Source" link + // Row 3: Spacer + // The extracted HTML is wrapped in a `
` + private extractZeroClickInfo(): ZeroClickInfo | null { + const tables = Array.from(this.document.querySelectorAll('table[border="0"]')); + + for (const table of tables) { + // Skip result tables — zero-click tables have no .result-link + if (table.querySelector('a.result-link')) continue; + + const rows = Array.from(table.querySelectorAll('tr')); + if (rows.length < 2) continue; + + const firstCell = rows[0].querySelector('td'); + if (!firstCell) continue; + + const firstText = firstCell.textContent || ''; + if (!firstText.includes('Zero-click info:')) continue; + + // Header: "Zero-click info: Title" + const headerHtml = this.unwrapLinksInHtml(firstCell); + + // Body: description text + "More atSource" link + const bodyCell = rows[1].querySelector('td'); + if (!bodyCell) continue; + + // Clone so we can safely remove the "More at" link + const cleanBody = bodyCell.cloneNode(true) as Element; + const links = Array.from(cleanBody.querySelectorAll('a')); + for (const link of links) { + if (link.textContent?.includes('More at')) { + link.remove(); + break; + } + } + + const bodyHtml = this.unwrapLinksInHtml(cleanBody); + + // Text-only version for the description variable + const textContent = cleanBody.textContent?.trim() || ''; + + return { header: headerHtml, body: bodyHtml, textContent }; + } + + return null; + } + + // Unwraps DuckDuckGo redirect URLs in all tags within the given element. + // Operates on the element's inner HTML string. + private unwrapLinksInHtml(el: Element): string { + const clone = el.cloneNode(true) as Element; + clone.querySelectorAll('a').forEach(a => { + const href = a.getAttribute('href') || ''; + const unwrapped = this.unwrapRedirectUrl(href); + if (unwrapped !== href) { + a.setAttribute('href', unwrapped); + } + }); + return serializeHTML(clone); + } + + // Iterates over the top-level result tables and collects each + // title / URL / snippet triplet into a structured array. + // + // The lite layout uses a flat table where each result is + // three consecutive rows: + // 1. title link + // 2. snippet + // 3. blank spacer + private extractResults(): SearchResult[] { + const results: SearchResult[] = []; + const tables = Array.from(this.document.querySelectorAll('table[border="0"]')); + + for (const table of tables) { + // Skip zero-click info tables (no .result-link) + if (!table.querySelector('a.result-link')) continue; + + const rows = Array.from(table.querySelectorAll('tr')); + let i = 0; + + while (i < rows.length) { + const row = rows[i]; + const linkEl = row.querySelector('a.result-link'); + + if (linkEl) { + const title = linkEl.textContent?.trim() || ''; + const href = linkEl.getAttribute('href') || ''; + const url = this.unwrapRedirectUrl(href); + + i++; + const snippetRow = rows[i]; + const snippetEl = snippetRow?.querySelector('.result-snippet'); + const snippet = snippetEl ? serializeHTML(snippetEl) : ''; + + // Skip the blank spacer row + i++; + i++; + + if (title && url) { + results.push({ title, url, snippet }); + } + } else { + i++; + } + } + } + + return results; + } + + + // Builds the final ordered list HTML. When there are no results a + // short "No results" message is returned instead of an empty string. + // Zero-click info, when present, is prepended as a blockquote. + private buildListingHtml(results: SearchResult[], zeroClick?: ZeroClickInfo | null): string { + let html = ''; + + if (zeroClick) { + html += ''; + html += `'; + } + + if (results.length === 0) { + html += '${zeroClick.header}
`; + if (zeroClick.body) { + html += `${zeroClick.body}
`; + } + html += 'No results found.
'; + return html; + } + + const items = results.map(result => { + let itemHtml = ''; + itemHtml += `${escapeHtml(result.title)}`; + + if (result.snippet) { + itemHtml += ` '; + return itemHtml; + }); + + html += `${result.snippet}
`; + } + + itemHtml += '${items.join('')}
`; + + return html; + } +} diff --git a/tests/expected/duckduckgo-lite.md b/tests/expected/duckduckgo-lite.md new file mode 100644 index 000000000..b77177aad --- /dev/null +++ b/tests/expected/duckduckgo-lite.md @@ -0,0 +1,19 @@ +```json +{ + "title": "test query", + "author": "", + "site": "DuckDuckGo Lite", + "published": "" +} +``` + +> Zero-click info: [Gasoline](https://en.wikipedia.org/wiki/Gasoline) +> +> Gasoline or petrol is a petrochemical product characterized as a transparent, yellowish and flammable liquid normally used as a fuel for spark-ignited internal combustion engines. + +1. [Example Result Title](https://example.com/test-page) + This is a **test** snippet describing the **query** result with some highlighted terms. +2. [Another Example Page](https://example.org/another-page?foo=bar) + More content about **test** and **query** functionality. +3. [Foo Example Dot Net](https://foo.example.net/page/) + A third result with different **query** content here. \ No newline at end of file diff --git a/tests/fixtures/duckduckgo-lite.html b/tests/fixtures/duckduckgo-lite.html new file mode 100644 index 000000000..0b93885b7 --- /dev/null +++ b/tests/fixtures/duckduckgo-lite.html @@ -0,0 +1,96 @@ + + + + + + +test query at DuckDuckGo + + ++
+ DuckDuckGo +++ + + +
+ +
+
+ ++ +Zero-click info: Gasoline ++ ++ Gasoline or petrol is a petrochemical product characterized as a transparent, yellowish and flammable liquid normally used as a fuel for spark-ignited internal combustion engines. + More at +Wikipedia++ ++ +
+ ++ ++ 1. + ++ Example Result Title + ++ ++ This is a test snippet describing the query result with some highlighted terms. ++ ++ + + ++ 2. + ++ Another Example Page + ++ ++ + More content about test and query functionality. + ++ ++ + + ++ 3. + ++ Foo Example Dot Net + ++ ++ + A third result with different query content here. + ++ ++ + + + +