cola/gitcmds.py

   1 """Provides commands and queries for Git."""
   2 from __future__ import division, absolute_import, unicode_literals
   3
   4 import re
   5 from io import StringIO
   6
   7 from cola import core
   8 from cola import gitcfg
   9 from cola import utils
  10 from cola import version
  11 from cola.git import git
  12 from cola.git import STDOUT
  13 from cola.i18n import N_
  14
  15
  16 EMPTY_TREE_SHA1 = '4b825dc642cb6eb9a060e54bf8d69288fbee4904'
  17
  18
  19 class InvalidRepositoryError(Exception):
  20     pass
  21
  22
  23 def default_remote(config=None):
  24     """Return the remote tracked by the current branch."""
  25     if config is None:
  26         config = gitcfg.current()
  27     return config.get('branch.%s.remote' % current_branch())
  28
  29
  30 def diff_index_filenames(ref):
  31     """Return a of filenames that have been modified relative to the index"""
  32     out = git.diff_index(ref, name_only=True, z=True)[STDOUT]
  33     return _parse_diff_filenames(out)
  34
  35
  36 def diff_filenames(*args):
  37     """Return a list of filenames that have been modified"""
  38     out = git.diff_tree(name_only=True, no_commit_id=True, r=True, z=True,
  39                         *args)[STDOUT]
  40     return _parse_diff_filenames(out)
  41
  42
  43 def diff(args):
  44     """Return a list of filenames for the given diff arguments
  45
  46     :param args: list of arguments to pass to "git diff --name-only"
  47
  48     """
  49     out = git.diff(name_only=True, z=True, *args)[STDOUT]
  50     return _parse_diff_filenames(out)
  51
  52
  53 def _parse_diff_filenames(out):
  54     if out:
  55         return out[:-1].split('\0')
  56     else:
  57         return []
  58
  59
  60 def tracked_files(*args):
  61     """Return the names of all files in the repository"""
  62     out = git.ls_files('--', *args, z=True)[STDOUT]
  63     if out:
  64         return sorted(out[:-1].split('\0'))
  65     else:
  66         return []
  67
  68
  69 def all_files(*args):
  70     """Returns a sorted list of all files, including untracked files."""
  71     ls_files = git.ls_files('--', *args,
  72                             z=True,
  73                             cached=True,
  74                             others=True,
  75                             exclude_standard=True)[STDOUT]
  76     return sorted([f for f in ls_files.split('\0') if f])
  77
  78
  79 class _current_branch:
  80     """Cache for current_branch()"""
  81     key = None
  82     value = None
  83
  84
  85 def reset():
  86     _current_branch.key = None
  87
  88
  89 def current_branch():
  90     """Return the current branch"""
  91     head = git.git_path('HEAD')
  92     try:
  93         key = core.stat(head).st_mtime
  94         if _current_branch.key == key:
  95             return _current_branch.value
  96     except OSError:
  97         # OSError means we can't use the stat cache
  98         key = 0
  99
 100     status, data, err = git.rev_parse('HEAD', symbolic_full_name=True)
 101     if status != 0:
 102         # git init -- read .git/HEAD.  We could do this unconditionally...
 103         data = _read_git_head(head)
 104
 105     for refs_prefix in ('refs/heads/', 'refs/remotes/', 'refs/tags/'):
 106         if data.startswith(refs_prefix):
 107             value = data[len(refs_prefix):]
 108             _current_branch.key = key
 109             _current_branch.value = value
 110             return value
 111     # Detached head
 112     return data
 113
 114
 115 def _read_git_head(head, default='master', git=git):
 116     """Pure-python .git/HEAD reader"""
 117     # Legacy .git/HEAD symlinks
 118     if core.islink(head):
 119         refs_heads = core.realpath(git.git_path('refs', 'heads'))
 120         path = core.abspath(head).replace('\\', '/')
 121         if path.startswith(refs_heads + '/'):
 122             return path[len(refs_heads)+1:]
 123
 124     # Common .git/HEAD "ref: refs/heads/master" file
 125     elif core.isfile(head):
 126         data = core.read(head).rstrip()
 127         ref_prefix = 'ref: '
 128         if data.startswith(ref_prefix):
 129             return data[len(ref_prefix):]
 130         # Detached head
 131         return data
 132
 133     return default
 134
 135
 136 def branch_list(remote=False):
 137     """
 138     Return a list of local or remote branches
 139
 140     This explicitly removes HEAD from the list of remote branches.
 141
 142     """
 143     if remote:
 144         return for_each_ref_basename('refs/remotes')
 145     else:
 146         return for_each_ref_basename('refs/heads')
 147
 148
 149 def for_each_ref_basename(refs, git=git):
 150     """Return refs starting with 'refs'."""
 151     out = git.for_each_ref(refs, format='%(refname)')[STDOUT]
 152     output = out.splitlines()
 153     non_heads = filter(lambda x: not x.endswith('/HEAD'), output)
 154     return list(map(lambda x: x[len(refs) + 1:], non_heads))
 155
 156
 157 def all_refs(split=False, git=git):
 158     """Return a tuple of (local branches, remote branches, tags)."""
 159     local_branches = []
 160     remote_branches = []
 161     tags = []
 162     triple = lambda x, y: (x, len(x) + 1, y)
 163     query = (triple('refs/tags', tags),
 164              triple('refs/heads', local_branches),
 165              triple('refs/remotes', remote_branches))
 166     out = git.for_each_ref(format='%(refname)')[STDOUT]
 167     for ref in out.splitlines():
 168         for prefix, prefix_len, dst in query:
 169             if ref.startswith(prefix) and not ref.endswith('/HEAD'):
 170                 dst.append(ref[prefix_len:])
 171                 continue
 172     if split:
 173         return local_branches, remote_branches, tags
 174     else:
 175         return local_branches + remote_branches + tags
 176
 177
 178 def tracked_branch(branch=None, config=None):
 179     """Return the remote branch associated with 'branch'."""
 180     if config is None:
 181         config = gitcfg.current()
 182     if branch is None:
 183         branch = current_branch()
 184     if branch is None:
 185         return None
 186     remote = config.get('branch.%s.remote' % branch)
 187     if not remote:
 188         return None
 189     merge_ref = config.get('branch.%s.merge' % branch)
 190     if not merge_ref:
 191         return None
 192     refs_heads = 'refs/heads/'
 193     if merge_ref.startswith(refs_heads):
 194         return remote + '/' + merge_ref[len(refs_heads):]
 195     return None
 196
 197
 198 def untracked_files(git=git, paths=None):
 199     """Returns a sorted list of untracked files."""
 200
 201     if paths is None:
 202         paths = []
 203     args = ['--'] + paths
 204     out = git.ls_files(z=True, others=True, exclude_standard=True,
 205                        *args)[STDOUT]
 206     if out:
 207         return out[:-1].split('\0')
 208     return []
 209
 210
 211 def tag_list():
 212     """Return a list of tags."""
 213     return list(reversed(for_each_ref_basename('refs/tags')))
 214
 215
 216 def log(git, *args, **kwargs):
 217     return git.log(no_color=True, no_abbrev_commit=True,
 218                    no_ext_diff=True, *args, **kwargs)[STDOUT]
 219
 220
 221 def commit_diff(sha1, git=git):
 222     return log(git, '-1', sha1, '--') + '\n\n' + sha1_diff(git, sha1)
 223
 224
 225 _diff_overrides = {}
 226 def update_diff_overrides(space_at_eol, space_change,
 227                           all_space, function_context):
 228     _diff_overrides['ignore_space_at_eol'] = space_at_eol
 229     _diff_overrides['ignore_space_change'] = space_change
 230     _diff_overrides['ignore_all_space'] = all_space
 231     _diff_overrides['function_context'] = function_context
 232
 233
 234 def common_diff_opts(config=None):
 235     if config is None:
 236         config = gitcfg.current()
 237     submodule = version.check('diff-submodule', version.git_version())
 238     opts = {
 239         'patience': True,
 240         'submodule': submodule,
 241         'no_color': True,
 242         'no_ext_diff': True,
 243         'unified': config.get('gui.diffcontext', 3),
 244         '_raw': True,
 245     }
 246     opts.update(_diff_overrides)
 247     return opts
 248
 249
 250 def _add_filename(args, filename):
 251     if filename:
 252         args.extend(['--', filename])
 253
 254
 255 def sha1_diff(git, sha1, filename=None):
 256     """Return the diff for a sha1"""
 257     # Naively "$sha1^!" is what we'd like to use but that doesn't
 258     # give the correct result for merges--the diff is reversed.
 259     # Be explicit and compare sha1 against its first parent.
 260     args = [sha1 + '~', sha1]
 261     opts = common_diff_opts()
 262     _add_filename(args, filename)
 263     status, out, err = git.diff(*args, **opts)
 264     if status != 0:
 265         # We probably don't have "$sha1~" because this is the root commit.
 266         # "git show" is clever enough to handle the root commit.
 267         args = [sha1 + '^!']
 268         _add_filename(args, filename)
 269         status, out, err = git.show(pretty='format:', *args, **opts)
 270         out = out.lstrip()
 271     return out
 272
 273
 274 def diff_info(sha1, git=git, filename=None):
 275     decoded = log(git, '-1', sha1, '--', pretty='format:%b').strip()
 276     if decoded:
 277         decoded += '\n\n'
 278     return decoded + sha1_diff(git, sha1, filename=filename)
 279
 280
 281 def diff_helper(commit=None,
 282                 ref=None,
 283                 endref=None,
 284                 filename=None,
 285                 cached=True,
 286                 deleted=False,
 287                 head=None,
 288                 amending=False,
 289                 with_diff_header=False,
 290                 suppress_header=True,
 291                 reverse=False,
 292                 git=git):
 293     "Invokes git diff on a filepath."
 294     if commit:
 295         ref, endref = commit+'^', commit
 296     argv = []
 297     if ref and endref:
 298         argv.append('%s..%s' % (ref, endref))
 299     elif ref:
 300         for r in utils.shell_split(ref.strip()):
 301             argv.append(r)
 302     elif head and amending and cached:
 303         argv.append(head)
 304
 305     encoding = None
 306     if filename:
 307         argv.append('--')
 308         if type(filename) is list:
 309             argv.extend(filename)
 310         else:
 311             argv.append(filename)
 312             cfg = gitcfg.current()
 313             encoding = cfg.file_encoding(filename)
 314
 315     status, out, err = git.diff(R=reverse, M=True, cached=cached,
 316                                 _encoding=encoding,
 317                                 *argv,
 318                                 **common_diff_opts())
 319     if status != 0:
 320         # git init
 321         if with_diff_header:
 322             return ('', '')
 323         else:
 324             return ''
 325
 326     return extract_diff_header(status, deleted,
 327                                with_diff_header, suppress_header, out)
 328
 329
 330 def extract_diff_header(status, deleted,
 331                         with_diff_header, suppress_header, diffoutput):
 332     headers = []
 333
 334     if diffoutput.startswith('Submodule'):
 335         if with_diff_header:
 336             return ('', diffoutput)
 337         else:
 338             return diffoutput
 339
 340     start = False
 341     del_tag = 'deleted file mode '
 342     output = StringIO()
 343
 344     diff = diffoutput.split('\n')
 345     for line in diff:
 346         if not start and '@@' == line[:2] and '@@' in line[2:]:
 347             start = True
 348         if start or (deleted and del_tag in line):
 349             output.write(line + '\n')
 350         else:
 351             if with_diff_header:
 352                 headers.append(line)
 353             elif not suppress_header:
 354                 output.write(line + '\n')
 355
 356     result = output.getvalue()
 357     output.close()
 358
 359     if with_diff_header:
 360         return('\n'.join(headers), result)
 361     else:
 362         return result
 363
 364
 365 def format_patchsets(to_export, revs, output='patches'):
 366     """
 367     Group contiguous revision selection into patchsets
 368
 369     Exists to handle multi-selection.
 370     Multiple disparate ranges in the revision selection
 371     are grouped into continuous lists.
 372
 373     """
 374
 375     outs = []
 376     errs = []
 377
 378     cur_rev = to_export[0]
 379     cur_master_idx = revs.index(cur_rev)
 380
 381     patches_to_export = [[cur_rev]]
 382     patchset_idx = 0
 383
 384     # Group the patches into continuous sets
 385     for idx, rev in enumerate(to_export[1:]):
 386         # Limit the search to the current neighborhood for efficiency
 387         master_idx = revs[cur_master_idx:].index(rev)
 388         master_idx += cur_master_idx
 389         if master_idx == cur_master_idx + 1:
 390             patches_to_export[ patchset_idx ].append(rev)
 391             cur_master_idx += 1
 392             continue
 393         else:
 394             patches_to_export.append([ rev ])
 395             cur_master_idx = master_idx
 396             patchset_idx += 1
 397
 398     # Export each patchsets
 399     status = 0
 400     for patchset in patches_to_export:
 401         stat, out, err = export_patchset(patchset[0],
 402                                          patchset[-1],
 403                                          output='patches',
 404                                          n=len(patchset) > 1,
 405                                          thread=True,
 406                                          patch_with_stat=True)
 407         outs.append(out)
 408         if err:
 409             errs.append(err)
 410         status = max(stat, status)
 411     return (status, '\n'.join(outs), '\n'.join(errs))
 412
 413
 414 def export_patchset(start, end, output='patches', **kwargs):
 415     """Export patches from start^ to end."""
 416     return git.format_patch('-o', output, start + '^..' + end, **kwargs)
 417
 418
 419 def unstage_paths(args, head='HEAD'):
 420     status, out, err = git.reset(head, '--', *set(args))
 421     if status == 128:
 422         # handle git init: we have to use 'git rm --cached'
 423         # detect this condition by checking if the file is still staged
 424         return untrack_paths(args, head=head)
 425     else:
 426         return (status, out, err)
 427
 428
 429 def untrack_paths(args, head='HEAD'):
 430     if not args:
 431         return (-1, N_('Nothing to do'), '')
 432     return git.update_index('--', force_remove=True, *set(args))
 433
 434
 435 def worktree_state(head='HEAD',
 436                    update_index=False,
 437                    display_untracked=True,
 438                    paths=None):
 439     """Return a dict of files in various states of being
 440
 441     :rtype: dict, keys are staged, unstaged, untracked, unmerged,
 442             changed_upstream, and submodule.
 443
 444     """
 445     if update_index:
 446         git.update_index(refresh=True)
 447
 448     staged, unmerged, staged_deleted, staged_submods = diff_index(head,
 449                                                                   paths=paths)
 450     modified, unstaged_deleted, modified_submods = diff_worktree(paths)
 451     untracked = display_untracked and untracked_files(paths=paths) or []
 452
 453     # Remove unmerged paths from the modified list
 454     if unmerged:
 455         unmerged_set = set(unmerged)
 456         modified = [path for path in modified if path not in unmerged_set]
 457
 458     # Look for upstream modified files if this is a tracking branch
 459     upstream_changed = diff_upstream(head)
 460
 461     # Keep stuff sorted
 462     staged.sort()
 463     modified.sort()
 464     unmerged.sort()
 465     untracked.sort()
 466     upstream_changed.sort()
 467
 468     return {'staged': staged,
 469             'modified': modified,
 470             'unmerged': unmerged,
 471             'untracked': untracked,
 472             'upstream_changed': upstream_changed,
 473             'staged_deleted': staged_deleted,
 474             'unstaged_deleted': unstaged_deleted,
 475             'submodules': staged_submods | modified_submods}
 476
 477
 478 def _parse_raw_diff(out):
 479     while out:
 480         info, path, out = out.split('\0', 2)
 481         status = info[-1]
 482         is_submodule = ('160000' in info[1:14])
 483         yield (path, status, is_submodule)
 484
 485
 486 def diff_index(head, cached=True, paths=None):
 487     staged = []
 488     unmerged = []
 489     deleted = set()
 490     submodules = set()
 491
 492     if paths is None:
 493         paths = []
 494     args = [head, '--'] + paths
 495     status, out, err = git.diff_index(cached=cached, z=True, *args)
 496     if status != 0:
 497         # handle git init
 498         args[0] = EMPTY_TREE_SHA1
 499         status, out, err = git.diff_index(cached=cached, z=True, *args)
 500
 501     for path, status, is_submodule in _parse_raw_diff(out):
 502         if is_submodule:
 503             submodules.add(path)
 504         if status in 'DAMT':
 505             staged.append(path)
 506             if status == 'D':
 507                 deleted.add(path)
 508         elif status == 'U':
 509             unmerged.append(path)
 510
 511     return staged, unmerged, deleted, submodules
 512
 513
 514 def diff_worktree(paths=None):
 515     modified = []
 516     deleted = set()
 517     submodules = set()
 518
 519     if paths is None:
 520         paths = []
 521     args = ['--'] + paths
 522     status, out, err = git.diff_files(z=True, *args)
 523     for path, status, is_submodule in _parse_raw_diff(out):
 524         if is_submodule:
 525             submodules.add(path)
 526         if status in 'DAMT':
 527             modified.append(path)
 528             if status == 'D':
 529                 deleted.add(path)
 530
 531     return modified, deleted, submodules
 532
 533
 534 def diff_upstream(head):
 535     tracked = tracked_branch()
 536     if not tracked:
 537         return []
 538     base = merge_base(head, tracked)
 539     return diff_filenames(base, tracked)
 540
 541
 542 def _branch_status(branch):
 543     """
 544     Returns a tuple of staged, unstaged, untracked, and unmerged files
 545
 546     This shows only the changes that were introduced in branch
 547
 548     """
 549     staged = diff_filenames(branch)
 550     return {'staged': staged,
 551             'upstream_changed': staged}
 552
 553
 554 def merge_base(head, ref):
 555     """Given `ref`, return $(git merge-base ref HEAD)..ref."""
 556     return git.merge_base(head, ref)[STDOUT]
 557
 558
 559 def merge_base_parent(branch):
 560     tracked = tracked_branch(branch=branch)
 561     if tracked:
 562         return tracked
 563     return 'HEAD'
 564
 565
 566 def parse_ls_tree(rev):
 567     """Return a list of(mode, type, sha1, path) tuples."""
 568     output = []
 569     lines = git.ls_tree(rev, r=True)[STDOUT].splitlines()
 570     regex = re.compile(r'^(\d+)\W(\w+)\W(\w+)[ \t]+(.*)$')
 571     for line in lines:
 572         match = regex.match(line)
 573         if match:
 574             mode = match.group(1)
 575             objtype = match.group(2)
 576             sha1 = match.group(3)
 577             filename = match.group(4)
 578             output.append((mode, objtype, sha1, filename,) )
 579     return output
 580
 581
 582 # A regex for matching the output of git(log|rev-list) --pretty=oneline
 583 REV_LIST_REGEX = re.compile(r'^([0-9a-f]{40}) (.*)$')
 584
 585 def parse_rev_list(raw_revs):
 586     """Parse `git log --pretty=online` output into (SHA-1, summary) pairs."""
 587     revs = []
 588     for line in raw_revs.splitlines():
 589         match = REV_LIST_REGEX.match(line)
 590         if match:
 591             rev_id = match.group(1)
 592             summary = match.group(2)
 593             revs.append((rev_id, summary,))
 594     return revs
 595
 596
 597 def log_helper(all=False, extra_args=None):
 598     """Return parallel arrays containing the SHA-1s and summaries."""
 599     revs = []
 600     summaries = []
 601     args = []
 602     if extra_args:
 603         args = extra_args
 604     output = log(git, pretty='oneline', all=all, *args)
 605     for line in output.splitlines():
 606         match = REV_LIST_REGEX.match(line)
 607         if match:
 608             revs.append(match.group(1))
 609             summaries.append(match.group(2))
 610     return (revs, summaries)
 611
 612
 613 def rev_list_range(start, end):
 614     """Return a (SHA-1, summary) pairs between start and end."""
 615     revrange = '%s..%s' % (start, end)
 616     out = git.rev_list(revrange, pretty='oneline')[STDOUT]
 617     return parse_rev_list(out)
 618
 619
 620 def commit_message_path():
 621     """Return the path to .git/GIT_COLA_MSG"""
 622     path = git.git_path("GIT_COLA_MSG")
 623     if core.exists(path):
 624         return path
 625     return None
 626
 627
 628 def merge_message_path():
 629     """Return the path to .git/MERGE_MSG or .git/SQUASH_MSG."""
 630     for basename in ('MERGE_MSG', 'SQUASH_MSG'):
 631         path = git.git_path(basename)
 632         if core.exists(path):
 633             return path
 634     return None
 635
 636
 637 def abort_merge():
 638     """Abort a merge by reading the tree at HEAD."""
 639     # Reset the worktree
 640     git.read_tree('HEAD', reset=True, u=True, v=True)
 641     # remove MERGE_HEAD
 642     merge_head = git.git_path('MERGE_HEAD')
 643     if core.exists(merge_head):
 644         core.unlink(merge_head)
 645     # remove MERGE_MESSAGE, etc.
 646     merge_msg_path = merge_message_path()
 647     while merge_msg_path:
 648         core.unlink(merge_msg_path)
 649         merge_msg_path = merge_message_path()
 650
 651
 652 def merge_message(revision):
 653     """Return a merge message for FETCH_HEAD."""
 654     fetch_head = git.git_path('FETCH_HEAD')
 655     if core.exists(fetch_head):
 656         return git.fmt_merge_msg('--file', fetch_head)[STDOUT]
 657     return "Merge branch '%s'" % revision