Class: MergeRequestDiff

Inherits:
ApplicationRecord show all
Includes:
BulkInsertableAssociations, EachBatch, FromUnion, Gitlab::Utils::StrongMemoize, Importable, ManualInverseAssociation, ObjectStorable, ShaAttribute, Sortable
Defined in:
app/models/merge_request_diff.rb

Constant Summary collapse

STORE_COLUMN =
:external_diff_store
COMMITS_SAFE_SIZE =

Don’t display more than 100 commits at once

100
BATCH_SIZE =
1000
EXTERNAL_DIFF_CUTOFF =

Applies to closed or merged MRs when determining whether to migrate their diffs to external storage

7.days.freeze
FILES_COUNT_SENTINEL =

The files_count column is a 2-byte signed integer. Look up the true value from the database if this sentinel is seen

(2**15) - 1
EXTERNAL_DIFFS_CACHE_TMPDIR =

External diff cache key used by diffs export

'project-%{project_id}-external-mr-%{mr_id}-diff-%{id}-cache'
EXTERNAL_DIFF_CACHE_CHUNK_SIZE =
8.megabytes

Constants inherited from ApplicationRecord

ApplicationRecord::MAX_PLUCK

Constants included from HasCheckConstraints

HasCheckConstraints::NOT_NULL_CHECK_PATTERN

Constants included from ResetOnColumnErrors

ResetOnColumnErrors::MAX_RESET_PERIOD

Instance Attribute Summary

Attributes included from Importable

#importing, #user_contributions

Class Method Summary collapse

Instance Method Summary collapse

Methods included from BulkInsertableAssociations

#bulk_insert_associations!, bulk_inserts_enabled?, with_bulk_insert

Methods inherited from ApplicationRecord

===, cached_column_list, #create_or_load_association, current_transaction, declarative_enum, default_select_columns, delete_all_returning, #deleted_from_database?, id_in, id_not_in, iid_in, nullable_column?, primary_key_in, #readable_by?, safe_ensure_unique, safe_find_or_create_by, safe_find_or_create_by!, #to_ability_name, underscore, where_exists, where_not_exists, with_fast_read_statement_timeout, without_order

Methods included from Organizations::Sharding

#sharding_organization

Methods included from ResetOnColumnErrors

#reset_on_union_error, #reset_on_unknown_attribute_error

Methods included from Gitlab::SensitiveSerializableHash

#serializable_hash

Class Method Details

.find_by_diff_refs(diff_refs) ⇒ Object



250
251
252
# File 'app/models/merge_request_diff.rb', line 250

def self.find_by_diff_refs(diff_refs)
  find_by(start_commit_sha: diff_refs.start_sha, head_commit_sha: diff_refs.head_sha, base_commit_sha: diff_refs.base_sha)
end

.ids_for_external_storage_migration(limit:) ⇒ Object



194
195
196
197
198
199
200
201
202
203
204
205
# File 'app/models/merge_request_diff.rb', line 194

def ids_for_external_storage_migration(limit:)
  return [] unless Gitlab.config.external_diffs.enabled

  case Gitlab.config.external_diffs.when
  when 'always'
    ids_for_external_storage_migration_strategy_always(limit: limit)
  when 'outdated'
    ids_for_external_storage_migration_strategy_outdated(limit: limit)
  else
    []
  end
end

.ids_for_external_storage_migration_strategy_always(limit:) ⇒ Object



207
208
209
# File 'app/models/merge_request_diff.rb', line 207

def ids_for_external_storage_migration_strategy_always(limit:)
  files_in_database.limit(limit).pluck(:id)
end

.ids_for_external_storage_migration_strategy_outdated(limit:) ⇒ Object



211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
# File 'app/models/merge_request_diff.rb', line 211

def ids_for_external_storage_migration_strategy_outdated(limit:)
  # Outdated is too complex to be a single SQL query, so split into three
  before = EXTERNAL_DIFF_CUTOFF.ago

  ids = files_in_database
    .old_merged_diffs(before)
    .limit(limit)
    .pluck(:id)

  return ids if ids.size >= limit

  ids += files_in_database
    .old_closed_diffs(before)
    .limit(limit - ids.size)
    .pluck(:id)

  return ids if ids.size >= limit

  ids + files_in_database
    .not_latest_diffs
    .limit(limit - ids.size)
    .pluck(:id)
end

Instance Method Details

#base_commitObject



398
399
400
401
402
# File 'app/models/merge_request_diff.rb', line 398

def base_commit
  return unless base_commit_sha

  project.commit_by(oid: base_commit_sha)
end

#cached_external_diffObject

Yields locally cached external diff if it’s externally stored. Used during Project Export to speed up externally stored merge request diffs export



701
702
703
704
705
706
707
708
709
# File 'app/models/merge_request_diff.rb', line 701

def cached_external_diff
  return yield(nil) unless stored_externally?

  cache_external_diff unless File.exist?(external_diff_cache_filepath)

  File.open(external_diff_cache_filepath) do |file|
    yield(file)
  end
end

#commit_shas(limit: nil, preload_metadata: false) ⇒ Object



416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
# File 'app/models/merge_request_diff.rb', line 416

def commit_shas(limit: nil, preload_metadata: false)
  if association(:merge_request_diff_commits).loaded?
    sorted_diff_commits = merge_request_diff_commits.sort_by { |diff_commit| [diff_commit.id, diff_commit.relative_order] }
    sorted_diff_commits = sorted_diff_commits.take(limit) if limit

    if  && diff_commits_dedup_enabled?
      # ActiveRecord::Associations::Preloader works with both arrays and relations
      (sorted_diff_commits)
    end

    sorted_diff_commits.map(&:sha)
  elsif diff_commits_dedup_enabled?
    (limit)
  else
    merge_request_diff_commits.limit(limit).pluck(:sha)
  end
end

#commits(limit: nil, load_from_gitaly: false, page: nil) ⇒ Object



380
381
382
383
384
# File 'app/models/merge_request_diff.rb', line 380

def commits(limit: nil, load_from_gitaly: false, page: nil)
  strong_memoize(:"commits_#{limit || 'all'}_#{load_from_gitaly}_page_#{page}") do
    load_commits(limit: limit, load_from_gitaly: load_from_gitaly, page: page)
  end
end

#compareObject



594
595
596
597
598
599
600
601
# File 'app/models/merge_request_diff.rb', line 594

def compare
  @compare ||=
    Gitlab::Git::Compare.new(
      repository.raw_repository,
      safe_start_commit_sha,
      head_commit_sha
    )
end

#diff_refsObject



452
453
454
455
456
457
458
459
460
# File 'app/models/merge_request_diff.rb', line 452

def diff_refs
  return unless start_commit_sha || base_commit_sha

  Gitlab::Diff::DiffRefs.new(
    base_sha: base_commit_sha,
    start_sha: start_commit_sha,
    head_sha: head_commit_sha
  )
end

#diff_refs=(new_diff_refs) ⇒ Object



446
447
448
449
450
# File 'app/models/merge_request_diff.rb', line 446

def diff_refs=(new_diff_refs)
  self.base_commit_sha = new_diff_refs&.base_sha
  self.start_commit_sha = new_diff_refs&.start_sha
  self.head_commit_sha = new_diff_refs&.head_sha
end

#diffs(diff_options = nil) ⇒ Object



568
569
570
571
572
573
574
575
576
577
578
579
580
# File 'app/models/merge_request_diff.rb', line 568

def diffs(diff_options = nil)
  fetching_repository_diffs(diff_options) do |comparison|
    # It should fetch the repository when diffs are cleaned by the system.
    # We don't keep these for storage overload purposes.
    # See https://gitlab.com/gitlab-org/gitlab-foss/issues/37639
    if comparison
      comparison.diffs(diff_options)
    else
      reorder_diff_files!
      diffs_collection(diff_options)
    end
  end
end

#diffs_collection(diff_options = nil) ⇒ Object

Should always return the DB persisted diffs collection (e.g. Gitlab::Diff::FileCollection::MergeRequestDiff. It’s useful when trying to invalidate old caches through FileCollection::MergeRequestDiff#clear_cache!



586
587
588
# File 'app/models/merge_request_diff.rb', line 586

def diffs_collection(diff_options = nil)
  Gitlab::Diff::FileCollection::MergeRequestDiff.new(self, diff_options: diff_options)
end

#diffs_for_streaming(diff_options = {}) ⇒ Object



485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
# File 'app/models/merge_request_diff.rb', line 485

def diffs_for_streaming(diff_options = {})
  fetching_repository_diffs(diff_options) do |comparison|
    reorder_diff_files!

    collection = Gitlab::Diff::FileCollection::MergeRequestDiffStream.new(
      self,
      diff_options: diff_options
    )

    if comparison
      # Delete the offset_index from options since we don't want to offset
      # the diffs we will request given that we are already requesting specific
      # paths
      diff_options.delete(:offset_index)
      diff_options[:generated_files] = comparison.generated_files
      diff_options[:paths] = collection.diff_paths

      comparison.diffs(diff_options)
    else
      collection
    end
  end
end

#diffs_in_batch(batch_page, batch_size, diff_options:) ⇒ Object



509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
# File 'app/models/merge_request_diff.rb', line 509

def diffs_in_batch(batch_page, batch_size, diff_options:)
  fetching_repository_diffs(diff_options) do |comparison|
    Gitlab::Metrics.measure(:diffs_reorder) do
      reorder_diff_files!
    end

    diffs_batch = Gitlab::Metrics.measure(:diffs_collection) do
      diffs_in_batch_collection(batch_page, batch_size, diff_options: diff_options)
    end

    if comparison
      if diff_options[:paths].blank? && !without_files?
        # Return the empty MergeRequestDiffBatch for an out of bound batch request
        break diffs_batch if diffs_batch.diff_paths.blank?

        diff_options.merge!(
          paths: diffs_batch.diff_paths,
          pagination_data: diffs_batch.pagination_data
        )
      end

      diff_options[:generated_files] = comparison.generated_files

      Gitlab::Metrics.measure(:diffs_comparison) do
        comparison.diffs(diff_options)
      end
    else
      diffs_batch
    end
  end
end

#ensure_commit_shasObject



319
320
321
322
323
324
325
326
327
328
329
330
331
# File 'app/models/merge_request_diff.rb', line 319

def ensure_commit_shas
  self.start_commit_sha ||= merge_request.target_branch_sha

  if merge_head? && merge_request.merge_ref_head.present?
    diff_refs = merge_request.merge_ref_head.diff_refs

    self.head_commit_sha  ||= diff_refs.head_sha
    self.base_commit_sha  ||= diff_refs.base_sha
  else
    self.head_commit_sha  ||= merge_request.source_branch_sha
    self.base_commit_sha  ||= find_base_sha
  end
end

#fallback_diff_refsObject

MRs created before 8.4 don’t store their true diff refs (start and base), but we need to get a commit SHA for the “View file @ …” link by a file, so we use an approximation of the diff refs if we can’t get the actual one.

These will not be the actual diff refs if the target branch was merged into the source branch after the merge request was created, but it is good enough for the specific purpose of linking to a commit.

It is not good enough for highlighting diffs, so we can’t simply pass these as diff_refs.



472
473
474
475
476
477
478
479
480
481
482
483
# File 'app/models/merge_request_diff.rb', line 472

def fallback_diff_refs
  real_refs = diff_refs
  return real_refs if real_refs

  likely_base_commit_sha = (first_commit&.parent || first_commit)&.sha

  Gitlab::Diff::DiffRefs.new(
    base_sha: likely_base_commit_sha,
    start_sha: safe_start_commit_sha,
    head_sha: head_commit_sha
  )
end

#files_countObject



343
344
345
346
347
348
349
350
351
352
# File 'app/models/merge_request_diff.rb', line 343

def files_count
  db_value = read_attribute(:files_count)

  case db_value
  when nil, FILES_COUNT_SENTINEL
    merge_request_diff_files.count
  else
    db_value
  end
end

#first_commitObject



390
391
392
# File 'app/models/merge_request_diff.rb', line 390

def first_commit
  commits(load_from_gitaly: true).last
end

#get_patch_id_shaObject



295
296
297
298
299
300
301
302
303
# File 'app/models/merge_request_diff.rb', line 295

def get_patch_id_sha
  return patch_id_sha if patch_id_sha.present?

  set_patch_id_sha

  return unless patch_id_sha.present?

  patch_id_sha
end

#has_encoded_file_paths?Boolean

Returns:

  • (Boolean)


720
721
722
# File 'app/models/merge_request_diff.rb', line 720

def has_encoded_file_paths?
  merge_request_diff_files.where(encoded_file_path: true).any?
end

#head_commitObject



410
411
412
413
414
# File 'app/models/merge_request_diff.rb', line 410

def head_commit
  return unless head_commit_sha

  project.commit_by(oid: head_commit_sha)
end

#head_commit_shaObject

Override head_commit_sha to keep compatibility with merge request diff created before version 8.4 that does not store head_commit_sha in separate db field.



335
336
337
338
339
340
341
# File 'app/models/merge_request_diff.rb', line 335

def head_commit_sha
  if persisted? && super.nil?
    last_commit_sha
  else
    super
  end
end

#includes_any_commits?(shas) ⇒ Boolean

Returns:

  • (Boolean)


434
435
436
437
438
439
440
441
442
443
444
# File 'app/models/merge_request_diff.rb', line 434

def includes_any_commits?(shas)
  return false if shas.blank?

  # when the number of shas is huge (1000+) we don't want
  # to pass them all as an SQL param, let's pass them in batches
  shas.each_slice(BATCH_SIZE).any? do |batched_shas|
    break true if diff_commits_dedup_enabled? && (batched_shas)

    merge_request_diff_commits.where(sha: batched_shas).exists?
  end
end

#last_commitObject



394
395
396
# File 'app/models/merge_request_diff.rb', line 394

def last_commit
  commits(load_from_gitaly: true).first
end

#last_commit_shaObject



386
387
388
# File 'app/models/merge_request_diff.rb', line 386

def last_commit_sha
  commit_shas(limit: 1).first
end

#latest?Boolean

Returns:

  • (Boolean)


603
604
605
# File 'app/models/merge_request_diff.rb', line 603

def latest?
  self.id == merge_request.latest_merge_request_diff_id
end

#lines_countObject



365
366
367
368
369
# File 'app/models/merge_request_diff.rb', line 365

def lines_count
  strong_memoize(:lines_count) do
    raw_diffs(limits: false).line_count
  end
end

#migrate_files_to_database!Object

Transactionally migrate the current merge_request_diff_files entries from external storage, back to the database. This is the rollback operation for migrate_files_to_external_storage!

If this diff isn’t in external storage, the method is a no-op.



680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
# File 'app/models/merge_request_diff.rb', line 680

def migrate_files_to_database!
  return unless stored_externally?
  return if files_count == 0

  rows = convert_external_diffs_to_database

  transaction do
    MergeRequestDiffFile.where(merge_request_diff_id: id).delete_all
    ApplicationRecord.legacy_bulk_insert('merge_request_diff_files', rows) # rubocop:disable Gitlab/BulkInsert
    update!(stored_externally: false)
  end

  # Only delete the external diff file after the contents have been saved to
  # the database
  remove_external_diff!
  merge_request_diff_files.reset
end

#migrate_files_to_external_storage!Object

Transactionally migrate the current merge_request_diff_files entries to external storage. If external storage isn’t an option for this diff, the method is a no-op.



655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
# File 'app/models/merge_request_diff.rb', line 655

def migrate_files_to_external_storage!
  return if stored_externally? || !use_external_diff? || files_count == 0

  rows = build_merge_request_diff_files(merge_request_diff_files)
  rows = build_external_merge_request_diff_files(rows)

  # Perform carrierwave activity before entering the database transaction.
  # This is safe as until the `external_diff_store` column is changed, we will
  # continue to consult the in-database content.
  self.external_diff.store!

  transaction do
    MergeRequestDiffFile.where(merge_request_diff_id: id).delete_all
    ApplicationRecord.legacy_bulk_insert('merge_request_diff_files', rows) # rubocop:disable Gitlab/BulkInsert
    save!
  end

  merge_request_diff_files.reset
end

#modified_paths(fallback_on_overflow: false) ⇒ Object



607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
# File 'app/models/merge_request_diff.rb', line 607

def modified_paths(fallback_on_overflow: false)
  if fallback_on_overflow && overflow?
    # This is an extremely slow means to find the modified paths for a given
    #   MergeRequestDiff. This should be avoided, except where the limit of
    #   1_000 (as of %12.10) entries returned by the default behavior is an
    #   issue.
    strong_memoize(:overflowed_modified_paths) do
      project.repository.diff_stats(
        base_commit_sha,
        head_commit_sha
      ).paths
    end
  else
    strong_memoize(:modified_paths) do
      merge_request_diff_files.pluck(:new_path, :old_path).flatten.compact.uniq
    end
  end
end

#opening_external_diffObject

If enabled, yields the external file containing the diff. Otherwise, yields nil. This method is not thread-safe, but it is re-entrant, which allows multiple merge_request_diff_files to load their data efficiently



635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
# File 'app/models/merge_request_diff.rb', line 635

def opening_external_diff
  return yield(nil) unless stored_externally?
  return yield(@external_diff_file) if @external_diff_file

  if use_external_diff? && Gitlab.config.external_diffs.object_store.enabled && external_diff_store == ObjectStorage::Store::LOCAL
    update_column(:external_diff_store, ObjectStorage::Store::REMOTE)
  end

  external_diff.open do |file|
    @external_diff_file = file

    yield(@external_diff_file)
  ensure
    @external_diff_file = nil
  end
end

#paginated_diffs(page, per_page, diff_options = {}) ⇒ Object



541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
# File 'app/models/merge_request_diff.rb', line 541

def paginated_diffs(page, per_page, diff_options = {})
  fetching_repository_diffs(diff_options) do |comparison|
    reorder_diff_files!

    collection = Gitlab::Diff::FileCollection::PaginatedMergeRequestDiff.new(
      self,
      page,
      per_page,
      diff_options
    )

    if comparison
      diff_options.merge!(
        generated_files: comparison.generated_files,
        paths: collection.diff_paths,
        page: collection.current_page,
        per_page: collection.limit_value,
        count: collection.total_count
      )

      comparison.diffs(diff_options)
    else
      collection
    end
  end
end

#preload_gitaly_dataObject



258
259
260
261
262
263
# File 'app/models/merge_request_diff.rb', line 258

def preload_gitaly_data
  ensure_commit_shas

  reversed_compare_commits_preloaded
  compare_diffs_preloaded
end

#projectObject



590
591
592
# File 'app/models/merge_request_diff.rb', line 590

def project
  merge_request.target_project
end

#raw_diffs(options = {}) ⇒ Object



371
372
373
374
375
376
377
378
# File 'app/models/merge_request_diff.rb', line 371

def raw_diffs(options = {})
  if options[:ignore_whitespace_change]
    @diffs_no_whitespace ||= compare.diffs(options)
  else
    @raw_diffs ||= {}
    @raw_diffs[options] ||= load_diffs(options)
  end
end

#remove_cached_external_diffObject



711
712
713
714
715
716
717
718
# File 'app/models/merge_request_diff.rb', line 711

def remove_cached_external_diff
  Gitlab::PathTraversal.check_path_traversal!(external_diff_cache_dir)
  Gitlab::PathTraversal.check_allowed_absolute_path!(external_diff_cache_dir, [Dir.tmpdir])

  return unless Dir.exist?(external_diff_cache_dir)

  FileUtils.rm_rf(external_diff_cache_dir)
end

#safe_start_commit_shaObject

This method will rely on repository branch sha in case start_commit_sha is nil. It’s necessary for old merge request diff created before version 8.4 to work



357
358
359
# File 'app/models/merge_request_diff.rb', line 357

def safe_start_commit_sha
  start_commit_sha || merge_request.target_branch_sha
end

#save_git_contentObject

Collect information about commits and diff from repository and save it to the database as serialized data



267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
# File 'app/models/merge_request_diff.rb', line 267

def save_git_content
  ensure_commit_shas

  save_commits
  save_diffs

  # Another set of `after_save` hooks will be called here when we update the record
  save
  # We need to reset so that dirty tracking is reset when running the original set
  # of `after_save` hooks that come after this `after_create` hook. Otherwise, the
  # hooks that run when an attribute was changed are run twice.
  reset

  keep_around_commits unless importing?
end

#set_as_latest_diffObject



305
306
307
308
309
310
311
312
313
# File 'app/models/merge_request_diff.rb', line 305

def set_as_latest_diff
  # Don't set merge_head diff as latest so it won't get considered as the
  # MergeRequest#merge_request_diff.
  return if merge_head?

  MergeRequest
    .where('id = ? AND COALESCE(latest_merge_request_diff_id, 0) < ?', self.merge_request_id, self.id)
    .update_all(latest_merge_request_diff_id: self.id)
end

#set_patch_id_shaObject



283
284
285
286
287
288
289
290
291
292
293
# File 'app/models/merge_request_diff.rb', line 283

def set_patch_id_sha
  return unless base_commit_sha && head_commit_sha
  return if base_commit_sha == head_commit_sha

  patch_id_sha = project.repository&.get_patch_id(
    base_commit_sha,
    head_commit_sha
  )

  update_column(:patch_id_sha, patch_id_sha)
end

#sizeObject



361
362
363
# File 'app/models/merge_request_diff.rb', line 361

def size
  real_size.presence || raw_diffs.size
end

#start_commitObject



404
405
406
407
408
# File 'app/models/merge_request_diff.rb', line 404

def start_commit
  return unless start_commit_sha

  project.commit_by(oid: start_commit_sha)
end

#trigger_diff_generated_subscriptionObject



315
316
317
# File 'app/models/merge_request_diff.rb', line 315

def trigger_diff_generated_subscription
  GraphqlTriggers.merge_request_diff_generated(merge_request)
end

#update_external_diff_storeObject



626
627
628
629
630
# File 'app/models/merge_request_diff.rb', line 626

def update_external_diff_store
  return unless saved_change_to_external_diff? || saved_change_to_stored_externally?

  update_column(:external_diff_store, external_diff.object_store)
end

#viewable?Boolean

Returns:

  • (Boolean)


254
255
256
# File 'app/models/merge_request_diff.rb', line 254

def viewable?
  collected? || without_files? || overflow?
end