feat: per-keyword resume — restart picks up from last completed keyword, not keyword 1

job_searcher.mjs

@@ -22,7 +22,7 @@ import { verifyLogin as wfLogin, searchWellfound } from './lib/wellfound.mjs';
 import { sendTelegram, formatSearchSummary } from './lib/notify.mjs';
 import { DEFAULT_FIRST_RUN_DAYS } from './lib/constants.mjs';
 import { generateKeywords } from './lib/keywords.mjs';
-import { initProgress, isCompleted, markComplete } from './lib/search_progress.mjs';
+import { initProgress, isCompleted, markComplete, getKeywordStart, markKeywordComplete } from './lib/search_progress.mjs';
 import { ensureLoggedIn } from './lib/session.mjs';
 
 async function main() {
@@ -112,7 +112,9 @@ async function main() {
           console.log(`  [${search.name}] ✓ already done, skipping`);
           continue;
         }
-        const effectiveSearch = { ...search, filters: { ...search.filters, posted_within_days: lookbackDays } };
+        const keywordStart = getKeywordStart('linkedin', search.name);
+        if (keywordStart > 0) console.log(`  [${search.name}] resuming from keyword ${keywordStart + 1}/${search.keywords.length}`);
+        const effectiveSearch = { ...search, keywords: search.keywords.slice(keywordStart), filters: { ...search.filters, posted_within_days: lookbackDays } };
         let queryFound = 0, queryAdded = 0;
         await searchLinkedIn(liBrowser.page, effectiveSearch, {
           onPage: (pageJobs) => {
@@ -122,6 +124,9 @@ async function main() {
             queryFound += pageJobs.length;
             queryAdded += added;
             process.stdout.write(`\r  [${search.name}] ${queryFound} found, ${queryAdded} new...`);
+          },
+          onKeyword: (ki) => {
+            markKeywordComplete('linkedin', search.name, keywordStart + ki);
           }
         });
         console.log(`\r  [${search.name}] ${queryFound} found, ${queryAdded} new`);

lib/linkedin.mjs

@@ -16,7 +16,8 @@ export async function verifyLogin(page) {
   return page.url().includes('/feed');
 }
 
-export async function searchLinkedIn(page, search, { onPage } = {}) {
+export async function searchLinkedIn(page, search, { onPage, onKeyword } = {}) {
+  const callbacks = { onPage, onKeyword };
   const jobs = [];
   const seenIds = new Set();
 
@@ -124,7 +125,7 @@ export async function searchLinkedIn(page, search, { onPage } = {}) {
         jobs.push(job);
       }
 
-      if (pageJobs.length > 0 && onPage) onPage(pageJobs);
+      if (pageJobs.length > 0 && callbacks.onPage) callbacks.onPage(pageJobs);
 
       const nextBtn = await page.$('button[aria-label="View next page"]');
       if (!nextBtn) break;
@@ -132,6 +133,8 @@ export async function searchLinkedIn(page, search, { onPage } = {}) {
       await page.waitForTimeout(PAGE_LOAD_WAIT);
       pageNum++;
     }
+    // Mark keyword complete after all its pages are done
+    callbacks.onKeyword?.(ki);
   }
 
   return jobs;

lib/search_progress.mjs

@@ -1,6 +1,6 @@
 /**
  * search_progress.mjs — Track which searches have completed
- * Enables resume on restart without re-running finished searches
+ * Enables resume on restart — skips completed tracks and completed keywords within a track
  */
 import { readFileSync, writeFileSync, existsSync, unlinkSync } from 'fs';
 
@@ -12,24 +12,22 @@ export function initProgress(dataDir, lookbackDays) {
 
   if (existsSync(progressPath)) {
     const saved = JSON.parse(readFileSync(progressPath, 'utf8'));
-    // Only resume if same lookback window
     if (saved.lookback_days === lookbackDays) {
       progress = saved;
-      const done = progress.completed.length;
+      const done = progress.completed?.length ?? 0;
       if (done > 0) {
-        console.log(`🔁 Resuming — skipping already-completed: ${progress.completed.join(', ')}\n`);
+        console.log(`🔁 Resuming — completed tracks: ${progress.completed.join(', ')}\n`);
       }
       return progress;
     }
     console.log(`🆕 New lookback window (${lookbackDays}d), starting fresh\n`);
   }
 
-  // Fresh start
   progress = {
     lookback_days: lookbackDays,
     started_at: Date.now(),
     completed: [],
-    pending: [],
+    keyword_progress: {},  // key: "platform:track" → last completed keyword index (0-based)
   };
   save();
   return progress;
@@ -40,11 +38,29 @@ export function isCompleted(platform, track) {
   return progress.completed.includes(`${platform}:${track}`);
 }
 
+/** Returns the index of the first keyword to run (skips already-completed ones) */
+export function getKeywordStart(platform, track) {
+  if (!progress) return 0;
+  const key = `${platform}:${track}`;
+  const last = progress.keyword_progress?.[key] ?? -1;
+  return last + 1; // resume from next keyword after last completed
+}
+
+/** Call after each keyword completes */
+export function markKeywordComplete(platform, track, keywordIndex) {
+  if (!progress) return;
+  const key = `${platform}:${track}`;
+  if (!progress.keyword_progress) progress.keyword_progress = {};
+  progress.keyword_progress[key] = keywordIndex;
+  save();
+}
+
 export function markComplete(platform, track, stats) {
   if (!progress) return;
   const key = `${platform}:${track}`;
-  progress.pending = progress.pending.filter(k => k !== key);
   if (!progress.completed.includes(key)) progress.completed.push(key);
+  // Clean up per-keyword progress for completed track
+  if (progress.keyword_progress) delete progress.keyword_progress[key];
   progress[`stats:${key}`] = { ...stats, completed_at: Date.now() };
   save();
 }