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     out = diff_tree(context, *args)[STDOUT]
  73     return _parse_diff_filenames(out)
  74
  75
  76 def changed_files(context, oid):
  77     """Return the list of filenames that changed in a given commit oid"""
  78     status, out, _ = diff_tree(context, oid + '~', oid)
  79     if status != 0:
  80         # git init
  81         status, out, _ = diff_tree(context, EMPTY_TREE_OID, oid)
  82     if status == 0:
  83         result = _parse_diff_filenames(out)
  84     else:
  85         result = []
  86     return result
  87
  88
  89 def diff_tree(context, *args):
  90     """Return a list of filenames that have been modified"""
  91     git = context.git
  92     return git_diff_tree(git, *args)
  93
  94
  95 def git_diff_tree(git, *args):
  96     return git.diff_tree(
  97         name_only=True, no_commit_id=True, r=True, z=True, _readonly=True, *args
  98     )
  99
 100
 101 def listdir(context, dirname, ref='HEAD'):
 102     """Get the contents of a directory according to Git
 103
 104     Query Git for the content of a directory, taking ignored
 105     files into account.
 106
 107     """
 108     dirs = []
 109     files = []
 110
 111     # first, parse git ls-tree to get the tracked files
 112     # in a list of (type, path) tuples
 113     entries = ls_tree(context, dirname, ref=ref)
 114     for entry in entries:
 115         if entry[0][0] == 't':  # tree
 116             dirs.append(entry[1])
 117         else:
 118             files.append(entry[1])
 119
 120     # gather untracked files
 121     untracked = untracked_files(context, paths=[dirname], directory=True)
 122     for path in untracked:
 123         if path.endswith('/'):
 124             dirs.append(path[:-1])
 125         else:
 126             files.append(path)
 127
 128     dirs.sort()
 129     files.sort()
 130
 131     return (dirs, files)
 132
 133
 134 def diff(context, args):
 135     """Return a list of filenames for the given diff arguments
 136
 137     :param args: list of arguments to pass to "git diff --name-only"
 138
 139     """
 140     git = context.git
 141     out = git.diff(name_only=True, z=True, _readonly=True, *args)[STDOUT]
 142     return _parse_diff_filenames(out)
 143
 144
 145 def _parse_diff_filenames(out):
 146     if out:
 147         return out[:-1].split('\0')
 148     return []
 149
 150
 151 def tracked_files(context, *args):
 152     """Return the names of all files in the repository"""
 153     git = context.git
 154     out = git.ls_files('--', *args, z=True, _readonly=True)[STDOUT]
 155     if out:
 156         return sorted(out[:-1].split('\0'))
 157     return []
 158
 159
 160 def all_files(context, *args):
 161     """Returns a sorted list of all files, including untracked files."""
 162     git = context.git
 163     ls_files = git.ls_files(
 164         '--',
 165         *args,
 166         z=True,
 167         cached=True,
 168         others=True,
 169         exclude_standard=True,
 170         _readonly=True,
 171     )[STDOUT]
 172     return sorted([f for f in ls_files.split('\0') if f])
 173
 174
 175 class CurrentBranchCache:
 176     """Cache for current_branch()"""
 177
 178     key = None
 179     value = None
 180
 181
 182 def reset():
 183     """Reset cached value in this module (e.g. the cached current branch)"""
 184     CurrentBranchCache.key = None
 185
 186
 187 def current_branch(context):
 188     """Return the current branch"""
 189     git = context.git
 190     head = git.git_path('HEAD')
 191     try:
 192         key = core.stat(head).st_mtime
 193         if CurrentBranchCache.key == key:
 194             return CurrentBranchCache.value
 195     except OSError:
 196         # OSError means we can't use the stat cache
 197         key = 0
 198
 199     status, data, _ = git.rev_parse('HEAD', symbolic_full_name=True, _readonly=True)
 200     if status != 0:
 201         # git init -- read .git/HEAD.  We could do this unconditionally...
 202         data = _read_git_head(context, head)
 203
 204     for refs_prefix in ('refs/heads/', 'refs/remotes/', 'refs/tags/'):
 205         if data.startswith(refs_prefix):
 206             value = data[len(refs_prefix) :]
 207             CurrentBranchCache.key = key
 208             CurrentBranchCache.value = value
 209             return value
 210     # Detached head
 211     return data
 212
 213
 214 def _read_git_head(context, head, default='main'):
 215     """Pure-python .git/HEAD reader"""
 216     # Common .git/HEAD "ref: refs/heads/main" files
 217     git = context.git
 218     islink = core.islink(head)
 219     if core.isfile(head) and not islink:
 220         data = core.read(head).rstrip()
 221         ref_prefix = 'ref: '
 222         if data.startswith(ref_prefix):
 223             return data[len(ref_prefix) :]
 224         # Detached head
 225         return data
 226     # Legacy .git/HEAD symlinks
 227     if islink:
 228         refs_heads = core.realpath(git.git_path('refs', 'heads'))
 229         path = core.abspath(head).replace('\\', '/')
 230         if path.startswith(refs_heads + '/'):
 231             return path[len(refs_heads) + 1 :]
 232
 233     return default
 234
 235
 236 def branch_list(context, remote=False):
 237     """
 238     Return a list of local or remote branches
 239
 240     This explicitly removes HEAD from the list of remote branches.
 241
 242     """
 243     if remote:
 244         return for_each_ref_basename(context, 'refs/remotes')
 245     return for_each_ref_basename(context, 'refs/heads')
 246
 247
 248 def _version_sort(context, key='version:refname'):
 249     if version.check_git(context, 'version-sort'):
 250         sort = key
 251     else:
 252         sort = False
 253     return sort
 254
 255
 256 def for_each_ref_basename(context, refs):
 257     """Return refs starting with 'refs'."""
 258     git = context.git
 259     sort = _version_sort(context)
 260     _, out, _ = git.for_each_ref(refs, format='%(refname)', sort=sort, _readonly=True)
 261     output = out.splitlines()
 262     non_heads = [x for x in output if not x.endswith('/HEAD')]
 263     offset = len(refs) + 1
 264     return [x[offset:] for x in non_heads]
 265
 266
 267 def _prefix_and_size(prefix, values):
 268     """Return a tuple of (prefix, len(prefix) + 1, y) for <prefix>/ stripping"""
 269     return (prefix, len(prefix) + 1, values)
 270
 271
 272 def all_refs(context, split=False, sort_key='version:refname'):
 273     """Return a tuple of (local branches, remote branches, tags)."""
 274     git = context.git
 275     local_branches = []
 276     remote_branches = []
 277     tags = []
 278     query = (
 279         _prefix_and_size('refs/tags', tags),
 280         _prefix_and_size('refs/heads', local_branches),
 281         _prefix_and_size('refs/remotes', remote_branches),
 282     )
 283     sort = _version_sort(context, key=sort_key)
 284     _, out, _ = git.for_each_ref(format='%(refname)', sort=sort, _readonly=True)
 285     for ref in out.splitlines():
 286         for prefix, prefix_len, dst in query:
 287             if ref.startswith(prefix) and not ref.endswith('/HEAD'):
 288                 dst.append(ref[prefix_len:])
 289                 continue
 290     tags.reverse()
 291     if split:
 292         return local_branches, remote_branches, tags
 293     return local_branches + remote_branches + tags
 294
 295
 296 def tracked_branch(context, branch=None):
 297     """Return the remote branch associated with 'branch'."""
 298     if branch is None:
 299         branch = current_branch(context)
 300     if branch is None:
 301         return None
 302     config = context.cfg
 303     remote = config.get('branch.%s.remote' % branch)
 304     if not remote:
 305         return None
 306     merge_ref = config.get('branch.%s.merge' % branch)
 307     if not merge_ref:
 308         return None
 309     refs_heads = 'refs/heads/'
 310     if merge_ref.startswith(refs_heads):
 311         return remote + '/' + merge_ref[len(refs_heads) :]
 312     return None
 313
 314
 315 def parse_remote_branch(branch):
 316     """Split a remote branch apart into (remote, name) components"""
 317     rgx = re.compile(r'^(?P<remote>[^/]+)/(?P<branch>.+)$')
 318     match = rgx.match(branch)
 319     remote = ''
 320     branch = ''
 321     if match:
 322         remote = match.group('remote')
 323         branch = match.group('branch')
 324     return (remote, branch)
 325
 326
 327 def untracked_files(context, paths=None, **kwargs):
 328     """Returns a sorted list of untracked files."""
 329     git = context.git
 330     if paths is None:
 331         paths = []
 332     args = ['--'] + paths
 333     out = git.ls_files(
 334         z=True, others=True, exclude_standard=True, _readonly=True, *args, **kwargs
 335     )[STDOUT]
 336     if out:
 337         return out[:-1].split('\0')
 338     return []
 339
 340
 341 def tag_list(context):
 342     """Return a list of tags."""
 343     result = for_each_ref_basename(context, 'refs/tags')
 344     result.reverse()
 345     return result
 346
 347
 348 def log(git, *args, **kwargs):
 349     return git.log(
 350         no_color=True,
 351         no_abbrev_commit=True,
 352         no_ext_diff=True,
 353         _readonly=True,
 354         *args,
 355         **kwargs,
 356     )[STDOUT]
 357
 358
 359 def commit_diff(context, oid):
 360     git = context.git
 361     return log(git, '-1', oid, '--') + '\n\n' + oid_diff(context, oid)
 362
 363
 364 _diff_overrides = {}
 365
 366
 367 def update_diff_overrides(space_at_eol, space_change, all_space, function_context):
 368     _diff_overrides['ignore_space_at_eol'] = space_at_eol
 369     _diff_overrides['ignore_space_change'] = space_change
 370     _diff_overrides['ignore_all_space'] = all_space
 371     _diff_overrides['function_context'] = function_context
 372
 373
 374 def common_diff_opts(context):
 375     config = context.cfg
 376     # Default to --patience when diff.algorithm is unset
 377     patience = not config.get('diff.algorithm', default='')
 378     submodule = version.check_git(context, 'diff-submodule')
 379     opts = {
 380         'patience': patience,
 381         'submodule': submodule,
 382         'no_color': True,
 383         'no_ext_diff': True,
 384         'unified': config.get('gui.diffcontext', default=3),
 385         '_raw': True,
 386         '_readonly': True,
 387     }
 388     opts.update(_diff_overrides)
 389     return opts
 390
 391
 392 def _add_filename(args, filename):
 393     if filename:
 394         args.extend(['--', filename])
 395
 396
 397 def oid_diff(context, oid, filename=None):
 398     """Return the diff for an oid"""
 399     # Naively "$oid^!" is what we'd like to use but that doesn't
 400     # give the correct result for merges--the diff is reversed.
 401     # Be explicit and compare oid against its first parent.
 402     return oid_diff_range(context, oid + '~', oid, filename=filename)
 403
 404
 405 def oid_diff_range(context, start, end, filename=None):
 406     """Return the diff for a commit range"""
 407     args = [start, end]
 408     git = context.git
 409     opts = common_diff_opts(context)
 410     _add_filename(args, filename)
 411     status, out, _ = git.diff(*args, **opts)
 412     if status != 0:
 413         # We probably don't have "$oid~" because this is the root commit.
 414         # "git show" is clever enough to handle the root commit.
 415         args = [end + '^!']
 416         _add_filename(args, filename)
 417         _, out, _ = git.show(pretty='format:', *args, **opts)
 418         out = out.lstrip()
 419     return out
 420
 421
 422 def diff_info(context, oid, filename=None):
 423     """Return the diff for the specified oid"""
 424     return diff_range(context, oid + '~', oid, filename=filename)
 425
 426
 427 def diff_range(context, start, end, filename=None):
 428     """Return the diff for the specified commit range"""
 429     git = context.git
 430     decoded = log(git, '-1', end, '--', pretty='format:%b').strip()
 431     if decoded:
 432         decoded += '\n\n'
 433     return decoded + oid_diff_range(context, start, end, filename=filename)
 434
 435
 436 def diff_helper(
 437     context,
 438     commit=None,
 439     ref=None,
 440     endref=None,
 441     filename=None,
 442     cached=True,
 443     deleted=False,
 444     head=None,
 445     amending=False,
 446     with_diff_header=False,
 447     suppress_header=True,
 448     reverse=False,
 449     untracked=False,
 450 ):
 451     """Invoke git diff on a path"""
 452     git = context.git
 453     cfg = context.cfg
 454     if commit:
 455         ref, endref = commit + '^', commit
 456     argv = []
 457     if ref and endref:
 458         argv.append(f'{ref}..{endref}')
 459     elif ref:
 460         argv.extend(utils.shell_split(ref.strip()))
 461     elif head and amending and cached:
 462         argv.append(head)
 463
 464     encoding = None
 465     if untracked:
 466         argv.append('--no-index')
 467         argv.append(os.devnull)
 468         argv.append(filename)
 469     elif filename:
 470         argv.append('--')
 471         if isinstance(filename, (list, tuple)):
 472             argv.extend(filename)
 473         else:
 474             argv.append(filename)
 475             encoding = cfg.file_encoding(filename)
 476
 477     status, out, _ = git.diff(
 478         R=reverse,
 479         M=True,
 480         cached=cached,
 481         _encoding=encoding,
 482         *argv,
 483         **common_diff_opts(context),
 484     )
 485
 486     success = status == 0
 487
 488     # Diff will return 1 when comparing untracked file and it has change,
 489     # therefore we will check for diff header from output to differentiate
 490     # from actual error such as file not found.
 491     if untracked and status == 1:
 492         try:
 493             _, second, _ = out.split('\n', 2)
 494         except ValueError:
 495             second = ''
 496         success = second.startswith('new file mode ')
 497
 498     if not success:
 499         # git init
 500         if with_diff_header:
 501             return ('', '')
 502         return ''
 503
 504     result = extract_diff_header(deleted, with_diff_header, suppress_header, out)
 505     return core.UStr(result, out.encoding)
 506
 507
 508 def extract_diff_header(deleted, with_diff_header, suppress_header, diffoutput):
 509     """Split a diff into a header section and payload section"""
 510
 511     if diffoutput.startswith('Submodule'):
 512         if with_diff_header:
 513             return ('', diffoutput)
 514         return diffoutput
 515
 516     start = False
 517     del_tag = 'deleted file mode '
 518
 519     output = StringIO()
 520     headers = StringIO()
 521
 522     for line in diffoutput.split('\n'):
 523         if not start and line[:2] == '@@' and '@@' in line[2:]:
 524             start = True
 525         if start or (deleted and del_tag in line):
 526             output.write(line + '\n')
 527         else:
 528             if with_diff_header:
 529                 headers.write(line + '\n')
 530             elif not suppress_header:
 531                 output.write(line + '\n')
 532
 533     output_text = output.getvalue()
 534     output.close()
 535
 536     headers_text = headers.getvalue()
 537     headers.close()
 538
 539     if with_diff_header:
 540         return (headers_text, output_text)
 541     return output_text
 542
 543
 544 def format_patchsets(context, to_export, revs, output='patches'):
 545     """
 546     Group contiguous revision selection into patch sets
 547
 548     Exists to handle multi-selection.
 549     Multiple disparate ranges in the revision selection
 550     are grouped into continuous lists.
 551
 552     """
 553
 554     outs = []
 555     errs = []
 556
 557     cur_rev = to_export[0]
 558     cur_rev_idx = revs.index(cur_rev)
 559
 560     patches_to_export = [[cur_rev]]
 561     patchset_idx = 0
 562
 563     # Group the patches into continuous sets
 564     for rev in to_export[1:]:
 565         # Limit the search to the current neighborhood for efficiency
 566         try:
 567             rev_idx = revs[cur_rev_idx:].index(rev)
 568             rev_idx += cur_rev_idx
 569         except ValueError:
 570             rev_idx = revs.index(rev)
 571
 572         if rev_idx == cur_rev_idx + 1:
 573             patches_to_export[patchset_idx].append(rev)
 574             cur_rev_idx += 1
 575         else:
 576             patches_to_export.append([rev])
 577             cur_rev_idx = rev_idx
 578             patchset_idx += 1
 579
 580     # Export each patch set
 581     status = 0
 582     for patchset in patches_to_export:
 583         stat, out, err = export_patchset(
 584             context,
 585             patchset[0],
 586             patchset[-1],
 587             output=output,
 588             n=len(patchset) > 1,
 589             thread=True,
 590             patch_with_stat=True,
 591         )
 592         outs.append(out)
 593         if err:
 594             errs.append(err)
 595         status = max(stat, status)
 596     return (status, '\n'.join(outs), '\n'.join(errs))
 597
 598
 599 def export_patchset(context, start, end, output='patches', **kwargs):
 600     """Export patches from start^ to end."""
 601     git = context.git
 602     return git.format_patch('-o', output, start + '^..' + end, **kwargs)
 603
 604
 605 def reset_paths(context, head, items):
 606     """Run "git reset" while preventing argument overflow"""
 607     items = list(set(items))
 608     func = context.git.reset
 609     status, out, err = utils.slice_func(items, lambda paths: func(head, '--', *paths))
 610     return (status, out, err)
 611
 612
 613 def unstage_paths(context, args, head='HEAD'):
 614     """Unstage paths while accounting for git init"""
 615     status, out, err = reset_paths(context, head, args)
 616     if status == 128:
 617         # handle git init: we have to use 'git rm --cached'
 618         # detect this condition by checking if the file is still staged
 619         return untrack_paths(context, args)
 620     return (status, out, err)
 621
 622
 623 def untrack_paths(context, args):
 624     if not args:
 625         return (-1, N_('Nothing to do'), '')
 626     git = context.git
 627     return git.update_index('--', force_remove=True, *set(args))
 628
 629
 630 def worktree_state(
 631     context, head='HEAD', update_index=False, display_untracked=True, paths=None
 632 ):
 633     """Return a dict of files in various states of being
 634
 635     :rtype: dict, keys are staged, unstaged, untracked, unmerged,
 636             changed_upstream, and submodule.
 637     """
 638     git = context.git
 639     if update_index:
 640         git.update_index(refresh=True)
 641
 642     staged, unmerged, staged_deleted, staged_submods = diff_index(
 643         context, head, paths=paths
 644     )
 645     modified, unstaged_deleted, modified_submods = diff_worktree(context, paths)
 646     if display_untracked:
 647         untracked = untracked_files(context, paths=paths)
 648     else:
 649         untracked = []
 650
 651     # Remove unmerged paths from the modified list
 652     if unmerged:
 653         unmerged_set = set(unmerged)
 654         modified = [path for path in modified if path not in unmerged_set]
 655
 656     # Look for upstream modified files if this is a tracking branch
 657     upstream_changed = diff_upstream(context, head)
 658
 659     # Keep stuff sorted
 660     staged.sort()
 661     modified.sort()
 662     unmerged.sort()
 663     untracked.sort()
 664     upstream_changed.sort()
 665
 666     return {
 667         'staged': staged,
 668         'modified': modified,
 669         'unmerged': unmerged,
 670         'untracked': untracked,
 671         'upstream_changed': upstream_changed,
 672         'staged_deleted': staged_deleted,
 673         'unstaged_deleted': unstaged_deleted,
 674         'submodules': staged_submods | modified_submods,
 675     }
 676
 677
 678 def _parse_raw_diff(out):
 679     while out:
 680         info, path, out = out.split('\0', 2)
 681         status = info[-1]
 682         is_submodule = '160000' in info[1:14]
 683         yield (path, status, is_submodule)
 684
 685
 686 def diff_index(context, head, cached=True, paths=None):
 687     git = context.git
 688     staged = []
 689     unmerged = []
 690     deleted = set()
 691     submodules = set()
 692
 693     if paths is None:
 694         paths = []
 695     args = [head, '--'] + paths
 696     status, out, _ = git.diff_index(cached=cached, z=True, _readonly=True, *args)
 697     if status != 0:
 698         # handle git init
 699         args[0] = EMPTY_TREE_OID
 700         status, out, _ = git.diff_index(cached=cached, z=True, _readonly=True, *args)
 701
 702     for path, status, is_submodule in _parse_raw_diff(out):
 703         if is_submodule:
 704             submodules.add(path)
 705         if status in 'DAMT':
 706             staged.append(path)
 707             if status == 'D':
 708                 deleted.add(path)
 709         elif status == 'U':
 710             unmerged.append(path)
 711
 712     return staged, unmerged, deleted, submodules
 713
 714
 715 def diff_worktree(context, paths=None):
 716     git = context.git
 717     ignore_submodules_value = context.cfg.get('diff.ignoresubmodules', 'none')
 718     ignore_submodules = ignore_submodules_value in {'all', 'dirty', 'untracked'}
 719     modified = []
 720     deleted = set()
 721     submodules = set()
 722
 723     if paths is None:
 724         paths = []
 725     args = ['--'] + paths
 726     status, out, _ = git.diff_files(z=True, _readonly=True, *args)
 727     for path, status, is_submodule in _parse_raw_diff(out):
 728         if is_submodule:
 729             submodules.add(path)
 730             if ignore_submodules:
 731                 continue
 732         if status in 'DAMT':
 733             modified.append(path)
 734             if status == 'D':
 735                 deleted.add(path)
 736
 737     return modified, deleted, submodules
 738
 739
 740 def diff_upstream(context, head):
 741     """Given `ref`, return $(git merge-base ref HEAD)..ref."""
 742     tracked = tracked_branch(context)
 743     if not tracked:
 744         return []
 745     base = merge_base(context, head, tracked)
 746     return diff_filenames(context, base, tracked)
 747
 748
 749 def list_submodule(context):
 750     """Return submodules in the format(state, sha_1, path, describe)"""
 751     git = context.git
 752     status, data, _ = git.submodule('status')
 753     ret = []
 754     if status == 0 and data:
 755         data = data.splitlines()
 756         # see git submodule status
 757         for line in data:
 758             state = line[0].strip()
 759             sha1 = line[1 : OID_LENGTH + 1]
 760             left_bracket = line.find('(', OID_LENGTH + 3)
 761             if left_bracket == -1:
 762                 left_bracket = len(line) + 1
 763             path = line[OID_LENGTH + 2 : left_bracket - 1]
 764             describe = line[left_bracket + 1 : -1]
 765             ret.append((state, sha1, path, describe))
 766     return ret
 767
 768
 769 def merge_base(context, head, ref):
 770     """Return the merge-base of head and ref"""
 771     git = context.git
 772     return git.merge_base(head, ref, _readonly=True)[STDOUT]
 773
 774
 775 def merge_base_parent(context, branch):
 776     tracked = tracked_branch(context, branch=branch)
 777     if tracked:
 778         return tracked
 779     return 'HEAD'
 780
 781
 782 def ls_tree(context, path, ref='HEAD'):
 783     """Return a parsed git ls-tree result for a single directory"""
 784     git = context.git
 785     result = []
 786     status, out, _ = git.ls_tree(
 787         ref, '--', path, z=True, full_tree=True, _readonly=True
 788     )
 789     if status == 0 and out:
 790         path_offset = 6 + 1 + 4 + 1 + OID_LENGTH + 1
 791         for line in out[:-1].split('\0'):
 792             #       1    1                                        1
 793             # .....6 ...4 ......................................40
 794             # 040000 tree c127cde9a0c644a3a8fef449a244f47d5272dfa6      relative
 795             # 100644 blob 139e42bf4acaa4927ec9be1ec55a252b97d3f1e2      relative/path
 796             # 0..... 7... 12......................................      53
 797             # path_offset = 6 + 1 + 4 + 1 + OID_LENGTH(40) + 1
 798             objtype = line[7:11]
 799             relpath = line[path_offset:]
 800             result.append((objtype, relpath))
 801
 802     return result
 803
 804
 805 # A regex for matching the output of git(log|rev-list) --pretty=oneline
 806 REV_LIST_REGEX = re.compile(r'^([0-9a-f]{40}) (.*)$')
 807
 808
 809 def parse_rev_list(raw_revs):
 810     """Parse `git log --pretty=online` output into (oid, summary) pairs."""
 811     revs = []
 812     for line in raw_revs.splitlines():
 813         match = REV_LIST_REGEX.match(line)
 814         if match:
 815             rev_id = match.group(1)
 816             summary = match.group(2)
 817             revs.append((
 818                 rev_id,
 819                 summary,
 820             ))
 821     return revs
 822
 823
 824 def log_helper(context, all=False, extra_args=None):
 825     """Return parallel arrays containing oids and summaries."""
 826     revs = []
 827     summaries = []
 828     args = []
 829     if extra_args:
 830         args = extra_args
 831     git = context.git
 832     output = log(git, pretty='oneline', all=all, *args)
 833     for line in output.splitlines():
 834         match = REV_LIST_REGEX.match(line)
 835         if match:
 836             revs.append(match.group(1))
 837             summaries.append(match.group(2))
 838     return (revs, summaries)
 839
 840
 841 def rev_list_range(context, start, end):
 842     """Return (oid, summary) pairs between start and end."""
 843     git = context.git
 844     revrange = f'{start}..{end}'
 845     out = git.rev_list(revrange, pretty='oneline', _readonly=True)[STDOUT]
 846     return parse_rev_list(out)
 847
 848
 849 def commit_message_path(context):
 850     """Return the path to .git/GIT_COLA_MSG"""
 851     git = context.git
 852     path = git.git_path('GIT_COLA_MSG')
 853     if core.exists(path):
 854         return path
 855     return None
 856
 857
 858 def merge_message_path(context):
 859     """Return the path to .git/MERGE_MSG or .git/SQUASH_MSG."""
 860     git = context.git
 861     for basename in ('MERGE_MSG', 'SQUASH_MSG'):
 862         path = git.git_path(basename)
 863         if core.exists(path):
 864             return path
 865     return None
 866
 867
 868 def read_merge_commit_message(context, path):
 869     """Read a merge commit message from disk while stripping commentary"""
 870     content = core.read(path)
 871     cleanup_mode = prefs.commit_cleanup(context)
 872     if cleanup_mode in ('verbatim', 'scissors', 'whitespace'):
 873         return content
 874     comment_char = prefs.comment_char(context)
 875     return '\n'.join(
 876         line for line in content.splitlines() if not line.startswith(comment_char)
 877     )
 878
 879
 880 def prepare_commit_message_hook(context):
 881     """Run the cola.preparecommitmessagehook to prepare the commit message"""
 882     config = context.cfg
 883     default_hook = config.hooks_path('cola-prepare-commit-msg')
 884     return config.get('cola.preparecommitmessagehook', default=default_hook)
 885
 886
 887 def cherry_pick(context, revs):
 888     """Cherry-picks each revision into the current branch.
 889
 890     Returns (0, out, err) where stdout and stderr across all "git cherry-pick"
 891     invocations are combined into single values when all cherry-picks succeed.
 892
 893     Returns a combined (status, out, err) of the first failing "git cherry-pick"
 894     in the event of a non-zero exit status.
 895     """
 896     if not revs:
 897         return []
 898     outs = []
 899     errs = []
 900     status = 0
 901     for rev in revs:
 902         status, out, err = context.git.cherry_pick(rev)
 903         if status != 0:
 904             details = N_(
 905                 'Hint: The "Actions > Abort Cherry-Pick" menu action can be used to '
 906                 'cancel the current cherry-pick.'
 907             )
 908             output = f'# git cherry-pick {rev}\n# {details}\n\n{out}'
 909             return (status, output, err)
 910         outs.append(out)
 911         errs.append(err)
 912     return (0, '\n'.join(outs), '\n'.join(errs))
 913
 914
 915 def abort_apply_patch(context):
 916     """Abort a "git am" session."""
 917     # Reset the worktree
 918     git = context.git
 919     status, out, err = git.am(abort=True)
 920     return status, out, err
 921
 922
 923 def abort_cherry_pick(context):
 924     """Abort a cherry-pick."""
 925     # Reset the worktree
 926     git = context.git
 927     status, out, err = git.cherry_pick(abort=True)
 928     return status, out, err
 929
 930
 931 def abort_merge(context):
 932     """Abort a merge"""
 933     # Reset the worktree
 934     git = context.git
 935     status, out, err = git.merge(abort=True)
 936     return status, out, err
 937
 938
 939 def strip_remote(remotes, remote_branch):
 940     """Get branch names with the "<remote>/" prefix removed"""
 941     for remote in remotes:
 942         prefix = remote + '/'
 943         if remote_branch.startswith(prefix):
 944             return remote_branch[len(prefix) :]
 945     return remote_branch.split('/', 1)[-1]
 946
 947
 948 def parse_refs(context, argv):
 949     """Parse command-line arguments into object IDs"""
 950     git = context.git
 951     status, out, _ = git.rev_parse(_readonly=True, *argv)
 952     if status == 0:
 953         oids = [oid for oid in out.splitlines() if oid]
 954     else:
 955         oids = argv
 956     return oids
 957
 958
 959 def prev_commitmsg(context, *args):
 960     """Queries git for the latest commit message."""
 961     git = context.git
 962     return git.log(
 963         '-1', no_color=True, pretty='format:%s%n%n%b', _readonly=True, *args
 964     )[STDOUT]
 965
 966
 967 def rev_parse(context, name):
 968     """Call git rev-parse and return the output"""
 969     git = context.git
 970     status, out, _ = git.rev_parse(name, _readonly=True)
 971     if status == 0:
 972         result = out.strip()
 973     else:
 974         result = name
 975     return result
 976
 977
 978 def write_blob(context, oid, filename):
 979     """Write a blob to a temporary file and return the path
 980
 981     Modern versions of Git allow invoking filters.  Older versions
 982     get the object content as-is.
 983
 984     """
 985     if version.check_git(context, 'cat-file-filters-path'):
 986         return cat_file_to_path(context, filename, oid)
 987     return cat_file_blob(context, filename, oid)
 988
 989
 990 def cat_file_blob(context, filename, oid):
 991     """Write a blob from git to the specified filename"""
 992     return cat_file(context, filename, 'blob', oid)
 993
 994
 995 def cat_file_to_path(context, filename, oid):
 996     """Extract a file from an commit ref and a write it to the specified filename"""
 997     return cat_file(context, filename, oid, path=filename, filters=True)
 998
 999
1000 def cat_file(context, filename, *args, **kwargs):
1001     """Redirect git cat-file output to a path"""
1002     result = None
1003     git = context.git
1004     # Use the original filename in the suffix so that the generated filename
1005     # has the correct extension, and so that it resembles the original name.
1006     basename = os.path.basename(filename)
1007     suffix = '-' + basename  # ensures the correct filename extension
1008     path = utils.tmp_filename('blob', suffix=suffix)
1009     with open(path, 'wb') as tmp_file:
1010         status, out, err = git.cat_file(
1011             _raw=True, _readonly=True, _stdout=tmp_file, *args, **kwargs
1012         )
1013         Interaction.command(N_('Error'), 'git cat-file', status, out, err)
1014         if status == 0:
1015             result = path
1016     if not result:
1017         core.unlink(path)
1018     return result
1019
1020
1021 def write_blob_path(context, head, oid, filename):
1022     """Use write_blob() when modern git is available"""
1023     if version.check_git(context, 'cat-file-filters-path'):
1024         return write_blob(context, oid, filename)
1025     return cat_file_blob(context, filename, head + ':' + filename)
1026
1027
1028 def annex_path(context, head, filename):
1029     """Return the git-annex path for a filename at the specified commit"""
1030     git = context.git
1031     path = None
1032     annex_info = {}
1033
1034     # unfortunately there's no way to filter this down to a single path
1035     # so we just have to scan all reported paths
1036     status, out, _ = git.annex('findref', '--json', head, _readonly=True)
1037     if status == 0:
1038         for line in out.splitlines():
1039             info = json.loads(line)
1040             try:
1041                 annex_file = info['file']
1042             except (ValueError, KeyError):
1043                 continue
1044             # we only care about this file so we can skip the rest
1045             if annex_file == filename:
1046                 annex_info = info
1047                 break
1048     key = annex_info.get('key', '')
1049     if key:
1050         status, out, _ = git.annex('contentlocation', key, _readonly=True)
1051         if status == 0 and os.path.exists(out):
1052             path = out
1053
1054     return path
1055
1056
1057 def is_binary(context, filename):
1058     """A heuristic to determine whether `filename` contains (non-text) binary content"""
1059     cfg_is_binary = context.cfg.is_binary(filename)
1060     if cfg_is_binary is not None:
1061         return cfg_is_binary
1062     # This is the same heuristic as xdiff-interface.c:buffer_is_binary().
1063     size = 8000
1064     try:
1065         result = core.read(filename, size=size, encoding='bytes')
1066     except OSError:
1067         result = b''
1068
1069     return b'\0' in result
1070
1071
1072 def is_valid_ref(context, ref):
1073     """Is the provided Git ref a valid refname?"""
1074     status, _, _ = context.git.rev_parse(ref, quiet=True, verify=True, _readonly=True)
1075     return status == 0