X-Git-Url: https://git.arvados.org/arvados.git/blobdiff_plain/d80be38b5be2c607d3aa0855c69b9f4e0c3fbed0..0e57453d2b637a3d105d4e3d67031f3915f9d302:/services/api/app/models/collection.rb diff --git a/services/api/app/models/collection.rb b/services/api/app/models/collection.rb index 4fab364ae4..cac25d1c53 100644 --- a/services/api/app/models/collection.rb +++ b/services/api/app/models/collection.rb @@ -9,9 +9,9 @@ class Collection < ArvadosModel serialize :properties, Hash before_validation :check_encoding + before_validation :log_invalid_manifest_format before_validation :check_signatures before_validation :strip_signatures_and_update_replication_confirmed - before_validation :set_portable_data_hash validate :ensure_pdh_matches_manifest_text before_save :set_file_names @@ -29,8 +29,6 @@ class Collection < ArvadosModel t.add :replication_confirmed_at end - LOCATOR_REGEXP = /^([[:xdigit:]]{32})(\+([[:digit:]]+))?(\+([[:upper:]][[:alnum:]+@_-]*))?$/ - def self.attributes_required_columns super.merge( # If we don't list manifest_text explicitly, the @@ -64,9 +62,11 @@ class Collection < ArvadosModel api_token: api_token, now: db_current_time.to_i, } - self.manifest_text.lines.each do |entry| - entry.split[1..-1].each do |tok| - if tok =~ FILE_TOKEN + self.manifest_text.each_line do |entry| + entry.split.each do |tok| + if tok == '.' or tok.starts_with? './' + # Stream name token. + elsif tok =~ FILE_TOKEN # This is a filename token, not a blob locator. Note that we # keep checking tokens after this, even though manifest # format dictates that all subsequent tokens will also be @@ -95,56 +95,53 @@ class Collection < ArvadosModel def strip_signatures_and_update_replication_confirmed if self.manifest_text_changed? in_old_manifest = {} - self.class.munge_manifest_locators(manifest_text_was) do |match| - in_old_manifest[match[1]] = true + if not self.replication_confirmed.nil? + self.class.each_manifest_locator(manifest_text_was) do |match| + in_old_manifest[match[1]] = true + end end - cleared_replication_confirmed = false - - # Remove any permission signatures from the manifest. - self[:manifest_text] = self.class.munge_manifest_locators(self[:manifest_text]) do |match| - if not cleared_replication_confirmed and not in_old_manifest[match[1]] + stripped_manifest = self.class.munge_manifest_locators(manifest_text) do |match| + if not self.replication_confirmed.nil? and not in_old_manifest[match[1]] + # If the new manifest_text contains locators whose hashes + # weren't in the old manifest_text, storage replication is no + # longer confirmed. self.replication_confirmed_at = nil self.replication_confirmed = nil - cleared_replication_confirmed = true end + + # Return the locator with all permission signatures removed, + # but otherwise intact. match[0].gsub(/\+A[^+]*/, '') end - end - @computed_pdh_for_manifest_text = self[:manifest_text] if @computed_pdh_for_manifest_text - true - end - def set_portable_data_hash - if (portable_data_hash.nil? or - portable_data_hash == "" or - (manifest_text_changed? and !portable_data_hash_changed?)) - self.portable_data_hash = computed_pdh - elsif portable_data_hash_changed? - begin - loc = Keep::Locator.parse!(self.portable_data_hash) - loc.strip_hints! - if loc.size - self.portable_data_hash = loc.to_s - else - self.portable_data_hash = "#{loc.hash}+#{portable_manifest_text.bytesize}" - end - rescue ArgumentError => e - errors.add(:portable_data_hash, "#{e}") - return false + if @computed_pdh_for_manifest_text == manifest_text + # If the cached PDH was valid before stripping, it is still + # valid after stripping. + @computed_pdh_for_manifest_text = stripped_manifest.dup end + + self[:manifest_text] = stripped_manifest end true end def ensure_pdh_matches_manifest_text - return true unless manifest_text_changed? or portable_data_hash_changed? - # No need verify it if :set_portable_data_hash just computed it! - expect_pdh = computed_pdh - if expect_pdh != portable_data_hash + if not manifest_text_changed? and not portable_data_hash_changed? + true + elsif portable_data_hash.nil? or not portable_data_hash_changed? + self.portable_data_hash = computed_pdh + elsif portable_data_hash !~ Keep::Locator::LOCATOR_REGEXP + errors.add(:portable_data_hash, "is not a valid locator") + false + elsif portable_data_hash[0..31] != computed_pdh[0..31] errors.add(:portable_data_hash, - "does not match computed hash #{expect_pdh}") - return false + "does not match computed hash #{computed_pdh}") + false + else + # Ignore the client-provided size part: always store + # computed_pdh in the database. + self.portable_data_hash = computed_pdh end end @@ -196,6 +193,15 @@ class Collection < ArvadosModel end end + def log_invalid_manifest_format + begin + Keep::Manifest.validate! manifest_text if manifest_text + rescue => e + logger.warn e + end + true + end + def signed_manifest_text if has_attribute? :manifest_text token = current_api_client_authorization.andand.api_token @@ -215,18 +221,20 @@ class Collection < ArvadosModel end def self.munge_manifest_locators manifest - # Given a manifest text and a block, yield each locator, - # and replace it with whatever the block returns. + # Given a manifest text and a block, yield the regexp MatchData + # for each locator. Return a new manifest in which each locator + # has been replaced by the block's return value. return nil if !manifest - return '' if !manifest.present? + return '' if manifest == '' new_lines = [] - lines = manifest.split("\n") - lines.each do |line| - words = line.split(' ') + manifest.each_line do |line| + line.rstrip! new_words = [] - words.each do |word| - if match = LOCATOR_REGEXP.match(word) + line.split(' ').each do |word| + if new_words.empty? + new_words << word + elsif match = Keep::Locator::LOCATOR_REGEXP.match(word) new_words << yield(match) else new_words << word @@ -234,8 +242,21 @@ class Collection < ArvadosModel end new_lines << new_words.join(' ') end + new_lines.join("\n") + "\n" + end - manifest = new_lines.join("\n") + "\n" + def self.each_manifest_locator manifest + # Given a manifest text and a block, yield the regexp match object + # for each locator. + manifest.each_line do |line| + # line will have a trailing newline, but the last token is never + # a locator, so it's harmless here. + line.split(' ').each do |word| + if match = Keep::Locator::LOCATOR_REGEXP.match(word) + yield(match) + end + end + end end def self.normalize_uuid uuid @@ -321,14 +342,13 @@ class Collection < ArvadosModel protected def portable_manifest_text - portable_manifest = self.class.munge_manifest_locators(manifest_text) do |match| + self.class.munge_manifest_locators(manifest_text) do |match| if match[2] # size match[1] + match[2] else match[1] end end - portable_manifest end def compute_pdh