Merge pull request #2842 from sparklemotion/2841-fix-html5lib-tests

gumbo: update adjusted foreign attributes
sparklemotion · Apr 10, 2023 · 6f0e3c4 · 6f0e3c4
2 parents d970981 + dc7c034
commit 6f0e3c4
Show file tree

Hide file tree

Showing 5 changed files with 28 additions and 20 deletions.
diff --git a/.gitmodules b/.gitmodules
@@ -1,4 +1,8 @@
+# [submodule "test/html5lib-tests"]
+# 	path = test/html5lib-tests
+# 	url = https://github.com/html5lib/html5lib-tests.git
+# 	branch = master
 [submodule "test/html5lib-tests"]
 	path = test/html5lib-tests
-	url = https://github.com/html5lib/html5lib-tests.git
-	branch = master
+	url = https://github.com/flavorjones/html5lib-tests.git
+	branch = flavorjones-fix-template-2023-04
diff --git a/gumbo-parser/src/foreign_attrs.c b/gumbo-parser/src/foreign_attrs.c
@@ -1,6 +1,6 @@
 /* ANSI-C code produced by gperf version 3.1 */
 /* Command-line: gperf -m100 -n src/foreign_attrs.gperf  */
-/* Computed positions: -k'2,8' */
+/* Computed positions: -k'8-9' */
 /* Filtered by: gperf-filter.sed */
 
 #include "replacement.h"
@@ -29,9 +29,9 @@ hash (register const char *str, register size_t len)
       11, 11, 11, 11, 11, 11, 11, 11, 11, 11,
       11, 11, 11, 11, 11, 11, 11, 11, 11, 11,
       11, 11, 11, 11, 11, 11, 11, 11, 11,  2,
-      11, 10, 11,  9,  7,  6, 11, 11,  1,  0,
-      11,  5, 11, 11,  4, 11, 11, 11, 11, 11,
-      11,  3, 11, 11, 11, 11, 11, 11, 11, 11,
+      11,  1, 11, 10,  4,  4, 11, 11,  3, 11,
+      11,  5,  3, 11,  0, 11,  2, 11, 11, 11,
+      11,  2, 11, 11, 11, 11, 11, 11, 11, 11,
       11, 11, 11, 11, 11, 11, 11, 11, 11, 11,
       11, 11, 11, 11, 11, 11, 11, 11, 11, 11,
       11, 11, 11, 11, 11, 11, 11, 11, 11, 11,
@@ -51,15 +51,14 @@ hash (register const char *str, register size_t len)
   switch (len)
     {
       default:
+        hval += asso_values[(unsigned char)str[8]];
+      /*FALLTHROUGH*/
+      case 8:
         hval += asso_values[(unsigned char)str[7]];
       /*FALLTHROUGH*/
       case 7:
       case 6:
       case 5:
-      case 4:
-      case 3:
-      case 2:
-        hval += asso_values[(unsigned char)str[1]];
         break;
     }
   return hval;
@@ -70,21 +69,21 @@ gumbo_get_foreign_attr_replacement (register const char *str, register size_t le
 {
   static const unsigned char lengthtable[] =
     {
-       5, 11,  9, 13, 10, 10, 10, 11, 10,  8,  8
+       5, 10, 13,  9, 13, 10, 11, 11, 10, 10,  8
     };
   static const ForeignAttrReplacement wordlist[] =
     {
       {"xmlns", "xmlns", GUMBO_ATTR_NAMESPACE_XMLNS},
-      {"xmlns:xlink", "xlink", GUMBO_ATTR_NAMESPACE_XMLNS},
+      {"xlink:href", "href", GUMBO_ATTR_NAMESPACE_XLINK},
+      {"xlink:arcrole", "arcrole", GUMBO_ATTR_NAMESPACE_XLINK},
       {"xml:space", "space", GUMBO_ATTR_NAMESPACE_XML},
       {"xlink:actuate", "actuate", GUMBO_ATTR_NAMESPACE_XLINK},
       {"xlink:type", "type", GUMBO_ATTR_NAMESPACE_XLINK},
-      {"xlink:href", "href", GUMBO_ATTR_NAMESPACE_XLINK},
-      {"xlink:role", "role", GUMBO_ATTR_NAMESPACE_XLINK},
       {"xlink:title", "title", GUMBO_ATTR_NAMESPACE_XLINK},
+      {"xmlns:xlink", "xlink", GUMBO_ATTR_NAMESPACE_XMLNS},
+      {"xlink:role", "role", GUMBO_ATTR_NAMESPACE_XLINK},
       {"xlink:show", "show", GUMBO_ATTR_NAMESPACE_XLINK},
-      {"xml:lang", "lang", GUMBO_ATTR_NAMESPACE_XML},
-      {"xml:base", "base", GUMBO_ATTR_NAMESPACE_XML}
+      {"xml:lang", "lang", GUMBO_ATTR_NAMESPACE_XML}
     };
 
   if (len <= MAX_WORD_LENGTH && len >= MIN_WORD_LENGTH)

diff --git a/gumbo-parser/src/foreign_attrs.gperf b/gumbo-parser/src/foreign_attrs.gperf
@@ -15,12 +15,12 @@ ForeignAttrReplacement;
 
 %%
 "xlink:actuate", "actuate", GUMBO_ATTR_NAMESPACE_XLINK
+"xlink:arcrole", "arcrole", GUMBO_ATTR_NAMESPACE_XLINK
 "xlink:href", "href", GUMBO_ATTR_NAMESPACE_XLINK
 "xlink:role", "role", GUMBO_ATTR_NAMESPACE_XLINK
 "xlink:show", "show", GUMBO_ATTR_NAMESPACE_XLINK
 "xlink:title", "title", GUMBO_ATTR_NAMESPACE_XLINK
 "xlink:type", "type", GUMBO_ATTR_NAMESPACE_XLINK
-"xml:base", "base", GUMBO_ATTR_NAMESPACE_XML
 "xml:lang", "lang", GUMBO_ATTR_NAMESPACE_XML
 "xml:space", "space", GUMBO_ATTR_NAMESPACE_XML
 "xmlns", "xmlns", GUMBO_ATTR_NAMESPACE_XMLNS

diff --git a/test/html5/test_tree_construction.rb b/test/html5/test_tree_construction.rb
@@ -30,7 +30,7 @@ def compare_nodes(node, ng_node)
         else
           attributes[attr[:name]].value
         end
-        assert_equal(attr[:value], value)
+        assert_equal(attr[:value], value, "expected #{attr}[:value] to equal #{value.inspect}")
       end
       assert_equal(
         node[:children].length,
@@ -117,7 +117,12 @@ def run_test
     end
 
     # Test the errors.
-    assert_equal(@test[:errors].length, doc.errors.length, "Wrong number of errors for #{@test[:data]}")
+    errpayload = doc.errors.map(&:to_s).join("\n")
+    assert_equal(
+      @test[:errors].length,
+      doc.errors.length,
+      "Expected #{@test[:errors].length} errors for #{@test[:data]}, found:\n#{errpayload}",
+    )
 
     # The new, standardized tokenizer errors live in @test[:new_errors]. Let's
     # match each one to exactly one error in doc.errors. Unfortunately, the

diff --git a/test/html5lib-tests b/test/html5lib-tests
+21 −0		.github/workflows/downstream.yml
+0 −8		tree-construction/foreign-fragment.dat
+0 −13		tree-construction/scriptdata01.dat
+24 −24		tree-construction/template.dat
+0 −31		tree-construction/tests1.dat
+0 −55		tree-construction/tests19.dat
+0 −13		tree-construction/tests20.dat
+0 −27		tree-construction/tests21.dat
+0 −44		tree-construction/tests_innerHTML_1.dat
+26 −0		tree-construction/webkit01.dat
+66 −0		tree-construction/webkit02.dat