From ee5e9298e7993ae516fb6f480599912329d3d7fd Mon Sep 17 00:00:00 2001
From: "J. King" <jking@jkingweb.ca>
Date: Wed, 31 Mar 2021 14:14:41 -0400
Subject: [PATCH] Tests and fixes for DOCTYPE serialization

---
 lib/DOM/traits/Serialize.php         |  5 ++++-
 tests/cases/TestSerializer.php       | 20 ++++++++++++--------
 tests/cases/serializer/menbeam02.dat | 28 ++++++++++++++++++++++++++++
 3 files changed, 44 insertions(+), 9 deletions(-)
 create mode 100644 tests/cases/serializer/menbeam02.dat

diff --git a/lib/DOM/traits/Serialize.php b/lib/DOM/traits/Serialize.php
index 3ee4685..e0ed9ff 100644
--- a/lib/DOM/traits/Serialize.php
+++ b/lib/DOM/traits/Serialize.php
@@ -57,7 +57,10 @@ trait Serialize {
                 # LETTER E), followed by a space (U+0020 SPACE), followed by the value of
                 # current node's name IDL attribute, followed by the literal string ">" (U+003E
                 # GREATER-THAN SIGN).
-                $s .= "<!DOCTYPE {$node->childNodes->item(0)->name}>";
+                // DEVIATION: The name is trimmed because PHP's DOM does not
+                //   accept the empty string as a DOCTYPE name
+                $name = trim($node->childNodes->item(0)->name, " ");
+                $s .= "<!DOCTYPE $name>";
                 $start = 1;
             }
 
diff --git a/tests/cases/TestSerializer.php b/tests/cases/TestSerializer.php
index b807f01..7072ab6 100644
--- a/tests/cases/TestSerializer.php
+++ b/tests/cases/TestSerializer.php
@@ -61,10 +61,10 @@ class TestSerializer extends \PHPUnit\Framework\TestCase {
                     $exp = [];
                     assert($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[$l + 1] ?? ""), ["#ddocument", "#fragment"])) {
+                        if ($lines[$l] === "" && in_array(($lines[$l + 1] ?? ""), ["#document", "#fragment"])) {
                             break;
                         }
-                        assert(preg_match('/^[^#]/', $lines[$l]) === 1, new \Exception("Test $file #$index contains unrecognized data after document at line ".($l + 1)));
+                        assert(preg_match('/^[^#]/', $lines[$l]) === 1, new \Exception("Test $file #$index contains unrecognized data after output at line ".($l + 1)));
                         $exp[] = $lines[$l];
                     }
                     $exp = implode("\n", $exp);
@@ -80,8 +80,12 @@ class TestSerializer extends \PHPUnit\Framework\TestCase {
 
     protected function buildTree(array $data, bool $fragment): \DOMNode {
         $document = new Document;
-        $document->appendChild($document->createElement("html"));
-        $out = $fragment ? $document->createDocumentFragment() : $document;
+        if ($fragment) {
+            $document->appendChild($document->createElement("html"));
+            $out = $document->createDocumentFragment();
+        } else {
+            $out = $document;
+        }
         $cur = $out;
         $pad = 2;
         // process each line in turn
@@ -99,11 +103,11 @@ class TestSerializer extends \PHPUnit\Framework\TestCase {
             if (preg_match('/^<!-- (.*?) -->$/', $d, $m)) {
                 // comment
                 $cur->appendChild($document->createComment($m[1]));
-            } elseif (preg_match('/^<!DOCTYPE ([^ >]*)(?: "([^"]*)" "([^"]*)")?>$/', $d, $m)) {
+            } elseif (preg_match('/^<!DOCTYPE(?: ([^ >]*)(?: "([^"]*)" "([^"]*)")?)?>$/', $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;
+                $name = strlen((string) ($m[1] ?? "")) ? $m[1] : " ";
+                $public = strlen((string) ($m[2] ?? "")) ? $m[2] : "";
+                $system = strlen((string) ($m[3] ?? "")) ? $m[3] : "";
                 $cur->appendChild($document->implementation->createDocumentType($name, $public, $system));
             } elseif (preg_match('/^<(?:([^ ]+) )?([^>]+)>$/', $d, $m)) {
                 // element
diff --git a/tests/cases/serializer/menbeam02.dat b/tests/cases/serializer/menbeam02.dat
new file mode 100644
index 0000000..4dffa11
--- /dev/null
+++ b/tests/cases/serializer/menbeam02.dat
@@ -0,0 +1,28 @@
+#document
+| <html>
+#output
+<html></html>
+
+#document
+| <!DOCTYPE html>
+| <html>
+#output
+<!DOCTYPE html><html></html>
+
+#document
+| <!DOCTYPE html "public" "system">
+| <html>
+#output
+<!DOCTYPE html><html></html>
+
+#document
+| <!DOCTYPE test>
+| <html>
+#output
+<!DOCTYPE test><html></html>
+
+#document
+| <!DOCTYPE>
+| <html>
+#output
+<!DOCTYPE ><html></html>