vaev-markup: create own buffer for peeking in html parser

Avoid mixup between compliant temporary buffer and our own workaround buffer
for "peeking"

src/web/vaev-markup/html.cpp

@@ -1778,27 +1778,28 @@ void HtmlLexer::consume(Rune rune, bool isEof) {
         // 13.2.5.42 MARK: Markup declaration open state
         // If the next few characters are:
 
-        _temp.append(rune);
+        peekerForSingleState.append(rune);
+
         // Two U+002D HYPHEN-MINUS characters (-)
         // Consume those two characters, create a comment token whose data
         // is the empty string, and switch to the comment start state.
-        if (auto r = startWith("--"s, _temp.str()); r != Match::NO) {
+        if (auto r = startWith("--"s, peekerForSingleState.str()); r != Match::NO) {
             if (r == Match::PARTIAL)
                 break;
 
-            _temp.clear();
+            peekerForSingleState.clear();
             _begin(HtmlToken::COMMENT);
             _switchTo(State::COMMENT_START);
         }
 
         // ASCII case-insensitive match for the word "DOCTYPE"
         // Consume those characters and switch to the DOCTYPE state.
 
-        else if (auto r = startWith("DOCTYPE"s, _temp.str()); r != Match::NO) {
+        else if (auto r = startWith("DOCTYPE"s, peekerForSingleState.str()); r != Match::NO) {
             if (r == Match::PARTIAL)
                 break;
 
-            _temp.clear();
+            peekerForSingleState.clear();
             _switchTo(State::DOCTYPE);
         }
 
@@ -1810,12 +1811,12 @@ void HtmlLexer::consume(Rune rune, bool isEof) {
         // error. Create a comment token whose data is the "[CDATA[" string.
         // Switch to the bogus comment state.
 
-        else if (auto r = startWith("[CDATA["s, _temp.str()); r != Match::NO) {
+        else if (auto r = startWith("[CDATA["s, peekerForSingleState.str()); r != Match::NO) {
             if (r == Match::PARTIAL)
                 break;
 
             // NOSPEC: This is in reallity more complicated
-            _temp.clear();
+            peekerForSingleState.clear();
             _switchTo(State::CDATA_SECTION);
         }
 
@@ -1824,7 +1825,7 @@ void HtmlLexer::consume(Rune rune, bool isEof) {
         // comment token whose data is the empty string. Switch to the bogus
         // comment state (don't consume anything in the current state).
         else {
-            _temp.clear();
+            peekerForSingleState.clear();
             _raise("incorrectly-opened-comment");
             _begin(HtmlToken::COMMENT);
             _reconsumeIn(State::BOGUS_COMMENT, rune);

src/web/vaev-markup/html.h

@@ -100,6 +100,7 @@ struct HtmlLexer {
     Rune _currChar = 0;
     StringBuilder _builder, _commentBuilder;
     StringBuilder _temp;
+    StringBuilder peekerForSingleState;
 
     HtmlToken &_begin(HtmlToken::Type type) {
         _token = HtmlToken{};

src/web/vaev-markup/tests/test-xhtml-parser.cpp

@@ -156,4 +156,33 @@ test$("parse-doctype") {
     return Ok();
 }
 
+test$("parse-title") {
+    auto s = Io::SScan("<title>the title</title>");
+    XmlParser p{};
+    auto dom = makeStrong<Markup::Document>(Mime::Url());
+    try$(p.parse(s, Vaev::HTML, *dom));
+    expect$(dom->title() == "the title");
+    return Ok();
+}
+
+test$("parse-comment-with-gt-symb") {
+    auto s = Io::SScan(
+        "<title>im a title!</title>"
+        "<!-- a b <meta> c d -->"
+    );
+    XmlParser p{};
+    auto dom = makeStrong<Markup::Document>(Mime::Url());
+    try$(p.parse(s, Vaev::HTML, *dom));
+
+    expect$(dom->hasChildren());
+    auto title = dom->firstChild();
+    expect$(title->nodeType() == NodeType::ELEMENT);
+
+    auto comment = title->nextSibling();
+    expect$(comment->nodeType() == NodeType::COMMENT);
+    expect$(try$(comment.cast<Comment>())->data == " a b <meta> c d ");
+
+    return Ok();
+}
+
 } // namespace Vaev::Markup::Tests