From 3114f3a9bbdd3f62e3093d98c686fcd41f738ec5 Mon Sep 17 00:00:00 2001
From: "J. King" <jking@jkingweb.ca>
Date: Mon, 29 Mar 2021 18:50:45 -0400
Subject: [PATCH] Fix most serializer test failures

Three remain related to xmlns attributes
---
 lib/DOM/Element.php                 |   2 +-
 lib/DOM/Text.php                    |   6 +-
 lib/DOM/traits/EscapeString.php     |  14 +--
 lib/DOM/traits/Serialize.php        |   2 +-
 tests/cases/TestSerializer.php      | 135 ++++++++++++++++++++++++++++
 tests/cases/TestTreeConstructor.php |   5 +-
 tests/cases/serializer/wpt01.dat    |  45 ++++++----
 tests/phpunit.dist.xml              |   3 +
 8 files changed, 180 insertions(+), 32 deletions(-)
 create mode 100644 tests/cases/TestSerializer.php

diff --git a/lib/DOM/Element.php b/lib/DOM/Element.php
index 4e3a0a0..160e551 100644
--- a/lib/DOM/Element.php
+++ b/lib/DOM/Element.php
@@ -323,7 +323,7 @@ class Element extends \DOMElement {
                 # If the attribute is in some other namespace
                 default:
                     # The attribute’s serialized name is the attribute’s qualified name.
-                    $name = $attr->name;
+                    $name = $attr->nodeName;
             }
 
             $value = $this->escapeString($attr->value, true);
diff --git a/lib/DOM/Text.php b/lib/DOM/Text.php
index 761dc2f..5f29dd9 100644
--- a/lib/DOM/Text.php
+++ b/lib/DOM/Text.php
@@ -14,8 +14,10 @@ class Text extends \DOMText {
         # noframes, or plaintext element, or if the parent of current node is a noscript
         # element and scripting is enabled for the node, then append the value of
         # current node’s data IDL attribute literally.
-        // DEVIATION: No scripting.
-
+        // DEVIATION: No scripting, so <noscript> is not included
+        if ($this->parentNode->namespaceURI === null && in_array($this->parentNode->nodeName, ['style', 'script', 'xmp', 'iframe', 'noembed', 'noframes', 'plaintext'])) {
+            return $this->data;
+        }
         # Otherwise, append the value of current node’s data IDL attribute, escaped as
         # described below.
         return $this->escapeString($this->data);
diff --git a/lib/DOM/traits/EscapeString.php b/lib/DOM/traits/EscapeString.php
index 8a0b1c6..0dfff10 100644
--- a/lib/DOM/traits/EscapeString.php
+++ b/lib/DOM/traits/EscapeString.php
@@ -13,16 +13,16 @@ trait EscapeString {
         # Escaping a string (for the purposes of the algorithm above) consists of
         # running the following steps:
 
-        # 1. Replace any occurrence of the "&amp;" character by the string "&amp;amp;".
+        # 1. Replace any occurrence of the "&" character by the string "&amp;".
         # 2. Replace any occurrences of the U+00A0 NO-BREAK SPACE character by the
-        # string "&amp;nbsp;".
-        $string = str_replace(['&amp;', chr(0x00A0)], ['&amp;amp;', '&amp;nbsp;'], $string);
+        # string "&nbsp;".
+        $string = str_replace(['&', "\u{A0}"], ['&amp;', '&nbsp;'], $string);
         # 3. If the algorithm was invoked in the attribute mode, replace any
-        # occurrences of the "&quot;" character by the string "&amp;quot;".
+        # occurrences of the """ character by the string "&quot;".
         # 4. If the algorithm was not invoked in the attribute mode, replace any
-        # occurrences of the "&lt;" character by the string "&amp;lt;", and any
-        # occurrences of the "&gt;" character by the string "&amp;gt;".
-        return ($attribute) ? str_replace('&quot;', '&amp;quot;', $string) : str_replace(['&lt;', '&gt;'], ['&amp;lt;', '&amp;gt;'], $string);
+        # occurrences of the "<" character by the string "&lt;", and any
+        # occurrences of the ">" character by the string "&gt;".
+        return ($attribute) ? str_replace('"', '&quot;', $string) : str_replace(['<', '>'], ['&lt;', '&gt;'], $string);
     }
 
     protected function coerceName(string $name): string {
diff --git a/lib/DOM/traits/Serialize.php b/lib/DOM/traits/Serialize.php
index d851f9c..3ee4685 100644
--- a/lib/DOM/traits/Serialize.php
+++ b/lib/DOM/traits/Serialize.php
@@ -7,7 +7,7 @@ declare(strict_types=1);
 namespace MensBeam\HTML;
 
 trait Serialize {
-    protected static $voidElements = [ 'area', 'base', 'basefont', 'bgsound', 'br', 'col', 'embed', 'hr', 'img', 'input', 'link', 'meta', 'param', 'source', 'track', 'wbr' ];
+    protected static $voidElements = [ 'area', 'base', 'basefont', 'bgsound', 'br', 'col', 'embed', 'frame', 'hr', 'img', 'input', 'keygen', 'link', 'meta', 'param', 'source', 'track', 'wbr' ];
 
     protected function serializesAsVoid(): bool {
         $name = $this->nodeName;
diff --git a/tests/cases/TestSerializer.php b/tests/cases/TestSerializer.php
new file mode 100644
index 0000000..d210b1a
--- /dev/null
+++ b/tests/cases/TestSerializer.php
@@ -0,0 +1,135 @@
+<?php
+/** @license MIT
+ * Copyright 2017 , Dustin Wilson, J. King et al.
+ * See LICENSE and AUTHORS files for details */
+
+declare(strict_types=1);
+namespace MensBeam\HTML\TestCase;
+
+use MensBeam\HTML\Document;
+use MensBeam\HTML\Parser;
+
+/** 
+ * @covers \MensBeam\HTML\Document
+ * @covers \MensBeam\HTML\DocumentFragment
+ * @covers \MensBeam\HTML\Element
+ * @covers \MensBeam\HTML\TemplateElement
+ * @covers \MensBeam\HTML\Comment
+ * @covers \MensBeam\HTML\Text
+ */
+class TestTreeConstructor extends \PHPUnit\Framework\TestCase {
+    use \MensBeam\HTML\EscapeString;
+
+    protected $out;
+    protected $depth;
+
+    /** @dataProvider provideStandardSerializerTests */
+    public function testStandardTreeTests(array $data, bool $fragment, string $exp): void {
+        $node = $this->buildTree($data, $fragment);
+        $this->assertSame($exp, (string) $node);
+    }
+
+    public function provideStandardSerializerTests(): iterable {
+        $blacklist = [];
+        $files = new \AppendIterator();
+        $files->append(new \GlobIterator(\MensBeam\HTML\BASE."tests/cases/serializer/*.dat", \FilesystemIterator::SKIP_DOTS | \FilesystemIterator::CURRENT_AS_PATHNAME));
+        foreach ($files as $file) {
+            $index = 0;
+            $l = 0;
+            if (!in_array(basename($file), $blacklist)) {
+                $lines = array_map(function($v) {
+                    return rtrim($v, "\n");
+                }, file($file));
+                while ($l < sizeof($lines)) {
+                    $pos = $l + 1;
+                    assert(in_array($lines[$l], ["#document", "#fragment"]), new \Exception("Test $file #$index does not start with #doocument or #fragment tag at line ".($l + 1)));
+                    $fragment = $lines[$l] === "#fragment";
+                    // collect the test input
+                    $data = [];
+                    for (++$l; $l < sizeof($lines); $l++) {
+                        if (preg_match('/^#(script-(on|off)|output)$/', $lines[$l])) {
+                            break;
+                        }
+                        $data[] = $lines[$l];
+                    }
+                    // set the script mode, if present
+                    assert(preg_match('/^#(script-(on|off)|output)$/', $lines[$l]) === 1, new \Exception("Test $file #$index follows data with something other than script flag or output at line ".($l + 1)));
+                    $script = null;
+                    if ($lines[$l] === "#script-off") {
+                        $script = false;
+                        $l++;
+                    } elseif ($lines[$l] === "#script-on") {
+                        $script = true;
+                        $l++;
+                    }
+                    // collect the output string
+                    $exp = [];
+                    assert($lines[$l] === "#output", new \Exception("Test $file #$index follows input with something other than output at line ".($l + 1)));
+                    for (++$l; $l < sizeof($lines); $l++) {
+                        if ($lines[$l] === "" && in_array(($lines[$l + 1] ?? ""), ["#ddocument", "#fragment"])) {
+                            break;
+                        }
+                        assert(preg_match('/^[^#]/', $lines[$l]) === 1, new \Exception("Test $file #$index contains unrecognized data after document at line ".($l + 1)));
+                        $exp[] = $lines[$l];
+                    }
+                    $exp = implode("\n", $exp);
+                    if (!$script) {
+                        yield basename($file)." #$index (line $pos)" => [$data, $fragment, $exp];
+                    }
+                    $l++;
+                    $index++;
+                }
+            }
+        }
+    }
+
+    protected function buildTree(array $data, bool $fragment): \DOMNode {
+        $document = new Document;
+        $out = $fragment ? $document->createDocumentFragment() : $document;
+        $cur = $out;
+        $pad = 2;
+        // process each line in turn
+        for ($l = 0; $l < sizeof($data); $l++) {
+            preg_match('/^(\|\s+)(.+)/', $data[$l], $m);
+            // pop any parents as long as the padding of the line is less than the expected padding
+            $p = strlen((string) $m[1]);
+            assert($p >= 2 && $p <= $pad && !($p % 2), new \Exception("Input data is invalid on line ".($l + 1)));
+            while ($p < $pad) {
+                $pad -= 2;
+                $cur = $cur->parentNode;
+            }
+            // act based upon what the rest of the line looks like
+            $d = $m[2];
+            if (preg_match('/^<!-- (.*?) -->$/', $d, $m)) {
+                // comment
+                $cur->appendChild($document->createComment($m[1]));
+            } elseif (preg_match('/^<!DOCTYPE ([^ >]*)(?: "([^"]*)" "([^"]*)")?>$/', $d, $m)) {
+                // doctype
+                $name = strlen((string) $m[1]) ? $m[1] : null;
+                $public = strlen((string) $m[2]) ? $m[2] : null;
+                $system = strlen((string) $m[3]) ? $m[3] : null;
+                $cur->appendChild($document->implementation->createDocumentType($name, $public, $system));
+            } elseif (preg_match('/^<(?:([^ ]+) )?([^>]+)>$/', $d, $m)) {
+                // element
+                $ns = strlen((string) $m[1]) ? (array_flip(Parser::NAMESPACE_MAP)[$m[1]] ?? null) : null;
+                $cur = $cur->appendChild($document->createElementNS($ns, $m[2]));
+                $pad += 2;
+            } elseif (preg_match('/^(?:([^" ]+) )?([^"=]+)="((?:[^"]|"(?!$))*)"$/', $d, $m)) {
+                // attribute
+                $ns = strlen((string) $m[1]) ? (array_flip(Parser::NAMESPACE_MAP)[$m[1]] ?? $m[1]) : "";
+                $cur->setAttributeNS($ns, $m[2], $m[3]);
+            } elseif (preg_match('/^"((?:[^"]|"(?!$))*)("?)$/', $d, $m)) {
+                // text
+                $t = $m[1];
+                while (!strlen((string) $m[2])) {
+                    preg_match('/^((?:[^"]|"(?!$))*)("?)$/', $data[++$l], $m);
+                    $t .= "\n".$m[1];
+                }
+                $cur->appendChild($document->createTextNode($t));
+            } else {
+                throw new \Exception("Input data is invalid on line ".($l + 1));
+            }
+        }
+        return $out;
+    }
+}
diff --git a/tests/cases/TestTreeConstructor.php b/tests/cases/TestTreeConstructor.php
index 8982f31..09a34c5 100644
--- a/tests/cases/TestTreeConstructor.php
+++ b/tests/cases/TestTreeConstructor.php
@@ -8,7 +8,6 @@ namespace MensBeam\HTML\TestCase;
 
 use MensBeam\HTML\Data;
 use MensBeam\HTML\Document;
-use MensBeam\HTML\EOFToken;
 use MensBeam\HTML\LoopException;
 use MensBeam\HTML\NotImplementedException;
 use MensBeam\HTML\OpenElementsStack;
@@ -286,7 +285,7 @@ class TestTreeConstructor extends \PHPUnit\Framework\TestCase {
     protected function balanceTree(array $act, array $exp): array {
         // makes sure that the actual tree contain the same number of lines as the expected tree
         // lines are inserted where the two trees diverge, until the end of the actual tree is reached
-        // this usuallyresults in clean PHPUnit comparison failure output
+        // this usually results in cleaner PHPUnit comparison failure output
         for ($a = 0; $a < sizeof($act) && sizeof($act) < sizeof($exp); $a++) {
             if (!isset($act[$a]) || $exp[$a] !== $act[$a]) {
                 array_splice($act, $a, 0, [""]);
@@ -431,7 +430,7 @@ class TestTreeConstructor extends \PHPUnit\Framework\TestCase {
                     }
                     // collect the output tree
                     $exp = [];
-                    assert($lines[$l] === "#document", new \Exception("Test $file #$index follows dociument fragment with something other than document at line ".($l + 1)));
+                    assert($lines[$l] === "#document", new \Exception("Test $file #$index follows document fragment with something other than document at line ".($l + 1)));
                     for (++$l; $l < sizeof($lines); $l++) {
                         if ($lines[$l] === "" && ($lines[$l + 1] ?? "") === "#data") {
                             break;
diff --git a/tests/cases/serializer/wpt01.dat b/tests/cases/serializer/wpt01.dat
index 30794a2..0074d36 100644
--- a/tests/cases/serializer/wpt01.dat
+++ b/tests/cases/serializer/wpt01.dat
@@ -26,7 +26,7 @@
 #fragment
 | <span>
 |   <a>
-|     b=" "
+|     b=" "
 #output
 <span><a b="&nbsp;"></a></span>
 
@@ -42,33 +42,33 @@
 |   <a>
 |     b="<"
 #output
-<span><a b="&lt;"></a></span>
+<span><a b="<"></a></span>
 
 #fragment
 | <span>
 |   <a>
 |     b=">"
 #output
-<span><a b="&gt;"></a></span>
+<span><a b=">"></a></span>
 
 #fragment
 | <span>
 |   <a>
 |     href="javascript:"<>""
 #output
-<span><a href="javascript:&quot;&lt;>&quot;"></a></span>
+<span><a href="javascript:&quot;<>&quot;"></a></span>
 
 #fragment
 | <span>
 |   <svg svg>
-|     xlink href="a"
+|     xlink xlink:href="a"
 #output
 <span><svg xlink:href="a"></svg></span>
 
 #fragment
 | <span>
 |   <svg svg>
-|     xmlns svg="test"
+|     xmlns xmlns:svg="test"
 #output
 <span><svg xmlns:svg="test"></svg></span>
 
@@ -86,7 +86,7 @@
 
 #fragment
 | <span>
-|   " "
+|   " "
 #output
 <span>&nbsp;</span>
 
@@ -106,7 +106,7 @@
 | <span>
 |   """
 #output
-<span>&quot;</span>
+<span>"</span>
 
 #fragment
 | <span>
@@ -162,6 +162,15 @@
 | <span>
 |   <noscript>
 |     "<&>"
+#script-off
+#output
+<span><noscript>&lt;&amp;&gt;</noscript></span>
+
+#fragment
+| <span>
+|   <noscript>
+|     "<&>"
+#script-on
 #output
 <span><noscript><&></noscript></span>
 
@@ -176,11 +185,11 @@
 |   <a>
 |     <b>
 |       <c>
-|   <d>
-|     "e"
-|   <f>
-|     <g>
-|       "h"
+|     <d>
+|       "e"
+|     <f>
+|       <g>
+|         "h"
 #output
 <span><a><b><c></c></b><d>e</d><f><g>h</g></f></a></span>
 
@@ -193,7 +202,7 @@
 #fragment
 | <span>
 |   <svg svg>
-|     xml foo="test"
+|     xml xml:foo="test"
 #output
 <span><svg xml:foo="test"></svg></span>
 
@@ -207,7 +216,7 @@
 #fragment
 | <span>
 |   <svg svg>
-|     xmlns foo="test"
+|     xmlns xmlns:foo="test"
 #output
 <span><svg xmlns:foo="test"></svg></span>
 
@@ -253,7 +262,7 @@
 #fragment
 | <span>
 |   <pre>
-|     "abc
+|     "a
 "
 #output
 <span><pre>a
@@ -287,7 +296,7 @@
 #fragment
 | <span>
 |   <textarea>
-|     "abc
+|     "a
 "
 #output
 <span><textarea>a
@@ -321,7 +330,7 @@
 #fragment
 | <span>
 |   <listing>
-|     "abc
+|     "a
 "
 #output
 <span><listing>a
diff --git a/tests/phpunit.dist.xml b/tests/phpunit.dist.xml
index 0e8d06e..ea81c63 100644
--- a/tests/phpunit.dist.xml
+++ b/tests/phpunit.dist.xml
@@ -25,5 +25,8 @@
     <testsuite name="Tree">
         <file>cases/TestTreeConstructor.php</file>
     </testsuite>
+    <testsuite name="Serializer">
+        <file>cases/TestSerializer.php</file>
+    </testsuite>
 </testsuites>
 </phpunit>