vipcare 0.6.1 → 0.6.3

package/bin/vip.js

@@ -64,14 +64,19 @@ function gatherData(person) {
 
   if (person.linkedinUrl) sources.push(person.linkedinUrl);
 
+  // Collect all search results into one consolidated file
+  const searchEntries = [];
+
   if (person.rawSnippets.length) {
     rawParts.push('=== Web Search Results ===');
-    rawParts.push(...person.rawSnippets);
-    // Save each snippet as a raw source file
     person.rawSnippets.forEach((snippet, i) => {
-      const url = person.otherUrls?.[i] || `search_result_${i}`;
-      const safeName = url.replace(/https?:\/\//, '').replace(/[^\w.-]/g, '_').substring(0, 60);
-      saveRawSource(personSlug, safeName, `# Search Result\n\nSource: ${url}\nFetched: ${new Date().toISOString()}\n\n${snippet}`);
+      if (!snippet.trim()) return; // skip empty
+      if (snippet.trim().length < 30) return; // too short to be useful
+      const lines = snippet.trim().split('\n');
+      if (lines.length >= 2 && lines[0].trim() === lines[1].trim()) return; // title == body
+      rawParts.push(snippet);
+      const url = person.otherUrls?.[i] || '';
+      if (url) searchEntries.push({ url, snippet });
     });
   }
 
@@ -79,13 +84,28 @@ function gatherData(person) {
     console.log(c.dim(`  Searching the web for ${person.name}...`));
     const results = searchPerson(person.name);
     for (const r of results) {
-      rawParts.push(`${r.title}\n${r.body}`);
+      if (!r.body?.trim() && !r.title?.trim()) continue; // skip empty
+      const snippet = `${r.title}\n${r.body}`;
+      if (snippet.trim().length < 30) continue; // too short to be useful
+      if (r.title?.trim() === r.body?.trim()) continue; // title == body (junk category)
+      if (!r.body?.trim() || r.body.trim().length < 20) continue; // no real body content
+      rawParts.push(snippet);
       if (!sources.includes(r.url)) sources.push(r.url);
-      const safeName = r.url.replace(/https?:\/\//, '').replace(/[^\w.-]/g, '_').substring(0, 60);
-      saveRawSource(personSlug, safeName, `# ${r.title}\n\nSource: ${r.url}\nFetched: ${new Date().toISOString()}\n\n${r.body}`);
+      searchEntries.push({ url: r.url, snippet });
     }
   }
 
+  // Save all search results in ONE file
+  if (searchEntries.length) {
+    const timestamp = new Date().toISOString();
+    const content = `# Web Search Results\n\nFetched: ${timestamp}\nQuery: ${person.name}\n\n` +
+      searchEntries
+        .filter(e => e.snippet.trim())
+        .map(e => `---\n\n**Source:** ${e.url}\n\n${e.snippet}`)
+        .join('\n\n');
+    saveRawSource(personSlug, 'web_search', content);
+  }
+
   console.log(c.dim(`  Raw data saved to ${path.join(getProfilesDir(), '.raw', personSlug)}/`));
   return [rawParts.join('\n\n'), sources];
 }
@@ -495,6 +515,7 @@ program.command('youtube-search')
 
 // --- card ---
 program.command('card')
+  .alias('open-cards')
   .description('Generate and serve H5 baseball card page')
   .option('-o, --output <path>', 'Output HTML file', path.join(os.homedir(), '.vip', 'cards', 'index.html'))
   .option('-p, --port <port>', 'Server port', '3000')

package/lib/card.js

@@ -19,8 +19,11 @@ export function generateCards(profiles, outputPath) {
     outputPath = path.join(os.homedir(), '.vip', 'cards', 'index.html');
   }
   const cards = [];
+  const seenSlugs = new Set();
 
   for (const p of profiles) {
+    if (seenSlugs.has(p.slug)) continue; // skip duplicates
+    seenSlugs.add(p.slug);
     const content = loadProfile(p.slug);
     if (!content) continue;
 

package/lib/fetchers/search.js

@@ -76,8 +76,15 @@ function searchViaDdgApi(query) {
     if (data.RelatedTopics) {
       for (const topic of data.RelatedTopics) {
         if (topic.Text && topic.FirstURL) {
+          // Skip DDG category pages (e.g. duckduckgo.com/c/SpaceX_people)
+          if (topic.FirstURL.includes('duckduckgo.com/c/')) continue;
+          const title = topic.Text.substring(0, 80);
+          // Skip entries where title and body are identical (just a category name)
+          if (title === topic.Text && topic.Text.length < 20) continue;
+          // Skip entries with body too short to be useful
+          if (topic.Text.length < 20) continue;
           results.push({
-            title: topic.Text.substring(0, 80),
+            title,
             url: topic.FirstURL,
             body: topic.Text,
           });
@@ -86,8 +93,15 @@ function searchViaDdgApi(query) {
         if (topic.Topics) {
           for (const sub of topic.Topics) {
             if (sub.Text && sub.FirstURL) {
+              // Skip DDG category pages
+              if (sub.FirstURL.includes('duckduckgo.com/c/')) continue;
+              const subTitle = sub.Text.substring(0, 80);
+              // Skip entries where title and body are identical
+              if (subTitle === sub.Text && sub.Text.length < 20) continue;
+              // Skip entries with body too short to be useful
+              if (sub.Text.length < 20) continue;
               results.push({
-                title: sub.Text.substring(0, 80),
+                title: subTitle,
                 url: sub.FirstURL,
                 body: sub.Text,
               });

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "vipcare",
-  "version": "0.6.1",
+  "version": "0.6.3",
   "description": "Auto-build VIP person profiles from Twitter/LinkedIn public data",
   "type": "module",
   "bin": {