14482: Allow unescaped " " on stream and file token regexes (WIP)

Arvados-DCO-1.1-Signed-off-by: Lucas Di Pentima <ldipentima@veritasgenetics.com>

diff --git a/sdk/ruby/lib/arvados/keep.rb b/sdk/ruby/lib/arvados/keep.rb
index d271610339f98aef135b996b6512e13579473a77..b8abf0f7cceb939c775cef64dd09ba59318bd81f 100644 (file)
--- a/sdk/ruby/lib/arvados/keep.rb
+++ b/sdk/ruby/lib/arvados/keep.rb
@@ -101,8 +101,8 @@ module Keep
   end
 
   class Manifest
   end
 
   class Manifest

-    STRICT_STREAM_TOKEN_REGEXP = /^(\.)(\/[^\/\s]+)*$/
-    STRICT_FILE_TOKEN_REGEXP = /^[[:digit:]]+:[[:digit:]]+:([^\s\/]+(\/[^\s\/]+)*)$/
+    STRICT_STREAM_TOKEN_REGEXP = /^(\.)(\/[^\/\t\v\n\r]+)*$/
+    STRICT_FILE_TOKEN_REGEXP = /^[[:digit:]]+:[[:digit:]]+:([^\t\v\n\r\/]+(\/[^\t\v\n\r\/]+)*)$/

     EMPTY_DOT_FILE_TOKEN_REGEXP = /^0:0:\.$/
 
     # Class to parse a manifest text and provide common views of that data.
     EMPTY_DOT_FILE_TOKEN_REGEXP = /^0:0:\.$/
 
     # Class to parse a manifest text and provide common views of that data.


@@ -132,15 +132,14 @@ module Keep
       end
     end
 
       end
     end
 

-    def self.unescape(s, except=[])
+    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 |_|
       return nil if s.nil?
 
       # Parse backslash escapes in a Keep manifest stream or file name.
       s.gsub(/\\(\\|[0-7]{3})/) do |_|

-        if $1 == '\\'
+        case $1
+        when '\\'

           '\\'
           '\\'

-        elsif except.include? $1
-          $1

         else
           $1.to_i(8).chr
         end
         else
           $1.to_i(8).chr
         end


@@ -261,7 +260,7 @@ module Keep
         count = 0
 
         word = words.shift
         count = 0
 
         word = words.shift

-        unescaped_word = unescape(word, except=["040"])
+        unescaped_word = unescape(word)

         count += 1 if unescaped_word =~ STRICT_STREAM_TOKEN_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 += 1 if unescaped_word =~ STRICT_STREAM_TOKEN_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
 


@@ -275,7 +274,7 @@ module Keep
 
         count = 0
         while unescape(word) =~ EMPTY_DOT_FILE_TOKEN_REGEXP or
 
         count = 0
         while unescape(word) =~ EMPTY_DOT_FILE_TOKEN_REGEXP or

-          (unescape(word, except=["040"]) =~ STRICT_FILE_TOKEN_REGEXP and ($~[1].split('/') & ['..', '.']).empty?)
+          (unescape(word) =~ STRICT_FILE_TOKEN_REGEXP and ($~[1].split('/') & ['..', '.']).empty?)

           word = words.shift
           count += 1
         end
           word = words.shift
           count += 1
         end

diff --git a/sdk/ruby/test/test_keep_manifest.rb b/sdk/ruby/test/test_keep_manifest.rb
index f29e978ab84cd230feea106a165272926372ee8e..637c998bd532f88445d418b7cf3b4780cedc4d39 100644 (file)
--- a/sdk/ruby/test/test_keep_manifest.rb
+++ b/sdk/ruby/test/test_keep_manifest.rb
@@ -448,7 +448,8 @@ class ManifestTest < Minitest::Test
     [false, ". d41d8cd98f00b204e9800998ecf8427e+0 0:0:\\057foo/bar\n",
       "Manifest invalid for stream 1: invalid file token \"0:0:\\\\057foo/bar\""],
     [true, ". d41d8cd98f00b204e9800998ecf8427e+0 0:0:\\134057foo/bar\n"],
     [false, ". d41d8cd98f00b204e9800998ecf8427e+0 0:0:\\057foo/bar\n",
       "Manifest invalid for stream 1: invalid file token \"0:0:\\\\057foo/bar\""],
     [true, ". d41d8cd98f00b204e9800998ecf8427e+0 0:0:\\134057foo/bar\n"],

-    [false, ". d41d8cd98f00b204e9800998ecf8427e+0 \\040:\\040:foo.txt\n"],
+    [false, ". d41d8cd98f00b204e9800998ecf8427e+0 \\040:\\040:foo.txt\n",
+      "Manifest invalid for stream 1: invalid file token \"\\\\040:\\\\040:foo.txt\""],

   ].each do |ok, manifest, expected_error=nil|
     define_method "test_validate manifest #{manifest.inspect}" do
       assert_equal ok, Keep::Manifest.valid?(manifest)
   ].each do |ok, manifest, expected_error=nil|
     define_method "test_validate manifest #{manifest.inspect}" do
       assert_equal ok, Keep::Manifest.valid?(manifest)