arvados 1.3.0.20181129194931 → 1.3.0.20190122164002

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA1:
-  metadata.gz: 4f6f937cde735c9c26cb22108d85f9ed128a945e
-  data.tar.gz: e3a07b466fe85bb5bb303b84ad77734babbab905
+  metadata.gz: 10109a46431c756a3146c97e6e7d64d8dbf72ef8
+  data.tar.gz: f52dd28f113f69ea56b53984e24f02f9b32656ac
 SHA512:
-  metadata.gz: 29803b2a219c13794b5b9f93705aa2115359a1d6e386bcd00bd64a6ebdf4e4be84ff1465a47bdfae735f3d7eda1adc9b4772646bd32e3e01f1a1b2f9087fdfa6
-  data.tar.gz: bbaf7d07f72c92499c0c000160c45508d19df285536fea41a54e230f0c2bd95b39c73c948808048d8be35a905440892ef4bd4c666c7b069f3512eea0b360f867
+  metadata.gz: 2d6cba2cd3e8ba606bbc5ce28695bb8129c2367ebdfa719257959fce83adaf967b694289383b2ee0ec24491df7199996906671eb3b5417dc68546792dfc4644c
+  data.tar.gz: d25379406f3ae7199598b87093afa2527387e90a082c564a93ecaec8f77e4d05177cc3c074369b26464e36ef40c0b8a2e46d309126fd7c239756135b0a550e7e

data/lib/arvados/keep.rb

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

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: arvados
 version: !ruby/object:Gem::Version
-  version: 1.3.0.20181129194931
+  version: 1.3.0.20190122164002
 platform: ruby
 authors:
 - Arvados Authors
 autorequire: 
 bindir: bin
 cert_chain: []
-date: 2018-11-29 00:00:00.000000000 Z
+date: 2019-01-22 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 0d22de74ea069003b49a9d52878d5f1ac04d71bb
+description: Arvados client library, git commit 7337b18bf7b6996a7fe4df0aba5356a03bda452d
 email: gem-dev@curoverse.com
 executables: []
 extensions: []