BigW Consortium Gitlab

repository.rb 29.7 KB
Newer Older
1 2
require 'securerandom'

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

7
  attr_accessor :path_with_namespace, :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
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
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
      cache_method_output(name, fallback: fallback, memoize_only: memoize_only) { __send__(original) }
52
    end
53
  end
54

55
  def initialize(path_with_namespace, project)
56
    @path_with_namespace = path_with_namespace
57
    @project = project
58
  end
59

60 61
  def raw_repository
    return nil unless path_with_namespace
62

63
    @raw_repository ||= initialize_raw_repository
64 65
  end

66
  # Return absolute path to repository
67
  def path_to_repo
68
    @path_to_repo ||= File.expand_path(
69
      File.join(repository_storage_path, path_with_namespace + ".git")
70
    )
71 72
  end

73 74 75 76 77 78 79 80 81 82
  #
  # 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?
83
    branch_count > 0
84 85
  end

86
  def commit(ref = 'HEAD')
87
    return nil unless exists?
88

89 90 91 92 93 94
    commit =
      if ref.is_a?(Gitlab::Git::Commit)
        ref
      else
        Gitlab::Git::Commit.find(raw_repository, ref)
      end
95

96
    commit = ::Commit.new(commit, @project) if commit
97
    commit
98
  rescue Rugged::OdbError, Rugged::TreeError
99
    nil
100 101
  end

102
  def commits(ref, path: nil, limit: nil, offset: nil, skip_merges: false, after: nil, before: nil)
103
    options = {
104 105 106 107 108
      repo: raw_repository,
      ref: ref,
      path: path,
      limit: limit,
      offset: offset,
109 110
      after: after,
      before: before,
111
      follow: Array(path).length == 1,
112
      skip_merges: skip_merges
113 114 115
    }

    commits = Gitlab::Git::Commit.where(options)
116
    commits = Commit.decorate(commits, @project) if commits.present?
117 118 119
    commits
  end

120 121
  def commits_between(from, to)
    commits = Gitlab::Git::Commit.between(raw_repository, from, to)
122
    commits = Commit.decorate(commits, @project) if commits.present?
123 124 125
    commits
  end

126
  # Gitaly migration: https://gitlab.com/gitlab-org/gitaly/issues/384
127
  def find_commits_by_message(query, ref = nil, path = nil, limit = 1000, offset = 0)
128 129 130 131
    unless exists? && has_visible_content? && query.present?
      return []
    end

132 133
    ref ||= root_ref

134 135 136 137
    args = %W(
      #{Gitlab.config.git.bin_path} log #{ref} --pretty=%H --skip #{offset}
      --max-count #{limit} --grep=#{query} --regexp-ignore-case
    )
138
    args = args.concat(%W(-- #{path})) if path.present?
139

140 141
    git_log_results = Gitlab::Popen.popen(args, path_to_repo).first.lines
    git_log_results.map { |c| commit(c.chomp) }.compact
142 143
  end

144 145 146 147 148 149 150
  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
151
    raw_repo = fresh_repo ? initialize_raw_repository : raw_repository
152 153

    raw_repo.find_branch(name)
154 155 156
  end

  def find_tag(name)
157
    tags.find { |tag| tag.name == name }
158 159
  end

160 161
  def add_branch(user, branch_name, ref)
    newrev = commit(ref).try(:sha)
162

163
    return false unless newrev
164

165
    GitOperationService.new(user, self).add_branch(branch_name, newrev)
166

167
    after_create_branch
168
    find_branch(branch_name)
169 170
  end

171
  def add_tag(user, tag_name, target, message = nil)
172
    newrev = commit(target).try(:id)
173 174
    options = { message: message, tagger: user_to_committer(user) } if message

175 176 177
    return false unless newrev

    GitOperationService.new(user, self).add_tag(tag_name, newrev, options)
178

179
    find_tag(tag_name)
180 181
  end

182
  def rm_branch(user, branch_name)
183
    before_remove_branch
184 185
    branch = find_branch(branch_name)

186
    GitOperationService.new(user, self).rm_branch(branch)
187

188
    after_remove_branch
189
    true
190 191
  end

192
  def rm_tag(user, tag_name)
193
    before_remove_tag
194
    tag = find_tag(tag_name)
195

196 197 198 199
    GitOperationService.new(user, self).rm_tag(tag)

    after_remove_tag
    true
200 201
  end

202 203 204 205
  def ref_names
    branch_names + tag_names
  end

206 207 208 209
  def branch_exists?(branch_name)
    branch_names.include?(branch_name)
  end

210 211
  def ref_exists?(ref)
    rugged.references.exist?(ref)
212 213
  rescue Rugged::ReferenceError
    false
214 215
  end

216 217 218 219
  # 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.
220 221 222 223 224
  def keep_around(sha)
    return unless sha && commit(sha)

    return if kept_around?(sha)

225 226 227 228 229
    # This will still fail if the file is corrupted (e.g. 0 bytes)
    begin
      rugged.references.create(keep_around_ref_name(sha), sha, force: true)
    rescue Rugged::ReferenceError => ex
      Rails.logger.error "Unable to create keep-around reference for repository #{path}: #{ex}"
230 231 232
    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}"
233
    end
234 235 236
  end

  def kept_around?(sha)
237
    ref_exists?(keep_around_ref_name(sha))
238
  end
239

240
  def diverging_commit_counts(branch)
241
    root_ref_hash = raw_repository.rev_parse_target(root_ref).oid
242
    cache.fetch(:"diverging_commit_counts_#{branch.name}") do
243 244
      # Rugged seems to throw a `ReferenceError` when given branch_names rather
      # than SHA-1 hashes
245 246
      number_commits_behind = raw_repository
        .count_commits_between(branch.dereferenced_target.sha, root_ref_hash)
247

248 249
      number_commits_ahead = raw_repository
        .count_commits_between(root_ref_hash, branch.dereferenced_target.sha)
250

251 252 253
      { behind: number_commits_behind, ahead: number_commits_ahead }
    end
  end
254

255 256 257
  def expire_tags_cache
    expire_method_caches(%i(tag_names tag_count))
    @tags = nil
258
  end
259

260 261 262
  def expire_branches_cache
    expire_method_caches(%i(branch_names branch_count))
    @local_branches = nil
263 264
  end

265 266
  def expire_statistics_caches
    expire_method_caches(%i(size commit_count))
267 268
  end

269 270
  def expire_all_method_caches
    expire_method_caches(CACHED_METHODS)
271 272
  end

273 274 275 276 277 278 279 280 281
  # 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
282 283
  end

284 285 286 287 288 289 290 291 292 293 294 295 296 297 298
  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
299
    end
300

301
    expire_method_caches(to_refresh)
302

303
    to_refresh.each { |method| send(method) }
304
  end
305

306 307 308 309 310 311 312
  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}")
313
        cache.expire(:"commit_count_#{branch.name}")
314 315 316 317 318
      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}")
319
      cache.expire(:"commit_count_#{branch_name}")
320
    end
Dmitriy Zaporozhets committed
321 322
  end

323
  def expire_root_ref_cache
324
    expire_method_caches(%i(root_ref))
325 326
  end

327 328
  # Expires the cache(s) used to determine if a repository is empty or not.
  def expire_emptiness_caches
329
    return unless empty?
330

331
    expire_method_caches(%i(empty?))
332 333
  end

334 335 336 337
  def lookup_cache
    @lookup_cache ||= {}
  end

338
  def expire_exists_cache
339
    expire_method_caches(%i(exists?))
340 341
  end

342 343 344 345 346 347 348
  # 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
349
    expire_statistics_caches
350 351 352 353 354
  end

  # Runs code after a repository has been created.
  def after_create
    expire_exists_cache
355 356
    expire_root_ref_cache
    expire_emptiness_caches
357 358

    repository_event(:create_repository)
359 360
  end

361 362
  # Runs code just before a repository is deleted.
  def before_delete
363
    expire_exists_cache
364 365
    expire_all_method_caches
    expire_branch_cache if exists?
366
    expire_content_cache
367 368

    repository_event(:remove_repository)
369 370 371 372 373 374 375
  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
376 377

    repository_event(:change_default_branch)
378 379
  end

380 381
  # Runs code before pushing (= creating or removing) a tag.
  def before_push_tag
382 383
    expire_statistics_caches
    expire_emptiness_caches
384
    expire_tags_cache
385 386

    repository_event(:push_tag)
387 388 389 390 391
  end

  # Runs code before removing a tag.
  def before_remove_tag
    expire_tags_cache
392
    expire_statistics_caches
393 394

    repository_event(:remove_tag)
395 396
  end

397 398 399 400 401
  # Runs code after removing a tag.
  def after_remove_tag
    expire_tags_cache
  end

402 403 404
  # Runs code after the HEAD of a repository is changed.
  def after_change_head
    expire_method_caches(METHOD_CACHES_FOR_FILE_TYPES.keys)
405 406
  end

407 408
  # Runs code after a repository has been forked/imported.
  def after_import
409
    expire_content_cache
410 411 412
  end

  # Runs code after a new commit has been pushed.
413 414 415
  def after_push_commit(branch_name)
    expire_statistics_caches
    expire_branch_cache(branch_name)
416 417

    repository_event(:push_commit, branch: branch_name)
418 419 420 421
  end

  # Runs code after a new branch has been created.
  def after_create_branch
422
    expire_branches_cache
423 424

    repository_event(:push_branch)
425 426
  end

427 428 429
  # Runs code before removing an existing branch.
  def before_remove_branch
    expire_branches_cache
430 431

    repository_event(:remove_branch)
432 433
  end

434 435
  # Runs code after an existing branch has been removed.
  def after_remove_branch
436
    expire_branches_cache
437 438
  end

439
  def method_missing(m, *args, &block)
440 441 442 443 444 445
    if m == :lookup && !block_given?
      lookup_cache[m] ||= {}
      lookup_cache[m][args.join(":")] ||= raw_repository.send(m, *args, &block)
    else
      raw_repository.send(m, *args, &block)
    end
446 447
  end

448 449
  def respond_to_missing?(method, include_private = false)
    raw_repository.respond_to?(method, include_private) || super
450
  end
451 452

  def blob_at(sha, path)
453
    unless Gitlab::Git.blank_ref?(sha)
Douwe Maan committed
454
      Blob.decorate(Gitlab::Git::Blob.find(self, sha, path), project)
455
    end
Douwe Maan committed
456 457
  rescue Gitlab::Git::Repository::NoRepository
    nil
458
  end
459

460 461 462 463
  def blob_by_oid(oid)
    Gitlab::Git::Blob.raw(self, oid)
  end

464 465 466 467 468 469 470
  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
471
  end
472
  cache_method :root_ref
473

474 475 476 477 478
  def exists?
    refs_directory_exists?
  end
  cache_method :exists?

Douwe Maan committed
479
  delegate :empty?, to: :raw_repository
480 481 482 483 484 485 486 487 488 489 490 491 492
  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

493
  def commit_count_for_ref(ref)
494
    return 0 unless exists?
495

496 497 498 499
    begin
      cache.fetch(:"commit_count_#{ref}") { raw_repository.commit_count(ref) }
    rescue Rugged::ReferenceError
      0
500 501 502
    end
  end

503
  delegate :branch_names, to: :raw_repository
504 505
  cache_method :branch_names, fallback: []

Douwe Maan committed
506
  delegate :tag_names, to: :raw_repository
507 508
  cache_method :tag_names, fallback: []

Kim "BKC" Carlbäcker committed
509
  delegate :branch_count, :tag_count, to: :raw_repository
510 511 512 513 514 515
  cache_method :branch_count, fallback: 0
  cache_method :tag_count, fallback: 0

  def avatar
    if tree = file_on_head(:avatar)
      tree.path
516 517
    end
  end
518
  cache_method :avatar
519

520
  def readme
521 522
    if readme = tree(:head)&.readme
      ReadmeBlob.new(readme, self)
523
    end
524 525
  end

526
  def rendered_readme
527
    MarkupHelper.markup_unsafe(readme.name, readme.data, project: project) if readme
528 529
  end
  cache_method :rendered_readme
530

531
  def contribution_guide
532
    file_on_head(:contributing)
533
  end
534
  cache_method :contribution_guide
535 536

  def changelog
537
    file_on_head(:changelog)
538
  end
539
  cache_method :changelog
540

541
  def license_blob
542
    file_on_head(:license)
543
  end
544
  cache_method :license_blob
545

546
  def license_key
547
    return unless exists?
548

549
    Licensee.license(path).try(:key)
550
  end
551
  cache_method :license_key
552

553 554
  def license
    return unless license_key
555

556
    Licensee::License.new(license_key)
557
  end
558
  cache_method :license, memoize_only: true
559 560

  def gitignore
561
    file_on_head(:gitignore)
562
  end
563
  cache_method :gitignore
564 565

  def koding_yml
566
    file_on_head(:koding)
567
  end
568
  cache_method :koding_yml
569

570
  def gitlab_ci_yml
571
    file_on_head(:gitlab_ci)
572
  end
573
  cache_method :gitlab_ci_yml
574

575
  def head_commit
576 577 578 579
    @head_commit ||= commit(self.root_ref)
  end

  def head_tree
580 581 582
    if head_commit
      @head_tree ||= Tree.new(self, head_commit.sha, nil)
    end
583 584
  end

585
  def tree(sha = :head, path = nil, recursive: false)
586
    if sha == :head
587 588
      return unless head_commit

589 590 591 592 593
      if path.nil?
        return head_tree
      else
        sha = head_commit.sha
      end
594 595
    end

596
    Tree.new(self, sha, path, recursive: recursive)
597
  end
598 599

  def blob_at_branch(branch_name, path)
600
    last_commit = commit(branch_name)
601

602 603 604 605 606
    if last_commit
      blob_at(last_commit.sha, path)
    else
      nil
    end
607
  end
608

609
  def last_commit_for_path(sha, path)
Hiroyuki Sato committed
610
    sha = last_commit_id_for_path(sha, path)
611
    commit(sha)
612
  end
613

614
  # Gitaly migration: https://gitlab.com/gitlab-org/gitaly/issues/383
Hiroyuki Sato committed
615 616
  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 committed
617

618
    cache.fetch(key) do
Hiroyuki Sato committed
619 620
      args = %W(#{Gitlab.config.git.bin_path} rev-list --max-count=1 #{sha} -- #{path})
      Gitlab::Popen.popen(args, path_to_repo).first.strip
621 622 623
    end
  end

624
  def next_branch(name, opts = {})
625 626 627
    branch_ids = self.branch_names.map do |n|
      next 1 if n == name
      result = n.match(/\A#{name}-([0-9]+)\z/)
628 629 630
      result[1].to_i if result
    end.compact

631
    highest_branch_id = branch_ids.max || 0
632

633 634 635
    return name if opts[:mild] && 0 == highest_branch_id

    "#{name}-#{highest_branch_id + 1}"
636 637
  end

638
  def branches_sorted_by(value)
639
    raw_repository.local_branches(sort_by: value)
640
  end
641

642 643 644
  def tags_sorted_by(value)
    case value
    when 'name'
645
      VersionSorter.rsort(tags) { |tag| tag.name }
646 647 648 649 650 651 652 653 654
    when 'updated_desc'
      tags_sorted_by_committed_date.reverse
    when 'updated_asc'
      tags_sorted_by_committed_date
    else
      tags
    end
  end

655
  def contributors
656
    commits = self.commits(nil, limit: 2000, offset: 0, skip_merges: true)
657

658
    commits.group_by(&:author_email).map do |email, commits|
659 660
      contributor = Gitlab::Contributor.new
      contributor.email = email
661

662
      commits.each do |commit|
663
        if contributor.name.blank?
664
          contributor.name = commit.author_name
665 666
        end

667
        contributor.commits += 1
668 669
      end

670 671
      contributor
    end
672
  end
673

674 675
  def refs_contains_sha(ref_type, sha)
    args = %W(#{Gitlab.config.git.bin_path} #{ref_type} --contains #{sha})
676 677 678 679 680 681 682 683 684 685 686 687 688 689
    names = Gitlab::Popen.popen(args, path_to_repo).first

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

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

      names
    else
      []
    end
  end
690

691 692 693
  def branch_names_contains(sha)
    refs_contains_sha('branch', sha)
  end
694

695 696
  def tag_names_contains(sha)
    refs_contains_sha('tag', sha)
697
  end
698

699
  def local_branches
700
    @local_branches ||= raw_repository.local_branches
701 702
  end

703 704
  alias_method :branches, :local_branches

705 706 707 708
  def tags
    @tags ||= raw_repository.tags
  end

Douwe Maan committed
709 710 711
  def create_dir(user, path, **options)
    options[:user] = user
    options[:actions] = [{ action: :create_dir, file_path: path }]
712

Douwe Maan committed
713
    multi_action(**options)
Stan Hu committed
714 715
  end

Douwe Maan committed
716 717 718
  def create_file(user, path, content, **options)
    options[:user] = user
    options[:actions] = [{ action: :create, file_path: path, content: content }]
719

Douwe Maan committed
720
    multi_action(**options)
Stan Hu committed
721
  end
722

Douwe Maan committed
723 724 725
  def update_file(user, path, content, **options)
    previous_path = options.delete(:previous_path)
    action = previous_path && previous_path != path ? :move : :update
726

Douwe Maan committed
727 728
    options[:user] = user
    options[:actions] = [{ action: action, file_path: path, previous_path: previous_path, content: content }]
729

Douwe Maan committed
730
    multi_action(**options)
731 732
  end

Douwe Maan committed
733 734 735
  def delete_file(user, path, **options)
    options[:user] = user
    options[:actions] = [{ action: :delete, file_path: path }]
736

Douwe Maan committed
737
    multi_action(**options)
738 739
  end

740
  # rubocop:disable Metrics/ParameterLists
741
  def multi_action(
742
    user:, branch_name:, message:, actions:,
743
    author_email: nil, author_name: nil,
744
    start_branch_name: nil, start_project: project)
745

746
    GitOperationService.new(user, self).with_branch(
747
      branch_name,
748 749
      start_branch_name: start_branch_name,
      start_project: start_project) do |start_commit|
750

751
      index = Gitlab::Git::Index.new(raw_repository)
752

753 754 755 756 757
      if start_commit
        index.read_tree(start_commit.raw_commit.tree)
        parents = [start_commit.sha]
      else
        parents = []
Marc Siegfriedt committed
758 759
      end

760
      actions.each do |options|
Douwe Maan committed
761
        index.public_send(options.delete(:action), options)
Marc Siegfriedt committed
762 763 764
      end

      options = {
765
        tree: index.write_tree,
Marc Siegfriedt committed
766 767 768 769 770
        message: message,
        parents: parents
      }
      options.merge!(get_committer_and_author(user, email: author_email, name: author_name))

771
      create_commit(options)
Marc Siegfriedt committed
772 773
    end
  end
774
  # rubocop:enable Metrics/ParameterLists
Marc Siegfriedt committed
775

776 777
  def get_committer_and_author(user, email: nil, name: nil)
    committer = user_to_committer(user)
778
    author = Gitlab::Git.committer_hash(email: email, name: name) || committer
779

780
    {
781 782
      author: author,
      committer: committer
783 784 785
    }
  end

786
  def user_to_committer(user)
787
    Gitlab::Git.committer_hash(email: user.email, name: user.name)
788 789
  end

790 791 792 793 794 795 796 797 798 799 800
  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

801
  def merge(user, source, merge_request, options = {})
802
    GitOperationService.new(user, self).with_branch(
803 804
      merge_request.target_branch) do |start_commit|
      our_commit = start_commit.sha
805
      their_commit = source
806

807 808
      raise 'Invalid merge target' unless our_commit
      raise 'Invalid merge source' unless their_commit
809

810 811
      merge_index = rugged.merge_commits(our_commit, their_commit)
      break if merge_index.conflicts?
812

813 814
      actual_options = options.merge(
        parents: [our_commit, their_commit],
815
        tree: merge_index.write_tree(rugged)
816
      )
817

818
      commit_id = create_commit(actual_options)
819 820
      merge_request.update(in_progress_merge_commit_sha: commit_id)
      commit_id
821
    end
822 823
  rescue Repository::CommitError # when merge_index.conflicts?
    false
824 825
  end

826
  def revert(
827
    user, commit, branch_name,
828
    start_branch_name: nil, start_project: project)
829
    GitOperationService.new(user, self).with_branch(
830
      branch_name,
831 832
      start_branch_name: start_branch_name,
      start_project: start_project) do |start_commit|
833

834 835
      revert_tree_id = check_revert_content(commit, start_commit.sha)
      unless revert_tree_id
836
        raise Repository::CreateTreeError.new('Failed to revert commit')
837
      end
838 839

      committer = user_to_committer(user)
840

841 842 843 844 845
      create_commit(message: commit.revert_message(user),
                    author: committer,
                    committer: committer,
                    tree: revert_tree_id,
                    parents: [start_commit.sha])
846
    end
847 848
  end

849
  def cherry_pick(
850
    user, commit, branch_name,
851
    start_branch_name: nil, start_project: project)
852
    GitOperationService.new(user, self).with_branch(
853
      branch_name,
854 855
      start_branch_name: start_branch_name,
      start_project: start_project) do |start_commit|
856

857 858
      cherry_pick_tree_id = check_cherry_pick_content(commit, start_commit.sha)
      unless cherry_pick_tree_id
859
        raise Repository::CreateTreeError.new('Failed to cherry-pick commit')
860
      end
861 862

      committer = user_to_committer(user)
863

864 865 866 867 868 869 870 871 872
      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])
873 874 875
    end
  end

876 877
  def resolve_conflicts(user, branch_name, params)
    GitOperationService.new(user, self).with_branch(branch_name) do
878 879
      committer = user_to_committer(user)

880
      create_commit(params.merge(author: committer, committer: committer))
881 882 883
    end
  end

884 885
  def check_revert_content(target_commit, source_sha)
    args = [target_commit.sha, source_sha]
886
    args << { mainline: 1 } if target_commit.merge_commit?
887 888 889 890 891 892 893 894 895 896

    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

897 898
  def check_cherry_pick_content(target_commit, source_sha)
    args = [target_commit.sha, source_sha]
899
    args << 1 if target_commit.merge_commit?
900 901 902 903 904 905 906 907 908 909

    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

910 911
  def diff_exists?(sha1, sha2)
    rugged.diff(sha1, sha2).size > 0
912 913
  end

914 915 916 917 918
  def merged_to_root_ref?(branch_name)
    branch_commit = commit(branch_name)
    root_ref_commit = commit(root_ref)

    if branch_commit
919 920
      same_head = branch_commit.id == root_ref_commit.id
      !same_head && is_ancestor?(branch_commit.id, root_ref_commit.id)
921 922 923 924 925
    else
      nil
    end
  end

926
  def merge_base(first_commit_id, second_commit_id)
927 928
    first_commit_id = commit(first_commit_id).try(:id) || first_commit_id
    second_commit_id = commit(second_commit_id).try(:id) || second_commit_id
929
    rugged.merge_base(first_commit_id, second_commit_id)
Douwe Maan committed
930 931
  rescue Rugged::ReferenceError
    nil
932 933
  end

934
  def is_ancestor?(ancestor_id, descendant_id)
935
    return false if ancestor_id.nil? || descendant_id.nil?
936

937 938 939 940 941 942 943
    Gitlab::GitalyClient.migrate(:is_ancestor) do |is_enabled|
      if is_enabled
        raw_repository.is_ancestor?(ancestor_id, descendant_id)
      else
        merge_base_commit(ancestor_id, descendant_id) == ancestor_id
      end
    end
944 945
  end

946 947 948 949 950 951
  def empty_repo?
    !exists? || !has_visible_content?
  end

  def search_files_by_content(query, ref)
    return [] if empty_repo? || query.blank?
Valery Sizov committed
952

953
    offset = 2
954
    args = %W(#{Gitlab.config.git.bin_path} grep -i -I -n --before-context #{offset} --after-context #{offset} -E -e #{Regexp.escape(query)} #{ref || root_ref})
955 956 957
    Gitlab::Popen.popen(args, path_to_repo).first.scrub.split(/^--$/)
  end

958 959
  def search_files_by_name(query, ref)
    return [] if empty_repo? || query.blank?
960

961 962
    args = %W(#{Gitlab.config.git.bin_path} ls-tree --full-tree -r #{ref || root_ref} --name-status | #{Regexp.escape(query)})
    Gitlab::Popen.popen(args, path_to_repo).first.lines.map(&:strip)
963 964
  end

965
  def with_repo_branch_commit(start_repository, start_branch_name)
966
    return yield(nil) if start_repository.empty_repo?
967

968
    branch_name_or_sha =
969 970
      if start_repository == self
        start_branch_name
971 972
      else
        tmp_ref = "refs/tmp/#{SecureRandom.hex}/head"
973

974
        fetch_ref(
975 976
          start_repository.path_to_repo,
          "#{Gitlab::Git::BRANCH_REF_PREFIX}#{start_branch_name}",
977 978
          tmp_ref
        )
979

980
        start_repository.commit(start_branch_name).sha
981
      end
982

983
    yield(commit(branch_name_or_sha))
984

985
  ensure
986
    rugged.references.delete(tmp_ref) if tmp_ref
987
  end
988

989 990 991 992 993 994 995 996 997 998 999 1000 1001 1002 1003
  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)
    gitlab_shell.fetch_remote(repository_storage_path, path_with_namespace, remote, forced: forced, no_tags: no_tags)
1004 1005
  end

1006
  def fetch_ref(source_path, source_ref, target_ref)
1007
    args = %W(#{Gitlab.config.git.bin_path} fetch --no-tags -f #{source_path} #{source_ref}:#{target_ref})
1008 1009
    Gitlab::Popen.popen(args, path_to_repo)
  end
1010

1011 1012
  def create_ref(ref, ref_path)
    fetch_ref(path_to_repo, ref, ref_path)
1013 1014
  end

1015 1016 1017 1018 1019
  def ls_files(ref)
    actual_ref = ref || root_ref
    raw_repository.ls_files(actual_ref)
  end

1020 1021 1022 1023
  def gitattribute(path, name)
    raw_repository.attributes(path)[name]
  end

1024 1025 1026 1027 1028 1029 1030 1031 1032 1033
  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

1034 1035 1036 1037 1038 1039 1040 1041 1042 1043 1044
  # 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.
1045
  def cache_method_output(key, fallback: nil, memoize_only: false, &block)
1046
    ivar = cache_instance_variable_name(key)
1047

1048 1049 1050 1051
    if instance_variable_defined?(ivar)
      instance_variable_get(ivar)
    else
      begin
1052 1053 1054 1055 1056 1057 1058
        value =
          if memoize_only
            yield
          else
            cache.fetch(key, &block)
          end
        instance_variable_set(ivar, value)
1059 1060 1061 1062
      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
1063 1064 1065
      end
    end
  end
1066

1067 1068 1069
  def cache_instance_variable_name(key)
    :"@#{key.to_s.tr('?!', '')}"
  end
1070

1071 1072
  def file_on_head(type)
    if head = tree(:head)
1073 1074
      head.blobs.find do |blob|
        Gitlab::FileDetector.type_of(blob.path) == type
1075 1076 1077 1078
      end
    end
  end

Douwe Maan committed
1079 1080 1081 1082
  def route_map_for(sha)
    blob_data_at(sha, '.gitlab/route-map.yml')
  end

1083 1084
  def gitlab_ci_yml_for(sha, path = '.gitlab-ci.yml')
    blob_data_at(sha, path)
Douwe Maan committed
1085 1086
  end

1087 1088
  private

Douwe Maan committed
1089 1090
  def blob_data_at(sha, path)
    blob = blob_at(sha, path)
1091
    return unless blob
1092

1093
    blob.load_all_data!
1094
    blob.data
1095
  end
1096

1097 1098 1099 1100
  def refs_directory_exists?
    return false unless path_with_namespace

    File.exist?(File.join(path_to_repo, 'refs'))
1101
  end
1102

1103
  def cache
1104
    @cache ||= RepositoryCache.new(path_with_namespace, @project.id)
1105
  end
1106 1107

  def tags_sorted_by_committed_date
1108 1109 1110 1111 1112 1113 1114 1115 1116 1117 1118 1119
    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
1120
  end
1121 1122 1123 1124

  def keep_around_ref_name(sha)
    "refs/keep-around/#{sha}"
  end
1125 1126 1127 1128

  def repository_event(event, tags = {})
    Gitlab::Metrics.add_event(event, { path: path_with_namespace }.merge(tags))
  end
1129

1130 1131 1132 1133 1134 1135
  def create_commit(params = {})
    params[:message].delete!("\r")

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

1136 1137 1138
  def repository_storage_path
    @project.repository_storage_path
  end
1139

1140
  def initialize_raw_repository
1141 1142
    Gitlab::Git::Repository.new(project.repository_storage, path_with_namespace + '.git')
  end
1143
end