From da0a4b12e36380d216772ffe372adb862b669360 Mon Sep 17 00:00:00 2001
From: "J. King" <jking@jkingweb.ca>
Date: Thu, 9 Mar 2023 17:17:56 -0500
Subject: [PATCH] Spec-related update

See https://github.com/whatwg/html/pull/8271

Also removed test patches which are no longer needed, and fixed a bug
exposed by new tests

The testsuite includes tests for the new <search> element, but this is
not included in the specification yet, so the tests have been suppressed
---
 lib/Parser/Tokenizer.php            |  2 +-
 lib/Parser/TreeConstructor.php      |  6 +++---
 tests/cases/TestTokenizer.php       | 26 --------------------------
 tests/cases/TestTreeConstructor.php | 21 +++++++++++++++++----
 4 files changed, 21 insertions(+), 34 deletions(-)

diff --git a/lib/Parser/Tokenizer.php b/lib/Parser/Tokenizer.php
index e7f2828..0a65f4b 100644
--- a/lib/Parser/Tokenizer.php
+++ b/lib/Parser/Tokenizer.php
@@ -276,7 +276,7 @@ class Tokenizer {
                     // DEVIATION: Character reference consumption implemented as a function
                     $outChar = $this->switchToCharacterReferenceState(self::DATA_STATE);
                     if (strspn($outChar, Data::WHITESPACE)) {
-                        yield new WhitespaceToken($outChar); // a character reference is either all whitespace is no whitespace
+                        yield new WhitespaceToken($outChar); // a character reference is either all whitespace or is not whitespace
                     } else {
                         yield new CharacterToken($outChar);
                     }
diff --git a/lib/Parser/TreeConstructor.php b/lib/Parser/TreeConstructor.php
index 55926d2..556b24d 100644
--- a/lib/Parser/TreeConstructor.php
+++ b/lib/Parser/TreeConstructor.php
@@ -1076,7 +1076,7 @@ class TreeConstructor {
                                 # 2. Generate implied end tags.
                                 $this->stack->generateImpliedEndTags();
                                 # 3. If the current node is not a form element, then this is a parse error.
-                                if (!$this->stack->currentNodeName !== 'form') {
+                                if ($this->stack->currentNodeName !== 'form') {
                                     $this->error(ParseError::UNEXPECTED_END_TAG, $token->name);
                                 }
                                 # 4. Pop elements from the stack of open elements until a form element has been
@@ -2036,8 +2036,8 @@ class TreeConstructor {
                     //   mode as it may have been turned on in a previous evluation
                     //   of this mode
                     $this->fosterParenting = false;
-                    # A character token, if the current node is table, tbody, tfoot, thead, or tr element
-                    if ($token instanceof CharacterToken && in_array($this->stack->currentNodeName, ["table", "tbody", "tfoot", "thead", "tr"])) {
+                    # A character token, if the current node is table, tbody, template, tfoot, thead, or tr element
+                    if ($token instanceof CharacterToken && in_array($this->stack->currentNodeName, ["table", "tbody", "template", "tfoot", "thead", "tr"])) {
                         # Let the pending table character tokens be an empty list of tokens.
                         $this->pendingTableCharacterTokens = [];
                         # Let the original insertion mode be the current insertion mode.
diff --git a/tests/cases/TestTokenizer.php b/tests/cases/TestTokenizer.php
index 90a3a70..4f4d5e3 100644
--- a/tests/cases/TestTokenizer.php
+++ b/tests/cases/TestTokenizer.php
@@ -235,32 +235,6 @@ class TestTokenizer extends \PHPUnit\Framework\TestCase {
             case ["<!\u{B}", ["Data state"]]:
                 $test['errors'] = array_reverse($test['errors']);
                 break;
-            // eof-in-<whatever> positions in some tests don't make sense
-            // https://github.com/html5lib/html5lib-tests/issues/125
-            case ["", ["CDATA section state"]]:
-                // there is no position 2
-                $test['errors'][0]['col']--;
-                break;
-            case ["\u{A}", ["CDATA section state"]]:
-                // the line break is, for some reason, not counted in the test
-                $test['errors'][0]['line']++;
-                $test['errors'][0]['col'] = 1;
-                break;
-            case ["<!----!\r\n>", ["Data state"]]:
-            case ["<!----!\n>", ["Data state"]]:
-            case ["<!----!\r>", ["Data state"]]:
-                // the line break is, for some reason, not counted in the test
-                $test['errors'][0]['line']++;
-                $test['errors'][0]['col'] = 2;
-                break;
-            case ["<!----! >", ["Data state"]]:
-                $test['errors'][0]['col']++;
-                break;
-            case [hex2bin("f4808080"), ["CDATA section state"]]:
-            case [hex2bin("3bf4808080"), ["CDATA section state"]]:
-                // malpaired surrogates count as two characters
-                $test['errors'][0]['col']++;
-                break;
         }
     }
 }
diff --git a/tests/cases/TestTreeConstructor.php b/tests/cases/TestTreeConstructor.php
index 86b0202..3e5eae5 100644
--- a/tests/cases/TestTreeConstructor.php
+++ b/tests/cases/TestTreeConstructor.php
@@ -33,13 +33,19 @@ class TestTreeConstructor extends \PHPUnit\Framework\TestCase {
     protected $depth;
     protected $ns;
 
+    protected static $passed = [];
+
     /** @dataProvider provideStandardTreeTests */
-    public function testStandardTreeTests(string $data, array $exp, array $errors, $fragment): void {
+    public function testStandardTreeTests(string $data, array $exp, array $errors, $fragment, string $id): void {
         $this->runTreeTest($data, $exp, $errors, $fragment, null);
+        self::$passed[$id] = true;
     }
 
     /** @dataProvider provideStandardTreeTests */
-    public function testStandardTreeTestsWithHtmlNamespace(string $data, array $exp, array $errors, $fragment): void {
+    public function testStandardTreeTestsWithHtmlNamespace(string $data, array $exp, array $errors, $fragment, string $id): void {
+        if (!isset(self::$passed[$id])) {
+            $this->markTestSkipped("Null-namespaced test failed or skipped.");
+        }
         $config = new Config;
         $config->htmlNamespace = true;
         $this->runTreeTest($data, $exp, $errors, $fragment, $config);
@@ -49,7 +55,13 @@ class TestTreeConstructor extends \PHPUnit\Framework\TestCase {
         $files = new \AppendIterator();
         $files->append(new \GlobIterator(\MensBeam\HTML\Parser\BASE."tests/html5lib-tests/tree-construction/*.dat", \FilesystemIterator::SKIP_DOTS | \FilesystemIterator::CURRENT_AS_PATHNAME));
         $files->append(new \GlobIterator(\MensBeam\HTML\Parser\BASE."tests/cases/tree-construction/mensbeam*.dat", \FilesystemIterator::SKIP_DOTS | \FilesystemIterator::CURRENT_AS_PATHNAME));
-        return $this->parseTreeTest($files);
+        // filter out tests for the prospective <search> element, which has not yet been added to HTML
+        $filtered = new class($files) extends \FilterIterator {
+            public function accept(): bool {
+                return !preg_match('/\bsearch-element.dat$/', parent::current());
+            }
+        };
+        return $this->parseTreeTest($filtered);
     }
 
     /** @dataProvider provideProcessingInstructionTreeTests */
@@ -312,7 +324,8 @@ class TestTreeConstructor extends \PHPUnit\Framework\TestCase {
                     if (!$script) {
                         // scripting-dependent tests are skipped entirely since we will not support scripting
                         $errors = ['old' => $errors, 'new' => $newErrors];
-                        yield basename($file)." #$index (line $pos)" => [$data, $exp, $errors, $fragment];
+                        $id = basename($file)." #$index (line $pos)";
+                        yield $id => [$data, $exp, $errors, $fragment, $id];
                     }
                     $l++;
                     $index++;