xmlserializer for doctype

sjorsdonkers · sjorsdonkers · commit 18463451c32e · 2025-06-02T08:21:03.000+02:00
diff --git a/src/browser/dump.zig b/src/browser/dump.zig
@@ -28,6 +28,32 @@ pub fn writeHTML(doc: *parser.Document, writer: anytype) !void {
     try writer.writeAll("\n");
 }
 
+// Spec: https://www.w3.org/TR/xml/#sec-prolog-dtd
+pub fn writeDocType(doc_type: *parser.DocumentType, writer: anytype) !void {
+    try writer.writeAll("<!DOCTYPE ");
+    try writer.writeAll(try parser.documentTypeGetName(doc_type));
+
+    const public_id = try parser.documentTypeGetPublicId(doc_type);
+    const system_id = try parser.documentTypeGetSystemId(doc_type);
+    if (public_id.len != 0 and system_id.len != 0) {
+        try writer.writeAll(" PUBLIC \"");
+        try writeEscapedAttributeValue(writer, public_id);
+        try writer.writeAll("\" \"");
+        try writeEscapedAttributeValue(writer, system_id);
+        try writer.writeAll("\"");
+    } else if (public_id.len != 0) {
+        try writer.writeAll(" PUBLIC \"");
+        try writeEscapedAttributeValue(writer, public_id);
+        try writer.writeAll("\"");
+    } else if (system_id.len != 0) {
+        try writer.writeAll(" SYSTEM \"");
+        try writeEscapedAttributeValue(writer, system_id);
+        try writer.writeAll("\"");
+    }
+    // Internal subset is not implemented
+    try writer.writeAll(">");
+}
+
 pub fn writeNode(node: *parser.Node, writer: anytype) anyerror!void {
     switch (try parser.nodeType(node)) {
         .element => {
@@ -88,7 +114,7 @@ pub fn writeNode(node: *parser.Node, writer: anytype) anyerror!void {
         .document_fragment => return,
         // document will never be called, but required for completeness.
         .document => return,
-        // done globally instead, but required for completeness.
+        // done globally instead, but required for completeness. Only the outer DOCTYPE should be written
         .document_type => return,
         // deprecated
         .attribute => return,
diff --git a/src/browser/xmlserializer/xmlserializer.zig b/src/browser/xmlserializer/xmlserializer.zig
@@ -35,10 +35,10 @@ pub const XMLSerializer = struct {
 
     pub fn _serializeToString(_: *const XMLSerializer, root: *parser.Node, page: *Page) ![]const u8 {
         var buf = std.ArrayList(u8).init(page.arena);
-        if (try parser.nodeType(root) == .document) {
-            try dump.writeHTML(@as(*parser.Document, @ptrCast(root)), buf.writer());
-        } else {
-            try dump.writeNode(root, buf.writer());
+        switch (try parser.nodeType(root)) {
+            .document => try dump.writeHTML(@as(*parser.Document, @ptrCast(root)), buf.writer()),
+            .document_type => try dump.writeDocType(@as(*parser.DocumentType, @ptrCast(root)), buf.writer()),
+            else => try dump.writeNode(root, buf.writer()),
         }
         return buf.items;
     }
@@ -54,3 +54,12 @@ test "Browser.XMLSerializer" {
         .{ "s.serializeToString(document.getElementById('para'))", "<p id=\"para\"> And</p>" },
     }, .{});
 }
+test "Browser.XMLSerializer with DOCTYPE" {
+    var runner = try testing.jsRunner(testing.tracking_allocator, .{ .html = "<!DOCTYPE html><html><head></head><body></body></html>" });
+    defer runner.deinit();
+
+    try runner.testCases(&.{
+        .{ "new XMLSerializer().serializeToString(document.doctype)", "<!DOCTYPE html>" },
+    }, .{});
+}
+test
