arvados 1.3.0.20190122164002 → 1.3.1.20181129194931

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA1:
-  metadata.gz: 10109a46431c756a3146c97e6e7d64d8dbf72ef8
-  data.tar.gz: f52dd28f113f69ea56b53984e24f02f9b32656ac
+  metadata.gz: ebedc440d70c4a447080e1e5c0c03e85e51b95c6
+  data.tar.gz: 684aa1208ea94e66445f6ef983b3de54162f7a32
 SHA512:
-  metadata.gz: 2d6cba2cd3e8ba606bbc5ce28695bb8129c2367ebdfa719257959fce83adaf967b694289383b2ee0ec24491df7199996906671eb3b5417dc68546792dfc4644c
-  data.tar.gz: d25379406f3ae7199598b87093afa2527387e90a082c564a93ecaec8f77e4d05177cc3c074369b26464e36ef40c0b8a2e46d309126fd7c239756135b0a550e7e
+  metadata.gz: 9eaace22f022dcd2c4b0b69785aa74cdb786eb93a93ca78c896dfec9b205f757ff96ebea3a3d9fcba3d299a50b6ce44bbcdb63f1782da1e22d98cf4dfaf76d40
+  data.tar.gz: 87921a34e55707b318e1ffbc76dd2dc54da80f8060c26d9a6194cc95d9bca631c1470a87857647c68aa70b0d8e31d3ea888c60653ec3dcb0e5a4094347a01297

data/lib/arvados/keep.rb

@@ -101,14 +101,8 @@ module Keep
   end
 
   class Manifest
-    STREAM_TOKEN_REGEXP = /^([^\000-\040\\]|\\[0-3][0-7][0-7])+$/
-    STREAM_NAME_REGEXP = /^(\.)(\/[^\/]+)*$/
-
-    EMPTY_DIR_TOKEN_REGEXP = /^0:0:\.$/ # The exception when a file can have '.' as a name
-    FILE_TOKEN_REGEXP = /^[[:digit:]]+:[[:digit:]]+:([^\000-\040\\]|\\[0-3][0-7][0-7])+$/
-    FILE_NAME_REGEXP = /^[[:digit:]]+:[[:digit:]]+:([^\/]+(\/[^\/]+)*)$/
-
-    NON_8BIT_ENCODED_CHAR = /[^\\]\\[4-7][0-7][0-7]/
+    STRICT_STREAM_TOKEN_REGEXP = /^(\.)(\/[^\/\s]+)*$/
+    STRICT_FILE_TOKEN_REGEXP = /^[[:digit:]]+:[[:digit:]]+:([^\s\/]+(\/[^\s\/]+)*)$/
 
     # Class to parse a manifest text and provide common views of that data.
     def initialize(manifest_text)
@@ -137,9 +131,7 @@ module Keep
       end
     end
 
-    def self.unescape(s)
-      return nil if s.nil?
-
+    def unescape(s)
       # Parse backslash escapes in a Keep manifest stream or file name.
       s.gsub(/\\(\\|[0-7]{3})/) do |_|
         case $1
@@ -151,10 +143,6 @@ module Keep
       end
     end
 
-    def unescape(s)
-      self.class.unescape(s)
-    end
-
     def split_file_token token
       start_pos, filesize, filename = token.split(':', 3)
       if filename.nil?
@@ -174,15 +162,15 @@ module Keep
           elsif in_file_tokens or not Locator.valid? token
             in_file_tokens = true
 
-            start_pos, file_size, file_name = split_file_token(token)
+            file_tokens = split_file_token(token)
             stream_name_adjuster = ''
-            if file_name.include?('/')                # '/' in filename
-              dirname, sep, basename = file_name.rpartition('/')
-              stream_name_adjuster = sep + dirname   # /dir_parts
-              file_name = basename
+            if file_tokens[2].include?('/')                # '/' in filename
+              parts = file_tokens[2].rpartition('/')
+              stream_name_adjuster = parts[1] + parts[0]   # /dir_parts
+              file_tokens[2] = parts[2]
             end
 
-            yield [stream_name + stream_name_adjuster, start_pos, file_size, file_name]
+            yield [stream_name + stream_name_adjuster] + file_tokens
           end
         end
       end
@@ -209,13 +197,10 @@ module Keep
       # files.  This can help you avoid parsing the entire manifest if you
       # just want to check if a small number of files are specified.
       if stop_after.nil? or not @files.nil?
-        # Avoid counting empty dir placeholders
-        return files.reject{|_, name, size| name == '.' and size == 0}.size
+        return files.size
       end
       seen_files = {}
-      each_file_spec do |streamname, _, filesize, filename|
-        # Avoid counting empty dir placeholders
-        next if filename == "." and filesize == 0
+      each_file_spec do |streamname, _, _, filename|
         seen_files[[streamname, filename]] = true
         return stop_after if (seen_files.size >= stop_after)
       end
@@ -265,9 +250,7 @@ module Keep
         count = 0
 
         word = words.shift
-        raise ArgumentError.new "Manifest invalid for stream #{line_count}: >8-bit encoded chars not allowed on stream token #{word.inspect}" if word =~ NON_8BIT_ENCODED_CHAR
-        unescaped_word = unescape(word)
-        count += 1 if word =~ STREAM_TOKEN_REGEXP and unescaped_word =~ STREAM_NAME_REGEXP and unescaped_word !~ /\/\.\.?(\/|$)/
+        count += 1 if word =~ STRICT_STREAM_TOKEN_REGEXP and word !~ /\/\.\.?(\/|$)/
         raise ArgumentError.new "Manifest invalid for stream #{line_count}: missing or invalid stream name #{word.inspect if word}" if count != 1
 
         count = 0
@@ -279,9 +262,7 @@ module Keep
         raise ArgumentError.new "Manifest invalid for stream #{line_count}: missing or invalid locator #{word.inspect if word}" if count == 0
 
         count = 0
-        raise ArgumentError.new "Manifest invalid for stream #{line_count}: >8-bit encoded chars not allowed on file token #{word.inspect}" if word =~ NON_8BIT_ENCODED_CHAR
-        while unescape(word) =~ EMPTY_DIR_TOKEN_REGEXP or
-          (word =~ FILE_TOKEN_REGEXP and unescape(word) =~ FILE_NAME_REGEXP and ($~[1].split('/') & ['..', '.']).empty?)
+        while word =~ STRICT_FILE_TOKEN_REGEXP and ($~[1].split('/') & ['..','.']).empty?
           word = words.shift
           count += 1
         end

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: arvados
 version: !ruby/object:Gem::Version
-  version: 1.3.0.20190122164002
+  version: 1.3.1.20181129194931
 platform: ruby
 authors:
 - Arvados Authors
 autorequire: 
 bindir: bin
 cert_chain: []
-date: 2019-01-22 00:00:00.000000000 Z
+date: 2018-11-29 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: activesupport
@@ -118,7 +118,7 @@ dependencies:
     - - ">="
       - !ruby/object:Gem::Version
         version: 0.1.5
-description: Arvados client library, git commit 7337b18bf7b6996a7fe4df0aba5356a03bda452d
+description: Arvados client library, git commit 0d22de74ea069003b49a9d52878d5f1ac04d71bb
 email: gem-dev@curoverse.com
 executables: []
 extensions: []