Merge pull request #3663 from lightpanda-io/markdown-anchor-no-href

Treat <a> without href as text in markdown and clutter strip

src/browser/RenderTree.zig

@@ -343,7 +343,7 @@ fn measureChild(self: *const RenderTree, child: Child, where: Measure.Where, m:
             const el = child.node.subtype(Element);
             const inner: Measure.Where = .{
                 .shell = where.shell or dump_html.isShellElement(el),
-                .link = where.link or el.getTag() == .anchor,
+                .link = where.link or (el.getTag() == .anchor and el.getAttributeInterned("href") != null),
             };
             if (el.is(Slot)) |slot| {
                 var it = self.slotted(slot);

src/browser/clutter.zig

@@ -286,9 +286,10 @@ const Pass = struct {
 
                 var inner = ctx;
                 switch (tag) {
-                    .anchor => {
+                    // Unlike readability: without href, <a> is a placeholder
+                    // (tag chips, JS buttons), and its text is not link text.
+                    .anchor => if (el.getAttributeSafe(comptime .wrap("href"))) |href| {
                         inner.in_link = true;
-                        const href = el.getAttributeSafe(comptime .wrap("href")) orelse "";
                         inner.hash_link = href.len > 0 and href[0] == '#';
                     },
                     .h1, .h2, .h3, .h4, .h5, .h6 => inner.in_heading = true,
@@ -921,6 +922,15 @@ test "clutter: qualifying sibling paragraphs come along" {
     try testing.expectEqual(null, std.mem.indexOf(u8, out, "[a]"));
 }
 
+test "clutter: anchors without href are not link text" {
+    const quote = "<div class=\"quote\"><span>" ++ prose ++ "</span><div class=\"tags\">Tags: <a class=\"tag\">bread</a></div></div>";
+    // A short quote with many tags: over 0.2 link density if tags counted.
+    const tagged = "<div class=\"quote\"><span>I have not failed. I've just found ten thousand ways that won't work.</span><span>by Thomas A. Edison</span>" ++
+        "<div class=\"tags\">Tags: <a class=\"tag\">edison</a> <a class=\"tag\">failure</a> <a class=\"tag\">inspirational</a> <a class=\"tag\">paraphrased</a></div></div>";
+    const out = try extract("<div class=\"quotes\">" ++ quote ++ quote ++ tagged ++ quote ++ "</div>");
+    try testing.expectEqual(true, std.mem.indexOf(u8, out, "ten thousand ways") != null);
+}
+
 test "clutter: the prune set lives in the caller's allocator" {
     const frame = try testing.createFrame();
     defer testing.test_session.closeAllPages();

src/browser/markdown.zig

@@ -82,9 +82,11 @@ const Context = struct {
     // what follows after the children
     const Epilogue = union(enum) {
         none,
+        // A standalone anchor without href: its own line, but no link syntax.
+        standalone_text,
         element: Element.Tag,
-        block_anchor: struct { href: ?[]const u8, label: ?[]const u8 },
-        inline_anchor: struct { href: ?[]const u8, standalone: bool },
+        block_anchor: struct { href: []const u8, label: ?[]const u8 },
+        inline_anchor: struct { href: []const u8, standalone: bool },
     };
 
     fn deinit(self: *Context) void {
@@ -259,15 +261,18 @@ const Context = struct {
             },
             .anchor => {
                 const frame = self.frame;
+                // Without href, <a> is a placeholder, not a hyperlink.
+                const href_raw = el.getAttributeInterned("href") orelse {
+                    if (!RenderTree.isStandaloneAnchor(el, frame)) {
+                        return self.open(.init(self.tree.content(el, boxed)), .none);
+                    }
+                    try self.ensureNewline();
+                    return self.open(.init(self.tree.content(el, boxed)), .standalone_text);
+                };
+
                 const info = RenderTree.analyzeContent(el.asNode(), frame);
                 const label = getAnchorLabel(el);
-                const href_raw = el.getAttributeInterned("href");
-
-                if (!info.has_visible and label == null and href_raw == null) {
-                    return;
-                }
-
-                const href = if (href_raw) |h| URL.resolve(frame.local_arena, frame.base(), h, .{ .encoding = frame.charset }) catch h else null;
+                const href = URL.resolve(frame.local_arena, frame.base(), href_raw, .{ .encoding = frame.charset }) catch href_raw;
 
                 if (info.has_block) {
                     return self.open(.init(self.tree.content(el, boxed)), .{ .block_anchor = .{
@@ -309,22 +314,23 @@ const Context = struct {
     fn close(self: *Context, epilogue: Epilogue) Error!void {
         const tag = switch (epilogue) {
             .none => return,
+            .standalone_text => {
+                try self.ensureNewline();
+                return;
+            },
             .element => |t| t,
             .block_anchor => |anchor| {
-                const href = anchor.href orelse return;
                 try self.ensureNewline();
                 try self.writer.writeByte('[');
-                try self.writer.writeAll(anchor.label orelse href);
+                try self.writer.writeAll(anchor.label orelse anchor.href);
                 try self.writer.writeAll("](");
-                try self.writer.writeAll(href);
+                try self.writer.writeAll(anchor.href);
                 try self.writer.writeAll(")\n");
                 return;
             },
             .inline_anchor => |anchor| {
                 try self.writer.writeAll("](");
-                if (anchor.href) |h| {
-                    try self.writer.writeAll(h);
-                }
+                try self.writer.writeAll(anchor.href);
                 try self.writer.writeByte(')');
                 if (anchor.standalone) {
                     try self.writer.writeByte('\n');
@@ -756,6 +762,20 @@ test "browser.markdown: anchor fallback label" {
     , "[](http://localhost/no-label)\n");
 }
 
+test "browser.markdown: anchor without href is plain text" {
+    try testMarkdownHTML(
+        \\<p>Read the <a class="term">glossary</a> first.</p>
+    , "\nRead the glossary first.\n");
+
+    try testMarkdownHTML(
+        \\<nav><a class="x">Home</a><a class="x"><b>About</b></a></nav>
+    , "Home\n**About**\n");
+
+    try testMarkdownHTML(
+        \\<a aria-label="Menu"><svg></svg></a>
+    , "");
+}
+
 test "browser.markdown: hidden elements are skipped" {
     try testMarkdownHTML(
         \\<p>before</p>