oopa

apps/web/scripts/scrape-training-dataset.ts

@@ -591,7 +591,9 @@ function countImagesInDir(classDir: string): number {
 function reconcileClassCount(classDir: string, progressCount: number): number {
   const fileCount = countImagesInDir(classDir);
   if (fileCount < progressCount) {
-    console.log(`    ↻ File count (${fileCount}) < progress count (${progressCount}) — reconciling`);
+    console.log(
+      `    ↻ File count (${fileCount}) < progress count (${progressCount}) — reconciling`,
+    );
     return fileCount;
   }
   return progressCount;
@@ -776,6 +778,13 @@ async function main() {
   mkdirSync(DATASET_DIR, { recursive: true });
   const progress = loadProgress();
 
+  // If all phases complete, exit early
+  if (progress.phase === 3) {
+    console.log("  ✓ All phases already complete. Delete .progress.json to re-run.");
+    await closeDb();
+    return;
+  }
+
   const startTime = Date.now();
 
   // ── Phase 1: Core set ──────────────────────────────────────────────────
@@ -786,7 +795,9 @@ async function main() {
 
   const coreStart = progress.phase === 0 ? progress.phaseIndex : 0;
   if (coreStart > 0) {
-    console.log(`  Resuming from disease #${coreStart + 1} (${((coreStart / coreDiseases.length) * 100).toFixed(0)}% done)`);
+    console.log(
+      `  Resuming from disease #${coreStart + 1} (${((coreStart / coreDiseases.length) * 100).toFixed(0)}% done)`,
+    );
   }
 
   for (let i = coreStart; i < coreDiseases.length; i++) {
@@ -814,7 +825,9 @@ async function main() {
 
   const fullStart = progress.phase === 1 ? progress.phaseIndex : 0;
   if (fullStart > 0) {
-    console.log(`  Resuming from disease #${fullStart + 1} (${((fullStart / fullDiseases.length) * 100).toFixed(0)}% done)`);
+    console.log(
+      `  Resuming from disease #${fullStart + 1} (${((fullStart / fullDiseases.length) * 100).toFixed(0)}% done)`,
+    );
   }
 
   for (let i = fullStart; i < fullDiseases.length; i++) {