Unescape codepoints in input rather than on terminals.

Automatically skip missing test manifests.

Author: gkellogg

lib/sparql/grammar/parser.rb

@@ -41,7 +41,7 @@ class Parser
     terminal(:BLANK_NODE_LABEL,     BLANK_NODE_LABEL) do |value, prod|
       bnode(value[2..-1])
     end
-    terminal(:IRIREF,               IRIREF, unescape: true) do |value, prod|
+    terminal(:IRIREF,               IRIREF) do |value, prod|
       begin
         iri(value[1..-2])
       rescue ArgumentError => e
@@ -2666,8 +2666,8 @@ class Parser
     # @return [SPARQL::Grammar::Parser]
     def initialize(input = nil, **options, &block)
       @input = case input
-      when IO, StringIO then input.read
-      else input.to_s.dup
+      when IO, StringIO then EBNF::Unescape.unescape_codepoints(input.read)
+      else EBNF::Unescape.unescape_codepoints(input.to_s)
       end
       @input.encode!(Encoding::UTF_8) if @input.respond_to?(:encode!)
       @options = {anon_base: "b0", validate: false}.merge(options)

lib/sparql/grammar/terminals.rb

@@ -15,8 +15,6 @@ module Terminals
     U_CHARS2         = Regexp.compile("\\u00B7|[\\u0300-\\u036F]|[\\u203F-\\u2040]").freeze
     IRI_RANGE        = Regexp.compile("[[^<>\"{}|^`\\\\]&&[^\\x00-\\x20]]").freeze
 
-    # 26
-    UCHAR                = EBNF::Unescape::UCHAR
     # 170s
     PERCENT              = /%[0-9A-Fa-f]{2}/.freeze
     # 172s
@@ -43,7 +41,7 @@ module Terminals
     # 149
     ECHAR                = /\\[tbnrf\\"']/.freeze
     # 18
-    IRIREF               = /<(?:#{IRI_RANGE}|#{UCHAR})*>/.freeze
+    IRIREF               = /<(?:#{IRI_RANGE})*>/.freeze
     # 129
     PNAME_NS             = /#{PN_PREFIX}?:/.freeze
     # 130
@@ -76,13 +74,13 @@ module Terminals
     # 143
     DOUBLE_NEGATIVE      = /(\-)([0-9]+\.[0-9]*#{EXPONENT}|\.?[0-9]+#{EXPONENT})/.freeze
     # 145
-    STRING_LITERAL1      = /'([^\'\\\n\r]|#{ECHAR}|#{UCHAR})*'/.freeze
+    STRING_LITERAL1      = /'([^\'\\\n\r]|#{ECHAR})*'/.freeze
     # 146
-    STRING_LITERAL2      = /"([^\"\\\n\r]|#{ECHAR}|#{UCHAR})*"/.freeze
+    STRING_LITERAL2      = /"([^\"\\\n\r]|#{ECHAR})*"/.freeze
     # 147
-    STRING_LITERAL_LONG1 = /'''((?:'|'')?(?:[^'\\]|#{ECHAR}|#{UCHAR}))*'''/m.freeze
+    STRING_LITERAL_LONG1 = /'''((?:'|'')?(?:[^'\\]|#{ECHAR}))*'''/m.freeze
     # 148
-    STRING_LITERAL_LONG2 = /"""((?:"|"")?(?:[^"\\]|#{ECHAR}|#{UCHAR}))*"""/m.freeze
+    STRING_LITERAL_LONG2 = /"""((?:"|"")?(?:[^"\\]|#{ECHAR}))*"""/m.freeze
 
     # 151
     WS                   = /(?:\s|(?:#[^\n\r]*))+/m.freeze

script/tc

@@ -47,8 +47,8 @@ def run_tc(tc, **options)
 
   id = tc.attributes['id'].
     sub('http://www.w3.org/2001/sw/DataAccess/tests/', '').
-    sub('http://www.w3.org/2009/sparql/docs/tests/', '').
-    sub('https://w3c.github.io/rdf-star/tests/', '').
+    sub('https://www.w3.org/2009/sparql/docs/tests/', '').
+    sub('https://w3c.github.io/rdf-tests/sparql/', '').
     sub('https://w3c.github.io/sparql-12/tests/', '')
   STDERR.write "run #{id}" unless options[:quiet]
 
@@ -413,6 +413,8 @@ manifests.each do |path|
         run_tc(tc, **options)
       end
     end
+  rescue IOError => e
+    STDERR.puts "skip #{path}: IOError"
   end
 
 STDERR.puts "" if options[:quiet]

spec/suite_helper.rb

@@ -226,7 +226,7 @@ def self.sparql_11_tests
   end
 
   def self.sparql_12_tests
-    %w(grouping syntax-triple-terms-negative syntax-triple-terms-positive).map do |partial|
+    %w(grouping syntax-escaping syntax-triple-terms-negative syntax-triple-terms-positive).map do |partial|
       "#{BASE}sparql12/#{partial}/manifest.ttl"
     end
   end

spec/suite_spec.rb

@@ -278,6 +278,8 @@
         it_behaves_like "SUITE", man.attributes['id'], man.label, man.comment, man.entries
         it_behaves_like "to_sparql", man.attributes['id'], man.label, man.comment, man.entries
       end
+    rescue IOError => e
+      skip(e.message)
     end
   end
 
@@ -287,6 +289,8 @@
         it_behaves_like "SUITE", man.attributes['id'], man.label, man.comment, man.entries
         it_behaves_like "to_sparql", man.attributes['id'], man.label, man.comment, man.entries
       end
+    rescue IOError => e
+      skip(e.message)
     end
   end
 end unless ENV['CI']