cola/gitcmds.py

   1 """Git commands and queries for Git"""
   2 import json
   3 import os
   4 import re
   5 from io import StringIO
   6
   7 from . import core
   8 from . import utils
   9 from . import version
  10 from .git import STDOUT
  11 from .git import EMPTY_TREE_OID
  12 from .git import OID_LENGTH
  13 from .i18n import N_
  14 from .interaction import Interaction
  15 from .models import prefs
  16
  17
  18 def add(context, items, u=False):
  19     """Run "git add" while preventing argument overflow"""
  20     git_add = context.git.add
  21     return utils.slice_func(
  22         items, lambda paths: git_add('--', force=True, verbose=True, u=u, *paths)
  23     )
  24
  25
  26 def apply_diff(context, filename):
  27     """Use "git apply" to apply the patch in `filename` to the staging area"""
  28     git = context.git
  29     return git.apply(filename, index=True, cached=True)
  30
  31
  32 def apply_diff_to_worktree(context, filename):
  33     """Use "git apply" to apply the patch in `filename` to the worktree"""
  34     git = context.git
  35     return git.apply(filename)
  36
  37
  38 def get_branch(context, branch):
  39     """Get the current branch"""
  40     if branch is None:
  41         branch = current_branch(context)
  42     return branch
  43
  44
  45 def upstream_remote(context, branch=None):
  46     """Return the remote associated with the specified branch"""
  47     config = context.cfg
  48     branch = get_branch(context, branch)
  49     return config.get('branch.%s.remote' % branch)
  50
  51
  52 def remote_url(context, remote, push=False):
  53     """Return the URL for the specified remote"""
  54     config = context.cfg
  55     url = config.get('remote.%s.url' % remote, '')
  56     if push:
  57         url = config.get('remote.%s.pushurl' % remote, url)
  58     return url
  59
  60
  61 def diff_index_filenames(context, ref):
  62     """
  63     Return a diff of filenames that have been modified relative to the index
  64     """
  65     git = context.git
  66     out = git.diff_index(ref, name_only=True, z=True, _readonly=True)[STDOUT]
  67     return _parse_diff_filenames(out)
  68
  69
  70 def diff_filenames(context, *args):
  71     """Return a list of filenames that have been modified"""
  72     git = context.git
  73     out = git.diff_tree(
  74         name_only=True, no_commit_id=True, r=True, z=True, _readonly=True, *args
  75     )[STDOUT]
  76     return _parse_diff_filenames(out)
  77
  78
  79 def listdir(context, dirname, ref='HEAD'):
  80     """Get the contents of a directory according to Git
  81
  82     Query Git for the content of a directory, taking ignored
  83     files into account.
  84
  85     """
  86     dirs = []
  87     files = []
  88
  89     # first, parse git ls-tree to get the tracked files
  90     # in a list of (type, path) tuples
  91     entries = ls_tree(context, dirname, ref=ref)
  92     for entry in entries:
  93         if entry[0][0] == 't':  # tree
  94             dirs.append(entry[1])
  95         else:
  96             files.append(entry[1])
  97
  98     # gather untracked files
  99     untracked = untracked_files(context, paths=[dirname], directory=True)
 100     for path in untracked:
 101         if path.endswith('/'):
 102             dirs.append(path[:-1])
 103         else:
 104             files.append(path)
 105
 106     dirs.sort()
 107     files.sort()
 108
 109     return (dirs, files)
 110
 111
 112 def diff(context, args):
 113     """Return a list of filenames for the given diff arguments
 114
 115     :param args: list of arguments to pass to "git diff --name-only"
 116
 117     """
 118     git = context.git
 119     out = git.diff(name_only=True, z=True, _readonly=True, *args)[STDOUT]
 120     return _parse_diff_filenames(out)
 121
 122
 123 def _parse_diff_filenames(out):
 124     if out:
 125         return out[:-1].split('\0')
 126     return []
 127
 128
 129 def tracked_files(context, *args):
 130     """Return the names of all files in the repository"""
 131     git = context.git
 132     out = git.ls_files('--', *args, z=True, _readonly=True)[STDOUT]
 133     if out:
 134         return sorted(out[:-1].split('\0'))
 135     return []
 136
 137
 138 def all_files(context, *args):
 139     """Returns a sorted list of all files, including untracked files."""
 140     git = context.git
 141     ls_files = git.ls_files(
 142         '--',
 143         *args,
 144         z=True,
 145         cached=True,
 146         others=True,
 147         exclude_standard=True,
 148         _readonly=True,
 149     )[STDOUT]
 150     return sorted([f for f in ls_files.split('\0') if f])
 151
 152
 153 class CurrentBranchCache:
 154     """Cache for current_branch()"""
 155
 156     key = None
 157     value = None
 158
 159
 160 def reset():
 161     """Reset cached value in this module (eg. the cached current branch)"""
 162     CurrentBranchCache.key = None
 163
 164
 165 def current_branch(context):
 166     """Return the current branch"""
 167     git = context.git
 168     head = git.git_path('HEAD')
 169     try:
 170         key = core.stat(head).st_mtime
 171         if CurrentBranchCache.key == key:
 172             return CurrentBranchCache.value
 173     except OSError:
 174         # OSError means we can't use the stat cache
 175         key = 0
 176
 177     status, data, _ = git.rev_parse('HEAD', symbolic_full_name=True, _readonly=True)
 178     if status != 0:
 179         # git init -- read .git/HEAD.  We could do this unconditionally...
 180         data = _read_git_head(context, head)
 181
 182     for refs_prefix in ('refs/heads/', 'refs/remotes/', 'refs/tags/'):
 183         if data.startswith(refs_prefix):
 184             value = data[len(refs_prefix) :]
 185             CurrentBranchCache.key = key
 186             CurrentBranchCache.value = value
 187             return value
 188     # Detached head
 189     return data
 190
 191
 192 def _read_git_head(context, head, default='main'):
 193     """Pure-python .git/HEAD reader"""
 194     # Common .git/HEAD "ref: refs/heads/main" files
 195     git = context.git
 196     islink = core.islink(head)
 197     if core.isfile(head) and not islink:
 198         data = core.read(head).rstrip()
 199         ref_prefix = 'ref: '
 200         if data.startswith(ref_prefix):
 201             return data[len(ref_prefix) :]
 202         # Detached head
 203         return data
 204     # Legacy .git/HEAD symlinks
 205     if islink:
 206         refs_heads = core.realpath(git.git_path('refs', 'heads'))
 207         path = core.abspath(head).replace('\\', '/')
 208         if path.startswith(refs_heads + '/'):
 209             return path[len(refs_heads) + 1 :]
 210
 211     return default
 212
 213
 214 def branch_list(context, remote=False):
 215     """
 216     Return a list of local or remote branches
 217
 218     This explicitly removes HEAD from the list of remote branches.
 219
 220     """
 221     if remote:
 222         return for_each_ref_basename(context, 'refs/remotes')
 223     return for_each_ref_basename(context, 'refs/heads')
 224
 225
 226 def _version_sort(context, key='version:refname'):
 227     if version.check_git(context, 'version-sort'):
 228         sort = key
 229     else:
 230         sort = False
 231     return sort
 232
 233
 234 def for_each_ref_basename(context, refs):
 235     """Return refs starting with 'refs'."""
 236     git = context.git
 237     sort = _version_sort(context)
 238     _, out, _ = git.for_each_ref(refs, format='%(refname)', sort=sort, _readonly=True)
 239     output = out.splitlines()
 240     non_heads = [x for x in output if not x.endswith('/HEAD')]
 241     offset = len(refs) + 1
 242     return [x[offset:] for x in non_heads]
 243
 244
 245 def _prefix_and_size(prefix, values):
 246     """Return a tuple of (prefix, len(prefix) + 1, y) for <prefix>/ stripping"""
 247     return (prefix, len(prefix) + 1, values)
 248
 249
 250 def all_refs(context, split=False, sort_key='version:refname'):
 251     """Return a tuple of (local branches, remote branches, tags)."""
 252     git = context.git
 253     local_branches = []
 254     remote_branches = []
 255     tags = []
 256     query = (
 257         _prefix_and_size('refs/tags', tags),
 258         _prefix_and_size('refs/heads', local_branches),
 259         _prefix_and_size('refs/remotes', remote_branches),
 260     )
 261     sort = _version_sort(context, key=sort_key)
 262     _, out, _ = git.for_each_ref(format='%(refname)', sort=sort, _readonly=True)
 263     for ref in out.splitlines():
 264         for prefix, prefix_len, dst in query:
 265             if ref.startswith(prefix) and not ref.endswith('/HEAD'):
 266                 dst.append(ref[prefix_len:])
 267                 continue
 268     tags.reverse()
 269     if split:
 270         return local_branches, remote_branches, tags
 271     return local_branches + remote_branches + tags
 272
 273
 274 def tracked_branch(context, branch=None):
 275     """Return the remote branch associated with 'branch'."""
 276     if branch is None:
 277         branch = current_branch(context)
 278     if branch is None:
 279         return None
 280     config = context.cfg
 281     remote = config.get('branch.%s.remote' % branch)
 282     if not remote:
 283         return None
 284     merge_ref = config.get('branch.%s.merge' % branch)
 285     if not merge_ref:
 286         return None
 287     refs_heads = 'refs/heads/'
 288     if merge_ref.startswith(refs_heads):
 289         return remote + '/' + merge_ref[len(refs_heads) :]
 290     return None
 291
 292
 293 def parse_remote_branch(branch):
 294     """Split a remote branch apart into (remote, name) components"""
 295     rgx = re.compile(r'^(?P<remote>[^/]+)/(?P<branch>.+)$')
 296     match = rgx.match(branch)
 297     remote = ''
 298     branch = ''
 299     if match:
 300         remote = match.group('remote')
 301         branch = match.group('branch')
 302     return (remote, branch)
 303
 304
 305 def untracked_files(context, paths=None, **kwargs):
 306     """Returns a sorted list of untracked files."""
 307     git = context.git
 308     if paths is None:
 309         paths = []
 310     args = ['--'] + paths
 311     out = git.ls_files(
 312         z=True, others=True, exclude_standard=True, _readonly=True, *args, **kwargs
 313     )[STDOUT]
 314     if out:
 315         return out[:-1].split('\0')
 316     return []
 317
 318
 319 def tag_list(context):
 320     """Return a list of tags."""
 321     result = for_each_ref_basename(context, 'refs/tags')
 322     result.reverse()
 323     return result
 324
 325
 326 def log(git, *args, **kwargs):
 327     return git.log(
 328         no_color=True,
 329         no_abbrev_commit=True,
 330         no_ext_diff=True,
 331         _readonly=True,
 332         *args,
 333         **kwargs,
 334     )[STDOUT]
 335
 336
 337 def commit_diff(context, oid):
 338     git = context.git
 339     return log(git, '-1', oid, '--') + '\n\n' + oid_diff(context, oid)
 340
 341
 342 _diff_overrides = {}
 343
 344
 345 def update_diff_overrides(space_at_eol, space_change, all_space, function_context):
 346     _diff_overrides['ignore_space_at_eol'] = space_at_eol
 347     _diff_overrides['ignore_space_change'] = space_change
 348     _diff_overrides['ignore_all_space'] = all_space
 349     _diff_overrides['function_context'] = function_context
 350
 351
 352 def common_diff_opts(context):
 353     config = context.cfg
 354     # Default to --patience when diff.algorithm is unset
 355     patience = not config.get('diff.algorithm', default='')
 356     submodule = version.check_git(context, 'diff-submodule')
 357     opts = {
 358         'patience': patience,
 359         'submodule': submodule,
 360         'no_color': True,
 361         'no_ext_diff': True,
 362         'unified': config.get('gui.diffcontext', default=3),
 363         '_raw': True,
 364         '_readonly': True,
 365     }
 366     opts.update(_diff_overrides)
 367     return opts
 368
 369
 370 def _add_filename(args, filename):
 371     if filename:
 372         args.extend(['--', filename])
 373
 374
 375 def oid_diff(context, oid, filename=None):
 376     """Return the diff for an oid"""
 377     # Naively "$oid^!" is what we'd like to use but that doesn't
 378     # give the correct result for merges--the diff is reversed.
 379     # Be explicit and compare oid against its first parent.
 380     return oid_diff_range(context, oid + '~', oid, filename=filename)
 381
 382
 383 def oid_diff_range(context, start, end, filename=None):
 384     """Reeturn the diff for a commit range"""
 385     args = [start, end]
 386     git = context.git
 387     opts = common_diff_opts(context)
 388     _add_filename(args, filename)
 389     status, out, _ = git.diff(*args, **opts)
 390     if status != 0:
 391         # We probably don't have "$oid~" because this is the root commit.
 392         # "git show" is clever enough to handle the root commit.
 393         args = [end + '^!']
 394         _add_filename(args, filename)
 395         _, out, _ = git.show(pretty='format:', *args, **opts)
 396         out = out.lstrip()
 397     return out
 398
 399
 400 def diff_info(context, oid, filename=None):
 401     """Return the diff for the specified oid"""
 402     return diff_range(context, oid + '~', oid, filename=filename)
 403
 404
 405 def diff_range(context, start, end, filename=None):
 406     """Return the diff for the specified commit range"""
 407     git = context.git
 408     decoded = log(git, '-1', end, '--', pretty='format:%b').strip()
 409     if decoded:
 410         decoded += '\n\n'
 411     return decoded + oid_diff_range(context, start, end, filename=filename)
 412
 413
 414 # pylint: disable=too-many-arguments
 415 def diff_helper(
 416     context,
 417     commit=None,
 418     ref=None,
 419     endref=None,
 420     filename=None,
 421     cached=True,
 422     deleted=False,
 423     head=None,
 424     amending=False,
 425     with_diff_header=False,
 426     suppress_header=True,
 427     reverse=False,
 428     untracked=False,
 429 ):
 430     'Invokes git diff on a filepath.'
 431     git = context.git
 432     cfg = context.cfg
 433     if commit:
 434         ref, endref = commit + '^', commit
 435     argv = []
 436     if ref and endref:
 437         argv.append(f'{ref}..{endref}')
 438     elif ref:
 439         argv.extend(utils.shell_split(ref.strip()))
 440     elif head and amending and cached:
 441         argv.append(head)
 442
 443     encoding = None
 444     if untracked:
 445         argv.append('--no-index')
 446         argv.append(os.devnull)
 447         argv.append(filename)
 448     elif filename:
 449         argv.append('--')
 450         if isinstance(filename, (list, tuple)):
 451             argv.extend(filename)
 452         else:
 453             argv.append(filename)
 454             encoding = cfg.file_encoding(filename)
 455
 456     status, out, _ = git.diff(
 457         R=reverse,
 458         M=True,
 459         cached=cached,
 460         _encoding=encoding,
 461         *argv,
 462         **common_diff_opts(context),
 463     )
 464
 465     success = status == 0
 466
 467     # Diff will return 1 when comparing untracked file and it has change,
 468     # therefore we will check for diff header from output to differentiate
 469     # from actual error such as file not found.
 470     if untracked and status == 1:
 471         try:
 472             _, second, _ = out.split('\n', 2)
 473         except ValueError:
 474             second = ''
 475         success = second.startswith('new file mode ')
 476
 477     if not success:
 478         # git init
 479         if with_diff_header:
 480             return ('', '')
 481         return ''
 482
 483     result = extract_diff_header(deleted, with_diff_header, suppress_header, out)
 484     return core.UStr(result, out.encoding)
 485
 486
 487 def extract_diff_header(deleted, with_diff_header, suppress_header, diffoutput):
 488     """Split a diff into a header section and payload section"""
 489
 490     if diffoutput.startswith('Submodule'):
 491         if with_diff_header:
 492             return ('', diffoutput)
 493         return diffoutput
 494
 495     start = False
 496     del_tag = 'deleted file mode '
 497
 498     output = StringIO()
 499     headers = StringIO()
 500
 501     for line in diffoutput.split('\n'):
 502         if not start and line[:2] == '@@' and '@@' in line[2:]:
 503             start = True
 504         if start or (deleted and del_tag in line):
 505             output.write(line + '\n')
 506         else:
 507             if with_diff_header:
 508                 headers.write(line + '\n')
 509             elif not suppress_header:
 510                 output.write(line + '\n')
 511
 512     output_text = output.getvalue()
 513     output.close()
 514
 515     headers_text = headers.getvalue()
 516     headers.close()
 517
 518     if with_diff_header:
 519         return (headers_text, output_text)
 520     return output_text
 521
 522
 523 def format_patchsets(context, to_export, revs, output='patches'):
 524     """
 525     Group contiguous revision selection into patchsets
 526
 527     Exists to handle multi-selection.
 528     Multiple disparate ranges in the revision selection
 529     are grouped into continuous lists.
 530
 531     """
 532
 533     outs = []
 534     errs = []
 535
 536     cur_rev = to_export[0]
 537     cur_rev_idx = revs.index(cur_rev)
 538
 539     patches_to_export = [[cur_rev]]
 540     patchset_idx = 0
 541
 542     # Group the patches into continuous sets
 543     for rev in to_export[1:]:
 544         # Limit the search to the current neighborhood for efficiency
 545         try:
 546             rev_idx = revs[cur_rev_idx:].index(rev)
 547             rev_idx += cur_rev_idx
 548         except ValueError:
 549             rev_idx = revs.index(rev)
 550
 551         if rev_idx == cur_rev_idx + 1:
 552             patches_to_export[patchset_idx].append(rev)
 553             cur_rev_idx += 1
 554         else:
 555             patches_to_export.append([rev])
 556             cur_rev_idx = rev_idx
 557             patchset_idx += 1
 558
 559     # Export each patchsets
 560     status = 0
 561     for patchset in patches_to_export:
 562         stat, out, err = export_patchset(
 563             context,
 564             patchset[0],
 565             patchset[-1],
 566             output=output,
 567             n=len(patchset) > 1,
 568             thread=True,
 569             patch_with_stat=True,
 570         )
 571         outs.append(out)
 572         if err:
 573             errs.append(err)
 574         status = max(stat, status)
 575     return (status, '\n'.join(outs), '\n'.join(errs))
 576
 577
 578 def export_patchset(context, start, end, output='patches', **kwargs):
 579     """Export patches from start^ to end."""
 580     git = context.git
 581     return git.format_patch('-o', output, start + '^..' + end, **kwargs)
 582
 583
 584 def reset_paths(context, head, items):
 585     """Run "git reset" while preventing argument overflow"""
 586     items = list(set(items))
 587     func = context.git.reset
 588     status, out, err = utils.slice_func(items, lambda paths: func(head, '--', *paths))
 589     return (status, out, err)
 590
 591
 592 def unstage_paths(context, args, head='HEAD'):
 593     """Unstage paths while accounting for git init"""
 594     status, out, err = reset_paths(context, head, args)
 595     if status == 128:
 596         # handle git init: we have to use 'git rm --cached'
 597         # detect this condition by checking if the file is still staged
 598         return untrack_paths(context, args)
 599     return (status, out, err)
 600
 601
 602 def untrack_paths(context, args):
 603     if not args:
 604         return (-1, N_('Nothing to do'), '')
 605     git = context.git
 606     return git.update_index('--', force_remove=True, *set(args))
 607
 608
 609 def worktree_state(
 610     context, head='HEAD', update_index=False, display_untracked=True, paths=None
 611 ):
 612     """Return a dict of files in various states of being
 613
 614     :rtype: dict, keys are staged, unstaged, untracked, unmerged,
 615             changed_upstream, and submodule.
 616
 617     """
 618     git = context.git
 619     if update_index:
 620         git.update_index(refresh=True)
 621
 622     staged, unmerged, staged_deleted, staged_submods = diff_index(
 623         context, head, paths=paths
 624     )
 625     modified, unstaged_deleted, modified_submods = diff_worktree(context, paths)
 626     if display_untracked:
 627         untracked = untracked_files(context, paths=paths)
 628     else:
 629         untracked = []
 630
 631     # Remove unmerged paths from the modified list
 632     if unmerged:
 633         unmerged_set = set(unmerged)
 634         modified = [path for path in modified if path not in unmerged_set]
 635
 636     # Look for upstream modified files if this is a tracking branch
 637     upstream_changed = diff_upstream(context, head)
 638
 639     # Keep stuff sorted
 640     staged.sort()
 641     modified.sort()
 642     unmerged.sort()
 643     untracked.sort()
 644     upstream_changed.sort()
 645
 646     return {
 647         'staged': staged,
 648         'modified': modified,
 649         'unmerged': unmerged,
 650         'untracked': untracked,
 651         'upstream_changed': upstream_changed,
 652         'staged_deleted': staged_deleted,
 653         'unstaged_deleted': unstaged_deleted,
 654         'submodules': staged_submods | modified_submods,
 655     }
 656
 657
 658 def _parse_raw_diff(out):
 659     while out:
 660         info, path, out = out.split('\0', 2)
 661         status = info[-1]
 662         is_submodule = '160000' in info[1:14]
 663         yield (path, status, is_submodule)
 664
 665
 666 def diff_index(context, head, cached=True, paths=None):
 667     git = context.git
 668     staged = []
 669     unmerged = []
 670     deleted = set()
 671     submodules = set()
 672
 673     if paths is None:
 674         paths = []
 675     args = [head, '--'] + paths
 676     status, out, _ = git.diff_index(cached=cached, z=True, _readonly=True, *args)
 677     if status != 0:
 678         # handle git init
 679         args[0] = EMPTY_TREE_OID
 680         status, out, _ = git.diff_index(cached=cached, z=True, _readonly=True, *args)
 681
 682     for path, status, is_submodule in _parse_raw_diff(out):
 683         if is_submodule:
 684             submodules.add(path)
 685         if status in 'DAMT':
 686             staged.append(path)
 687             if status == 'D':
 688                 deleted.add(path)
 689         elif status == 'U':
 690             unmerged.append(path)
 691
 692     return staged, unmerged, deleted, submodules
 693
 694
 695 def diff_worktree(context, paths=None):
 696     git = context.git
 697     ignore_submodules_value = context.cfg.get('diff.ignoresubmodules', 'none')
 698     ignore_submodules = ignore_submodules_value in {'all', 'dirty', 'untracked'}
 699     modified = []
 700     deleted = set()
 701     submodules = set()
 702
 703     if paths is None:
 704         paths = []
 705     args = ['--'] + paths
 706     status, out, _ = git.diff_files(z=True, _readonly=True, *args)
 707     for path, status, is_submodule in _parse_raw_diff(out):
 708         if is_submodule:
 709             submodules.add(path)
 710             if ignore_submodules:
 711                 continue
 712         if status in 'DAMT':
 713             modified.append(path)
 714             if status == 'D':
 715                 deleted.add(path)
 716
 717     return modified, deleted, submodules
 718
 719
 720 def diff_upstream(context, head):
 721     """Given `ref`, return $(git merge-base ref HEAD)..ref."""
 722     tracked = tracked_branch(context)
 723     if not tracked:
 724         return []
 725     base = merge_base(context, head, tracked)
 726     return diff_filenames(context, base, tracked)
 727
 728
 729 def list_submodule(context):
 730     """Return submodules in the format(state, sha1, path, describe)"""
 731     git = context.git
 732     status, data, _ = git.submodule('status')
 733     ret = []
 734     if status == 0 and data:
 735         data = data.splitlines()
 736         # see git submodule status
 737         for line in data:
 738             state = line[0].strip()
 739             sha1 = line[1 : OID_LENGTH + 1]
 740             left_bracket = line.find('(', OID_LENGTH + 3)
 741             if left_bracket == -1:
 742                 left_bracket = len(line) + 1
 743             path = line[OID_LENGTH + 2 : left_bracket - 1]
 744             describe = line[left_bracket + 1 : -1]
 745             ret.append((state, sha1, path, describe))
 746     return ret
 747
 748
 749 def merge_base(context, head, ref):
 750     """Return the merge-base of head and ref"""
 751     git = context.git
 752     return git.merge_base(head, ref, _readonly=True)[STDOUT]
 753
 754
 755 def merge_base_parent(context, branch):
 756     tracked = tracked_branch(context, branch=branch)
 757     if tracked:
 758         return tracked
 759     return 'HEAD'
 760
 761
 762 def ls_tree(context, path, ref='HEAD'):
 763     """Return a parsed git ls-tree result for a single directory"""
 764     git = context.git
 765     result = []
 766     status, out, _ = git.ls_tree(
 767         ref, '--', path, z=True, full_tree=True, _readonly=True
 768     )
 769     if status == 0 and out:
 770         path_offset = 6 + 1 + 4 + 1 + OID_LENGTH + 1
 771         for line in out[:-1].split('\0'):
 772             #       1    1                                        1
 773             # .....6 ...4 ......................................40
 774             # 040000 tree c127cde9a0c644a3a8fef449a244f47d5272dfa6      relative
 775             # 100644 blob 139e42bf4acaa4927ec9be1ec55a252b97d3f1e2      relative/path
 776             # 0..... 7... 12......................................      53
 777             # path_offset = 6 + 1 + 4 + 1 + OID_LENGTH(40) + 1
 778             objtype = line[7:11]
 779             relpath = line[path_offset:]
 780             result.append((objtype, relpath))
 781
 782     return result
 783
 784
 785 # A regex for matching the output of git(log|rev-list) --pretty=oneline
 786 REV_LIST_REGEX = re.compile(r'^([0-9a-f]{40}) (.*)$')
 787
 788
 789 def parse_rev_list(raw_revs):
 790     """Parse `git log --pretty=online` output into (oid, summary) pairs."""
 791     revs = []
 792     for line in raw_revs.splitlines():
 793         match = REV_LIST_REGEX.match(line)
 794         if match:
 795             rev_id = match.group(1)
 796             summary = match.group(2)
 797             revs.append((
 798                 rev_id,
 799                 summary,
 800             ))
 801     return revs
 802
 803
 804 # pylint: disable=redefined-builtin
 805 def log_helper(context, all=False, extra_args=None):
 806     """Return parallel arrays containing oids and summaries."""
 807     revs = []
 808     summaries = []
 809     args = []
 810     if extra_args:
 811         args = extra_args
 812     git = context.git
 813     output = log(git, pretty='oneline', all=all, *args)
 814     for line in output.splitlines():
 815         match = REV_LIST_REGEX.match(line)
 816         if match:
 817             revs.append(match.group(1))
 818             summaries.append(match.group(2))
 819     return (revs, summaries)
 820
 821
 822 def rev_list_range(context, start, end):
 823     """Return (oid, summary) pairs between start and end."""
 824     git = context.git
 825     revrange = f'{start}..{end}'
 826     out = git.rev_list(revrange, pretty='oneline', _readonly=True)[STDOUT]
 827     return parse_rev_list(out)
 828
 829
 830 def commit_message_path(context):
 831     """Return the path to .git/GIT_COLA_MSG"""
 832     git = context.git
 833     path = git.git_path('GIT_COLA_MSG')
 834     if core.exists(path):
 835         return path
 836     return None
 837
 838
 839 def merge_message_path(context):
 840     """Return the path to .git/MERGE_MSG or .git/SQUASH_MSG."""
 841     git = context.git
 842     for basename in ('MERGE_MSG', 'SQUASH_MSG'):
 843         path = git.git_path(basename)
 844         if core.exists(path):
 845             return path
 846     return None
 847
 848
 849 def read_merge_commit_message(context, path):
 850     """Read a merge commit message from disk while stripping commentary"""
 851     content = core.read(path)
 852     cleanup_mode = prefs.commit_cleanup(context)
 853     if cleanup_mode in ('verbatim', 'scissors', 'whitespace'):
 854         return content
 855     comment_char = prefs.comment_char(context)
 856     return '\n'.join(
 857         (line for line in content.splitlines() if not line.startswith(comment_char))
 858     )
 859
 860
 861 def prepare_commit_message_hook(context):
 862     """Run the cola.preparecommitmessagehook to prepare the commit message"""
 863     config = context.cfg
 864     default_hook = config.hooks_path('cola-prepare-commit-msg')
 865     return config.get('cola.preparecommitmessagehook', default=default_hook)
 866
 867
 868 def cherry_pick(context, revs):
 869     """Cherry-picks each revision into the current branch.
 870
 871     Returns (0, out, err) where stdout and stderr across all "git cherry-pick"
 872     invocations are combined into single values when all cherry-picks succeed.
 873
 874     Returns a combined (status, out, err) of the first failing "git cherry-pick"
 875     in the event of a non-zero exit status.
 876     """
 877     if not revs:
 878         return []
 879     outs = []
 880     errs = []
 881     status = 0
 882     for rev in revs:
 883         status, out, err = context.git.cherry_pick(rev)
 884         if status != 0:
 885             details = N_(
 886                 'Hint: The "Actions > Abort Cherry-Pick" menu action can be used to '
 887                 'cancel the current cherry-pick.'
 888             )
 889             output = f'# git cherry-pick {rev}\n# {details}\n\n{out}'
 890             return (status, output, err)
 891         outs.append(out)
 892         errs.append(err)
 893     return (0, '\n'.join(outs), '\n'.join(errs))
 894
 895
 896 def abort_apply_patch(context):
 897     """Abort a "git am" session."""
 898     # Reset the worktree
 899     git = context.git
 900     status, out, err = git.am(abort=True)
 901     return status, out, err
 902
 903
 904 def abort_cherry_pick(context):
 905     """Abort a cherry-pick."""
 906     # Reset the worktree
 907     git = context.git
 908     status, out, err = git.cherry_pick(abort=True)
 909     return status, out, err
 910
 911
 912 def abort_merge(context):
 913     """Abort a merge"""
 914     # Reset the worktree
 915     git = context.git
 916     status, out, err = git.merge(abort=True)
 917     return status, out, err
 918
 919
 920 def strip_remote(remotes, remote_branch):
 921     """Get branch names with the "<remote>/" prefix removed"""
 922     for remote in remotes:
 923         prefix = remote + '/'
 924         if remote_branch.startswith(prefix):
 925             return remote_branch[len(prefix) :]
 926     return remote_branch.split('/', 1)[-1]
 927
 928
 929 def parse_refs(context, argv):
 930     """Parse command-line arguments into object IDs"""
 931     git = context.git
 932     status, out, _ = git.rev_parse(_readonly=True, *argv)
 933     if status == 0:
 934         oids = [oid for oid in out.splitlines() if oid]
 935     else:
 936         oids = argv
 937     return oids
 938
 939
 940 def prev_commitmsg(context, *args):
 941     """Queries git for the latest commit message."""
 942     git = context.git
 943     return git.log(
 944         '-1', no_color=True, pretty='format:%s%n%n%b', _readonly=True, *args
 945     )[STDOUT]
 946
 947
 948 def rev_parse(context, name):
 949     """Call git rev-parse and return the output"""
 950     git = context.git
 951     status, out, _ = git.rev_parse(name, _readonly=True)
 952     if status == 0:
 953         result = out.strip()
 954     else:
 955         result = name
 956     return result
 957
 958
 959 def write_blob(context, oid, filename):
 960     """Write a blob to a temporary file and return the path
 961
 962     Modern versions of Git allow invoking filters.  Older versions
 963     get the object content as-is.
 964
 965     """
 966     if version.check_git(context, 'cat-file-filters-path'):
 967         return cat_file_to_path(context, filename, oid)
 968     return cat_file_blob(context, filename, oid)
 969
 970
 971 def cat_file_blob(context, filename, oid):
 972     """Write a blob from git to the specified filename"""
 973     return cat_file(context, filename, 'blob', oid)
 974
 975
 976 def cat_file_to_path(context, filename, oid):
 977     """Extract a file from an commit ref and a write it to the specified filename"""
 978     return cat_file(context, filename, oid, path=filename, filters=True)
 979
 980
 981 def cat_file(context, filename, *args, **kwargs):
 982     """Redirect git cat-file output to a path"""
 983     result = None
 984     git = context.git
 985     # Use the original filename in the suffix so that the generated filename
 986     # has the correct extension, and so that it resembles the original name.
 987     basename = os.path.basename(filename)
 988     suffix = '-' + basename  # ensures the correct filename extension
 989     path = utils.tmp_filename('blob', suffix=suffix)
 990     with open(path, 'wb') as tmp_file:
 991         status, out, err = git.cat_file(
 992             _raw=True, _readonly=True, _stdout=tmp_file, *args, **kwargs
 993         )
 994         Interaction.command(N_('Error'), 'git cat-file', status, out, err)
 995         if status == 0:
 996             result = path
 997     if not result:
 998         core.unlink(path)
 999     return result
1000
1001
1002 def write_blob_path(context, head, oid, filename):
1003     """Use write_blob() when modern git is available"""
1004     if version.check_git(context, 'cat-file-filters-path'):
1005         return write_blob(context, oid, filename)
1006     return cat_file_blob(context, filename, head + ':' + filename)
1007
1008
1009 def annex_path(context, head, filename):
1010     """Return the git-annex path for a filename at the specified commit"""
1011     git = context.git
1012     path = None
1013     annex_info = {}
1014
1015     # unfortunately there's no way to filter this down to a single path
1016     # so we just have to scan all reported paths
1017     status, out, _ = git.annex('findref', '--json', head, _readonly=True)
1018     if status == 0:
1019         for line in out.splitlines():
1020             info = json.loads(line)
1021             try:
1022                 annex_file = info['file']
1023             except (ValueError, KeyError):
1024                 continue
1025             # we only care about this file so we can skip the rest
1026             if annex_file == filename:
1027                 annex_info = info
1028                 break
1029     key = annex_info.get('key', '')
1030     if key:
1031         status, out, _ = git.annex('contentlocation', key, _readonly=True)
1032         if status == 0 and os.path.exists(out):
1033             path = out
1034
1035     return path
1036
1037
1038 def is_binary(context, filename):
1039     """A heustic to determine whether `filename` contains (non-text) binary content"""
1040     cfg_is_binary = context.cfg.is_binary(filename)
1041     if cfg_is_binary is not None:
1042         return cfg_is_binary
1043     # This is the same heuristic as xdiff-interface.c:buffer_is_binary().
1044     size = 8000
1045     try:
1046         result = core.read(filename, size=size, encoding='bytes')
1047     except OSError:
1048         result = b''
1049
1050     return b'\0' in result
1051
1052
1053 def is_valid_ref(context, ref):
1054     """Is the provided Git ref a valid refname?"""
1055     status, _, _ = context.git.rev_parse(ref, quiet=True, verify=True, _readonly=True)
1056     return status == 0