cola/gitcmds.py

   1 """Provides commands and queries for Git."""
   2 import re
   3 from cStringIO import StringIO
   4
   5 from cola import core
   6 from cola import gitcfg
   7 from cola import utils
   8 from cola import version
   9 from cola.compat import set
  10 from cola.git import git
  11 from cola.git import STDOUT
  12 from cola.i18n import N_
  13
  14 config = gitcfg.instance()
  15
  16
  17 class InvalidRepositoryError(StandardError):
  18     pass
  19
  20
  21 def default_remote(config=None):
  22     """Return the remote tracked by the current branch."""
  23     if config is None:
  24         config = gitcfg.instance()
  25     return config.get('branch.%s.remote' % current_branch())
  26
  27
  28 def diff_index_filenames(ref):
  29     """Return a of filenames that have been modified relative to the index"""
  30     out = git.diff_index(ref, name_only=True, z=True)[STDOUT]
  31     return _parse_diff_filenames(out)
  32
  33
  34 def diff_filenames(*args):
  35     """Return a list of filenames that have been modified"""
  36     out = git.diff_tree(name_only=True, no_commit_id=True, r=True, z=True,
  37                         *args)[STDOUT]
  38     return _parse_diff_filenames(out)
  39
  40
  41 def diff(args):
  42     """Return a list of filenames for the given diff arguments
  43
  44     :param args: list of arguments to pass to "git diff --name-only"
  45
  46     """
  47     out = git.diff(name_only=True, z=True, *args)[STDOUT]
  48     return _parse_diff_filenames(out)
  49
  50
  51 def _parse_diff_filenames(out):
  52     if out:
  53         return out[:-1].split('\0')
  54     else:
  55         return []
  56
  57
  58 def all_files():
  59     """Return the names of all files in the repository"""
  60     out = git.ls_files(z=True)[STDOUT]
  61     if out:
  62         return out[:-1].split('\0')
  63     else:
  64         return []
  65
  66
  67 class _current_branch:
  68     """Cache for current_branch()"""
  69     key = None
  70     value = None
  71
  72
  73 def clear_cache():
  74     _current_branch.key = None
  75
  76
  77 def current_branch():
  78     """Return the current branch"""
  79     head = git.git_path('HEAD')
  80     try:
  81         key = core.stat(head).st_mtime
  82         if _current_branch.key == key:
  83             return _current_branch.value
  84     except OSError:
  85         pass
  86     status, data, err = git.rev_parse('HEAD', symbolic_full_name=True)
  87     if status != 0:
  88         # git init -- read .git/HEAD.  We could do this unconditionally...
  89         data = _read_git_head(head)
  90
  91     for refs_prefix in ('refs/heads/', 'refs/remotes/', 'refs/tags/'):
  92         if data.startswith(refs_prefix):
  93             value = data[len(refs_prefix):]
  94             _current_branch.key = key
  95             _current_branch.value = value
  96             return value
  97     # Detached head
  98     return data
  99
 100
 101 def _read_git_head(head, default='master', git=git):
 102     """Pure-python .git/HEAD reader"""
 103     # Legacy .git/HEAD symlinks
 104     if core.islink(head):
 105         refs_heads = core.realpath(git.git_path('refs', 'heads'))
 106         path = core.abspath(head).replace('\\', '/')
 107         if path.startswith(refs_heads + '/'):
 108             return path[len(refs_heads)+1:]
 109
 110     # Common .git/HEAD "ref: refs/heads/master" file
 111     elif core.isfile(head):
 112         data = core.read(head).rstrip()
 113         ref_prefix = 'ref: '
 114         if data.startswith(ref_prefix):
 115             return data[len(ref_prefix):]
 116         # Detached head
 117         return data
 118
 119     return default
 120
 121
 122 def branch_list(remote=False):
 123     """
 124     Return a list of local or remote branches
 125
 126     This explicitly removes HEAD from the list of remote branches.
 127
 128     """
 129     if remote:
 130         return for_each_ref_basename('refs/remotes')
 131     else:
 132         return for_each_ref_basename('refs/heads')
 133
 134
 135 def for_each_ref_basename(refs, git=git):
 136     """Return refs starting with 'refs'."""
 137     out = git.for_each_ref(refs, format='%(refname)')[STDOUT]
 138     output = out.splitlines()
 139     non_heads = filter(lambda x: not x.endswith('/HEAD'), output)
 140     return map(lambda x: x[len(refs) + 1:], non_heads)
 141
 142
 143 def all_refs(split=False, git=git):
 144     """Return a tuple of (local branches, remote branches, tags)."""
 145     local_branches = []
 146     remote_branches = []
 147     tags = []
 148     triple = lambda x, y: (x, len(x) + 1, y)
 149     query = (triple('refs/tags', tags),
 150              triple('refs/heads', local_branches),
 151              triple('refs/remotes', remote_branches))
 152     out = git.for_each_ref(format='%(refname)')[STDOUT]
 153     for ref in out.splitlines():
 154         for prefix, prefix_len, dst in query:
 155             if ref.startswith(prefix) and not ref.endswith('/HEAD'):
 156                 dst.append(ref[prefix_len:])
 157                 continue
 158     if split:
 159         return local_branches, remote_branches, tags
 160     else:
 161         return local_branches + remote_branches + tags
 162
 163
 164 def tracked_branch(branch=None, config=None):
 165     """Return the remote branch associated with 'branch'."""
 166     if config is None:
 167         config = gitcfg.instance()
 168     if branch is None:
 169         branch = current_branch()
 170     if branch is None:
 171         return None
 172     remote = config.get('branch.%s.remote' % branch)
 173     if not remote:
 174         return None
 175     merge_ref = config.get('branch.%s.merge' % branch)
 176     if not merge_ref:
 177         return None
 178     refs_heads = 'refs/heads/'
 179     if merge_ref.startswith(refs_heads):
 180         return remote + '/' + merge_ref[len(refs_heads):]
 181     return None
 182
 183
 184 def untracked_files(git=git):
 185     """Returns a sorted list of untracked files."""
 186     out = git.ls_files(z=True, others=True, exclude_standard=True)[STDOUT]
 187     if out:
 188         return out[:-1].split('\0')
 189     return []
 190
 191
 192 def tag_list():
 193     """Return a list of tags."""
 194     tags = for_each_ref_basename('refs/tags')
 195     tags.reverse()
 196     return tags
 197
 198
 199 def log(git, *args, **kwargs):
 200     return git.log(no_color=True, no_ext_diff=True, *args, **kwargs)[STDOUT]
 201
 202
 203 def commit_diff(sha1, git=git):
 204     return log(git, '-1', sha1, '--') + '\n\n' + sha1_diff(git, sha1)
 205
 206
 207 _diff_overrides = {}
 208 def update_diff_overrides(space_at_eol, space_change,
 209                           all_space, function_context):
 210     _diff_overrides['ignore_space_at_eol'] = space_at_eol
 211     _diff_overrides['ignore_space_change'] = space_change
 212     _diff_overrides['ignore_all_space'] = all_space
 213     _diff_overrides['function_context'] = function_context
 214
 215
 216 def common_diff_opts(config=config):
 217     submodule = version.check('diff-submodule', version.git_version())
 218     opts = {
 219         'patience': True,
 220         'submodule': submodule,
 221         'no_color': True,
 222         'no_ext_diff': True,
 223         'unified': config.get('gui.diffcontext', 3),
 224         '_raw': True,
 225     }
 226     opts.update(_diff_overrides)
 227     return opts
 228
 229
 230 def sha1_diff(git, sha1):
 231     return git.diff(sha1+'^!', **common_diff_opts())[STDOUT]
 232
 233
 234 def diff_info(sha1, git=git):
 235     decoded = log(git, '-1', sha1, '--', pretty='format:%b').strip()
 236     if decoded:
 237         decoded += '\n\n'
 238     return decoded + sha1_diff(git, sha1)
 239
 240
 241 def diff_helper(commit=None,
 242                 ref=None,
 243                 endref=None,
 244                 filename=None,
 245                 cached=True,
 246                 head=None,
 247                 amending=False,
 248                 with_diff_header=False,
 249                 suppress_header=True,
 250                 reverse=False,
 251                 git=git):
 252     "Invokes git diff on a filepath."
 253     if commit:
 254         ref, endref = commit+'^', commit
 255     argv = []
 256     if ref and endref:
 257         argv.append('%s..%s' % (ref, endref))
 258     elif ref:
 259         for r in utils.shell_split(ref.strip()):
 260             argv.append(r)
 261     elif head and amending and cached:
 262         argv.append(head)
 263
 264     encoding = None
 265     if filename:
 266         argv.append('--')
 267         if type(filename) is list:
 268             argv.extend(filename)
 269         else:
 270             argv.append(filename)
 271             encoding = config.file_encoding(filename)
 272
 273     if filename is not None:
 274         deleted = cached and not core.exists(filename)
 275     else:
 276         deleted = False
 277
 278     status, out, err = git.diff(R=reverse, M=True, cached=cached,
 279                                 _encoding=encoding,
 280                                 *argv,
 281                                 **common_diff_opts())
 282     if status != 0:
 283         # git init
 284         if with_diff_header:
 285             return ('', '')
 286         else:
 287             return ''
 288
 289     return extract_diff_header(status, deleted,
 290                                with_diff_header, suppress_header, out)
 291
 292
 293 def extract_diff_header(status, deleted,
 294                         with_diff_header, suppress_header, diffoutput):
 295     encode = core.encode
 296     headers = []
 297
 298     if diffoutput.startswith('Submodule'):
 299         if with_diff_header:
 300             return ('', diffoutput)
 301         else:
 302             return diffoutput
 303
 304     start = False
 305     del_tag = 'deleted file mode '
 306     output = StringIO()
 307
 308     diff = diffoutput.split('\n')
 309     for line in diff:
 310         if not start and '@@' == line[:2] and '@@' in line[2:]:
 311             start = True
 312         if start or (deleted and del_tag in line):
 313             output.write(encode(line) + '\n')
 314         else:
 315             if with_diff_header:
 316                 headers.append(line)
 317             elif not suppress_header:
 318                 output.write(encode(line) + '\n')
 319
 320     result = core.decode(output.getvalue())
 321     output.close()
 322
 323     if with_diff_header:
 324         return('\n'.join(headers), result)
 325     else:
 326         return result
 327
 328
 329 def format_patchsets(to_export, revs, output='patches'):
 330     """
 331     Group contiguous revision selection into patchsets
 332
 333     Exists to handle multi-selection.
 334     Multiple disparate ranges in the revision selection
 335     are grouped into continuous lists.
 336
 337     """
 338
 339     outs = []
 340     errs = []
 341
 342     cur_rev = to_export[0]
 343     cur_master_idx = revs.index(cur_rev)
 344
 345     patches_to_export = [[cur_rev]]
 346     patchset_idx = 0
 347
 348     # Group the patches into continuous sets
 349     for idx, rev in enumerate(to_export[1:]):
 350         # Limit the search to the current neighborhood for efficiency
 351         master_idx = revs[cur_master_idx:].index(rev)
 352         master_idx += cur_master_idx
 353         if master_idx == cur_master_idx + 1:
 354             patches_to_export[ patchset_idx ].append(rev)
 355             cur_master_idx += 1
 356             continue
 357         else:
 358             patches_to_export.append([ rev ])
 359             cur_master_idx = master_idx
 360             patchset_idx += 1
 361
 362     # Export each patchsets
 363     status = 0
 364     for patchset in patches_to_export:
 365         stat, out, err = export_patchset(patchset[0],
 366                                          patchset[-1],
 367                                          output='patches',
 368                                          n=len(patchset) > 1,
 369                                          thread=True,
 370                                          patch_with_stat=True)
 371         outs.append(out)
 372         if err:
 373             errs.append(err)
 374         status = max(stat, status)
 375     return (status, '\n'.join(outs), '\n'.join(errs))
 376
 377
 378 def export_patchset(start, end, output='patches', **kwargs):
 379     """Export patches from start^ to end."""
 380     return git.format_patch('-o', output, start + '^..' + end, **kwargs)
 381
 382
 383 def unstage_paths(args, head='HEAD'):
 384     status, out, err = git.reset(head, '--', *set(args))
 385     if status == 128:
 386         # handle git init: we have to use 'git rm --cached'
 387         # detect this condition by checking if the file is still staged
 388         return untrack_paths(args, head=head)
 389     else:
 390         return (status, out, err)
 391
 392
 393 def untrack_paths(args, head='HEAD'):
 394     if not args:
 395         return (-1, N_('Nothing to do'), '')
 396     return git.update_index('--', force_remove=True, *set(args))
 397
 398
 399 def worktree_state(head='HEAD'):
 400     """Return a tuple of files in various states of being
 401
 402     Can be staged, unstaged, untracked, unmerged, or changed
 403     upstream.
 404
 405     """
 406     state = worktree_state_dict(head=head)
 407     return(state.get('staged', []),
 408            state.get('modified', []),
 409            state.get('unmerged', []),
 410            state.get('untracked', []),
 411            state.get('upstream_changed', []))
 412
 413
 414 def worktree_state_dict(head='HEAD', update_index=False, display_untracked=True):
 415     """Return a dict of files in various states of being
 416
 417     :rtype: dict, keys are staged, unstaged, untracked, unmerged,
 418             changed_upstream, and submodule.
 419
 420     """
 421     if update_index:
 422         git.update_index(refresh=True)
 423
 424     staged, unmerged, staged_submods = diff_index(head)
 425     modified, modified_submods = diff_worktree()
 426     untracked = display_untracked and untracked_files() or []
 427
 428     # Remove unmerged paths from the modified list
 429     unmerged_set = set(unmerged)
 430     modified_set = set(modified)
 431     modified_unmerged = modified_set.intersection(unmerged_set)
 432     for path in modified_unmerged:
 433         modified.remove(path)
 434
 435     # All submodules
 436     submodules = staged_submods.union(modified_submods)
 437
 438     # Only include the submodule in the staged list once it has
 439     # been staged.  Otherwise, we'll see the submodule as being
 440     # both modified and staged.
 441     modified_submods = modified_submods.difference(staged_submods)
 442
 443     # Add submodules to the staged and unstaged lists
 444     staged.extend(list(staged_submods))
 445     modified.extend(list(modified_submods))
 446
 447     # Look for upstream modified files if this is a tracking branch
 448     upstream_changed = diff_upstream(head)
 449
 450     # Keep stuff sorted
 451     staged.sort()
 452     modified.sort()
 453     unmerged.sort()
 454     untracked.sort()
 455     upstream_changed.sort()
 456
 457     return {'staged': staged,
 458             'modified': modified,
 459             'unmerged': unmerged,
 460             'untracked': untracked,
 461             'upstream_changed': upstream_changed,
 462             'submodules': submodules}
 463
 464
 465 def diff_index(head, cached=True):
 466     submodules = set()
 467     staged = []
 468     unmerged = []
 469
 470     status, out, err = git.diff_index(head, '--', cached=cached, z=True)
 471     if status != 0:
 472         # handle git init
 473         return all_files(), unmerged, submodules
 474
 475     while out:
 476         rest, out = out.split('\0', 1)
 477         name, out = out.split('\0', 1)
 478         status = rest[-1]
 479         if '160000' in rest[1:14]:
 480             submodules.add(name)
 481         elif status  in 'DAMT':
 482             staged.append(name)
 483         elif status == 'U':
 484             unmerged.append(name)
 485
 486     return staged, unmerged, submodules
 487
 488
 489 def diff_worktree():
 490     modified = []
 491     submodules = set()
 492
 493     status, out, err = git.diff_files(z=True)
 494     if status != 0:
 495         # handle git init
 496         out = git.ls_files(modified=True, z=True)[STDOUT]
 497         if out:
 498             modified = out[:-1].split('\0')
 499         return modified, submodules
 500
 501     while out:
 502         rest, out = out.split('\0', 1)
 503         name, out = out.split('\0', 1)
 504         status = rest[-1]
 505         if '160000' in rest[1:14]:
 506             submodules.add(name)
 507         elif status in 'DAMT':
 508             modified.append(name)
 509
 510     return modified, submodules
 511
 512
 513 def diff_upstream(head):
 514     tracked = tracked_branch()
 515     if not tracked:
 516         return []
 517     base = merge_base(head, tracked)
 518     return diff_filenames(base, tracked)
 519
 520
 521 def _branch_status(branch):
 522     """
 523     Returns a tuple of staged, unstaged, untracked, and unmerged files
 524
 525     This shows only the changes that were introduced in branch
 526
 527     """
 528     staged = diff_filenames(branch)
 529     return {'staged': staged,
 530             'upstream_changed': staged}
 531
 532
 533 def merge_base(head, ref):
 534     """Given `ref`, return $(git merge-base ref HEAD)..ref."""
 535     return git.merge_base(head, ref)[STDOUT]
 536
 537
 538 def merge_base_parent(branch):
 539     tracked = tracked_branch(branch=branch)
 540     if tracked:
 541         return tracked
 542     return 'HEAD'
 543
 544
 545 def parse_ls_tree(rev):
 546     """Return a list of(mode, type, sha1, path) tuples."""
 547     output = []
 548     lines = git.ls_tree(rev, r=True)[STDOUT].splitlines()
 549     regex = re.compile(r'^(\d+)\W(\w+)\W(\w+)[ \t]+(.*)$')
 550     for line in lines:
 551         match = regex.match(line)
 552         if match:
 553             mode = match.group(1)
 554             objtype = match.group(2)
 555             sha1 = match.group(3)
 556             filename = match.group(4)
 557             output.append((mode, objtype, sha1, filename,) )
 558     return output
 559
 560
 561 # A regex for matching the output of git(log|rev-list) --pretty=oneline
 562 REV_LIST_REGEX = re.compile(r'^([0-9a-f]{40}) (.*)$')
 563
 564 def parse_rev_list(raw_revs):
 565     """Parse `git log --pretty=online` output into (SHA-1, summary) pairs."""
 566     revs = []
 567     for line in raw_revs.splitlines():
 568         match = REV_LIST_REGEX.match(line)
 569         if match:
 570             rev_id = match.group(1)
 571             summary = match.group(2)
 572             revs.append((rev_id, summary,))
 573     return revs
 574
 575
 576 def log_helper(all=False, extra_args=None):
 577     """Return parallel arrays containing the SHA-1s and summaries."""
 578     revs = []
 579     summaries = []
 580     args = []
 581     if extra_args:
 582         args = extra_args
 583     output = log(git, pretty='oneline', all=all, *args)
 584     for line in output.splitlines():
 585         match = REV_LIST_REGEX.match(line)
 586         if match:
 587             revs.append(match.group(1))
 588             summaries.append(match.group(2))
 589     return (revs, summaries)
 590
 591
 592 def rev_list_range(start, end):
 593     """Return a (SHA-1, summary) pairs between start and end."""
 594     revrange = '%s..%s' % (start, end)
 595     out = git.rev_list(revrange, pretty='oneline')[STDOUT]
 596     return parse_rev_list(out)
 597
 598
 599 def commit_message_path():
 600     """Return the path to .git/GIT_COLA_MSG"""
 601     path = git.git_path("GIT_COLA_MSG")
 602     if core.exists(path):
 603         return path
 604     return None
 605
 606
 607 def merge_message_path():
 608     """Return the path to .git/MERGE_MSG or .git/SQUASH_MSG."""
 609     for basename in ('MERGE_MSG', 'SQUASH_MSG'):
 610         path = git.git_path(basename)
 611         if core.exists(path):
 612             return path
 613     return None
 614
 615
 616 def abort_merge():
 617     """Abort a merge by reading the tree at HEAD."""
 618     # Reset the worktree
 619     git.read_tree('HEAD', reset=True, u=True, v=True)
 620     # remove MERGE_HEAD
 621     merge_head = git.git_path('MERGE_HEAD')
 622     if core.exists(merge_head):
 623         core.unlink(merge_head)
 624     # remove MERGE_MESSAGE, etc.
 625     merge_msg_path = merge_message_path()
 626     while merge_msg_path:
 627         core.unlink(merge_msg_path)
 628         merge_msg_path = merge_message_path()
 629
 630
 631 def merge_message(revision):
 632     """Return a merge message for FETCH_HEAD."""
 633     fetch_head = git.git_path('FETCH_HEAD')
 634     if core.exists(fetch_head):
 635         return git.fmt_merge_msg('--file', fetch_head)[STDOUT]
 636     return "Merge branch '%s'" % revision