✨feature: First push to git

src/parsing/generic.ts

@@ -0,0 +1,42 @@
+import * as cheerio from 'cheerio';
+import { ExtractedLink, ParserInput, ParserPlugin } from './types.js';
+
+function nearestSection($: cheerio.CheerioAPI, element: any): string | undefined {
+  const previous = $(element).prevAll('h1,h2,h3,h4,h5,h6,strong,b').first().text().trim();
+  if (previous) {
+    return previous;
+  }
+  const parentPrevious = $(element)
+    .parent()
+    .prevAll('h1,h2,h3,h4,h5,h6,p,tr')
+    .first()
+    .text()
+    .trim();
+  return parentPrevious || undefined;
+}
+
+export const genericParser: ParserPlugin = {
+  name: 'generic',
+  matches: () => true,
+  parse(input: ParserInput): ExtractedLink[] {
+    const $ = cheerio.load(input.html);
+    return $('a[href]')
+      .toArray()
+      .map((element) => {
+        const anchor = $(element);
+        const title = anchor.text().replace(/\s+/g, ' ').trim() || anchor.attr('aria-label') || '';
+        const url = anchor.attr('href') ?? '';
+        const context = anchor.closest('p,li,td,div').text().replace(/\s+/g, ' ').trim();
+
+        return {
+          url,
+          title,
+          description: context && context !== title ? context : '',
+          sourceText: title,
+          section: nearestSection($, element),
+          context
+        };
+      })
+      .filter((link) => Boolean(link.url));
+  }
+};