repository.rb 32.1 KB
Newer Older
1 2
require 'securerandom'

3
class Repository
4
  include Gitlab::ShellAdapter
5
  include RepositoryMirroring
6

7
  attr_accessor :full_path, :disk_path, :project
8

9 10
  delegate :ref_name_for_sha, to: :raw_repository

11
  CommitError = Class.new(StandardError)
12
  CreateTreeError = Class.new(StandardError)
13

14 15 16 17 18 19
  # Methods that cache data from the Git repository.
  #
  # Each entry in this Array should have a corresponding method with the exact
  # same name. The cache key used by those methods must also match method's
  # name.
  #
20 21 22
  # For example, for entry `:commit_count` there's a method called `commit_count` which
  # stores its data in the `commit_count` cache key.
  CACHED_METHODS = %i(size commit_count rendered_readme contribution_guide
23 24
                      changelog license_blob license_key gitignore koding_yml
                      gitlab_ci_yml branch_names tag_names branch_count
Douwe Maan's avatar
Douwe Maan committed
25
                      tag_count avatar exists? empty? root_ref).freeze
26 27 28 29 30

  # Certain method caches should be refreshed when certain types of files are
  # changed. This Hash maps file types (as returned by Gitlab::FileDetector) to
  # the corresponding methods to call for refreshing caches.
  METHOD_CACHES_FOR_FILE_TYPES = {
31
    readme: :rendered_readme,
32
    changelog: :changelog,
33
    license: %i(license_blob license_key license),
34 35 36 37 38
    contributing: :contribution_guide,
    gitignore: :gitignore,
    koding: :koding_yml,
    gitlab_ci: :gitlab_ci_yml,
    avatar: :avatar
Douwe Maan's avatar
Douwe Maan committed
39
  }.freeze
40 41 42 43 44

  # Wraps around the given method and caches its output in Redis and an instance
  # variable.
  #
  # This only works for methods that do not take any arguments.
45
  def self.cache_method(name, fallback: nil, memoize_only: false)
46
    original = :"_uncached_#{name}"
47

48
    alias_method(original, name)
49

50
    define_method(name) do
51 52 53
      cache_method_output(name, fallback: fallback, memoize_only: memoize_only) do
        __send__(original) # rubocop:disable GitlabSecurity/PublicSend
      end
54
    end
55
  end
56

57
  def initialize(full_path, project, disk_path: nil)
58
    @full_path = full_path
59
    @disk_path = disk_path || full_path
60
    @project = project
61
  end
62

63
  def ==(other)
64 65 66
    @disk_path == other.disk_path
  end

67
  def raw_repository
68
    return nil unless full_path
69

70
    @raw_repository ||= initialize_raw_repository
71 72
  end

73 74
  alias_method :raw, :raw_repository

75
  # Return absolute path to repository
76
  def path_to_repo
77
    @path_to_repo ||= File.expand_path(
78
      File.join(repository_storage_path, disk_path + '.git')
79
    )
80 81
  end

82 83 84 85
  def inspect
    "#<#{self.class.name}:#{@disk_path}>"
  end

86 87 88 89 90 91 92 93 94 95
  #
  # Git repository can contains some hidden refs like:
  #   /refs/notes/*
  #   /refs/git-as-svn/*
  #   /refs/pulls/*
  # This refs by default not visible in project page and not cloned to client side.
  #
  # This method return true if repository contains some content visible in project page.
  #
  def has_visible_content?
96
    branch_count > 0
97 98
  end

99
  def commit(ref = 'HEAD')
100
    return nil unless exists?
101

102 103 104 105 106 107
    commit =
      if ref.is_a?(Gitlab::Git::Commit)
        ref
      else
        Gitlab::Git::Commit.find(raw_repository, ref)
      end
108

109
    commit = ::Commit.new(commit, @project) if commit
110
    commit
111
  rescue Rugged::OdbError, Rugged::TreeError
112
    nil
113 114
  end

115
  def commits(ref, path: nil, limit: nil, offset: nil, skip_merges: false, after: nil, before: nil)
116
    options = {
117 118 119 120 121
      repo: raw_repository,
      ref: ref,
      path: path,
      limit: limit,
      offset: offset,
122 123
      after: after,
      before: before,
124
      follow: Array(path).length == 1,
125
      skip_merges: skip_merges
126 127 128
    }

    commits = Gitlab::Git::Commit.where(options)
129
    commits = Commit.decorate(commits, @project) if commits.present?
130 131 132
    commits
  end

133 134
  def commits_between(from, to)
    commits = Gitlab::Git::Commit.between(raw_repository, from, to)
135
    commits = Commit.decorate(commits, @project) if commits.present?
136 137 138
    commits
  end

Jacob Vosmaer's avatar
Jacob Vosmaer committed
139
  # Gitaly migration: https://gitlab.com/gitlab-org/gitaly/issues/384
140
  def find_commits_by_message(query, ref = nil, path = nil, limit = 1000, offset = 0)
141 142 143 144
    unless exists? && has_visible_content? && query.present?
      return []
    end

145 146 147 148 149 150 151
    raw_repository.gitaly_migrate(:commits_by_message) do |is_enabled|
      if is_enabled
        find_commits_by_message_by_gitaly(query, ref, path, limit, offset)
      else
        find_commits_by_message_by_shelling_out(query, ref, path, limit, offset)
      end
    end
152 153
  end

154 155 156 157 158 159 160
  def find_branch(name, fresh_repo: true)
    # Since the Repository object may have in-memory index changes, invalidating the memoized Repository object may
    # cause unintended side effects. Because finding a branch is a read-only operation, we can safely instantiate
    # a new repo here to ensure a consistent state to avoid a libgit2 bug where concurrent access (e.g. via git gc)
    # may cause the branch to "disappear" erroneously or have the wrong SHA.
    #
    # See: https://github.com/libgit2/libgit2/issues/1534 and https://gitlab.com/gitlab-org/gitlab-ce/issues/15392
161
    raw_repo = fresh_repo ? initialize_raw_repository : raw_repository
162 163

    raw_repo.find_branch(name)
164 165 166
  end

  def find_tag(name)
167
    tags.find { |tag| tag.name == name }
168 169
  end

170 171
  def add_branch(user, branch_name, ref)
    newrev = commit(ref).try(:sha)
172

173
    return false unless newrev
174

175
    GitOperationService.new(user, self).add_branch(branch_name, newrev)
176

177
    after_create_branch
178
    find_branch(branch_name)
179 180
  end

181
  def add_tag(user, tag_name, target, message = nil)
182
    newrev = commit(target).try(:id)
183 184
    options = { message: message, tagger: user_to_committer(user) } if message

185 186 187
    return false unless newrev

    GitOperationService.new(user, self).add_tag(tag_name, newrev, options)
188

189
    find_tag(tag_name)
190 191
  end

192
  def rm_branch(user, branch_name)
193
    before_remove_branch
194 195
    branch = find_branch(branch_name)

196
    GitOperationService.new(user, self).rm_branch(branch)
197

198
    after_remove_branch
199
    true
200 201
  end

Lin Jen-Shin's avatar
Lin Jen-Shin committed
202
  def rm_tag(user, tag_name)
203
    before_remove_tag
Lin Jen-Shin's avatar
Lin Jen-Shin committed
204
    tag = find_tag(tag_name)
205

Lin Jen-Shin's avatar
Lin Jen-Shin committed
206 207 208 209
    GitOperationService.new(user, self).rm_tag(tag)

    after_remove_tag
    true
210 211
  end

212 213 214 215
  def ref_names
    branch_names + tag_names
  end

216
  def branch_exists?(branch_name)
217 218 219 220 221 222 223
    return false unless raw_repository

    @branch_exists_memo ||= Hash.new do |hash, key|
      hash[key] = raw_repository.branch_exists?(key)
    end

    @branch_exists_memo[branch_name]
224 225
  end

226
  def ref_exists?(ref)
227 228
    !!raw_repository&.ref_exists?(ref)
  rescue ArgumentError
229
    false
230 231
  end

232 233 234 235
  # Makes sure a commit is kept around when Git garbage collection runs.
  # Git GC will delete commits from the repository that are no longer in any
  # branches or tags, but we want to keep some of these commits around, for
  # example if they have comments or CI builds.
236 237 238 239 240
  def keep_around(sha)
    return unless sha && commit(sha)

    return if kept_around?(sha)

241 242
    # This will still fail if the file is corrupted (e.g. 0 bytes)
    begin
243
      write_ref(keep_around_ref_name(sha), sha)
244 245
    rescue Rugged::ReferenceError => ex
      Rails.logger.error "Unable to create keep-around reference for repository #{path}: #{ex}"
246 247 248
    rescue Rugged::OSError => ex
      raise unless ex.message =~ /Failed to create locked file/ && ex.message =~ /File exists/
      Rails.logger.error "Unable to create keep-around reference for repository #{path}: #{ex}"
249
    end
250 251 252
  end

  def kept_around?(sha)
253
    ref_exists?(keep_around_ref_name(sha))
254
  end
255

256 257 258 259
  def write_ref(ref_path, sha)
    rugged.references.create(ref_path, sha, force: true)
  end

260
  def diverging_commit_counts(branch)
261
    root_ref_hash = raw_repository.rev_parse_target(root_ref).oid
Jeff Stubler's avatar
Jeff Stubler committed
262
    cache.fetch(:"diverging_commit_counts_#{branch.name}") do
263 264
      # Rugged seems to throw a `ReferenceError` when given branch_names rather
      # than SHA-1 hashes
265 266
      number_commits_behind = raw_repository
        .count_commits_between(branch.dereferenced_target.sha, root_ref_hash)
267

268 269
      number_commits_ahead = raw_repository
        .count_commits_between(root_ref_hash, branch.dereferenced_target.sha)
270

271 272 273
      { behind: number_commits_behind, ahead: number_commits_ahead }
    end
  end
274

275 276 277
  def expire_tags_cache
    expire_method_caches(%i(tag_names tag_count))
    @tags = nil
278
  end
279

280 281 282
  def expire_branches_cache
    expire_method_caches(%i(branch_names branch_count))
    @local_branches = nil
283
    @branch_exists_memo = nil
284 285
  end

286 287
  def expire_statistics_caches
    expire_method_caches(%i(size commit_count))
288 289
  end

290 291
  def expire_all_method_caches
    expire_method_caches(CACHED_METHODS)
292 293
  end

294 295 296 297 298 299 300 301 302
  # Expires the caches of a specific set of methods
  def expire_method_caches(methods)
    methods.each do |key|
      cache.expire(key)

      ivar = cache_instance_variable_name(key)

      remove_instance_variable(ivar) if instance_variable_defined?(ivar)
    end
303 304
  end

305 306 307 308 309 310 311 312 313 314 315 316 317 318 319
  def expire_avatar_cache
    expire_method_caches(%i(avatar))
  end

  # Refreshes the method caches of this repository.
  #
  # types - An Array of file types (e.g. `:readme`) used to refresh extra
  #         caches.
  def refresh_method_caches(types)
    to_refresh = []

    types.each do |type|
      methods = METHOD_CACHES_FOR_FILE_TYPES[type.to_sym]

      to_refresh.concat(Array(methods)) if methods
320
    end
321

322
    expire_method_caches(to_refresh)
323

324
    to_refresh.each { |method| send(method) } # rubocop:disable GitlabSecurity/PublicSend
325
  end
326

327 328 329 330 331 332 333
  def expire_branch_cache(branch_name = nil)
    # When we push to the root branch we have to flush the cache for all other
    # branches as their statistics are based on the commits relative to the
    # root branch.
    if !branch_name || branch_name == root_ref
      branches.each do |branch|
        cache.expire(:"diverging_commit_counts_#{branch.name}")
334
        cache.expire(:"commit_count_#{branch.name}")
335 336 337 338 339
      end
    # In case a commit is pushed to a non-root branch we only have to flush the
    # cache for said branch.
    else
      cache.expire(:"diverging_commit_counts_#{branch_name}")
340
      cache.expire(:"commit_count_#{branch_name}")
341
    end
Dmitriy Zaporozhets's avatar
Dmitriy Zaporozhets committed
342 343
  end

344
  def expire_root_ref_cache
345
    expire_method_caches(%i(root_ref))
346 347
  end

348 349
  # Expires the cache(s) used to determine if a repository is empty or not.
  def expire_emptiness_caches
350
    return unless empty?
351

352
    expire_method_caches(%i(empty?))
353 354
  end

355 356 357 358
  def lookup_cache
    @lookup_cache ||= {}
  end

359
  def expire_exists_cache
360
    expire_method_caches(%i(exists?))
361 362
  end

363 364 365 366 367 368 369
  # expire cache that doesn't depend on repository data (when expiring)
  def expire_content_cache
    expire_tags_cache
    expire_branches_cache
    expire_root_ref_cache
    expire_emptiness_caches
    expire_exists_cache
370
    expire_statistics_caches
371 372 373 374 375
  end

  # Runs code after a repository has been created.
  def after_create
    expire_exists_cache
376 377
    expire_root_ref_cache
    expire_emptiness_caches
Yorick Peterse's avatar
Yorick Peterse committed
378 379

    repository_event(:create_repository)
380 381
  end

382 383
  # Runs code just before a repository is deleted.
  def before_delete
384
    expire_exists_cache
385 386
    expire_all_method_caches
    expire_branch_cache if exists?
387
    expire_content_cache
Yorick Peterse's avatar
Yorick Peterse committed
388 389

    repository_event(:remove_repository)
390 391 392 393 394 395 396
  end

  # Runs code just before the HEAD of a repository is changed.
  def before_change_head
    # Cached divergent commit counts are based on repository head
    expire_branch_cache
    expire_root_ref_cache
Yorick Peterse's avatar
Yorick Peterse committed
397 398

    repository_event(:change_default_branch)
399 400
  end

401 402
  # Runs code before pushing (= creating or removing) a tag.
  def before_push_tag
403 404
    expire_statistics_caches
    expire_emptiness_caches
405
    expire_tags_cache
Yorick Peterse's avatar
Yorick Peterse committed
406 407

    repository_event(:push_tag)
408 409 410 411 412
  end

  # Runs code before removing a tag.
  def before_remove_tag
    expire_tags_cache
413
    expire_statistics_caches
Yorick Peterse's avatar
Yorick Peterse committed
414 415

    repository_event(:remove_tag)
416 417
  end

Lin Jen-Shin's avatar
Lin Jen-Shin committed
418 419 420 421 422
  # Runs code after removing a tag.
  def after_remove_tag
    expire_tags_cache
  end

423 424 425
  # Runs code after the HEAD of a repository is changed.
  def after_change_head
    expire_method_caches(METHOD_CACHES_FOR_FILE_TYPES.keys)
426 427
  end

428 429
  # Runs code after a repository has been forked/imported.
  def after_import
430
    expire_content_cache
431 432 433
  end

  # Runs code after a new commit has been pushed.
434 435 436
  def after_push_commit(branch_name)
    expire_statistics_caches
    expire_branch_cache(branch_name)
Yorick Peterse's avatar
Yorick Peterse committed
437 438

    repository_event(:push_commit, branch: branch_name)
439 440 441 442
  end

  # Runs code after a new branch has been created.
  def after_create_branch
443
    expire_branches_cache
Yorick Peterse's avatar
Yorick Peterse committed
444 445

    repository_event(:push_branch)
446 447
  end

448 449 450
  # Runs code before removing an existing branch.
  def before_remove_branch
    expire_branches_cache
Yorick Peterse's avatar
Yorick Peterse committed
451 452

    repository_event(:remove_branch)
453 454
  end

455 456
  # Runs code after an existing branch has been removed.
  def after_remove_branch
457
    expire_branches_cache
458 459
  end

460
  def method_missing(m, *args, &block)
461 462
    if m == :lookup && !block_given?
      lookup_cache[m] ||= {}
463
      lookup_cache[m][args.join(":")] ||= raw_repository.__send__(m, *args, &block) # rubocop:disable GitlabSecurity/PublicSend
464
    else
465
      raw_repository.__send__(m, *args, &block) # rubocop:disable GitlabSecurity/PublicSend
466
    end
467 468
  end

469 470
  def respond_to_missing?(method, include_private = false)
    raw_repository.respond_to?(method, include_private) || super
471
  end
Dmitriy Zaporozhets's avatar
Dmitriy Zaporozhets committed
472 473

  def blob_at(sha, path)
474
    unless Gitlab::Git.blank_ref?(sha)
Douwe Maan's avatar
Douwe Maan committed
475
      Blob.decorate(Gitlab::Git::Blob.find(self, sha, path), project)
476
    end
Douwe Maan's avatar
Douwe Maan committed
477 478
  rescue Gitlab::Git::Repository::NoRepository
    nil
Dmitriy Zaporozhets's avatar
Dmitriy Zaporozhets committed
479
  end
480

481 482 483 484 485 486 487
  def root_ref
    if raw_repository
      raw_repository.root_ref
    else
      # When the repo does not exist we raise this error so no data is cached.
      raise Rugged::ReferenceError
    end
488
  end
489
  cache_method :root_ref
490

491
  # Gitaly migration: https://gitlab.com/gitlab-org/gitaly/issues/314
492
  def exists?
493
    return false unless full_path
494 495 496 497 498 499 500 501

    Gitlab::GitalyClient.migrate(:repository_exists) do |enabled|
      if enabled
        raw_repository.exists?
      else
        refs_directory_exists?
      end
    end
502 503 504
  end
  cache_method :exists?

Douwe Maan's avatar
Douwe Maan committed
505
  delegate :empty?, to: :raw_repository
506 507 508 509 510 511 512 513 514 515 516 517 518
  cache_method :empty?

  # The size of this repository in megabytes.
  def size
    exists? ? raw_repository.size : 0.0
  end
  cache_method :size, fallback: 0.0

  def commit_count
    root_ref ? raw_repository.commit_count(root_ref) : 0
  end
  cache_method :commit_count, fallback: 0

519
  def commit_count_for_ref(ref)
520
    return 0 unless exists?
521

522 523 524 525
    begin
      cache.fetch(:"commit_count_#{ref}") { raw_repository.commit_count(ref) }
    rescue Rugged::ReferenceError
      0
526 527 528
    end
  end

529
  delegate :branch_names, to: :raw_repository
530 531
  cache_method :branch_names, fallback: []

Douwe Maan's avatar
Douwe Maan committed
532
  delegate :tag_names, to: :raw_repository
533 534
  cache_method :tag_names, fallback: []

Kim "BKC" Carlbäcker's avatar
Kim "BKC" Carlbäcker committed
535
  delegate :branch_count, :tag_count, to: :raw_repository
536 537 538 539 540 541
  cache_method :branch_count, fallback: 0
  cache_method :tag_count, fallback: 0

  def avatar
    if tree = file_on_head(:avatar)
      tree.path
542 543
    end
  end
544
  cache_method :avatar
545

546
  def readme
547 548
    if readme = tree(:head)&.readme
      ReadmeBlob.new(readme, self)
549
    end
550 551
  end

552
  def rendered_readme
553
    MarkupHelper.markup_unsafe(readme.name, readme.data, project: project) if readme
554 555
  end
  cache_method :rendered_readme
556

557
  def contribution_guide
558
    file_on_head(:contributing)
559
  end
560
  cache_method :contribution_guide
561 562

  def changelog
563
    file_on_head(:changelog)
564
  end
565
  cache_method :changelog
566

567
  def license_blob
568
    file_on_head(:license)
569
  end
570
  cache_method :license_blob
571

572
  def license_key
573
    return unless exists?
574

575
    Licensee.license(path).try(:key)
576
  end
577
  cache_method :license_key
578

579 580
  def license
    return unless license_key
581

582
    Licensee::License.new(license_key)
583
  end
584
  cache_method :license, memoize_only: true
585 586

  def gitignore
587
    file_on_head(:gitignore)
588
  end
589
  cache_method :gitignore
590 591

  def koding_yml
592
    file_on_head(:koding)
593
  end
594
  cache_method :koding_yml
595

596
  def gitlab_ci_yml
597
    file_on_head(:gitlab_ci)
598
  end
599
  cache_method :gitlab_ci_yml
600

601
  def head_commit
602 603 604 605
    @head_commit ||= commit(self.root_ref)
  end

  def head_tree
606 607 608
    if head_commit
      @head_tree ||= Tree.new(self, head_commit.sha, nil)
    end
609 610
  end

611
  def tree(sha = :head, path = nil, recursive: false)
612
    if sha == :head
613 614
      return unless head_commit

615 616 617 618 619
      if path.nil?
        return head_tree
      else
        sha = head_commit.sha
      end
620 621
    end

622
    Tree.new(self, sha, path, recursive: recursive)
623
  end
624 625

  def blob_at_branch(branch_name, path)
626
    last_commit = commit(branch_name)
627

628 629 630 631 632
    if last_commit
      blob_at(last_commit.sha, path)
    else
      nil
    end
633
  end
634

635
  def last_commit_for_path(sha, path)
636 637 638 639 640 641 642
    raw_repository.gitaly_migrate(:last_commit_for_path) do |is_enabled|
      if is_enabled
        last_commit_for_path_by_gitaly(sha, path)
      else
        last_commit_for_path_by_rugged(sha, path)
      end
    end
643
  end
644

Hiroyuki Sato's avatar
Hiroyuki Sato committed
645 646
  def last_commit_id_for_path(sha, path)
    key = path.blank? ? "last_commit_id_for_path:#{sha}" : "last_commit_id_for_path:#{sha}:#{Digest::SHA1.hexdigest(path)}"
Hiroyuki Sato's avatar
Hiroyuki Sato committed
647

648
    cache.fetch(key) do
649 650 651 652 653 654 655
      raw_repository.gitaly_migrate(:last_commit_for_path) do |is_enabled|
        if is_enabled
          last_commit_for_path_by_gitaly(sha, path).id
        else
          last_commit_id_for_path_by_shelling_out(sha, path)
        end
      end
656 657 658
    end
  end

659
  def next_branch(name, opts = {})
660 661 662
    branch_ids = self.branch_names.map do |n|
      next 1 if n == name
      result = n.match(/\A#{name}-([0-9]+)\z/)
663 664 665
      result[1].to_i if result
    end.compact

666
    highest_branch_id = branch_ids.max || 0
667

668 669 670
    return name if opts[:mild] && 0 == highest_branch_id

    "#{name}-#{highest_branch_id + 1}"
671 672
  end

673
  def branches_sorted_by(value)
674
    raw_repository.local_branches(sort_by: value)
675
  end
676

677 678 679
  def tags_sorted_by(value)
    case value
    when 'name'
680
      VersionSorter.rsort(tags) { |tag| tag.name }
681 682 683 684 685 686 687 688 689
    when 'updated_desc'
      tags_sorted_by_committed_date.reverse
    when 'updated_asc'
      tags_sorted_by_committed_date
    else
      tags
    end
  end

690
  def contributors
691
    commits = self.commits(nil, limit: 2000, offset: 0, skip_merges: true)
692

Dmitriy Zaporozhets's avatar
Dmitriy Zaporozhets committed
693
    commits.group_by(&:author_email).map do |email, commits|
694 695
      contributor = Gitlab::Contributor.new
      contributor.email = email
696

Dmitriy Zaporozhets's avatar
Dmitriy Zaporozhets committed
697
      commits.each do |commit|
698
        if contributor.name.blank?
Dmitriy Zaporozhets's avatar
Dmitriy Zaporozhets committed
699
          contributor.name = commit.author_name
700 701
        end

702
        contributor.commits += 1
703 704
      end

705 706
      contributor
    end
707
  end
Dmitriy Zaporozhets's avatar
Dmitriy Zaporozhets committed
708

709
  def refs_contains_sha(ref_type, sha)
710 711
    args = %W(#{ref_type} --contains #{sha})
    names = run_git(args).first
712 713 714 715 716 717 718 719 720 721 722 723 724

    if names.respond_to?(:split)
      names = names.split("\n").map(&:strip)

      names.each do |name|
        name.slice! '* '
      end

      names
    else
      []
    end
  end
725

726 727 728
  def branch_names_contains(sha)
    refs_contains_sha('branch', sha)
  end
729

730 731
  def tag_names_contains(sha)
    refs_contains_sha('tag', sha)
732
  end
733

734
  def local_branches
735
    @local_branches ||= raw_repository.local_branches
736 737
  end

738 739
  alias_method :branches, :local_branches

740 741 742 743
  def tags
    @tags ||= raw_repository.tags
  end

Douwe Maan's avatar
Douwe Maan committed
744 745 746
  def create_dir(user, path, **options)
    options[:user] = user
    options[:actions] = [{ action: :create_dir, file_path: path }]
747

Douwe Maan's avatar
Douwe Maan committed
748
    multi_action(**options)
Stan Hu's avatar
Stan Hu committed
749 750
  end

Douwe Maan's avatar
Douwe Maan committed
751 752 753
  def create_file(user, path, content, **options)
    options[:user] = user
    options[:actions] = [{ action: :create, file_path: path, content: content }]
754

Douwe Maan's avatar
Douwe Maan committed
755
    multi_action(**options)
Stan Hu's avatar
Stan Hu committed
756
  end
757

Douwe Maan's avatar
Douwe Maan committed
758 759 760
  def update_file(user, path, content, **options)
    previous_path = options.delete(:previous_path)
    action = previous_path && previous_path != path ? :move : :update
761

Douwe Maan's avatar
Douwe Maan committed
762 763
    options[:user] = user
    options[:actions] = [{ action: action, file_path: path, previous_path: previous_path, content: content }]
764

Douwe Maan's avatar
Douwe Maan committed
765
    multi_action(**options)
766 767
  end

Douwe Maan's avatar
Douwe Maan committed
768 769 770
  def delete_file(user, path, **options)
    options[:user] = user
    options[:actions] = [{ action: :delete, file_path: path }]
771

Douwe Maan's avatar
Douwe Maan committed
772
    multi_action(**options)
773 774
  end

775
  # rubocop:disable Metrics/ParameterLists
Lin Jen-Shin's avatar
Lin Jen-Shin committed
776
  def multi_action(
777
    user:, branch_name:, message:, actions:,
778
    author_email: nil, author_name: nil,
779
    start_branch_name: nil, start_project: project)
780

781
    GitOperationService.new(user, self).with_branch(
782
      branch_name,
783 784
      start_branch_name: start_branch_name,
      start_project: start_project) do |start_commit|
785

786
      index = Gitlab::Git::Index.new(raw_repository)
787

788
      if start_commit
789
        index.read_tree(start_commit.rugged_commit.tree)
790 791 792
        parents = [start_commit.sha]
      else
        parents = []
Marc Siegfriedt's avatar
Marc Siegfriedt committed
793 794
      end

795
      actions.each do |options|
796
        index.public_send(options.delete(:action), options) # rubocop:disable GitlabSecurity/PublicSend
Marc Siegfriedt's avatar
Marc Siegfriedt committed
797 798 799
      end

      options = {
800
        tree: index.write_tree,
Marc Siegfriedt's avatar
Marc Siegfriedt committed
801 802 803 804 805
        message: message,
        parents: parents
      }
      options.merge!(get_committer_and_author(user, email: author_email, name: author_name))

806
      create_commit(options)
Marc Siegfriedt's avatar
Marc Siegfriedt committed
807 808
    end
  end
809
  # rubocop:enable Metrics/ParameterLists
Marc Siegfriedt's avatar
Marc Siegfriedt committed
810

811 812
  def get_committer_and_author(user, email: nil, name: nil)
    committer = user_to_committer(user)
Douwe Maan's avatar
Douwe Maan committed
813
    author = Gitlab::Git.committer_hash(email: email, name: name) || committer
814

815
    {
816 817
      author: author,
      committer: committer
818 819 820
    }
  end

821
  def user_to_committer(user)
822
    Gitlab::Git.committer_hash(email: user.email, name: user.name)
823 824
  end

825 826 827 828 829 830 831 832 833 834 835
  def can_be_merged?(source_sha, target_branch)
    our_commit = rugged.branches[target_branch].target
    their_commit = rugged.lookup(source_sha)

    if our_commit && their_commit
      !rugged.merge_commits(our_commit, their_commit).conflicts?
    else
      false
    end
  end

836
  def merge(user, source, merge_request, options = {})
837
    GitOperationService.new(user, self).with_branch(
838 839
      merge_request.target_branch) do |start_commit|
      our_commit = start_commit.sha
840
      their_commit = source
841

842 843
      raise 'Invalid merge target' unless our_commit
      raise 'Invalid merge source' unless their_commit
844

845 846
      merge_index = rugged.merge_commits(our_commit, their_commit)
      break if merge_index.conflicts?
847

848 849
      actual_options = options.merge(
        parents: [our_commit, their_commit],
850
        tree: merge_index.write_tree(rugged)
851
      )
852

853
      commit_id = create_commit(actual_options)
854 855
      merge_request.update(in_progress_merge_commit_sha: commit_id)
      commit_id
856
    end
857 858
  rescue Repository::CommitError # when merge_index.conflicts?
    false
859 860
  end

861
  def revert(
862
    user, commit, branch_name,
863
    start_branch_name: nil, start_project: project)
864
    GitOperationService.new(user, self).with_branch(
865
      branch_name,
866 867
      start_branch_name: start_branch_name,
      start_project: start_project) do |start_commit|
868

869 870
      revert_tree_id = check_revert_content(commit, start_commit.sha)
      unless revert_tree_id
871
        raise Repository::CreateTreeError.new('Failed to revert commit')
872
      end
873 874

      committer = user_to_committer(user)
875

876 877 878 879 880
      create_commit(message: commit.revert_message(user),
                    author: committer,
                    committer: committer,
                    tree: revert_tree_id,
                    parents: [start_commit.sha])
881
    end
882 883
  end

884
  def cherry_pick(
885
    user, commit, branch_name,
886
    start_branch_name: nil, start_project: project)
887
    GitOperationService.new(user, self).with_branch(
888
      branch_name,
889 890
      start_branch_name: start_branch_name,
      start_project: start_project) do |start_commit|
891

892 893
      cherry_pick_tree_id = check_cherry_pick_content(commit, start_commit.sha)
      unless cherry_pick_tree_id
894
        raise Repository::CreateTreeError.new('Failed to cherry-pick commit')
895
      end
896 897

      committer = user_to_committer(user)
898

899 900 901 902 903 904 905 906 907
      create_commit(message: commit.message,
                    author: {
                        email: commit.author_email,
                        name: commit.author_name,
                        time: commit.authored_date
                    },
                    committer: committer,
                    tree: cherry_pick_tree_id,
                    parents: [start_commit.sha])
908 909 910
    end
  end

911 912
  def resolve_conflicts(user, branch_name, params)
    GitOperationService.new(user, self).with_branch(branch_name) do
913 914
      committer = user_to_committer(user)

915
      create_commit(params.merge(author: committer, committer: committer))
916 917 918
    end
  end

919 920
  def check_revert_content(target_commit, source_sha)
    args = [target_commit.sha, source_sha]
921
    args << { mainline: 1 } if target_commit.merge_commit?
922 923 924 925 926 927 928 929 930 931

    revert_index = rugged.revert_commit(*args)
    return false if revert_index.conflicts?

    tree_id = revert_index.write_tree(rugged)
    return false unless diff_exists?(source_sha, tree_id)

    tree_id
  end

932 933
  def check_cherry_pick_content(target_commit, source_sha)
    args = [target_commit.sha, source_sha]
934
    args << 1 if target_commit.merge_commit?
935 936 937 938 939 940 941 942 943 944

    cherry_pick_index = rugged.cherrypick_commit(*args)
    return false if cherry_pick_index.conflicts?

    tree_id = cherry_pick_index.write_tree(rugged)
    return false unless diff_exists?(source_sha, tree_id)

    tree_id
  end

945 946
  def diff_exists?(sha1, sha2)
    rugged.diff(sha1, sha2).size > 0
947 948
  end

949 950 951 952 953
  def merged_to_root_ref?(branch_name)
    branch_commit = commit(branch_name)
    root_ref_commit = commit(root_ref)

    if branch_commit
954 955
      same_head = branch_commit.id == root_ref_commit.id
      !same_head && is_ancestor?(branch_commit.id, root_ref_commit.id)
956 957 958 959 960
    else
      nil
    end
  end

961
  def merge_base(first_commit_id, second_commit_id)
962 963
    first_commit_id = commit(first_commit_id).try(:id) || first_commit_id
    second_commit_id = commit(second_commit_id).try(:id) || second_commit_id
964
    rugged.merge_base(first_commit_id, second_commit_id)
Douwe Maan's avatar
Douwe Maan committed
965 966
  rescue Rugged::ReferenceError
    nil
967 968
  end

969
  def is_ancestor?(ancestor_id, descendant_id)
970
    return false if ancestor_id.nil? || descendant_id.nil?
971

972 973 974 975
    Gitlab::GitalyClient.migrate(:is_ancestor) do |is_enabled|
      if is_enabled
        raw_repository.is_ancestor?(ancestor_id, descendant_id)
      else
976
        rugged_is_ancestor?(ancestor_id, descendant_id)
977 978
      end
    end
979 980
  end

981 982 983 984 985 986
  def empty_repo?
    !exists? || !has_visible_content?
  end

  def search_files_by_content(query, ref)
    return [] if empty_repo? || query.blank?
Valery Sizov's avatar
Valery Sizov committed
987

988
    offset = 2
989 990 991
    args = %W(grep -i -I -n --before-context #{offset} --after-context #{offset} -E -e #{Regexp.escape(query)} #{ref || root_ref})

    run_git(args).first.scrub.split(/^--$/)
992 993
  end

994 995
  def search_files_by_name(query, ref)
    return [] if empty_repo? || query.blank?
996

997 998 999
    args = %W(ls-tree --full-tree -r #{ref || root_ref} --name-status | #{Regexp.escape(query)})

    run_git(args).first.lines.map(&:strip)
1000 1001
  end

1002
  def with_repo_branch_commit(start_repository, start_branch_name)
1003 1004
    tmp_ref = nil
    return yield nil if start_repository.empty_repo?
1005

1006
    branch_commit =
1007
      if start_repository == self
1008
        commit(start_branch_name)
1009
      else
1010 1011 1012 1013 1014 1015 1016 1017 1018 1019 1020
        sha = start_repository.find_branch(start_branch_name).target
        commit(sha) ||
          begin
            tmp_ref = fetch_ref(
              start_repository.path_to_repo,
              "#{Gitlab::Git::BRANCH_REF_PREFIX}#{start_branch_name}",
              "refs/tmp/#{SecureRandom.hex}/head"
            )

            commit(start_repository.commit(start_branch_name).sha)
          end
1021
      end
1022

1023
    yield branch_commit
1024
  ensure
1025
    rugged.references.delete(tmp_ref) if tmp_ref
1026
  end
1027

1028 1029 1030 1031 1032 1033 1034 1035 1036 1037 1038 1039 1040 1041
  def add_remote(name, url)
    raw_repository.remote_add(name, url)
  rescue Rugged::ConfigError
    raw_repository.remote_update(name, url: url)
  end

  def remove_remote(name)
    raw_repository.remote_delete(name)
    true
  rescue Rugged::ConfigError
    false
  end

  def fetch_remote(remote, forced: false, no_tags: false)
1042
    gitlab_shell.fetch_remote(repository_storage_path, disk_path, remote, forced: forced, no_tags: no_tags)
1043 1044
  end

1045
  def fetch_ref(source_path, source_ref, target_ref)
1046
    args = %W(fetch --no-tags -f #{source_path} #{source_ref}:#{target_ref})
1047
    message, status = run_git(args)
1048 1049

    # Make sure ref was created, and raise Rugged::ReferenceError when not
1050
    raise Rugged::ReferenceError, message if status != 0
1051 1052

    target_ref
1053
  end
1054

1055 1056
  def create_ref(ref, ref_path)
    fetch_ref(path_to_repo, ref, ref_path)
1057 1058
  end

1059 1060 1061 1062 1063
  def ls_files(ref)
    actual_ref = ref || root_ref
    raw_repository.ls_files(actual_ref)
  end

1064 1065 1066 1067
  def gitattribute(path, name)
    raw_repository.attributes(path)[name]
  end

1068 1069 1070 1071 1072 1073 1074 1075 1076 1077
  def copy_gitattributes(ref)
    actual_ref = ref || root_ref
    begin
      raw_repository.copy_gitattributes(actual_ref)
      true
    rescue Gitlab::Git::Repository::InvalidRef
      false
    end
  end

1078 1079 1080 1081 1082 1083 1084 1085 1086 1087 1088
  # Caches the supplied block both in a cache and in an instance variable.
  #
  # The cache key and instance variable are named the same way as the value of
  # the `key` argument.
  #
  # This method will return `nil` if the corresponding instance variable is also
  # set to `nil`. This ensures we don't keep yielding the block when it returns
  # `nil`.
  #
  # key - The name of the key to cache the data in.
  # fallback - A value to fall back to in the event of a Git error.
1089
  def cache_method_output(key, fallback: nil, memoize_only: false, &block)
1090
    ivar = cache_instance_variable_name(key)
1091

1092 1093 1094 1095
    if instance_variable_defined?(ivar)
      instance_variable_get(ivar)
    else
      begin
1096 1097 1098 1099 1100 1101 1102
        value =
          if memoize_only
            yield
          else
            cache.fetch(key, &block)
          end
        instance_variable_set(ivar, value)
1103 1104 1105 1106
      rescue Rugged::ReferenceError, Gitlab::Git::Repository::NoRepository
        # if e.g. HEAD or the entire repository doesn't exist we want to
        # gracefully handle this and not cache anything.
        fallback
1107 1108 1109
      end
    end
  end
1110

1111 1112 1113
  def cache_instance_variable_name(key)
    :"@#{key.to_s.tr('?!', '')}"
  end
1114

1115 1116
  def file_on_head(type)
    if head = tree(:head)
1117 1118
      head.blobs.find do |blob|
        Gitlab::FileDetector.type_of(blob.path) == type
1119 1120 1121 1122
      end
    end
  end

Douwe Maan's avatar
Douwe Maan committed
1123 1124 1125 1126
  def route_map_for(sha)
    blob_data_at(sha, '.gitlab/route-map.yml')
  end

1127 1128
  def gitlab_ci_yml_for(sha, path = '.gitlab-ci.yml')
    blob_data_at(sha, path)
Douwe Maan's avatar
Douwe Maan committed
1129 1130
  end

1131 1132
  private

1133 1134 1135 1136 1137 1138
  def run_git(args)
    circuit_breaker.perform do
      Gitlab::Popen.popen([Gitlab.config.git.bin_path, *args], path_to_repo)
    end
  end

Douwe Maan's avatar
Douwe Maan committed
1139 1140
  def blob_data_at(sha, path)
    blob = blob_at(sha, path)
1141
    return unless blob
1142

1143
    blob.load_all_data!
1144
    blob.data
1145
  end
1146

1147
  def refs_directory_exists?
1148 1149 1150
    circuit_breaker.perform do
      File.exist?(File.join(path_to_repo, 'refs'))
    end
1151
  end
1152

1153
  def cache
1154 1155
    # TODO: should we use UUIDs here? We could move repositories without clearing this cache
    @cache ||= RepositoryCache.new(full_path, @project.id)
1156
  end
1157 1158

  def tags_sorted_by_committed_date
1159 1160 1161 1162 1163 1164 1165 1166 1167 1168 1169 1170
    tags.sort_by do |tag|
      # Annotated tags can point to any object (e.g. a blob), but generally
      # tags point to a commit. If we don't have a commit, then just default
      # to putting the tag at the end of the list.
      target = tag.dereferenced_target

      if target
        target.committed_date
      else
        Time.now
      end
    end
1171
  end
1172 1173 1174 1175

  def keep_around_ref_name(sha)
    "refs/keep-around/#{sha}"
  end
Yorick Peterse's avatar
Yorick Peterse committed
1176 1177

  def repository_event(event, tags = {})
1178
    Gitlab::Metrics.add_event(event, { path: full_path }.merge(tags))
Yorick Peterse's avatar
Yorick Peterse committed
1179
  end
1180

1181 1182 1183 1184 1185 1186
  def create_commit(params = {})
    params[:message].delete!("\r")

    Rugged::Commit.create(rugged, params)
  end

1187 1188 1189 1190 1191 1192 1193 1194 1195 1196 1197
  def last_commit_for_path_by_gitaly(sha, path)
    c = raw_repository.gitaly_commit_client.last_commit_for_path(sha, path)
    commit(c)
  end

  def last_commit_for_path_by_rugged(sha, path)
    sha = last_commit_id_for_path_by_shelling_out(sha, path)
    commit(sha)
  end

  def last_commit_id_for_path_by_shelling_out(sha, path)
1198 1199
    args = %W(rev-list --max-count=1 #{sha} -- #{path})
    run_git(args).first.strip
1200 1201
  end

1202 1203 1204
  def repository_storage_path
    @project.repository_storage_path
  end
1205

1206
  def initialize_raw_repository
1207
    Gitlab::Git::Repository.new(project.repository_storage, disk_path + '.git', Gitlab::GlRepository.gl_repository(project, false))
1208
  end
1209 1210 1211 1212

  def circuit_breaker
    @circuit_breaker ||= Gitlab::Git::Storage::CircuitBreaker.for_storage(project.repository_storage)
  end
1213 1214 1215 1216 1217 1218 1219 1220 1221 1222 1223 1224 1225 1226 1227 1228 1229 1230 1231 1232 1233

  def find_commits_by_message_by_shelling_out(query, ref, path, limit, offset)
    ref ||= root_ref

    args = %W(
      log #{ref} --pretty=%H --skip #{offset}
      --max-count #{limit} --grep=#{query} --regexp-ignore-case
    )
    args = args.concat(%W(-- #{path})) if path.present?

    git_log_results = run_git(args).first.lines

    git_log_results.map { |c| commit(c.chomp) }.compact
  end

  def find_commits_by_message_by_gitaly(query, ref, path, limit, offset)
    raw_repository
      .gitaly_commit_client
      .commits_by_message(query, revision: ref, path: path, limit: limit, offset: offset)
      .map { |c| commit(c) }
  end
1234
end