Merge pull request #934 from lightpanda-io/with-base

karlseguin · web-flow · commit bede244598a9 · 2025-08-09T07:37:32.000+08:00
add a --with_base option to fetch
diff --git a/src/browser/dom/node.zig b/src/browser/dom/node.zig
@@ -111,6 +111,9 @@ pub const Node = struct {
     // --------
 
     // Read-only attributes
+    pub fn get_baseURI(_: *parser.Node, page: *Page) ![]const u8 {
+        return page.url.raw;
+    }
 
     pub fn get_firstChild(self: *parser.Node) !?Union {
         const res = try parser.nodeFirstChild(self);
@@ -737,6 +740,10 @@ test "Browser.DOM.node" {
         .{ "link.normalize()", "undefined" },
     }, .{});
 
+    try runner.testCases(&.{
+        .{ "link.baseURI", "https://lightpanda.io/opensource-browser/" },
+    }, .{});
+
     try runner.testCases(&.{
         .{ "content.removeChild(append) !== undefined", "true" },
         .{ "last_child.__proto__.constructor.name !== 'HTMLHeadingElement'", "true" },
diff --git a/src/browser/page.zig b/src/browser/page.zig
@@ -141,16 +141,47 @@ pub const Page = struct {
         repeat_delay.* = 100 * std.time.ns_per_ms;
     }
 
+    pub const DumpOpts = struct {
+        exclude_scripts: bool = false,
+        with_base: bool = false,
+    };
+
     // dump writes the page content into the given file.
-    pub fn dump(self: *const Page, opts: Dump.Opts, out: std.fs.File) !void {
+    pub fn dump(self: *const Page, opts: DumpOpts, out: std.fs.File) !void {
         if (self.raw_data) |raw_data| {
             // raw_data was set if the document was not HTML, dump the data content only.
             return try out.writeAll(raw_data);
         }
 
         // if the page has a pointer to a document, dumps the HTML.
         const doc = parser.documentHTMLToDocument(self.window.document);
-        try Dump.writeHTML(doc, opts, out);
+
+        // if the base si requested, add the base's node in the document's headers.
+        if (opts.with_base) {
+            try self.addDOMTreeBase();
+        }
+
+        try Dump.writeHTML(doc, .{
+            .exclude_scripts = opts.exclude_scripts,
+        }, out);
+    }
+
+    // addDOMTreeBase modifies the page's document to add a <base> tag after
+    // <head>.
+    // If <head> is missing, the function returns silently.
+    fn addDOMTreeBase(self: *const Page) !void {
+        const doc = parser.documentHTMLToDocument(self.window.document);
+        std.debug.assert(doc.is_html);
+
+        // find <head> tag
+        const list = try parser.documentGetElementsByTagName(doc, "head");
+        const head = try parser.nodeListItem(list, 0) orelse return;
+
+        const base = try parser.documentCreateElement(doc, "base");
+        try parser.elementSetAttribute(base, "href", self.url.raw);
+
+        const Node = @import("dom/node.zig").Node;
+        try Node.prepend(head, &[_]Node.NodeOrText{.{ .node = parser.elementToNode(base) }});
     }
 
     pub fn fetchModuleSource(ctx: *anyopaque, src: []const u8) !?[]const u8 {
diff --git a/src/main.zig b/src/main.zig
@@ -134,7 +134,10 @@ fn run(alloc: Allocator) !void {
 
             // dump
             if (opts.dump) {
-                try page.dump(.{ .exclude_scripts = opts.noscript }, std.io.getStdOut());
+                try page.dump(.{
+                    .exclude_scripts = opts.noscript,
+                    .with_base = opts.withbase,
+                }, std.io.getStdOut());
             }
         },
         else => unreachable,
@@ -213,6 +216,7 @@ const Command = struct {
         dump: bool = false,
         common: Common,
         noscript: bool = false,
+        withbase: bool = false,
     };
 
     const Common = struct {
@@ -277,6 +281,7 @@ const Command = struct {
             \\--dump          Dumps document to stdout.
             \\                Defaults to false.
             \\--noscript      Exclude <script> tags in dump. Defaults to false.
+            \\--with_base     Add a <base> tag in dump. Defaults to false.
             \\
         ++ common_options ++
             \\
@@ -351,13 +356,19 @@ fn inferMode(opt: []const u8) ?App.RunMode {
         return .serve;
     }
 
+    if (std.mem.startsWith(u8, opt, "--") == false) {
+        return .fetch;
+    }
+
     if (std.mem.eql(u8, opt, "--dump")) {
         return .fetch;
     }
+
     if (std.mem.eql(u8, opt, "--noscript")) {
         return .fetch;
     }
-    if (std.mem.startsWith(u8, opt, "--") == false) {
+
+    if (std.mem.eql(u8, opt, "--with_base")) {
         return .fetch;
     }
 
@@ -442,7 +453,8 @@ fn parseFetchArgs(
     args: *std.process.ArgIterator,
 ) !Command.Fetch {
     var dump: bool = false;
-    var noscript: bool = true;
+    var noscript: bool = false;
+    var withbase: bool = false;
     var url: ?[]const u8 = null;
     var common: Command.Common = .{};
 
@@ -457,6 +469,11 @@ fn parseFetchArgs(
             continue;
         }
 
+        if (std.mem.eql(u8, "--with_base", opt)) {
+            withbase = true;
+            continue;
+        }
+
         if (try parseCommonArg(allocator, opt, args, &common)) {
             continue;
         }
@@ -483,6 +500,7 @@ fn parseFetchArgs(
         .dump = dump,
         .common = common,
         .noscript = noscript,
+        .withbase = withbase,
     };
 }
 
