hh-shipments 0.3.26 → 0.3.27

package/package.json

@@ -19,7 +19,7 @@
     "test": "node test/run-extract-tests.js && node test/run-tests.js && node test/run-carriers-tests.js"
   },
   "type": "commonjs",
-  "version": "0.3.26",
+  "version": "0.3.27",
   "bin": {
     "hh-shipments": "server.js"
   },

package/public/app.js

@@ -400,13 +400,15 @@ function loadZxing() {
   return zxingPromise;
 }
 
-// OCR just the top strip of a label (sender block) for the seller name.
+// OCR the top strip of a label (sender block) for the seller name.
 // Works from the ORIGINAL full-resolution image, upscaled — small text
-// reads far better than on the shrunk copy. Returns text or null.
+// reads far better than on the shrunk copy. Top 45%: covers the sender
+// block on USPS labels (~35-40%) as well as UPS/FedEx (top). Returns text
+// or null.
 async function ocrTopStrip(origFile, statusEl) {
   try {
     const bmp = await createImageBitmap(origFile);
-    const sw = bmp.width, sh = Math.max(60, Math.round(bmp.height * 0.35));
+    const sw = bmp.width, sh = Math.max(60, Math.round(bmp.height * 0.45));
     const targetW = 1400, scale = targetW / sw;
     const canvas = document.createElement('canvas');
     canvas.width = targetW;
@@ -537,9 +539,14 @@ async function handlePastedImageFile(f) {
       if (r && !r.added.length && !r.duplicates.length && !r.needs_review.some((c) => c.carrier && c.carrier !== 'unknown')) showOcrRead(target, text);
       return;
     }
-    // 2) Fallback: OCR the printed text
+    // 2) Fallback: OCR the printed text (+ the sender strip for the name,
+    // in parallel — the number crop usually excludes the sender block)
     if (live) live.textContent = '📖 Reading screenshot…';
-    text = await ocrFile(f, $('ocr-status-paste'));
+    const [ocrText, topTextFb] = await Promise.all([
+      ocrFile(f, $('ocr-status-paste')),
+      ocrTopStrip(orig, $('ocr-status-paste')),
+    ]);
+    text = ocrText;
     if (live) live.textContent = '⬆️ Uploading…';
     const fd = new FormData();
     fd.append('image', f);
@@ -549,7 +556,7 @@ async function handlePastedImageFile(f) {
     if (!up.ok) throw new Error('upload failed');
     const upId2 = await up.json().then((j) => j.id).catch(() => null);
     if (live) live.textContent = '🔍 Finding tracking number…';
-    const r = await runQuick(text, target);
+    const r = await runQuick(text, target, topTextFb ? { labelText: topTextFb } : undefined);
     // Paste flow already handled this image — don't leave it nagging in the review queue.
     if (upId2) api.patch('/api/intake/' + upId2, { state: 'reviewed' }).catch(() => {});
     if (looksLikeAppScreenshot(text)) target.insertAdjacentHTML('afterbegin', '<p class="warn">⚠️ This looks like a screenshot of the app itself — paste the label photo directly for best results.</p>');

package/src/extract.js

@@ -66,24 +66,43 @@
       var m2 = text.match(/(?:from|seller)\s*:\s*([A-Z][a-z]+(?:\s+[A-Z][a-z]+)?)/);
       if (m2) out.seller_name = m2[1].trim();
     }
-    // Shipping labels: the sender name is the first or second line
-    // ("MOHAMMED SALEM"). Only take it when it's clearly a name — never
-    // digits, never address fragments, never OCR garbage like "EE SS",
-    // and never the app's own UI words when a screenshot of the app itself
-    // gets pasted ("Back Shipment").
+    // Shipping-label name test, shared by the address-anchor rule and the
+    // first-two-lines fallback below.
+    var labelWords = function (s) { return String(s).trim().split(/\s+/); };
+    var looksLikeLabelName = function (s) {
+      var words = labelWords(s);
+      if (words.length < 2 || words.length > 3) return false;
+      // Real shipping labels print names in ALL CAPS; app UI chrome
+      // ("Back Shipment") and OCR fragments ("Eg NTT") don't.
+      if (s !== s.toUpperCase()) return false;
+      // First word 4+ letters (a real first name — rejects "EE SS");
+      // later words 4+ letters or a single initial ("MOHAMED A").
+      if (!/^[A-Z]{4,}$/.test(words[0])) return false;
+      for (var wi = 1; wi < words.length; wi++) if (!/^([A-Z]{4,}|[A-Z])$/.test(words[wi])) return false;
+      if (/\b(UPS|FEDEX|USPS|DHL|GROUND|EXPRESS|PRIORITY|BILLING|SHIP\b|FROM|TRACKING|POSTAGE|LABEL|BACK|SHIPMENT|HOME|ARCHIVE|SETTINGS|SCREENSHOT|READING|FINDING|UPLOADING|EXTRACT|PASTE|DETAILS|SAVE|TAP|CLIPBOARD|NUMBER|HUNTER|HILLS)\b/.test(s)) return false;
+      return true;
+    };
+    // Shipping labels: the sender name sits on the line directly above the
+    // street-address line ("MOHAMED A" above "2507 GLACIER ST"). Find that
+    // anchor anywhere in the text — far more reliable than line position,
+    // since OCR crops don't always start at the label's first line. The
+    // first such block wins (sender prints above recipient).
+    if (!out.seller_name) {
+      var alines = String(text).split(/\r?\n/);
+      for (var ai = 0; ai < alines.length; ai++) {
+        var addr = alines[ai].trim();
+        if (!/^\d{1,6}\s+[A-Z0-9][A-Z0-9 .'-]*$/.test(addr)) continue;
+        if (!/\b(ST|STREET|AVE|AVENUE|RD|ROAD|DR|DRIVE|LN|LANE|BLVD|PL|PLACE|CT|COURT|WAY|TER|TERRACE|PKWY|PARKWAY)\b/.test(addr)) continue;
+        if (ai === 0) continue;
+        var cand = alines[ai - 1].replace(/[^A-Za-z ]/g, ' ').replace(/\s+/g, ' ').trim();
+        if (looksLikeLabelName(cand)) { out.seller_name = cand; break; }
+      }
+    }
+    // Fallback: sender name is the first or second line ("MOHAMMED SALEM").
+    // Only take it when it's clearly a name — never digits, never address
+    // fragments, never OCR garbage like "EE SS", and never the app's own UI
+    // words when a screenshot of the app itself gets pasted ("Back Shipment").
     if (!out.seller_name) {
-      var labelWords = function (s) { return String(s).trim().split(/\s+/); };
-      var looksLikeLabelName = function (s) {
-        var words = labelWords(s);
-        if (words.length < 2 || words.length > 3) return false;
-        // Real shipping labels print names in ALL CAPS; app UI chrome
-        // ("Back Shipment") and OCR fragments ("Eg NTT") don't.
-        if (s !== s.toUpperCase()) return false;
-        // every word 4+ letters: OCR garbage ("EE SS") never qualifies.
-        for (var i = 0; i < words.length; i++) if (!/^[A-Z]{4,}$/.test(words[i])) return false;
-        if (/\b(UPS|FEDEX|USPS|DHL|GROUND|EXPRESS|PRIORITY|BILLING|SHIP\b|FROM|TRACKING|POSTAGE|LABEL|BACK|SHIPMENT|HOME|ARCHIVE|SETTINGS|SCREENSHOT|READING|FINDING|UPLOADING|EXTRACT|PASTE|DETAILS|SAVE|TAP|CLIPBOARD|NUMBER)\b/.test(s)) return false;
-        return true;
-      };
       var lines = String(text).split(/\r?\n/);
       for (var li = 0; li < Math.min(2, lines.length); li++) {
         var cleaned = lines[li].replace(/[^A-Za-z ]/g, ' ').replace(/\s+/g, ' ').trim();