Skip to content

Lightpanda: the headless browser designed for AI and automation

markdown: write a text/markdown document as-is

Jay Hemnani68dbf42ef3dc
diff --git a/src/TestHTTPServer.zig b/src/TestHTTPServer.zig
index 1d920dbe8..1633cc382 100644
--- a/src/TestHTTPServer.zig
+++ b/src/TestHTTPServer.zig
@@ -166,6 +166,10 @@ fn getContentType(file_path: []const u8) []const u8 {
return "application/xhtml+xml";
}
+ if (std.mem.endsWith(u8, file_path, ".md")) {
+ return "text/markdown";
+ }
+
if (std.mem.endsWith(u8, file_path, ".xml")) {
// some wpt tests do this
return "text/xml";
diff --git a/src/browser/markdown.zig b/src/browser/markdown.zig
index 585fc18f2..d4ab05239 100644
--- a/src/browser/markdown.zig
+++ b/src/browser/markdown.zig
@@ -24,6 +24,7 @@ const RenderTree = @import("RenderTree.zig");
const StyleManager = @import("StyleManager.zig");
const Node = @import("webapi/Node.zig");
+const Document = @import("webapi/Document.zig");
const Element = @import("webapi/Element.zig");
const Slot = @import("webapi/element/html/Slot.zig");
@@ -477,8 +478,29 @@ const Context = struct {
}
};
+/// A text/markdown response is loaded as a text document, its body a single
+/// `<pre>`. Converting that would fence the whole file, so the document's
+/// text, which is the markdown the server sent, is written as-is.
+fn markdownSource(node: *Node) ?*Node {
+ const doc = node.is(Document) orelse return null;
+ if (!std.mem.eql(u8, doc.getContentType(), "text/markdown")) return null;
+ const root = doc.getDocumentElement() orelse return null;
+ return root.asNode();
+}
+
+fn writeSource(source: *Node, opts: Opts, writer: *std.Io.Writer) !void {
+ var lw = LimitedWriter.init(writer, opts.max_bytes);
+ source.getTextContent(&lw.writer) catch |err| {
+ if (!lw.truncated) return err;
+ try writer.writeAll(truncation_marker);
+ };
+}
+
pub fn dump(state: RenderTree.State, opts: Opts, writer: *std.Io.Writer, frame: *Frame) !void {
const node = state.root;
+ if (markdownSource(node)) |source| {
+ return writeSource(source, opts, writer);
+ }
if (opts.max_bytes) |limit| {
var lw = LimitedWriter.init(writer, limit);
var ctx: Context = .{
@@ -956,6 +978,21 @@ test "browser.markdown: max_bytes truncates with marker" {
try testing.expect(out.len <= 50 + truncation_marker.len);
}
+test "browser.markdown: a text/markdown document is written as-is" {
+ var page = try testing.pageTest("markdown_source.md", .{});
+ defer page.close();
+ const frame = page.frame().?;
+ const source = @embedFile("tests/markdown_source.md");
+
+ var aw: std.Io.Writer.Allocating = .init(testing.arena_allocator);
+ try dump(.{ .root = frame.window._document.asNode() }, .{}, &aw.writer, frame);
+ try testing.expectString(source, aw.written());
+
+ var capped: std.Io.Writer.Allocating = .init(testing.arena_allocator);
+ try dump(.{ .root = frame.window._document.asNode() }, .{ .max_bytes = 7 }, &capped.writer, frame);
+ try testing.expectString("# Title" ++ truncation_marker, capped.written());
+}
+
// Builds a shadow host <div>, populates its light DOM with `light` and its
// (open) shadow tree with `shadow`, then dumps the host. Declarative shadow DOM
// parsing isn't implemented, so the shadow tree is attached imperatively.
diff --git a/src/browser/tests/markdown_source.md b/src/browser/tests/markdown_source.md
new file mode 100644
index 000000000..69a67e088
--- /dev/null
+++ b/src/browser/tests/markdown_source.md
@@ -0,0 +1,7 @@
+# Title
+
+Some *emphasis*, a [link](/x) & a <tag>.
+
+```zig
+const x = 1;
+```