fix(duckduckgo): preserve words around highlighted matches (#111460)

2026-08-12 21:53:00 -06:00 · 2026-07-20 10:09:09 +08:00
parent 27f05c8993
commit c5bc6c3d2d
2 changed files with 18 additions and 1 deletions
@@ -53,7 +53,9 @@ function decodeHtmlEntities(text: string): string {
 }

 function stripHtml(html: string): string {
-  return html
+  // DuckDuckGo match highlights can occur inside words, so remove them without adding whitespace.
+  const withoutMatchHighlights = html.replace(/<\/?b\b[^>]*>/gi, "");
+  return withoutMatchHighlights
    .replace(/<[^>]+>/g, " ")
    .replace(/\s+/g, " ")
    .trim();
@@ -254,6 +254,21 @@ describe("duckduckgo web search provider", () => {
    ]);
  });

+  it("keeps inline result markup from splitting words", () => {
+    const html = `
+      <a class="result__a" href="https://example.com/cafe">Caf<b>é</b> guide</a>
+      <a class="result__snippet">Find the best caf<b>é</b> near you.</a>
+    `;
+
+    expect(ddgClientTesting.parseDuckDuckGoHtml(html)).toEqual([
+      {
+        title: "Café guide",
+        url: "https://example.com/cafe",
+        snippet: "Find the best café near you.",
+      },
+    ]);
+  });
+
  it("detects bot challenge pages without flagging ordinary result snippets", () => {
    const challengeHtml = `
      <html>