cola/gitcmds.py

   1 """Git commands and queries for Git"""
   2 from __future__ import division, absolute_import, unicode_literals
   3 import json
   4 import os
   5 import re
   6 from io import StringIO
   7
   8 from . import core
   9 from . import utils
  10 from . import version
  11 from .git import STDOUT
  12 from .git import EMPTY_TREE_OID
  13 from .git import OID_LENGTH
  14 from .i18n import N_
  15 from .interaction import Interaction
  16
  17
  18 class InvalidRepositoryError(Exception):
  19     pass
  20
  21
  22 def add(context, items, u=False):
  23     """Run "git add" while preventing argument overflow"""
  24     fn = context.git.add
  25     return utils.slice_fn(
  26         items, lambda paths: fn('--', force=True, verbose=True, u=u, *paths)
  27     )
  28
  29
  30 def apply_diff(context, filename):
  31     git = context.git
  32     return git.apply(filename, index=True, cached=True)
  33
  34
  35 def apply_diff_to_worktree(context, filename):
  36     git = context.git
  37     return git.apply(filename)
  38
  39
  40 def get_branch(context, branch):
  41     if branch is None:
  42         branch = current_branch(context)
  43     return branch
  44
  45
  46 def upstream_remote(context, branch=None):
  47     """Return the remote associated with the specified branch"""
  48     config = context.cfg
  49     branch = get_branch(context, branch)
  50     return config.get('branch.%s.remote' % branch)
  51
  52
  53 def remote_url(context, remote, push=False):
  54     """Return the URL for the specified remote"""
  55     config = context.cfg
  56     url = config.get('remote.%s.url' % remote, '')
  57     if push:
  58         url = config.get('remote.%s.pushurl' % remote, url)
  59     return url
  60
  61
  62 def diff_index_filenames(context, ref):
  63     """
  64     Return a diff of filenames that have been modified relative to the index
  65     """
  66     git = context.git
  67     out = git.diff_index(ref, name_only=True, z=True)[STDOUT]
  68     return _parse_diff_filenames(out)
  69
  70
  71 def diff_filenames(context, *args):
  72     """Return a list of filenames that have been modified"""
  73     git = context.git
  74     out = git.diff_tree(
  75         name_only=True, no_commit_id=True, r=True, z=True, _readonly=True, *args
  76     )[STDOUT]
  77     return _parse_diff_filenames(out)
  78
  79
  80 def listdir(context, dirname, ref='HEAD'):
  81     """Get the contents of a directory according to Git
  82
  83     Query Git for the content of a directory, taking ignored
  84     files into account.
  85
  86     """
  87     dirs = []
  88     files = []
  89
  90     # first, parse git ls-tree to get the tracked files
  91     # in a list of (type, path) tuples
  92     entries = ls_tree(context, dirname, ref=ref)
  93     for entry in entries:
  94         if entry[0][0] == 't':  # tree
  95             dirs.append(entry[1])
  96         else:
  97             files.append(entry[1])
  98
  99     # gather untracked files
 100     untracked = untracked_files(context, paths=[dirname], directory=True)
 101     for path in untracked:
 102         if path.endswith('/'):
 103             dirs.append(path[:-1])
 104         else:
 105             files.append(path)
 106
 107     dirs.sort()
 108     files.sort()
 109
 110     return (dirs, files)
 111
 112
 113 def diff(context, args):
 114     """Return a list of filenames for the given diff arguments
 115
 116     :param args: list of arguments to pass to "git diff --name-only"
 117
 118     """
 119     git = context.git
 120     out = git.diff(name_only=True, z=True, *args)[STDOUT]
 121     return _parse_diff_filenames(out)
 122
 123
 124 def _parse_diff_filenames(out):
 125     if out:
 126         return out[:-1].split('\0')
 127     return []
 128
 129
 130 def tracked_files(context, *args):
 131     """Return the names of all files in the repository"""
 132     git = context.git
 133     out = git.ls_files('--', *args, z=True)[STDOUT]
 134     if out:
 135         return sorted(out[:-1].split('\0'))
 136     return []
 137
 138
 139 def all_files(context, *args):
 140     """Returns a sorted list of all files, including untracked files."""
 141     git = context.git
 142     ls_files = git.ls_files(
 143         '--', *args, z=True, cached=True, others=True, exclude_standard=True
 144     )[STDOUT]
 145     return sorted([f for f in ls_files.split('\0') if f])
 146
 147
 148 class _current_branch(object):
 149     """Cache for current_branch()"""
 150
 151     key = None
 152     value = None
 153
 154
 155 def reset():
 156     _current_branch.key = None
 157
 158
 159 def current_branch(context):
 160     """Return the current branch"""
 161     git = context.git
 162     head = git.git_path('HEAD')
 163     try:
 164         key = core.stat(head).st_mtime
 165         if _current_branch.key == key:
 166             return _current_branch.value
 167     except OSError:
 168         # OSError means we can't use the stat cache
 169         key = 0
 170
 171     status, data, _ = git.rev_parse('HEAD', symbolic_full_name=True)
 172     if status != 0:
 173         # git init -- read .git/HEAD.  We could do this unconditionally...
 174         data = _read_git_head(context, head)
 175
 176     for refs_prefix in ('refs/heads/', 'refs/remotes/', 'refs/tags/'):
 177         if data.startswith(refs_prefix):
 178             value = data[len(refs_prefix) :]
 179             _current_branch.key = key
 180             _current_branch.value = value
 181             return value
 182     # Detached head
 183     return data
 184
 185
 186 def _read_git_head(context, head, default='main'):
 187     """Pure-python .git/HEAD reader"""
 188     # Common .git/HEAD "ref: refs/heads/main" files
 189     git = context.git
 190     islink = core.islink(head)
 191     if core.isfile(head) and not islink:
 192         data = core.read(head).rstrip()
 193         ref_prefix = 'ref: '
 194         if data.startswith(ref_prefix):
 195             return data[len(ref_prefix) :]
 196         # Detached head
 197         return data
 198     # Legacy .git/HEAD symlinks
 199     elif islink:
 200         refs_heads = core.realpath(git.git_path('refs', 'heads'))
 201         path = core.abspath(head).replace('\\', '/')
 202         if path.startswith(refs_heads + '/'):
 203             return path[len(refs_heads) + 1 :]
 204
 205     return default
 206
 207
 208 def branch_list(context, remote=False):
 209     """
 210     Return a list of local or remote branches
 211
 212     This explicitly removes HEAD from the list of remote branches.
 213
 214     """
 215     if remote:
 216         return for_each_ref_basename(context, 'refs/remotes')
 217     return for_each_ref_basename(context, 'refs/heads')
 218
 219
 220 def _version_sort(context, key='version:refname'):
 221     if version.check_git(context, 'version-sort'):
 222         sort = key
 223     else:
 224         sort = False
 225     return sort
 226
 227
 228 def for_each_ref_basename(context, refs):
 229     """Return refs starting with 'refs'."""
 230     git = context.git
 231     sort = _version_sort(context)
 232     _, out, _ = git.for_each_ref(refs, format='%(refname)', sort=sort, _readonly=True)
 233     output = out.splitlines()
 234     non_heads = [x for x in output if not x.endswith('/HEAD')]
 235     offset = len(refs) + 1
 236     return [x[offset:] for x in non_heads]
 237
 238
 239 def _triple(x, y):
 240     return (x, len(x) + 1, y)
 241
 242
 243 def all_refs(context, split=False, sort_key='version:refname'):
 244     """Return a tuple of (local branches, remote branches, tags)."""
 245     git = context.git
 246     local_branches = []
 247     remote_branches = []
 248     tags = []
 249     triple = _triple
 250     query = (
 251         triple('refs/tags', tags),
 252         triple('refs/heads', local_branches),
 253         triple('refs/remotes', remote_branches),
 254     )
 255     sort = _version_sort(context, key=sort_key)
 256     _, out, _ = git.for_each_ref(format='%(refname)', sort=sort, _readonly=True)
 257     for ref in out.splitlines():
 258         for prefix, prefix_len, dst in query:
 259             if ref.startswith(prefix) and not ref.endswith('/HEAD'):
 260                 dst.append(ref[prefix_len:])
 261                 continue
 262     tags.reverse()
 263     if split:
 264         return local_branches, remote_branches, tags
 265     return local_branches + remote_branches + tags
 266
 267
 268 def tracked_branch(context, branch=None):
 269     """Return the remote branch associated with 'branch'."""
 270     if branch is None:
 271         branch = current_branch(context)
 272     if branch is None:
 273         return None
 274     config = context.cfg
 275     remote = config.get('branch.%s.remote' % branch)
 276     if not remote:
 277         return None
 278     merge_ref = config.get('branch.%s.merge' % branch)
 279     if not merge_ref:
 280         return None
 281     refs_heads = 'refs/heads/'
 282     if merge_ref.startswith(refs_heads):
 283         return remote + '/' + merge_ref[len(refs_heads) :]
 284     return None
 285
 286
 287 def parse_remote_branch(branch):
 288     """Split a remote branch apart into (remote, name) components"""
 289     rgx = re.compile(r'^(?P<remote>[^/]+)/(?P<branch>.+)$')
 290     match = rgx.match(branch)
 291     remote = ''
 292     branch = ''
 293     if match:
 294         remote = match.group('remote')
 295         branch = match.group('branch')
 296     return (remote, branch)
 297
 298
 299 def untracked_files(context, paths=None, **kwargs):
 300     """Returns a sorted list of untracked files."""
 301     git = context.git
 302     if paths is None:
 303         paths = []
 304     args = ['--'] + paths
 305     out = git.ls_files(z=True, others=True, exclude_standard=True, *args, **kwargs)[
 306         STDOUT
 307     ]
 308     if out:
 309         return out[:-1].split('\0')
 310     return []
 311
 312
 313 def tag_list(context):
 314     """Return a list of tags."""
 315     result = for_each_ref_basename(context, 'refs/tags')
 316     result.reverse()
 317     return result
 318
 319
 320 def log(git, *args, **kwargs):
 321     return git.log(
 322         no_color=True,
 323         no_abbrev_commit=True,
 324         no_ext_diff=True,
 325         _readonly=True,
 326         *args,
 327         **kwargs
 328     )[STDOUT]
 329
 330
 331 def commit_diff(context, oid):
 332     git = context.git
 333     return log(git, '-1', oid, '--') + '\n\n' + oid_diff(context, oid)
 334
 335
 336 _diff_overrides = {}
 337
 338
 339 def update_diff_overrides(space_at_eol, space_change, all_space, function_context):
 340     _diff_overrides['ignore_space_at_eol'] = space_at_eol
 341     _diff_overrides['ignore_space_change'] = space_change
 342     _diff_overrides['ignore_all_space'] = all_space
 343     _diff_overrides['function_context'] = function_context
 344
 345
 346 def common_diff_opts(context):
 347     config = context.cfg
 348     # Default to --patience when diff.algorithm is unset
 349     patience = not config.get('diff.algorithm', default='')
 350     submodule = version.check_git(context, 'diff-submodule')
 351     opts = {
 352         'patience': patience,
 353         'submodule': submodule,
 354         'no_color': True,
 355         'no_ext_diff': True,
 356         'unified': config.get('gui.diffcontext', default=3),
 357         '_raw': True,
 358     }
 359     opts.update(_diff_overrides)
 360     return opts
 361
 362
 363 def _add_filename(args, filename):
 364     if filename:
 365         args.extend(['--', filename])
 366
 367
 368 def oid_diff(context, oid, filename=None):
 369     """Return the diff for an oid"""
 370     # Naively "$oid^!" is what we'd like to use but that doesn't
 371     # give the correct result for merges--the diff is reversed.
 372     # Be explicit and compare oid against its first parent.
 373     git = context.git
 374     args = [oid + '~', oid]
 375     opts = common_diff_opts(context)
 376     _add_filename(args, filename)
 377     status, out, _ = git.diff(*args, **opts)
 378     if status != 0:
 379         # We probably don't have "$oid~" because this is the root commit.
 380         # "git show" is clever enough to handle the root commit.
 381         args = [oid + '^!']
 382         _add_filename(args, filename)
 383         _, out, _ = git.show(pretty='format:', _readonly=True, *args, **opts)
 384         out = out.lstrip()
 385     return out
 386
 387
 388 def diff_info(context, oid, filename=None):
 389     git = context.git
 390     decoded = log(git, '-1', oid, '--', pretty='format:%b').strip()
 391     if decoded:
 392         decoded += '\n\n'
 393     return decoded + oid_diff(context, oid, filename=filename)
 394
 395
 396 # pylint: disable=too-many-arguments
 397 def diff_helper(
 398     context,
 399     commit=None,
 400     ref=None,
 401     endref=None,
 402     filename=None,
 403     cached=True,
 404     deleted=False,
 405     head=None,
 406     amending=False,
 407     with_diff_header=False,
 408     suppress_header=True,
 409     reverse=False,
 410     untracked=False,
 411 ):
 412     "Invokes git diff on a filepath."
 413     git = context.git
 414     cfg = context.cfg
 415     if commit:
 416         ref, endref = commit + '^', commit
 417     argv = []
 418     if ref and endref:
 419         argv.append('%s..%s' % (ref, endref))
 420     elif ref:
 421         argv.extend(context, utils.shell_split(ref.strip()))
 422     elif head and amending and cached:
 423         argv.append(head)
 424
 425     encoding = None
 426     if untracked:
 427         argv.append('--no-index')
 428         argv.append(os.devnull)
 429         argv.append(filename)
 430     elif filename:
 431         argv.append('--')
 432         if isinstance(filename, (list, tuple)):
 433             argv.extend(filename)
 434         else:
 435             argv.append(filename)
 436             encoding = cfg.file_encoding(filename)
 437
 438     status, out, _ = git.diff(
 439         R=reverse,
 440         M=True,
 441         cached=cached,
 442         _encoding=encoding,
 443         *argv,
 444         **common_diff_opts(context)
 445     )
 446
 447     success = status == 0
 448
 449     # Diff will return 1 when comparing untracked file and it has change,
 450     # therefore we will check for diff header from output to differentiate
 451     # from actual error such as file not found.
 452     if untracked and status == 1:
 453         try:
 454             _, second, _ = out.split('\n', 2)
 455         except ValueError:
 456             second = ''
 457         success = second.startswith('new file mode ')
 458
 459     if not success:
 460         # git init
 461         if with_diff_header:
 462             return ('', '')
 463         return ''
 464
 465     result = extract_diff_header(deleted, with_diff_header, suppress_header, out)
 466     return core.UStr(result, out.encoding)
 467
 468
 469 def extract_diff_header(deleted, with_diff_header, suppress_header, diffoutput):
 470     """Split a diff into a header section and payload section"""
 471
 472     if diffoutput.startswith('Submodule'):
 473         if with_diff_header:
 474             return ('', diffoutput)
 475         return diffoutput
 476
 477     start = False
 478     del_tag = 'deleted file mode '
 479
 480     output = StringIO()
 481     headers = StringIO()
 482
 483     for line in diffoutput.split('\n'):
 484         if not start and line[:2] == '@@' and '@@' in line[2:]:
 485             start = True
 486         if start or (deleted and del_tag in line):
 487             output.write(line + '\n')
 488         else:
 489             if with_diff_header:
 490                 headers.write(line + '\n')
 491             elif not suppress_header:
 492                 output.write(line + '\n')
 493
 494     output_text = output.getvalue()
 495     output.close()
 496
 497     headers_text = headers.getvalue()
 498     headers.close()
 499
 500     if with_diff_header:
 501         return (headers_text, output_text)
 502     return output_text
 503
 504
 505 def format_patchsets(context, to_export, revs, output='patches'):
 506     """
 507     Group contiguous revision selection into patchsets
 508
 509     Exists to handle multi-selection.
 510     Multiple disparate ranges in the revision selection
 511     are grouped into continuous lists.
 512
 513     """
 514
 515     outs = []
 516     errs = []
 517
 518     cur_rev = to_export[0]
 519     cur_rev_idx = revs.index(cur_rev)
 520
 521     patches_to_export = [[cur_rev]]
 522     patchset_idx = 0
 523
 524     # Group the patches into continuous sets
 525     for rev in to_export[1:]:
 526         # Limit the search to the current neighborhood for efficiency
 527         try:
 528             rev_idx = revs[cur_rev_idx:].index(rev)
 529             rev_idx += cur_rev_idx
 530         except ValueError:
 531             rev_idx = revs.index(rev)
 532
 533         if rev_idx == cur_rev_idx + 1:
 534             patches_to_export[patchset_idx].append(rev)
 535             cur_rev_idx += 1
 536         else:
 537             patches_to_export.append([rev])
 538             cur_rev_idx = rev_idx
 539             patchset_idx += 1
 540
 541     # Export each patchsets
 542     status = 0
 543     for patchset in patches_to_export:
 544         stat, out, err = export_patchset(
 545             context,
 546             patchset[0],
 547             patchset[-1],
 548             output=output,
 549             n=len(patchset) > 1,
 550             thread=True,
 551             patch_with_stat=True,
 552         )
 553         outs.append(out)
 554         if err:
 555             errs.append(err)
 556         status = max(stat, status)
 557     return (status, '\n'.join(outs), '\n'.join(errs))
 558
 559
 560 def export_patchset(context, start, end, output='patches', **kwargs):
 561     """Export patches from start^ to end."""
 562     git = context.git
 563     return git.format_patch('-o', output, start + '^..' + end, **kwargs)
 564
 565
 566 def reset_paths(context, head, items):
 567     """Run "git reset" while preventing argument overflow"""
 568     items = list(set(items))
 569     fn = context.git.reset
 570     status, out, err = utils.slice_fn(items, lambda paths: fn(head, '--', *paths))
 571     return (status, out, err)
 572
 573
 574 def unstage_paths(context, args, head='HEAD'):
 575     """Unstage paths while accounting for git init"""
 576     status, out, err = reset_paths(context, head, args)
 577     if status == 128:
 578         # handle git init: we have to use 'git rm --cached'
 579         # detect this condition by checking if the file is still staged
 580         return untrack_paths(context, args)
 581     return (status, out, err)
 582
 583
 584 def untrack_paths(context, args):
 585     if not args:
 586         return (-1, N_('Nothing to do'), '')
 587     git = context.git
 588     return git.update_index('--', force_remove=True, *set(args))
 589
 590
 591 def worktree_state(
 592     context, head='HEAD', update_index=False, display_untracked=True, paths=None
 593 ):
 594     """Return a dict of files in various states of being
 595
 596     :rtype: dict, keys are staged, unstaged, untracked, unmerged,
 597             changed_upstream, and submodule.
 598
 599     """
 600     git = context.git
 601     if update_index:
 602         git.update_index(refresh=True)
 603
 604     staged, unmerged, staged_deleted, staged_submods = diff_index(
 605         context, head, paths=paths
 606     )
 607     modified, unstaged_deleted, modified_submods = diff_worktree(context, paths)
 608     if display_untracked:
 609         untracked = untracked_files(context, paths=paths)
 610     else:
 611         untracked = []
 612
 613     # Remove unmerged paths from the modified list
 614     if unmerged:
 615         unmerged_set = set(unmerged)
 616         modified = [path for path in modified if path not in unmerged_set]
 617
 618     # Look for upstream modified files if this is a tracking branch
 619     upstream_changed = diff_upstream(context, head)
 620
 621     # Keep stuff sorted
 622     staged.sort()
 623     modified.sort()
 624     unmerged.sort()
 625     untracked.sort()
 626     upstream_changed.sort()
 627
 628     return {
 629         'staged': staged,
 630         'modified': modified,
 631         'unmerged': unmerged,
 632         'untracked': untracked,
 633         'upstream_changed': upstream_changed,
 634         'staged_deleted': staged_deleted,
 635         'unstaged_deleted': unstaged_deleted,
 636         'submodules': staged_submods | modified_submods,
 637     }
 638
 639
 640 def _parse_raw_diff(out):
 641     while out:
 642         info, path, out = out.split('\0', 2)
 643         status = info[-1]
 644         is_submodule = '160000' in info[1:14]
 645         yield (path, status, is_submodule)
 646
 647
 648 def diff_index(context, head, cached=True, paths=None):
 649     git = context.git
 650     staged = []
 651     unmerged = []
 652     deleted = set()
 653     submodules = set()
 654
 655     if paths is None:
 656         paths = []
 657     args = [head, '--'] + paths
 658     status, out, _ = git.diff_index(cached=cached, z=True, *args)
 659     if status != 0:
 660         # handle git init
 661         args[0] = EMPTY_TREE_OID
 662         status, out, _ = git.diff_index(cached=cached, z=True, *args)
 663
 664     for path, status, is_submodule in _parse_raw_diff(out):
 665         if is_submodule:
 666             submodules.add(path)
 667         if status in 'DAMT':
 668             staged.append(path)
 669             if status == 'D':
 670                 deleted.add(path)
 671         elif status == 'U':
 672             unmerged.append(path)
 673
 674     return staged, unmerged, deleted, submodules
 675
 676
 677 def diff_worktree(context, paths=None):
 678     git = context.git
 679     modified = []
 680     deleted = set()
 681     submodules = set()
 682
 683     if paths is None:
 684         paths = []
 685     args = ['--'] + paths
 686     status, out, _ = git.diff_files(z=True, *args)
 687     for path, status, is_submodule in _parse_raw_diff(out):
 688         if is_submodule:
 689             submodules.add(path)
 690         if status in 'DAMT':
 691             modified.append(path)
 692             if status == 'D':
 693                 deleted.add(path)
 694
 695     return modified, deleted, submodules
 696
 697
 698 def diff_upstream(context, head):
 699     """Given `ref`, return $(git merge-base ref HEAD)..ref."""
 700     tracked = tracked_branch(context)
 701     if not tracked:
 702         return []
 703     base = merge_base(context, head, tracked)
 704     return diff_filenames(context, base, tracked)
 705
 706
 707 def list_submodule(context):
 708     """Return submodules in the format(state, sha1, path, describe)"""
 709     git = context.git
 710     status, data, _ = git.submodule('status')
 711     ret = []
 712     if status == 0 and data:
 713         data = data.splitlines()
 714         # see git submodule status
 715         # TODO better separation
 716         for line in data:
 717             state = line[0].strip()
 718             sha1 = line[1 : OID_LENGTH + 1]
 719             left_bracket = line.find('(', OID_LENGTH + 3)
 720             if left_bracket == -1:
 721                 left_bracket = len(line) + 1
 722             path = line[OID_LENGTH + 2 : left_bracket - 1]
 723             describe = line[left_bracket + 1 : -1]
 724             ret.append((state, sha1, path, describe))
 725     return ret
 726
 727
 728 def merge_base(context, head, ref):
 729     """Return the merge-base of head and ref"""
 730     git = context.git
 731     return git.merge_base(head, ref, _readonly=True)[STDOUT]
 732
 733
 734 def merge_base_parent(context, branch):
 735     tracked = tracked_branch(context, branch=branch)
 736     if tracked:
 737         return tracked
 738     return 'HEAD'
 739
 740
 741 # TODO Unused?
 742 def parse_ls_tree(context, rev):
 743     """Return a list of (mode, type, oid, path) tuples."""
 744     output = []
 745     git = context.git
 746     lines = git.ls_tree(rev, r=True, _readonly=True)[STDOUT].splitlines()
 747     regex = re.compile(r'^(\d+)\W(\w+)\W(\w+)[ \t]+(.*)$')
 748     for line in lines:
 749         match = regex.match(line)
 750         if match:
 751             mode = match.group(1)
 752             objtype = match.group(2)
 753             oid = match.group(3)
 754             filename = match.group(4)
 755             output.append(
 756                 (
 757                     mode,
 758                     objtype,
 759                     oid,
 760                     filename,
 761                 )
 762             )
 763     return output
 764
 765
 766 # TODO unused?
 767 def ls_tree(context, path, ref='HEAD'):
 768     """Return a parsed git ls-tree result for a single directory"""
 769     git = context.git
 770     result = []
 771     status, out, _ = git.ls_tree(ref, '--', path, z=True, full_tree=True)
 772     if status == 0 and out:
 773         path_offset = 6 + 1 + 4 + 1 + OID_LENGTH + 1
 774         for line in out[:-1].split('\0'):
 775             #       1    1                                        1
 776             # .....6 ...4 ......................................40
 777             # 040000 tree c127cde9a0c644a3a8fef449a244f47d5272dfa6      relative
 778             # 100644 blob 139e42bf4acaa4927ec9be1ec55a252b97d3f1e2      relative/path
 779             # 0..... 7... 12......................................      53
 780             # path_offset = 6 + 1 + 4 + 1 + OID_LENGTH(40) + 1
 781             objtype = line[7:11]
 782             relpath = line[path_offset:]
 783             result.append((objtype, relpath))
 784
 785     return result
 786
 787
 788 # A regex for matching the output of git(log|rev-list) --pretty=oneline
 789 REV_LIST_REGEX = re.compile(r'^([0-9a-f]{40}) (.*)$')
 790
 791
 792 def parse_rev_list(raw_revs):
 793     """Parse `git log --pretty=online` output into (oid, summary) pairs."""
 794     revs = []
 795     for line in raw_revs.splitlines():
 796         match = REV_LIST_REGEX.match(line)
 797         if match:
 798             rev_id = match.group(1)
 799             summary = match.group(2)
 800             revs.append(
 801                 (
 802                     rev_id,
 803                     summary,
 804                 )
 805             )
 806     return revs
 807
 808
 809 # pylint: disable=redefined-builtin
 810 def log_helper(context, all=False, extra_args=None):
 811     """Return parallel arrays containing oids and summaries."""
 812     revs = []
 813     summaries = []
 814     args = []
 815     if extra_args:
 816         args = extra_args
 817     git = context.git
 818     output = log(git, pretty='oneline', all=all, *args)
 819     for line in output.splitlines():
 820         match = REV_LIST_REGEX.match(line)
 821         if match:
 822             revs.append(match.group(1))
 823             summaries.append(match.group(2))
 824     return (revs, summaries)
 825
 826
 827 def rev_list_range(context, start, end):
 828     """Return (oid, summary) pairs between start and end."""
 829     git = context.git
 830     revrange = '%s..%s' % (start, end)
 831     out = git.rev_list(revrange, pretty='oneline')[STDOUT]
 832     return parse_rev_list(out)
 833
 834
 835 def commit_message_path(context):
 836     """Return the path to .git/GIT_COLA_MSG"""
 837     git = context.git
 838     path = git.git_path('GIT_COLA_MSG')
 839     if core.exists(path):
 840         return path
 841     return None
 842
 843
 844 def merge_message_path(context):
 845     """Return the path to .git/MERGE_MSG or .git/SQUASH_MSG."""
 846     git = context.git
 847     for basename in ('MERGE_MSG', 'SQUASH_MSG'):
 848         path = git.git_path(basename)
 849         if core.exists(path):
 850             return path
 851     return None
 852
 853
 854 def prepare_commit_message_hook(context):
 855     """Run the cola.preparecommitmessagehook to prepare the commit message"""
 856     config = context.cfg
 857     default_hook = config.hooks_path('cola-prepare-commit-msg')
 858     return config.get('cola.preparecommitmessagehook', default=default_hook)
 859
 860
 861 def abort_merge(context):
 862     """Abort a merge by reading the tree at HEAD."""
 863     # Reset the worktree
 864     git = context.git
 865     status, out, err = git.read_tree('HEAD', reset=True, u=True, v=True)
 866     # remove MERGE_HEAD
 867     merge_head = git.git_path('MERGE_HEAD')
 868     if core.exists(merge_head):
 869         core.unlink(merge_head)
 870     # remove MERGE_MESSAGE, etc.
 871     merge_msg_path = merge_message_path(context)
 872     while merge_msg_path:
 873         core.unlink(merge_msg_path)
 874         merge_msg_path = merge_message_path(context)
 875     return status, out, err
 876
 877
 878 def strip_remote(remotes, remote_branch):
 879     for remote in remotes:
 880         prefix = remote + '/'
 881         if remote_branch.startswith(prefix):
 882             return remote_branch[len(prefix) :]
 883     return remote_branch.split('/', 1)[-1]
 884
 885
 886 def parse_refs(context, argv):
 887     """Parse command-line arguments into object IDs"""
 888     git = context.git
 889     status, out, _ = git.rev_parse(*argv)
 890     if status == 0:
 891         oids = [oid for oid in out.splitlines() if oid]
 892     else:
 893         oids = argv
 894     return oids
 895
 896
 897 def prev_commitmsg(context, *args):
 898     """Queries git for the latest commit message."""
 899     git = context.git
 900     return git.log('-1', no_color=True, pretty='format:%s%n%n%b', *args)[STDOUT]
 901
 902
 903 def rev_parse(context, name):
 904     """Call git rev-parse and return the output"""
 905     git = context.git
 906     status, out, _ = git.rev_parse(name)
 907     if status == 0:
 908         result = out.strip()
 909     else:
 910         result = name
 911     return result
 912
 913
 914 def write_blob(context, oid, filename):
 915     """Write a blob to a temporary file and return the path
 916
 917     Modern versions of Git allow invoking filters.  Older versions
 918     get the object content as-is.
 919
 920     """
 921     if version.check_git(context, 'cat-file-filters-path'):
 922         return cat_file_to_path(context, filename, oid)
 923     return cat_file_blob(context, filename, oid)
 924
 925
 926 def cat_file_blob(context, filename, oid):
 927     return cat_file(context, filename, 'blob', oid)
 928
 929
 930 def cat_file_to_path(context, filename, oid):
 931     return cat_file(context, filename, oid, path=filename, filters=True)
 932
 933
 934 def cat_file(context, filename, *args, **kwargs):
 935     """Redirect git cat-file output to a path"""
 936     result = None
 937     git = context.git
 938     # Use the original filename in the suffix so that the generated filename
 939     # has the correct extension, and so that it resembles the original name.
 940     basename = os.path.basename(filename)
 941     suffix = '-' + basename  # ensures the correct filename extension
 942     path = utils.tmp_filename('blob', suffix=suffix)
 943     with open(path, 'wb') as fp:
 944         status, out, err = git.cat_file(
 945             _raw=True, _readonly=True, _stdout=fp, *args, **kwargs
 946         )
 947         Interaction.command(N_('Error'), 'git cat-file', status, out, err)
 948         if status == 0:
 949             result = path
 950     if not result:
 951         core.unlink(path)
 952     return result
 953
 954
 955 def write_blob_path(context, head, oid, filename):
 956     """Use write_blob() when modern git is available"""
 957     if version.check_git(context, 'cat-file-filters-path'):
 958         return write_blob(context, oid, filename)
 959     return cat_file_blob(context, filename, head + ':' + filename)
 960
 961
 962 def annex_path(context, head, filename):
 963     """Return the git-annex path for a filename at the specified commit"""
 964     git = context.git
 965     path = None
 966     annex_info = {}
 967
 968     # unfortunately there's no way to filter this down to a single path
 969     # so we just have to scan all reported paths
 970     status, out, _ = git.annex('findref', '--json', head)
 971     if status == 0:
 972         for line in out.splitlines():
 973             info = json.loads(line)
 974             try:
 975                 annex_file = info['file']
 976             except (ValueError, KeyError):
 977                 continue
 978             # we only care about this file so we can skip the rest
 979             if annex_file == filename:
 980                 annex_info = info
 981                 break
 982     key = annex_info.get('key', '')
 983     if key:
 984         status, out, _ = git.annex('contentlocation', key)
 985         if status == 0 and os.path.exists(out):
 986             path = out
 987
 988     return path
 989
 990
 991 def is_binary(context, filename):
 992     cfg_is_binary = context.cfg.is_binary(filename)
 993     if cfg_is_binary is not None:
 994         return cfg_is_binary
 995     # This is the same heuristic as xdiff-interface.c:buffer_is_binary().
 996     size = 8000
 997     try:
 998         result = core.read(filename, size=size, encoding='bytes')
 999     except (IOError, OSError):
1000         result = b''
1001
1002     return b'\0' in result
1003
1004
1005 def is_valid_ref(context, ref):
1006     """Is the provided Git ref a valid refname?"""
1007     status, _, _ = context.git.rev_parse(ref, quiet=True, verify=True)
1008     return status == 0