web-xml: Added doctype parsing.

skift-org · Apr 16, 2024 · 0ed35a0 · 0ed35a0
1 parent bf16ef4
commit 0ed35a0
Show file tree

Hide file tree

Showing 4 changed files with 78 additions and 1 deletion.
diff --git a/src/web/web-dom/document-type.h b/src/web/web-dom/document-type.h
@@ -10,6 +10,8 @@ struct DocumentType : public Node {
     String publicId;
     String systemId;
 
+    DocumentType() = default;
+
     DocumentType(String name, String publicId, String systemId)
         : name(name), publicId(publicId), systemId(systemId) {
     }

diff --git a/src/web/web-xml/parser.cpp b/src/web/web-xml/parser.cpp
@@ -249,12 +249,44 @@ Res<> Parser::_parseProlog(Io::SScan &s, Dom::Node &parent) {
     while (_parseMisc(s, parent) and not s.ended())
         ;
 
-    // TODO: Parse doctype declaration
+    if (auto doctype = _parseDoctype(s)) {
+        parent.appendChild(doctype.unwrap());
+        while (_parseMisc(s, parent) and not s.ended())
+            ;
+    }
 
     rollback.disarm();
     return Ok();
 }
 
+static constexpr auto RE_DOCTYPE_START = "<!DOCTYPE"_re;
+
+Res<Strong<Dom::DocumentType>> Parser::_parseDoctype(Io::SScan &s) {
+    // doctypedecl ::= '<!DOCTYPE' S Name (S ExternalID)? S? ('[' intSubset ']' S?)? '>'
+    auto rollback = s.rollbackPoint();
+
+    logDebug("Parsing doctype declaration");
+
+    if (not s.skip(RE_DOCTYPE_START))
+        return Error::invalidData("expected '<!DOCTYPE'");
+
+    auto docType = makeStrong<Dom::DocumentType>();
+
+    try$(_parseS(s));
+
+    docType->name = try$(_parseName(s));
+
+    try$(_parseS(s));
+    (void)_parseExternalId(s, *docType);
+
+    try$(_parseS(s));
+    if (not s.skip('>'))
+        return Error::invalidData("expected '>'");
+
+    rollback.disarm();
+    return Ok(docType);
+}
+
 // 2.9 MARK: Standalone Document Declaration
 // https://www.w3.org/TR/xml/#sec-rmd
 
@@ -530,4 +562,32 @@ Res<Rune> Parser::_parseReference(Io::SScan &s) {
         return Error::invalidData("expected reference");
 }
 
+// 4.2 MARK: Entity Declarations
+// https://www.w3.org/TR/xml/#sec-entity-decl
+
+Res<> Parser::_parseExternalId(Io::SScan &s, Dom::DocumentType &docType) {
+    // ExternalID ::= 'SYSTEM' S SystemLiteral | 'PUBLIC' S PubidLiteral S SystemLiteral
+    logDebug("Parsing external ID");
+
+    auto rollback = s.rollbackPoint();
+
+    if (s.skip("SYSTEM"_re)) {
+        try$(_parseS(s));
+        // NOSPEC: We are parsing the system literal as att value
+        docType.systemId = try$(_parseAttValue(s));
+        rollback.disarm();
+        return Ok();
+    } else if (s.skip("PUBLIC"_re)) {
+        // NOSPEC: We are parsing the public and system literals as att values
+        try$(_parseS(s));
+        docType.publicId = try$(_parseAttValue(s));
+        try$(_parseS(s));
+        docType.systemId = try$(_parseAttValue(s));
+        rollback.disarm();
+        return Ok();
+    } else {
+        return Error::invalidData("expected 'SYSTEM' or 'PUBLIC'");
+    }
+}
+
 } // namespace Web::Xml
diff --git a/src/web/web-xml/parser.h b/src/web/web-xml/parser.h
@@ -1,6 +1,7 @@
 #pragma once
 
 #include <web-dom/comment.h>
+#include <web-dom/document-type.h>
 #include <web-dom/document.h>
 #include <web-dom/element.h>
 #include <web-dom/text.h>
@@ -44,6 +45,8 @@ struct Parser {
 
     Res<> _parseProlog(Io::SScan &s, Dom::Node &parent);
 
+    Res<Strong<Dom::DocumentType>> _parseDoctype(Io::SScan &s);
+
     Res<Strong<Dom::Element>> _parseElement(Io::SScan &s, Ns ns);
 
     Res<Strong<Dom::Element>> _parseStartTag(Io::SScan &s, Ns ns);
@@ -65,6 +68,8 @@ struct Parser {
     Res<Rune> _parseEntityRef(Io::SScan &s);
 
     Res<Rune> _parseReference(Io::SScan &s);
+
+    Res<> _parseExternalId(Io::SScan &s, Dom::DocumentType &docType);
 };
 
 } // namespace Web::Xml
diff --git a/src/web/web-xml/tests/test-parser.cpp b/src/web/web-xml/tests/test-parser.cpp
@@ -82,4 +82,14 @@ test$("parse-comment") {
     return Ok();
 }
 
+test$("parse-doctype") {
+    auto s = Io::SScan("<!DOCTYPE html><html></html>");
+    auto p = Parser();
+    auto doc = try$(p.parse(s, Web::HTML));
+    auto first = doc->firstChild();
+    auto doctype = try$(first.cast<Dom::DocumentType>());
+    expect$(doctype->name == "html");
+    return Ok();
+}
+
 } // namespace Web::Xml::Tests