diff options
Diffstat (limited to 'lib/gitlab/git')
| -rw-r--r-- | lib/gitlab/git/blob.rb | 45 | ||||
| -rw-r--r-- | lib/gitlab/git/commit.rb | 2 | ||||
| -rw-r--r-- | lib/gitlab/git/diff.rb | 17 | ||||
| -rw-r--r-- | lib/gitlab/git/gitmodules_parser.rb | 77 | ||||
| -rw-r--r-- | lib/gitlab/git/repository.rb | 52 |
5 files changed, 145 insertions, 48 deletions
diff --git a/lib/gitlab/git/blob.rb b/lib/gitlab/git/blob.rb index 33a7624e303..a7aceab4c14 100644 --- a/lib/gitlab/git/blob.rb +++ b/lib/gitlab/git/blob.rb @@ -14,6 +14,51 @@ module Gitlab class << self def find(repository, sha, path) + 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) + end + end + end + + def find_by_gitaly(repository, sha, path) + path = path.sub(/\A\/*/, '') + path = '/' if path.empty? + name = File.basename(path) + entry = Gitlab::GitalyClient::Commit.new(repository).tree_entry(sha, path, MAX_DATA_DISPLAY_SIZE) + return unless entry + + case entry.type + when :COMMIT + new( + id: entry.oid, + name: name, + size: 0, + data: '', + path: path, + commit_id: sha + ) + when :BLOB + # EncodingDetector checks the first 1024 * 1024 bytes for NUL byte, libgit2 checks + # only the first 8000 (https://github.com/libgit2/libgit2/blob/2ed855a9e8f9af211e7274021c2264e600c0f86b/src/filter.h#L15), + # which is what we use below to keep a consistent behavior. + detect = CharlockHolmes::EncodingDetector.new(8000).detect(entry.data) + 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: detect && detect[:type] == :binary + ) + end + end + + def find_by_rugged(repository, sha, path) commit = repository.lookup(sha) root_tree = commit.tree diff --git a/lib/gitlab/git/commit.rb b/lib/gitlab/git/commit.rb index bb04731f08c..d5d149f1423 100644 --- a/lib/gitlab/git/commit.rb +++ b/lib/gitlab/git/commit.rb @@ -4,7 +4,7 @@ module Gitlab class Commit include Gitlab::EncodingHelper - attr_accessor :raw_commit, :head, :refs + attr_accessor :raw_commit, :head SERIALIZE_KEYS = [ :id, :message, :parent_ids, diff --git a/lib/gitlab/git/diff.rb b/lib/gitlab/git/diff.rb index 4b689f0e94f..f825568f194 100644 --- a/lib/gitlab/git/diff.rb +++ b/lib/gitlab/git/diff.rb @@ -16,11 +16,11 @@ module Gitlab alias_method :renamed_file?, :renamed_file attr_accessor :expanded + attr_writer :too_large alias_method :expanded?, :expanded - # We need this accessor because of `to_hash` and `init_from_hash` - attr_accessor :too_large + SERIALIZE_KEYS = %i(diff new_path old_path a_mode b_mode new_file renamed_file deleted_file too_large).freeze class << self # The maximum size of a diff to display. @@ -231,16 +231,10 @@ module Gitlab end end - def serialize_keys - @serialize_keys ||= %i(diff new_path old_path a_mode b_mode new_file renamed_file deleted_file too_large) - end - def to_hash hash = {} - keys = serialize_keys - - keys.each do |key| + SERIALIZE_KEYS.each do |key| hash[key] = send(key) end @@ -267,6 +261,9 @@ module Gitlab end end + # This is used by `to_hash` and `init_from_hash`. + alias_method :too_large, :too_large? + def too_large! @diff = '' @line_count = 0 @@ -315,7 +312,7 @@ module Gitlab def init_from_hash(hash) raw_diff = hash.symbolize_keys - serialize_keys.each do |key| + SERIALIZE_KEYS.each do |key| send(:"#{key}=", raw_diff[key.to_sym]) end end diff --git a/lib/gitlab/git/gitmodules_parser.rb b/lib/gitlab/git/gitmodules_parser.rb new file mode 100644 index 00000000000..f4e3b5e5129 --- /dev/null +++ b/lib/gitlab/git/gitmodules_parser.rb @@ -0,0 +1,77 @@ +module Gitlab + module Git + class GitmodulesParser + def initialize(content) + @content = content + end + + # Parses the contents of a .gitmodules file and returns a hash of + # submodule information, indexed by path. + def parse + reindex_by_path(get_submodules_by_name) + end + + private + + class State + def initialize + @result = {} + @current_submodule = nil + end + + def start_section(section) + # In some .gitmodules files (e.g. nodegit's), a header + # with the same name appears multiple times; we want to + # accumulate the configs across these + @current_submodule = @result[section] || { 'name' => section } + @result[section] = @current_submodule + end + + def set_attribute(attr, value) + @current_submodule[attr] = value + end + + def section_started? + !@current_submodule.nil? + end + + def submodules_by_name + @result + end + end + + def get_submodules_by_name + iterator = State.new + + @content.split("\n").each_with_object(iterator) do |text, iterator| + next if text =~ /^\s*#/ + + if text =~ /\A\[submodule "(?<name>[^"]+)"\]\z/ + iterator.start_section($~[:name]) + else + next unless iterator.section_started? + + next unless text =~ /\A\s*(?<key>\w+)\s*=\s*(?<value>.*)\z/ + + value = $~[:value].chomp + iterator.set_attribute($~[:key], value) + end + end + + iterator.submodules_by_name + end + + def reindex_by_path(submodules_by_name) + # Convert from an indexed by name to an array indexed by path + # If a submodule doesn't have a path, it is considered bogus + # and is ignored + submodules_by_name.each_with_object({}) do |(name, data), results| + path = data.delete 'path' + next unless path + + results[path] = data + end + end + end + end +end diff --git a/lib/gitlab/git/repository.rb b/lib/gitlab/git/repository.rb index 85695d0a4df..c1f942f931a 100644 --- a/lib/gitlab/git/repository.rb +++ b/lib/gitlab/git/repository.rb @@ -617,9 +617,9 @@ module Gitlab # # Ex. # { - # "rack" => { + # "current_path/rack" => { + # "name" => "original_path/rack", # "id" => "c67be4624545b4263184c4a0e8f887efd0a66320", - # "path" => "rack", # "url" => "git://github.com/chneukirchen/rack.git" # }, # "encoding" => { @@ -637,7 +637,8 @@ module Gitlab return {} end - parse_gitmodules(commit, content) + parser = GitmodulesParser.new(content) + fill_submodule_ids(commit, parser.parse) end # Return total commits count accessible from passed ref @@ -998,42 +999,19 @@ module Gitlab end end - # Parses the contents of a .gitmodules file and returns a hash of - # submodule information. - def parse_gitmodules(commit, content) - modules = {} - - name = nil - content.each_line do |line| - case line.strip - when /\A\[submodule "(?<name>[^"]+)"\]\z/ # Submodule header - name = $~[:name] - modules[name] = {} - when /\A(?<key>\w+)\s*=\s*(?<value>.*)\z/ # Key/value pair - key = $~[:key] - value = $~[:value].chomp - - next unless name && modules[name] - - modules[name][key] = value - - if key == 'path' - begin - modules[name]['id'] = blob_content(commit, value) - rescue InvalidBlobName - # The current entry is invalid - modules.delete(name) - name = nil - end - end - when /\A#/ # Comment - next - else # Invalid line - name = nil + # Fill in the 'id' field of a submodule hash from its values + # as-of +commit+. Return a Hash consisting only of entries + # from the submodule hash for which the 'id' field is filled. + def fill_submodule_ids(commit, submodule_data) + submodule_data.each do |path, data| + id = begin + blob_content(commit, path) + rescue InvalidBlobName + nil end + data['id'] = id end - - modules + submodule_data.select { |path, data| data['id'] } end # Returns true if +commit+ introduced changes to +path+, using commit |