fix(fetch): preserve details and discard empty SVG placeholders

2026-09-26 16:01:30 +00:00 · 2026-09-20 20:50:03 +08:00
parent b75e26fca2
commit 7985fcb7bc
4 changed files with 63 additions and 3 deletions
@@ -234,7 +234,20 @@ impl<'a, D: Dom + ?Sized> Machine<'a, D> {
                    .attribute(node, "src")
                    .filter(|src| !src.is_empty())
                {
-                    self.writer().image(alt, src, title);
+                    // Lazy-image placeholders carry no visible image. Keep
+                    // accessible alternate text without emitting a data URI.
+                    if is_empty_svg_placeholder(src) {
+                        if !alt.is_empty()
+                            && !self
+                                .dom
+                                .attribute(node, "aria-hidden")
+                                .is_some_and(|value| value.trim().eq_ignore_ascii_case("true"))
+                        {
+                            self.writer().text(alt);
+                        }
+                    } else {
+                        self.writer().image(alt, src, title);
+                    }
                }
                return;
            }
@@ -428,6 +441,13 @@ impl<'a, D: Dom + ?Sized> Machine<'a, D> {
    }
 }

+fn is_empty_svg_placeholder(src: &str) -> bool {
+    let lower = src.to_ascii_lowercase();
+    lower.starts_with("data:image/svg+xml,%3csvg")
+        && lower.ends_with("%3c/svg%3e")
+        && lower.matches("%3c").count() == 2
+}
+
 fn class_language(class: Option<&str>) -> Option<&str> {
    class?
        .split_ascii_whitespace()
@@ -225,6 +225,21 @@ fn images_without_sources_do_not_emit_placeholder_markdown() {
    }
 }

+#[test]
+fn empty_svg_lazy_image_placeholders_do_not_emit_data_uris() {
+    let placeholder = "data:image/svg+xml,%3Csvg%20xmlns='http://www.w3.org/2000/svg'%20viewBox='0%200%20720%20960'%3E%3C/svg%3E";
+    let html = format!("before<img alt='Recipe photo' src=\"{placeholder}\">after");
+    assert_eq!(markdown(&html, false), "beforeRecipe photoafter");
+
+    let html =
+        format!("before<img alt='Recipe photo' aria-hidden='true' src=\"{placeholder}\">after");
+    assert_eq!(markdown(&html, false), "beforeafter");
+
+    let real_svg = "data:image/svg+xml,%3Csvg%3E%3Cpath%20d='M0%200'/%3E%3C/svg%3E";
+    let html = format!("<img alt='Logo' src=\"{real_svg}\">");
+    assert!(markdown(&html, false).contains(real_svg));
+}
+
 #[test]
 fn list_paragraphs_and_code_blocks_require_blank_lines_between_items() {
    for (html, expected) in [
@@ -157,8 +157,6 @@ fn apply_strip_options(dom: &mut NativeDom, strip: RendererPageDumpStripOptions)
                        | "dialog"
                        | "menu"
                        | "menuitem"
-                        | "details"
-                        | "summary"
                ));
        if should_remove {
            remove.push(node_id);
@@ -129,6 +129,33 @@ async fn render_page_dump_default_html_uses_renderer_live_serialize() -> Result<
    Ok(())
 }

+#[tokio::test]
+async fn strip_ui_keeps_content_inside_details() -> Result<()> {
+    let (_browser, mut page, http_server) = load_page(
+        "<!doctype html><html><body><nav>Menu</nav><main><details><summary>Product details</summary><p>Critical product description</p></details></main></body></html>",
+    )
+    .await?;
+    let rendered = render_page_dump_with_options_async(
+        &mut page,
+        DumpFormat::Markdown,
+        StripOptions {
+            js: false,
+            css: false,
+            ui: true,
+        },
+        false,
+        false,
+        false,
+        None,
+    )
+    .await?;
+    assert!(!rendered.contains("Menu"));
+    assert!(rendered.contains("Product details"));
+    assert!(rendered.contains("Critical product description"));
+    http_server.abort();
+    Ok(())
+}
+
 #[tokio::test]
 async fn render_page_dump_postprocessed_html_uses_renderer_live_dump() -> Result<()> {
    let (_browser, mut page, http_server) = load_page(