Fix undef boolean attributes

.github/workflows/linux.yml

@@ -6,6 +6,7 @@ on:
     tags-ignore:
       - '*'
   pull_request:
+  workflow_dispatch:
 jobs:
   perl:
     runs-on: ubuntu-latest
@@ -14,18 +15,19 @@ jobs:
       matrix:
         perl-version:
           - '5.38'
-          # - '5.34'
-          # - '5.32'
-          # - '5.30'
-          # - '5.28'
-          # - '5.26'
-          # - '5.24'
-          # - '5.22'
-          # - '5.20'
+          - '5.36'
+          - '5.34'
+          - '5.32'
+          - '5.30'
+          - '5.28'
+          - '5.26'
+          - '5.24'
+          - '5.22'
+          - '5.20'
           - '5.18'
-          # - '5.16'
-          # - '5.14'
-          # - '5.12'
+          - '5.16'
+          - '5.14'
+          - '5.12'
           - '5.10'
     container:
       image: perl:${{ matrix.perl-version }}

.gitignore

@@ -9,4 +9,4 @@ Makefile
 MYMETA.*
 .build/
 HTML-Parser-*/
-
+local/

hparser.c

@@ -85,6 +85,40 @@ static const char * const argname[] = {
     /* ARG_FLAG_FLAT_ARRAY */
 };
 
+/* https://meiert.com/en/blog/boolean-attributes-of-html/ */
+const static struct boolean_attribute {
+    int len;
+    const char* str;
+}
+boolean_attributes[] = {
+    {15, "allowfullscreen"},
+    {19, "allowpaymentrequest"},
+    {5,  "async"},
+    {9,  "autofocus"},
+    {8,  "autoplay"},
+    {7,  "checked"},
+    {8,  "controls"},
+    {7,  "default"},
+    {8,  "disabled"},
+    {14, "formnovalidate"},
+    {6,  "hidden"},
+    {5,  "ismap"},
+    {9,  "itemscope"},
+    {4,  "loop"},
+    {8,  "multiple"},
+    {5,  "muted"},
+    {8,  "nomodule"},
+    {10, "novalidate"},
+    {4,  "open"},
+    {11, "playsinline"},
+    {8,  "readonly"},
+    {8,  "required"},
+    {8,  "reversed"},
+    {8,  "selected"},
+    {9,  "truespeed"},
+    {0,  0}
+};
+
 #define CASE_SENSITIVE(p_state) \
          ((p_state)->xml_mode || (p_state)->case_sensitive)
 #define STRICT_NAMES(p_state) \
@@ -438,8 +472,8 @@ report_event(PSTATE* p_state,
 		}
 
 		for (i = 1; i < num_tokens; i += 2) {
-		    SV* attrname = newSVpvn(tokens[i].beg,
-					    tokens[i].end-tokens[i].beg);
+            int attrlen = tokens[i].end-tokens[i].beg;
+		    SV* attrname = newSVpvn(tokens[i].beg, attrlen);
 		    SV* attrval;
 
 		    if (utf8)
@@ -465,11 +499,34 @@ report_event(PSTATE* p_state,
 			}
 		    }
 		    else { /* boolean */
-			if (p_state->bool_attr_val)
-			    attrval = newSVsv(p_state->bool_attr_val);
-			else
-			    attrval = newSVsv(attrname);
-		    }
+			int i;
+			int found = 0;
+			for ( i = 0; boolean_attributes[i].len; i++ ) {
+			if( attrlen == boolean_attributes[i].len ) {
+			char *attrname_s = SvPVbyte_nolen(attrname);
+			const char *t = boolean_attributes[i].str;
+			int len = attrlen;
+			while(len) {
+				if(toLOWER(*attrname_s) != *t)
+					break;
+				attrname_s++;
+				t++;
+				if(!--len) {
+					/* this is a boolean attribute */
+					if (p_state->bool_attr_val)
+						attrval = newSVsv(p_state->bool_attr_val);
+					else
+						attrval = newSVsv(attrname);
+					}
+					found = 1;
+			}
+			}
+			}
+			/* no matches were found, so set attr to undef */
+			if (!found)
+				attrval = newSV(0);
+
+			}
 
 		    if (!CASE_SENSITIVE(p_state))
 			sv_lower(aTHX_ attrname);

t/cases.t

@@ -24,7 +24,7 @@ my @result;
         my ($self, $tag, $attr) = @_;
         push @result, "START[$tag]";
         for (sort keys %$attr) {
-            push @result, "\t$_: " . $attr->{$_};
+            push @result, "\t$_: " . ( defined $attr->{$_} ? $attr->{$_} : '<undef>' );
         }
         $start++;
     }
@@ -57,10 +57,10 @@ my @result;
 }
 
 my @tests = (
-    '<a ">'                => ['START[a]', "\t\": \""],
+    '<a ">'                => ['START[a]', "\t\": <undef>"],
     '<a/>'                 => ['START[a/]',],
-    '<a />'                => ['START[a]', "\t/: /"],
-    '<a a/>'               => ['START[a]', "\ta/: a/"],
+    '<a />'                => ['START[a]', "\t/: <undef>"],
+    '<a a/>'               => ['START[a]', "\ta/: <undef>"],
     '<a a/=/>'             => ['START[a]', "\ta/: /"],
     '<a x="foo&nbsp;bar">' => ['START[a]', "\tx: foo\xA0bar"],
     '<a x="foo&nbspbar">'  => ['START[a]', "\tx: foo&nbspbar"],
@@ -73,7 +73,7 @@ my @tests = (
     "2 <a href='foo bar'> 2" =>
         ['TEXT[2 ]', 'START[a]', "\thref: foo bar", 'TEXT[ 2]'],
     '2 <a href=foo bar> 2' =>
-        ['TEXT[2 ]', 'START[a]', "\tbar: bar", "\thref: foo", 'TEXT[ 2]'],
+        ['TEXT[2 ]', 'START[a]', "\tbar: <undef>", "\thref: foo", 'TEXT[ 2]'],
     '2 <a href="foo bar"> 2' =>
         ['TEXT[2 ]', 'START[a]', "\thref: foo bar", 'TEXT[ 2]'],
     '2 <a href="foo\'bar"> 2' =>
@@ -84,7 +84,7 @@ my @tests = (
         ['TEXT[2 ]', 'START[a]', "\thref: foo\"bar", 'TEXT[ 2]'],
     '2 <a.b> 2' => ['TEXT[2 ]', 'START[a.b]', 'TEXT[ 2]'],
     '2 <a.b-12 a.b = 2 a> 2' =>
-        ['TEXT[2 ]', 'START[a.b-12]', "\ta: a", "\ta.b: 2", 'TEXT[ 2]'],
+        ['TEXT[2 ]', 'START[a.b-12]', "\ta: <undef>", "\ta.b: 2", 'TEXT[ 2]'],
     '2 <a_b> 2' => ['TEXT[2 ]', 'START[a_b]', 'TEXT[ 2]'],
     '<!ENTITY nbsp   CDATA "&#160;" -- no-break space -->' =>
         ['DECLARATION[ENTITY nbsp   CDATA "&#160;" -- no-break space --]'],
@@ -94,6 +94,7 @@ my @tests = (
     '<!-- comment <!-- not comment --> comment -->' =>
         ['COMMENT[ comment <!]', 'COMMENT[> comment ]'],
     '<!-- <a href="foo"> -->' => ['COMMENT[ <a href="foo"> ]'],
+    '<input type="checkbox" checked disabled foo>' => ['START[input]', "\tchecked: checked", "\tdisabled: disabled", "\tfoo: <undef>", "\ttype: checkbox" ],
 );
 
 plan tests => @tests / 2;

t/msie-compat.t

@@ -62,7 +62,7 @@ $p->eof;
 
 is($TEXT, <<'EOT');
 [start_document,<undef>,,]
-[start,a,<a name=`foo bar`>,name:`foo:bar`:bar`]
+[start,a,<a name=`foo bar`>,name:`foo:bar`:<undef>]
 [end_document,<undef>,,]
 EOT
 

t/parser.t

@@ -71,7 +71,7 @@ HTML
 
     sub start {
         my ($self, $tag, $attr) = @_;
-        $attr = join("/", map "$_=$attr->{$_}", sort keys %$attr);
+        $attr = join("/", map { "$_=" . ( defined $attr->{$_} ? $attr->{$_} : '<undef>' ) }  sort keys %$attr);
         $attr = "/$attr" if length $attr;
         $OUT .= "<<$tag$attr>>|";
     }