Merge branch 'master' into github-3408-production-datamanager

[arvados.git] / services / api / app / models / collection.rb

diff --git a/services/api/app/models/collection.rb b/services/api/app/models/collection.rb
index c4740f96eb4373552ef15130187750fc3b6812ce..cac25d1c537981a433db526f730957f943c0d487 100644 (file)
--- a/services/api/app/models/collection.rb
+++ b/services/api/app/models/collection.rb
@@ -9,6 +9,7 @@ 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
   validate :ensure_pdh_matches_manifest_text
@@ -61,9 +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
@@ -128,8 +131,8 @@ class Collection < ArvadosModel
       true
     elsif portable_data_hash.nil? or not portable_data_hash_changed?
       self.portable_data_hash = computed_pdh
-    elsif portable_data_hash !~ /^[0-9a-f]{32}(\+\d+)?$/
-      errors.add(:portable_data_hash, "is not a valid hash or hash+size")
+    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,
@@ -190,6 +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
@@ -216,13 +228,13 @@ class Collection < ArvadosModel
     return '' if manifest == ''
 
     new_lines = []
-    lines = manifest.split("\n")
     manifest.each_line do |line|
       line.rstrip!
-      words = line.split(' ')
       new_words = []
-      words.each do |word|
-        if match = Keep::Locator::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
@@ -230,17 +242,16 @@ class Collection < ArvadosModel
       end
       new_lines << new_words.join(' ')
     end
-
-    manifest = new_lines.join("\n") + "\n"
+    new_lines.join("\n") + "\n"
   end
 
   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.rstrip!
-      words = line.split(' ')
-      words.each do |word|
+      # 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