diff options
author | Lin Jen-Shin <godfat@godfat.org> | 2017-11-06 21:44:57 +0800 |
---|---|---|
committer | Lin Jen-Shin <godfat@godfat.org> | 2017-11-06 21:44:57 +0800 |
commit | fc6aad0b4442c58fde1ac924cb2dd73823273537 (patch) | |
tree | 3f4a46a5b649cf623ab5e8e42eaa2e06cb2b20cf /lib/gitlab/git/blob.rb | |
parent | 239332eed3fa870fd41be83864882c0f389840d8 (diff) | |
parent | cfc932cad10b1d6c494222e9d91aa75583b56145 (diff) | |
download | gitlab-ce-fc6aad0b4442c58fde1ac924cb2dd73823273537.tar.gz |
Merge remote-tracking branch 'upstream/master' into no-ivar-in-modules
* upstream/master: (1723 commits)
Resolve "Editor icons"
Refactor issuable destroy action
Ignore routes matching legacy_*_redirect in route specs
Gitlab::Git::RevList and LfsChanges use lazy popen
Gitlab::Git::Popen can lazily hand output to a block
Merge branch 'master-i18n' into 'master'
Remove unique validation from external_url in Environment
Expose `duration` in Job API entity
Add TimeCop freeze for DST and Regular time
Harcode project visibility
update a changelog
Put a condition to old migration that adds fast_forward column to MRs
Expose project visibility as CI variable
fix flaky tests by removing unneeded clicks and focus actions
fix flaky test in gfm_autocomplete_spec.rb
Use Gitlab::Git operations for repository mirroring
Encapsulate git operations for mirroring in Gitlab::Git
Create a Wiki Repository's raw_repository properly
Add `Gitlab::Git::Repository#fetch` command
Fix Gitlab::Metrics::System#real_time and #monotonic_time doc
...
Diffstat (limited to 'lib/gitlab/git/blob.rb')
-rw-r--r-- | lib/gitlab/git/blob.rb | 57 |
1 files changed, 46 insertions, 11 deletions
diff --git a/lib/gitlab/git/blob.rb b/lib/gitlab/git/blob.rb index a4336facee5..cc6c7609ec7 100644 --- a/lib/gitlab/git/blob.rb +++ b/lib/gitlab/git/blob.rb @@ -12,6 +12,12 @@ module Gitlab # blob data should use load_all_data!. MAX_DATA_DISPLAY_SIZE = 10.megabytes + # These limits are used as a heuristic to ignore files which can't be LFS + # pointers. The format of these is described in + # https://github.com/git-lfs/git-lfs/blob/master/docs/spec.md#the-pointer + LFS_POINTER_MIN_SIZE = 120.bytes + LFS_POINTER_MAX_SIZE = 200.bytes + attr_accessor :name, :path, :size, :data, :mode, :id, :commit_id, :loaded_size, :binary class << self @@ -30,16 +36,7 @@ module Gitlab if is_enabled Gitlab::GitalyClient::BlobService.new(repository).get_blob(oid: sha, limit: MAX_DATA_DISPLAY_SIZE) else - blob = repository.lookup(sha) - - next unless blob.is_a?(Rugged::Blob) - - new( - id: blob.oid, - size: blob.size, - data: blob.content(MAX_DATA_DISPLAY_SIZE), - binary: blob.binary? - ) + rugged_raw(repository, sha, limit: MAX_DATA_DISPLAY_SIZE) end end end @@ -59,10 +56,25 @@ module Gitlab end end + # Find LFS blobs given an array of sha ids + # Returns array of Gitlab::Git::Blob + # Does not guarantee blob data will be set + def batch_lfs_pointers(repository, blob_ids) + blob_ids.lazy + .select { |sha| possible_lfs_blob?(repository, sha) } + .map { |sha| rugged_raw(repository, sha, limit: LFS_POINTER_MAX_SIZE) } + .select(&:lfs_pointer?) + .force + end + def binary?(data) EncodingHelper.detect_libgit2_binary?(data) end + def size_could_be_lfs?(size) + size.between?(LFS_POINTER_MIN_SIZE, LFS_POINTER_MAX_SIZE) + end + private # Recursive search of blob id by path @@ -167,6 +179,29 @@ module Gitlab end end end + + def rugged_raw(repository, sha, limit:) + blob = repository.lookup(sha) + + return unless blob.is_a?(Rugged::Blob) + + new( + id: blob.oid, + size: blob.size, + data: blob.content(limit), + binary: blob.binary? + ) + end + + # Efficient lookup to determine if object size + # and type make it a possible LFS blob without loading + # blob content into memory with repository.lookup(sha) + def possible_lfs_blob?(repository, sha) + object_header = repository.rugged.read_header(sha) + + object_header[:type] == :blob && + size_could_be_lfs?(object_header[:len]) + end end def initialize(options) @@ -226,7 +261,7 @@ module Gitlab # size # see https://github.com/github/git-lfs/blob/v1.1.0/docs/spec.md#the-pointer def lfs_pointer? - has_lfs_version_key? && lfs_oid.present? && lfs_size.present? + self.class.size_could_be_lfs?(size) && has_lfs_version_key? && lfs_oid.present? && lfs_size.present? end def lfs_oid |