+require 'arvados/keep'
+
class Collection < ArvadosModel
include HasUuid
include KindAndEtag
include CommonApiTemplate
+ before_validation :check_encoding
before_validation :check_signatures
before_validation :strip_manifest_text
before_validation :set_portable_data_hash
validate :ensure_hash_matches_manifest_text
+ before_save :set_file_names
+
+ # Query only undeleted collections by default.
+ default_scope where("expires_at IS NULL or expires_at > CURRENT_TIMESTAMP")
api_accessible :user, extend: :common do |t|
- t.add :data_size
- t.add :files
t.add :name
t.add :description
t.add :properties
t.add :portable_data_hash
- t.add :manifest_text
+ t.add :signed_manifest_text, as: :manifest_text
end
def self.attributes_required_columns
- super.merge({ "data_size" => ["manifest_text"],
- "files" => ["manifest_text"],
- })
+ # If we don't list this explicitly, the params[:select] code gets
+ # confused by the way we expose signed_manifest_text as
+ # manifest_text in the API response, and never let clients select
+ # the manifest_text column.
+ super.merge('manifest_text' => ['manifest_text'])
end
def check_signatures
return true if current_user.andand.is_admin
+ # Provided the manifest_text hasn't changed materially since an
+ # earlier validation, it's safe to pass this validation on
+ # subsequent passes without checking any signatures. This is
+ # important because the signatures have probably been stripped off
+ # by the time we get to a second validation pass!
+ return true if @signatures_checked and @signatures_checked == compute_pdh
+
if self.manifest_text_changed?
# Check permissions on the collection manifest.
# If any signature cannot be verified, raise PermissionDeniedError
# filenames. Safety first!
elsif Blob.verify_signature tok, signing_opts
# OK.
- elsif Locator.parse(tok).andand.signature
+ elsif Keep::Locator.parse(tok).andand.signature
# Signature provided, but verify_signature did not like it.
logger.warn "Invalid signature on locator #{tok}"
raise ArvadosModel::PermissionDeniedError
end
end
end
- true
+ @signatures_checked = compute_pdh
end
def strip_manifest_text
if self.manifest_text_changed?
# Remove any permission signatures from the manifest.
- Collection.munge_manifest_locators(self[:manifest_text]) do |loc|
+ self.class.munge_manifest_locators!(self[:manifest_text]) do |loc|
loc.without_signature.to_s
end
end
end
def set_portable_data_hash
- if (self.portable_data_hash.nil? or (self.portable_data_hash == "") or (manifest_text_changed? and !portable_data_hash_changed?))
- self.portable_data_hash = "#{Digest::MD5.hexdigest(manifest_text)}+#{manifest_text.length}"
+ if (portable_data_hash.nil? or
+ portable_data_hash == "" or
+ (manifest_text_changed? and !portable_data_hash_changed?))
+ @need_pdh_validation = false
+ self.portable_data_hash = compute_pdh
elsif portable_data_hash_changed?
+ @need_pdh_validation = true
begin
- loc = Locator.parse!(self.portable_data_hash)
+ loc = Keep::Locator.parse!(self.portable_data_hash)
loc.strip_hints!
- self.portable_data_hash = loc.to_s
+ 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
end
def ensure_hash_matches_manifest_text
- if manifest_text_changed? or portable_data_hash_changed?
- computed_hash = "#{Digest::MD5.hexdigest(manifest_text)}+#{manifest_text.length}"
- unless computed_hash == portable_data_hash
- logger.debug "(computed) '#{computed_hash}' != '#{portable_data_hash}' (provided)"
- errors.add(:portable_data_hash, "does not match hash of manifest_text")
- return false
- end
+ return true unless manifest_text_changed? or portable_data_hash_changed?
+ # No need verify it if :set_portable_data_hash just computed it!
+ return true if not @need_pdh_validation
+ expect_pdh = compute_pdh
+ if expect_pdh != portable_data_hash
+ errors.add(:portable_data_hash,
+ "does not match computed hash #{expect_pdh}")
+ return false
+ end
+ end
+
+ def set_file_names
+ if self.manifest_text_changed?
+ self.file_names = manifest_files
end
true
end
+ def manifest_files
+ names = ''
+ if self.manifest_text
+ self.manifest_text.scan(/ \d+:\d+:(\S+)/) do |name|
+ names << name.first.gsub('\040',' ') + "\n"
+ break if names.length > 2**13
+ end
+ end
+
+ if self.manifest_text and names.length < 2**13
+ self.manifest_text.scan(/^\.\/(\S+)/m) do |stream_name|
+ names << stream_name.first.gsub('\040',' ') + "\n"
+ break if names.length > 2**13
+ end
+ end
+
+ names[0,2**13]
+ end
+
+ def check_encoding
+ if manifest_text.encoding.name == 'UTF-8' and manifest_text.valid_encoding?
+ true
+ else
+ begin
+ # If Ruby thinks the encoding is something else, like 7-bit
+ # ASCII, but its stored bytes are equal to the (valid) UTF-8
+ # encoding of the same string, we declare it to be a UTF-8
+ # string.
+ utf8 = manifest_text
+ utf8.force_encoding Encoding::UTF_8
+ if utf8.valid_encoding? and utf8 == manifest_text.encode(Encoding::UTF_8)
+ manifest_text = utf8
+ return true
+ end
+ rescue
+ end
+ errors.add :manifest_text, "must use UTF-8 encoding"
+ false
+ end
+ end
+
def redundancy_status
if redundancy_confirmed_as.nil?
'unconfirmed'
end
end
- def data_size
- inspect_manifest_text if @data_size.nil? or manifest_text_changed?
- @data_size
- end
-
- def files
- inspect_manifest_text if @files.nil? or manifest_text_changed?
- @files
- end
-
- def inspect_manifest_text
- if !manifest_text
- @data_size = false
- @files = []
- return
- end
-
- @data_size = 0
- tmp = {}
-
- manifest_text.split("\n").each do |stream|
- toks = stream.split(" ")
-
- stream = toks[0].gsub /\\(\\|[0-7]{3})/ do |escape_sequence|
- case $1
- when '\\' '\\'
- else $1.to_i(8).chr
- end
- end
-
- toks[1..-1].each do |tok|
- if (re = tok.match /^[0-9a-f]{32}/)
- blocksize = nil
- tok.split('+')[1..-1].each do |hint|
- if !blocksize and hint.match /^\d+$/
- blocksize = hint.to_i
- end
- if (re = hint.match /^GS(\d+)$/)
- blocksize = re[1].to_i
- end
- end
- @data_size = false if !blocksize
- @data_size += blocksize if @data_size
- else
- if (re = tok.match /^(\d+):(\d+):(\S+)$/)
- filename = re[3].gsub /\\(\\|[0-7]{3})/ do |escape_sequence|
- case $1
- when '\\' '\\'
- else $1.to_i(8).chr
- end
- end
- fn = stream + '/' + filename
- i = re[2].to_i
- if tmp[fn]
- tmp[fn] += i
- else
- tmp[fn] = i
- end
- end
- end
- end
+ def signed_manifest_text
+ if has_attribute? :manifest_text
+ token = current_api_client_authorization.andand.api_token
+ @signed_manifest_text = self.class.sign_manifest manifest_text, token
end
+ end
- @files = []
- tmp.each do |k, v|
- re = k.match(/^(.+)\/(.+)/)
- @files << [re[1], re[2], v]
+ def self.sign_manifest manifest, token
+ signing_opts = {
+ key: Rails.configuration.blob_signing_key,
+ api_token: token,
+ ttl: Rails.configuration.blob_signing_ttl,
+ }
+ m = manifest.dup
+ munge_manifest_locators!(m) do |loc|
+ Blob.sign_locator(loc.to_s, signing_opts)
end
+ return m
end
- def self.munge_manifest_locators(manifest)
+ def self.munge_manifest_locators! manifest
# Given a manifest text and a block, yield each locator,
# and replace it with whatever the block returns.
manifest.andand.gsub!(/ [[:xdigit:]]{32}(\+[[:digit:]]+)?(\+\S+)/) do |word|
- if loc = Locator.parse(word.strip)
+ if loc = Keep::Locator.parse(word.strip)
" " + yield(loc)
else
" " + word
# If the search term is a Collection locator that contains one file
# that looks like a Docker image, return it.
- if loc = Locator.parse(search_term)
+ if loc = Keep::Locator.parse(search_term)
loc.strip_hints!
coll_match = readable_by(*readers).where(portable_data_hash: loc.to_s).limit(1).first
- if coll_match and (coll_match.files.size == 1) and
- (coll_match.files[0][1] =~ /^[0-9A-Fa-f]{64}\.tar$/)
- return [coll_match]
+ if coll_match
+ # Check if the Collection contains exactly one file whose name
+ # looks like a saved Docker image.
+ manifest = Keep::Manifest.new(coll_match.manifest_text)
+ if manifest.exact_file_count?(1) and
+ (manifest.files[0][1] =~ /^[0-9A-Fa-f]{64}\.tar$/)
+ return [coll_match]
+ end
end
end
def self.for_latest_docker_image(search_term, search_tag=nil, readers=nil)
find_all_for_docker_image(search_term, search_tag, readers).first
end
+
+ def self.searchable_columns operator
+ super - ["manifest_text"]
+ end
+
+ protected
+ def portable_manifest_text
+ portable_manifest = self[:manifest_text].dup
+ self.class.munge_manifest_locators!(portable_manifest) do |loc|
+ loc.hash + '+' + loc.size.to_s
+ end
+ portable_manifest
+ end
+
+ def compute_pdh
+ portable_manifest = portable_manifest_text
+ (Digest::MD5.hexdigest(portable_manifest) +
+ '+' +
+ portable_manifest.bytesize.to_s)
+ end
end
projects / arvados.git / blobdiff