debian-mirror-gitlab/lib/gitlab/git/blob.rb

340 lines
10 KiB
Ruby
Raw Normal View History

2017-09-10 17:25:29 +05:30
# Gitaly note: JV: seems to be completely migrated (behind feature flags).
2017-08-17 22:00:37 +05:30
module Gitlab
module Git
class Blob
include Linguist::BlobHelper
2017-09-10 17:25:29 +05:30
include Gitlab::EncodingHelper
2017-08-17 22:00:37 +05:30
# This number is the maximum amount of data that we want to display to
# the user. We load as much as we can for encoding detection
# (Linguist) and LFS pointer parsing. All other cases where we need full
# blob data should use load_all_data!.
MAX_DATA_DISPLAY_SIZE = 10.megabytes
2018-03-17 18:26:18 +05:30
# 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
2017-08-17 22:00:37 +05:30
attr_accessor :name, :path, :size, :data, :mode, :id, :commit_id, :loaded_size, :binary
class << self
def find(repository, sha, path)
2017-09-10 17:25:29 +05:30
Gitlab::GitalyClient.migrate(:project_raw_show) do |is_enabled|
if is_enabled
find_by_gitaly(repository, sha, path)
else
find_by_rugged(repository, sha, path, limit: MAX_DATA_DISPLAY_SIZE)
end
end
end
2017-08-17 22:00:37 +05:30
2017-09-10 17:25:29 +05:30
def raw(repository, sha)
Gitlab::GitalyClient.migrate(:git_blob_raw) do |is_enabled|
if is_enabled
2018-03-17 18:26:18 +05:30
repository.gitaly_blob_client.get_blob(oid: sha, limit: MAX_DATA_DISPLAY_SIZE)
2017-09-10 17:25:29 +05:30
else
2018-03-17 18:26:18 +05:30
rugged_raw(repository, sha, limit: MAX_DATA_DISPLAY_SIZE)
2017-08-17 22:00:37 +05:30
end
end
end
2017-09-10 17:25:29 +05:30
# Returns an array of Blob instances, specified in blob_references as
# [[commit_sha, path], [commit_sha, path], ...]. If blob_size_limit < 0 then the
# full blob contents are returned. If blob_size_limit >= 0 then each blob will
# contain no more than limit bytes in its data attribute.
2018-03-17 18:26:18 +05:30
#
2017-09-10 17:25:29 +05:30
# Keep in mind that this method may allocate a lot of memory. It is up
# to the caller to limit the number of blobs and blob_size_limit.
#
2018-03-17 18:26:18 +05:30
# Gitaly migration issue: https://gitlab.com/gitlab-org/gitaly/issues/798
def batch(repository, blob_references, blob_size_limit: MAX_DATA_DISPLAY_SIZE)
Gitlab::GitalyClient.migrate(:list_blobs_by_sha_path) do |is_enabled|
if is_enabled
repository.gitaly_blob_client.get_blobs(blob_references, blob_size_limit).to_a
else
blob_references.map do |sha, path|
find_by_rugged(repository, sha, path, limit: blob_size_limit)
end
end
end
end
2018-10-15 14:42:47 +05:30
# Returns an array of Blob instances just with the metadata, that means
# the data attribute has no content.
def batch_metadata(repository, blob_references)
batch(repository, blob_references, blob_size_limit: 0)
end
2018-03-17 18:26:18 +05:30
# 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)
repository.gitaly_migrate(:batch_lfs_pointers) do |is_enabled|
if is_enabled
repository.gitaly_blob_client.batch_lfs_pointers(blob_ids.to_a)
else
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
2017-09-10 17:25:29 +05:30
end
2017-08-17 22:00:37 +05:30
end
2018-03-17 18:26:18 +05:30
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
2017-09-10 17:25:29 +05:30
private
2017-08-17 22:00:37 +05:30
# Recursive search of blob id by path
#
# Ex.
# blog/ # oid: 1a
# app/ # oid: 2a
# models/ # oid: 3a
# file.rb # oid: 4a
#
#
2018-10-15 14:42:47 +05:30
# Blob.find_entry_by_path(repo, '1a', 'blog', 'app', 'file.rb') # => '4a'
2017-08-17 22:00:37 +05:30
#
2018-10-15 14:42:47 +05:30
def find_entry_by_path(repository, root_id, *path_parts)
2017-08-17 22:00:37 +05:30
root_tree = repository.lookup(root_id)
entry = root_tree.find do |entry|
2018-10-15 14:42:47 +05:30
entry[:name] == path_parts[0]
2017-08-17 22:00:37 +05:30
end
return nil unless entry
2018-10-15 14:42:47 +05:30
if path_parts.size > 1
2017-08-17 22:00:37 +05:30
return nil unless entry[:type] == :tree
2018-03-17 18:26:18 +05:30
2018-10-15 14:42:47 +05:30
path_parts.shift
find_entry_by_path(repository, entry[:oid], *path_parts)
2017-08-17 22:00:37 +05:30
else
[:blob, :commit].include?(entry[:type]) ? entry : nil
end
end
def submodule_blob(blob_entry, path, sha)
new(
id: blob_entry[:oid],
name: blob_entry[:name],
2017-09-10 17:25:29 +05:30
size: 0,
2017-08-17 22:00:37 +05:30
data: '',
path: path,
2017-09-10 17:25:29 +05:30
commit_id: sha
2017-08-17 22:00:37 +05:30
)
end
2017-09-10 17:25:29 +05:30
2018-03-17 18:26:18 +05:30
def find_by_gitaly(repository, sha, path, limit: MAX_DATA_DISPLAY_SIZE)
return unless path
path = path.sub(%r{\A/*}, '')
2017-09-10 17:25:29 +05:30
path = '/' if path.empty?
name = File.basename(path)
2018-03-17 18:26:18 +05:30
# Gitaly will think that setting the limit to 0 means unlimited, while
# the client might only need the metadata and thus set the limit to 0.
# In this method we'll then set the limit to 1, but clear the byte of data
# that we got back so for the outside world it looks like the limit was
# actually 0.
req_limit = limit == 0 ? 1 : limit
entry = Gitlab::GitalyClient::CommitService.new(repository).tree_entry(sha, path, req_limit)
2017-09-10 17:25:29 +05:30
return unless entry
2018-03-17 18:26:18 +05:30
entry.data = "" if limit == 0
2017-09-10 17:25:29 +05:30
case entry.type
when :COMMIT
new(
id: entry.oid,
name: name,
size: 0,
data: '',
path: path,
commit_id: sha
)
when :BLOB
new(
id: entry.oid,
name: name,
size: entry.size,
data: entry.data.dup,
mode: entry.mode.to_s(8),
path: path,
commit_id: sha,
binary: binary?(entry.data)
)
end
end
def find_by_rugged(repository, sha, path, limit:)
2018-03-17 18:26:18 +05:30
return unless path
2018-10-15 14:42:47 +05:30
# Strip any leading / characters from the path
path = path.sub(%r{\A/*}, '')
2018-03-17 18:26:18 +05:30
rugged_commit = repository.lookup(sha)
root_tree = rugged_commit.tree
2017-09-10 17:25:29 +05:30
2018-10-15 14:42:47 +05:30
blob_entry = find_entry_by_path(repository, root_tree.oid, *path.split('/'))
2017-09-10 17:25:29 +05:30
return nil unless blob_entry
if blob_entry[:type] == :commit
submodule_blob(blob_entry, path, sha)
else
blob = repository.lookup(blob_entry[:oid])
if blob
new(
id: blob.oid,
name: blob_entry[:name],
size: blob.size,
# Rugged::Blob#content is expensive; don't call it if we don't have to.
data: limit.zero? ? '' : blob.content(limit),
mode: blob_entry[:filemode].to_s(8),
path: path,
commit_id: sha,
binary: blob.binary?
)
end
end
2018-03-17 18:26:18 +05:30
rescue Rugged::ReferenceError
nil
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])
2017-09-10 17:25:29 +05:30
end
2017-08-17 22:00:37 +05:30
end
def initialize(options)
%w(id name path size data mode commit_id binary).each do |key|
2018-03-17 18:26:18 +05:30
self.__send__("#{key}=", options[key.to_sym]) # rubocop:disable GitlabSecurity/PublicSend
2017-08-17 22:00:37 +05:30
end
# Retain the actual size before it is encoded
@loaded_size = @data.bytesize if @data
2018-03-27 19:54:05 +05:30
@loaded_all_data = @loaded_size == size
2017-08-17 22:00:37 +05:30
end
def binary?
@binary.nil? ? super : @binary == true
end
def data
encode! @data
end
# Load all blob data (not just the first MAX_DATA_DISPLAY_SIZE bytes) into
# memory as a Ruby string.
def load_all_data!(repository)
return if @data == '' # don't mess with submodule blobs
2018-03-27 19:54:05 +05:30
# Even if we return early, recalculate wether this blob is binary in
# case a blob was initialized as text but the full data isn't
@binary = nil
return if @loaded_all_data
@data = Gitlab::GitalyClient.migrate(:git_blob_load_all_data) do |is_enabled|
begin
2017-09-10 17:25:29 +05:30
if is_enabled
2018-03-17 18:26:18 +05:30
repository.gitaly_blob_client.get_blob(oid: id, limit: -1).data
2017-09-10 17:25:29 +05:30
else
repository.lookup(id).content
end
end
end
2017-08-17 22:00:37 +05:30
@loaded_all_data = true
@loaded_size = @data.bytesize
end
def name
encode! @name
end
2017-09-10 17:25:29 +05:30
def path
encode! @path
end
2017-08-17 22:00:37 +05:30
def truncated?
size && (size > loaded_size)
end
# Valid LFS object pointer is a text file consisting of
# version
# oid
# size
# see https://github.com/github/git-lfs/blob/v1.1.0/docs/spec.md#the-pointer
def lfs_pointer?
2018-03-17 18:26:18 +05:30
self.class.size_could_be_lfs?(size) && has_lfs_version_key? && lfs_oid.present? && lfs_size.present?
2017-08-17 22:00:37 +05:30
end
def lfs_oid
if has_lfs_version_key?
oid = data.match(/(?<=sha256:)([0-9a-f]{64})/)
return oid[1] if oid
end
nil
end
def lfs_size
if has_lfs_version_key?
size = data.match(/(?<=size )([0-9]+)/)
return size[1].to_i if size
end
nil
end
def external_storage
return unless lfs_pointer?
:lfs
end
alias_method :external_size, :lfs_size
private
def has_lfs_version_key?
!empty? && text? && data.start_with?("version https://git-lfs.github.com/spec")
end
end
end
end