full_name = True):
171 """Return the files known to GIT or raise an error otherwise. It also
172 converts the file to the full path relative the the .git directory.
179 args.append('--with-tree=%s' % tree)
181 args.append('--full-name')
185 # use a set to avoid file names duplication due to different stages
186 fileset = set(GRun('ls-files', '--error-unmatch', *args).output_lines())
187 except GitRunException:
188 # just hide the details of the 'git ls-files' command we use
189 raise GitException, \
190 'Some of the given paths are either missing or not known to GIT'
193 def parse_git_ls(output):
195 for line in output.split('\0'):
197 # There's a zero byte at the end of the output, which
198 # gives us an empty string as the last "line".
201 mode_a, mode_b, sha1_a, sha1_b, t = line.split(' ')
206 def tree_status(files = None, tree_id = 'HEAD', unknown = False,
207 noexclude = True, verbose = False, diff_flags = []):
208 """Get the status of all changed files, or of a selected set of
209 files. Returns a list of pairs - (status, filename).
211 If 'not files', it will check all files, and optionally all
212 unknown files. If 'files' is a list, it will only check the files
215 assert not files or not unknown
218 out.start('Checking for changes in the working directory')
228 cmd = ['ls-files', '-z', '--others', '--directory',
229 '--no-empty-directory']
231 cmd += ['--exclude=%s' % s for s in
232 ['*.[ao]', '*.pyc', '.*', '*~', '#*', 'TAGS', 'tags']]
233 cmd += ['--exclude-per-directory=.gitignore']
234 cmd += ['--exclude-from=%s' % fn
235 for fn in exclude_files()
236 if os.path.exists(fn)]
238 lines = GRun(*cmd).raw_output().split('\0')
239 cache_files += [('?', line) for line in lines if line]
242 conflicts = get_conflicts()
243 cache_files += [('C', filename) for filename in conflicts
244 if not files or filename in files]
245 reported_files = set(conflicts)
246 files_left = [f for f in files if f not in reported_files]
248 # files in the index. Only execute this code if no files were
249 # specified when calling the function (i.e. report all files) or
250 # files were specified but already found in the previous step
251 if not files or files_left:
252 args = diff_flags + [tree_id]
254 args += ['--'] + files_left
255 for t, fn in parse_git_ls(GRun('diff-index', '-z', *args).raw_output()):
256 # the condition is needed in case files is emtpy and
257 # diff-index lists those already reported
258 if not fn in reported_files:
259 cache_files.append((t, fn))
260 reported_files.add(fn)
261 files_left = [f for f in files if f not in reported_files]
263 # files in the index but changed on (or removed from) disk. Only
264 # execute this code if no files were specified when calling the
265 # function (i.e. report all files) or files were specified but
266 # already found in the previous step
267 if not files or files_left:
268 args = list(diff_flags)
270 args += ['--'] + files_left
271 for t, fn in parse_git_ls(GRun('diff-files', '-z', *args).raw_output()):
272 # the condition is needed in case files is empty and
273 # diff-files lists those already reported
274 if not fn in reported_files:
275 cache_files.append((t, fn))
276 reported_files.add(fn)
283 def local_changes(verbose = True):
284 """Return true if there are local changes in the tree
286 return len(tree_status(verbose = verbose)) != 0
290 hr = re.compile(r'^[0-9a-f]{40} refs/heads/(.+)$')
291 for line in GRun('show-ref', '--heads').output_lines():
293 heads.append(m.group(1))
300 """Verifies the HEAD and returns the SHA1 id that represents it
305 __head = rev_parse('HEAD')
308 class DetachedHeadException(GitException):
310 GitException.__init__(self, 'Not on any branch')
313 """Return the name of the file pointed to by the HEAD symref.
314 Throw an exception if HEAD is detached."""
317 'refs/heads/', GRun('symbolic-ref', '-q', 'HEAD'
319 except GitRunException:
320 raise DetachedHeadException()
322 def set_head_file(ref):
323 """Resets HEAD to point to a new ref
325 # head cache flushing is needed since we might have a different value
329 GRun('symbolic-ref', 'HEAD', 'refs/heads/%s' % ref).run()
330 except GitRunException:
331 raise GitException, 'Could not set head to "%s"' % ref
333 def set_ref(ref, val):
334 """Point ref at a new commit object."""
336 GRun('update-ref', ref, val).run()
337 except GitRunException:
338 raise GitException, 'Could not update %s to "%s".' % (ref, val)
340 def set_branch(branch, val):
341 set_ref('refs/heads/%s' % branch, val)
344 """Sets the HEAD value
348 if not __head or __head != val:
352 # only allow SHA1 hashes
353 assert(len(__head) == 40)
355 def __clear_head_cache():
356 """Sets the __head to None so that a re-read is forced
363 """Refresh index with stat() information from the working directory.
365 GRun('update-index', '-q', '--unmerged', '--refresh').run()
367 def rev_parse(git_id):
368 """Parse the string and return a verified SHA1 id
371 return GRun('rev-parse', '--verify', git_id
372 ).discard_stderr().output_one_line()
373 except GitRunException:
374 raise GitException, 'Unknown revision: %s' % git_id
383 def branch_exists(branch):
384 return ref_exists('refs/heads/%s' % branch)
386 def create_branch(new_branch, tree_id = None):
387 """Create a new branch in the git repository
389 if branch_exists(new_branch):
390 raise GitException, 'Branch "%s" already exists' % new_branch
392 current_head_file = get_head_file()
393 current_head = get_head()
394 set_head_file(new_branch)
395 __set_head(current_head)
397 # a checkout isn't needed if new branch points to the current head
402 # Tree switching failed. Revert the head file
403 set_head_file(current_head_file)
404 delete_branch(new_branch)
407 if os.path.isfile(os.path.join(basedir.get(), 'MERGE_HEAD')):
408 os.remove(os.path.join(basedir.get(), 'MERGE_HEAD'))
410 def switch_branch(new_branch):
411 """Switch to a git branch
415 if not branch_exists(new_branch):
416 raise GitException, 'Branch "%s" does not exist' % new_branch
418 tree_id = rev_parse('refs/heads/%s^{commit}' % new_branch)
419 if tree_id != get_head():
422 GRun('read-tree', '-u', '-m', get_head(), tree_id).run()
423 except GitRunException:
424 raise GitException, 'read-tree failed (local changes maybe?)'
426 set_head_file(new_branch)
428 if os.path.isfile(os.path.join(basedir.get(), 'MERGE_HEAD')):
429 os.remove(os.path.join(basedir.get(), 'MERGE_HEAD'))
432 if not ref_exists(ref):
433 raise GitException, '%s does not exist' % ref
434 sha1 = GRun('show-ref', '-s', ref).output_one_line()
436 GRun('update-ref', '-d', ref, sha1).run()
437 except GitRunException:
438 raise GitException, 'Failed to delete ref %s' % ref
440 def delete_branch(name):
441 delete_ref('refs/heads/%s' % name)
443 def rename_ref(from_ref, to_ref):
444 if not ref_exists(from_ref):
445 raise GitException, '"%s" does not exist' % from_ref
446 if ref_exists(to_ref):
447 raise GitException, '"%s" already exists' % to_ref
449 sha1 = GRun('show-ref', '-s', from_ref).output_one_line()
451 GRun('update-ref', to_ref, sha1, '0'*40).run()
452 except GitRunException:
453 raise GitException, 'Failed to create new ref %s' % to_ref
455 GRun('update-ref', '-d', from_ref, sha1).run()
456 except GitRunException:
457 raise GitException, 'Failed to delete ref %s' % from_ref
459 def rename_branch(from_name, to_name):
460 """Rename a git branch."""
461 rename_ref('refs/heads/%s' % from_name, 'refs/heads/%s' % to_name)
463 if get_head_file() == from_name:
464 set_head_file(to_name)
465 except DetachedHeadException:
466 pass # detached HEAD, so the renamee can't be the current branch
467 reflog_dir = os.path.join(basedir.get(), 'logs', 'refs', 'heads')
468 if os.path.exists(reflog_dir) \
469 and os.path.exists(os.path.join(reflog_dir, from_name)):
470 rename(reflog_dir, from_name, to_name)
478 """Return the user information.
482 name=config.get('user.name')
483 email=config.get('user.email')
484 __user = Person(name, email)
488 """Return the author information.
493 # the environment variables take priority over config
495 date = os.environ['GIT_AUTHOR_DATE']
498 __author = Person(os.environ['GIT_AUTHOR_NAME'],
499 os.environ['GIT_AUTHOR_EMAIL'],
506 """Return the author information.
511 # the environment variables take priority over config
513 date = os.environ['GIT_COMMITTER_DATE']
516 __committer = Person(os.environ['GIT_COMMITTER_NAME'],
517 os.environ['GIT_COMMITTER_EMAIL'],
523 def update_cache(files = None, force = False):
524 """Update the cache information for the given files
526 cache_files = tree_status(files, verbose = False)
528 # everything is up-to-date
529 if len(cache_files) == 0:
532 # check for unresolved conflicts
533 if not force and [x for x in cache_files
534 if x[0] not in ['M', 'N', 'A', 'D']]:
535 raise GitException, 'Updating cache failed: unresolved conflicts'
538 add_files = [x[1] for x in cache_files if x[0] in ['N', 'A']]
539 rm_files = [x[1] for x in cache_files if x[0] in ['D']]
540 m_files = [x[1] for x in cache_files if x[0] in ['M']]
542 GRun('update-index', '--add', '--').xargs(add_files)
543 GRun('update-index', '--force-remove', '--').xargs(rm_files)
544 GRun('update-index', '--').xargs(m_files)
548 def commit(message, files = None, parents = None, allowempty = False,
549 cache_update = True, tree_id = None, set_head = False,
550 author_name = None, author_email = None, author_date = None,
551 committer_name = None, committer_email = None):
552 """Commit the current tree to repository
557 # Get the tree status
558 if cache_update and parents != []:
559 changes = update_cache(files)
560 if not changes and not allowempty:
561 raise GitException, 'No changes to commit'
563 # get the commit message
566 elif message[-1:] != '\n':
569 # write the index to repository
571 tree_id = GRun('write-tree').output_one_line()
577 env['GIT_AUTHOR_NAME'] = author_name
579 env['GIT_AUTHOR_EMAIL'] = author_email
581 env['GIT_AUTHOR_DATE'] = author_date
583 env['GIT_COMMITTER_NAME'] = committer_name
585 env['GIT_COMMITTER_EMAIL'] = committer_email
586 commit_id = GRun('commit-tree', tree_id,
587 *sum([['-p', p] for p in parents], [])
588 ).env(env).raw_input(message).output_one_line()
590 __set_head(commit_id)
594 def apply_diff(rev1, rev2, check_index = True, files = None):
595 """Apply the diff between rev1 and rev2 onto the current
596 index. This function doesn't need to raise an exception since it
597 is only used for fast-pushing a patch. If this operation fails,
598 the pushing would fall back to the three-way merge.
601 index_opt = ['--index']
608 diff_str = diff(files, rev1, rev2)
611 GRun('apply', *index_opt).raw_input(
612 diff_str).discard_stderr().no_output()
613 except GitRunException:
618 stages_re = re.compile('^([0-7]+) ([0-9a-f]{40}) ([1-3])\t(.*)$', re.S)
620 def merge_recursive(base, head1, head2):
621 """Perform a 3-way merge between base, head1 and head2 into the
625 p = GRun('merge-recursive', base, '--', head1, head2).env(
626 { 'GITHEAD_%s' % base: 'ancestor',
627 'GITHEAD_%s' % head1: 'current',
628 'GITHEAD_%s' % head2: 'patched'}).returns([0, 1])
629 output = p.output_lines()
631 # There were conflicts
632 conflicts = [l.strip() for l in output if l.startswith('CONFLICT')]
635 # try the interactive merge or stage checkout (if enabled)
636 for filename in get_conflicts():
637 if (gitmergeonefile.merge(filename)):
638 # interactive merge succeeded
641 # any conflicts left unsolved?
642 cn = len(get_conflicts())
644 raise GitException, "%d conflict(s)" % cn
646 def diff(files = None, rev1 = 'HEAD', rev2 = None, diff_flags = [],
648 """Show the diff between rev1 and rev2
652 if binary and '--binary' not in diff_flags:
653 diff_flags = diff_flags + ['--binary']
656 return GRun('diff-tree', '-p',
657 *(diff_flags + [rev1, rev2, '--'] + files)).raw_output()
661 return GRun('diff-index', '-p', '-R',
662 *(diff_flags + [rev2, '--'] + files)).raw_output()
664 return GRun('diff-index', '-p',
665 *(diff_flags + [rev1, '--'] + files)).raw_output()
670 """Return the diffstat of the supplied diff."""
671 return GRun('apply', '--stat', '--summary').raw_input(diff).raw_output()
673 def files(rev1, rev2, diff_flags = []):
674 """Return the files modified between rev1 and rev2
678 for line in GRun('diff-tree', *(diff_flags + ['-r', rev1, rev2])
680 result.append('%s %s' % tuple(line.split(' ', 4)[-1].split('\t', 1)))
682 return '\n'.join(result)
684 def barefiles(rev1, rev2):
685 """Return the files modified between rev1 and rev2, without status info
689 for line in GRun('diff-tree', '-r', rev1, rev2).output_lines():
690 result.append(line.split(' ', 4)[-1].split('\t', 1)[-1])
692 return '\n'.join(result)
694 def pretty_commit(commit_id = 'HEAD', flags = []):
695 """Return a given commit (log + diff)
697 return GRun('show', *(flags + [commit_id])).raw_output()
699 def checkout(files = None, tree_id = None, force = False):
700 """Check out the given or all files
704 GRun('read-tree', '--reset', tree_id).run()
705 except GitRunException:
706 raise GitException, 'Failed "git read-tree" --reset %s' % tree_id
708 cmd = ['checkout-index', '-q', '-u']
712 GRun(*(cmd + ['--'])).xargs(files)
714 GRun(*(cmd + ['-a'])).run()
716 def switch(tree_id, keep = False):
717 """Switch the tree to the given id
720 # only update the index while keeping the local changes
721 GRun('read-tree', tree_id).run()
725 GRun('read-tree', '-u', '-m', get_head(), tree_id).run()
726 except GitRunException:
727 raise GitException, 'read-tree failed (local changes maybe?)'
731 def reset(files = None, tree_id = None, check_out = True):
732 """Revert the tree changes relative to the given tree_id. It removes
739 cache_files = tree_status(files, tree_id)
740 # files which were added but need to be removed
741 rm_files = [x[1] for x in cache_files if x[0] in ['A']]
743 checkout(files, tree_id, True)
744 # checkout doesn't remove files
745 map(os.remove, rm_files)
747 # if the reset refers to the whole tree, switch the HEAD as well
751 def resolved(filenames, reset = None):
753 stage = {'ancestor': 1, 'current': 2, 'patched': 3}[reset]
754 GRun('checkout-index', '--no-create', '--stage=%d' % stage,
755 '--stdin', '-z').input_nulterm(filenames).no_output()
756 GRun('update-index', '--add', '--').xargs(filenames)
757 for filename in filenames:
758 gitmergeonefile.clean_up(filename)
759 # update the access and modificatied times
760 os.utime(filename, None)
762 def fetch(repository = 'origin', refspec = None):
763 """Fetches changes from the remote repository, using 'git fetch'
773 command = config.get('branch.%s.stgit.fetchcmd' % get_head_file()) or \
774 config.get('stgit.fetchcmd')
775 Run(*(command.split() + args)).run()
777 def pull(repository = 'origin', refspec = None):
778 """Fetches changes from the remote repository, using 'git pull'
788 command = config.get('branch.%s.stgit.pullcmd' % get_head_file()) or \
789 config.get('stgit.pullcmd')
790 Run(*(command.split() + args)).run()
792 def rebase(tree_id = None):
793 """Rebase the current tree to the give tree_id. The tree_id
794 argument may be something other than a GIT id if an external
797 command = config.get('branch.%s.stgit.rebasecmd' % get_head_file()) \
798 or config.get('stgit.rebasecmd')
804 raise GitException, 'Default rebasing requires a commit id'
806 # clear the HEAD cache as the custom rebase command will update it
808 Run(*(command.split() + args)).run()
811 reset(tree_id = tree_id)
814 """Repack all objects into a single pack
816 GRun('repack', '-a', '-d', '-f').run()
818 def apply_patch(filename = None, diff = None, base = None,
820 """Apply a patch onto the current or given index. There must not
821 be any local changes in the tree, otherwise the command fails
833 orig_head = get_head()
839 GRun('apply', '--index').raw_input(diff).no_output()
840 except GitRunException:
844 # write the failed diff to a file
845 f = file('.stgit-failed.patch', 'w+')
848 out.warn('Diff written to the .stgit-failed.patch file')
853 top = commit(message = 'temporary commit used for applying a patch',
856 merge_recursive(base, orig_head, top)
858 def clone(repository, local_dir):
859 """Clone a remote repository. At the moment, just use the
862 GRun('clone', repository, local_dir).run()
864 def modifying_revs(files, base_rev, head_rev):
865 """Return the revisions from the list modifying the given files."""
866 return GRun('rev-list', '%s..%s' % (base_rev, head_rev), '--', *files
869 def refspec_localpart(refspec):
870 m = re.match('^[^:]*:([^:]*)$', refspec)
874 raise GitException, 'Cannot parse refspec "%s"' % line
876 def refspec_remotepart(refspec):
877 m = re.match('^([^:]*):[^:]*$', refspec)
881 raise GitException, 'Cannot parse refspec "%s"' % line
883 def __remotes_from_config():
884 return config.sections_matching(r'remote\.(.*)\.url')
886 def __remotes_from_dir(dir):
887 d = os.path.join(basedir.get(), dir)
888 if os.path.exists(d):
894 """Return the list of remotes in the repository
896 return (set(__remotes_from_config())
897 | set(__remotes_from_dir('remotes'))
898 | set(__remotes_from_dir('branches')))
900 def remotes_local_branches(remote):
901 """Returns the list of local branches fetched from given remote
905 if remote in __remotes_from_config():
906 for line in config.getall('remote.%s.fetch' % remote):
907 branches.append(refspec_localpart(line))
908 elif remote in __remotes_from_dir('remotes'):
909 stream = open(os.path.join(basedir.get(), 'remotes', remote), 'r')
911 # Only consider Pull lines
912 m = re.match('^Pull: (.*)\n$', line)
914 branches.append(refspec_localpart(m.group(1)))
916 elif remote in __remotes_from_dir('branches'):
917 # old-style branches only declare one branch
918 branches.append('refs/heads/'+remote);
920 raise GitException, 'Unknown remote "%s"' % remote
924 def identify_remote(branchname):
925 """Return the name for the remote to pull the given branchname
926 from, or None if we believe it is a local branch.
929 for remote in remotes_list():
930 if branchname in remotes_local_branches(remote):
933 # if we get here we've found nothing, the branch is a local one
937 """Return the git id for the tip of the parent branch as left by
942 stream = open(os.path.join(basedir.get(), 'FETCH_HEAD'), "r")
944 # Only consider lines not tagged not-for-merge
945 m = re.match('^([^\t]*)\t\t', line)
948 raise GitException, 'StGit does not support multiple FETCH_HEAD'
950 fetch_head=m.group(1)
954 out.warn('No for-merge remote head found in FETCH_HEAD')
956 # here we are sure to have a single fetch_head
960 """Return a list of all refs in the current repository.
963 return [line.split()[1] for line in GRun('show-ref').output_lines()]