hh-shipments 0.3.7 → 0.3.9

package/package.json

@@ -18,7 +18,7 @@
     "test": "node test/run-tests.js && node test/run-carriers-tests.js"
   },
   "type": "commonjs",
-  "version": "0.3.7",
+  "version": "0.3.9",
   "bin": {
     "hh-shipments": "server.js"
   },

package/public/app.js

@@ -111,7 +111,10 @@ async function loadDash() {
 async function runQuick(text, targetEl, preset) {
   targetEl.innerHTML = '<p class="muted">Extracting…</p>';
   try {
-    const r = await api.post('/api/intake/quick', { text, seller_name: preset && preset.seller, device: preset && preset.device });
+    const r = await Promise.race([
+      api.post('/api/intake/quick', { text, seller_name: preset && preset.seller, device: preset && preset.device }),
+      new Promise((_, rej) => setTimeout(() => rej(new Error('extract timed out — check your connection and try again')), 45000)),
+    ]);
     let html = '';
     r.added.forEach((a) => { html += `<div class="review">✅ Added <b>${esc(a.number)}</b> (${esc(a.carrier.toUpperCase())})${r.simulated ? ' <span class="sim-badge">SIMULATED</span>' : ''}</div>`; });
     r.duplicates.forEach((n) => { html += `<div class="review">⛔ <b>${esc(n)}</b> is already tracked — skipped.</div>`; });
@@ -171,13 +174,16 @@ async function ocrFile(file, statusEl) {
 async function handlePhotoFile(f) {
   const target = $('photo-result');
   try {
+    f = await shrinkImage(f);
     const url = URL.createObjectURL(f);
     target.innerHTML = `<div class="review"><img src="${url}" style="max-width:100%;border-radius:10px"><p class="muted">📖 Reading screenshot…</p></div>`;
     const text = await ocrFile(f, $('ocr-status'));
     // Upload original for reference, then run the quick flow on OCR text
     const fd = new FormData();
     fd.append('image', f);
-    const up = await fetch('/api/intake/image', { method: 'POST', body: fd });
+    let up;
+    try { up = await fetchTimeout('/api/intake/image', { method: 'POST', body: fd }, 60000); }
+    catch (e) { throw new Error('upload timed out — check your connection and try again'); }
     if (!up.ok) throw new Error('upload failed');
     const r = await runQuick(text, $('photo-result'));
     if (r && !r.added.length && !r.needs_review.length && !r.duplicates.length) showOcrRead($('photo-result'), text);
@@ -206,21 +212,46 @@ function pastedImageFiles(e) {
   return files;
 }
 
-// Paste-text tab: pasting a screenshot runs OCR and extracts automatically —
-// no "Extract" tap needed. Plain text paste still lands in the textarea.
+// Shrink huge phone photos before OCR/upload: much faster to read and send,
+// and less exposed to a stalled connection. Falls back to the original file.
+async function shrinkImage(file, maxDim = 1600) {
+  try {
+    const bmp = await createImageBitmap(file);
+    const scale = Math.min(1, maxDim / Math.max(bmp.width, bmp.height));
+    if (scale >= 1) { bmp.close(); return file; }
+    const canvas = document.createElement('canvas');
+    canvas.width = Math.round(bmp.width * scale);
+    canvas.height = Math.round(bmp.height * scale);
+    canvas.getContext('2d').drawImage(bmp, 0, 0, canvas.width, canvas.height);
+    bmp.close();
+    const blob = await new Promise((res) => canvas.toBlob(res, 'image/jpeg', 0.85));
+    return blob ? new File([blob], 'label.jpg', { type: 'image/jpeg' }) : file;
+  } catch (e) { return file; }
+}
+
+async function fetchTimeout(url, opts = {}, ms = 60000) {
+  const c = new AbortController();
+  const t = setTimeout(() => c.abort(), ms);
+  try { return await fetch(url, { ...opts, signal: c.signal }); }
+  finally { clearTimeout(t); }
+}
 async function handlePastedImageFile(f) {
   const target = $('quick-result');
   try {
     // Instant feedback: show a thumbnail + progress the moment the image lands
+    f = await shrinkImage(f);
     const url = URL.createObjectURL(f);
     target.innerHTML = `<div class="review"><img src="${url}" style="max-width:100%;border-radius:10px"><p class="muted" id="ocr-live">📖 Reading screenshot…</p></div>`;
     const text = await ocrFile(f, $('ocr-status-paste'));
     const live = $('ocr-live');
-    if (live) live.textContent = '🔍 Finding tracking number…';
+    if (live) live.textContent = '⬆️ Uploading…';
     const fd = new FormData();
     fd.append('image', f);
-    const up = await fetch('/api/intake/image', { method: 'POST', body: fd });
+    let up;
+    try { up = await fetchTimeout('/api/intake/image', { method: 'POST', body: fd }, 60000); }
+    catch (e) { throw new Error('upload timed out — check your connection and try again'); }
     if (!up.ok) throw new Error('upload failed');
+    if (live) live.textContent = '🔍 Finding tracking number…';
     const r = await runQuick(text, target);
     if (r && !r.added.length && !r.needs_review.length && !r.duplicates.length) showOcrRead(target, text);
   } catch (err) { $('ocr-status-paste').textContent = ''; target.innerHTML = `<p class="error">${esc(err.message)}</p>`; }

package/src/extract.js

@@ -76,7 +76,13 @@
     var m, re = /\b(?:1Z|12|IZ)[\s-]?(?:[0-9A-Z][\s-]?){16}\b/gi;
     while ((m = re.exec(text)) !== null) {
       var ups = normalize(m[0]);
-      var num = /^1Z[0-9A-Z]{16}$/.test(ups) ? ups : repairUpsZ(ups);
+      var num = null;
+      if (/^1Z[0-9A-Z]{6}[0-9]{10}$/.test(ups)) num = ups; // fully strict
+      else if (/^1Z[0-9A-Z]{16}$/.test(ups)) {
+        // letters sitting in the digit zone are OCR confusions — repair them
+        num = repairUpsTail(ups.slice(2)) || ups;
+      }
+      else num = repairUpsZ(ups) || repairUpsTail(ups.slice(2));
       if (num && !seen[num]) {
         seen[num] = true;
         found.push({ number: num, carrier: 'ups', confidence: conf, ocr_fixed: num !== ups });
@@ -84,6 +90,39 @@
     }
   }
 
+  // Tesseract also mangles punctuation on labels: ")" / "]" / "}" for J,
+  // "(" / "[" / "{" for C, "|" / "!" for 1 (e.g. "1Z 1X) 539 ..." for 1XJ).
+  // The UPS structure is strict (1Z + 6-char shipper + 2-digit service +
+  // 8-digit serial), so a mangled tail can be repaired and accepted only
+  // when the repaired result validates strictly — never invented.
+  var OCR_PUNCT_FIX = { ')': 'J', ']': 'J', '}': 'J', '(': 'C', '[': 'C', '{': 'C', '|': '1', '!': '1' };
+  var OCR_TAIL_DIGIT_FIX = { 'O': '0', 'Q': '0', 'I': '1', 'l': '1', '|': '1', '!': '1', 'Z': '2', 'S': '5', 'B': '8' };
+  function repairUpsTail(tail) {
+    // tail: 16 chars after the 1Z anchor (6-char shipper + 2-digit service +
+    // 8-digit serial), possibly with OCR confusions.
+    var ship = tail.slice(0, 6), digits = tail.slice(6, 16);
+    var fixShip = ship.replace(/[)\]}(\[{|!]/g, function (c) { return OCR_PUNCT_FIX[c]; });
+    var fixDigits = digits.replace(/[OQIil|!ZSB]/g, function (c) { return OCR_TAIL_DIGIT_FIX[c]; });
+    var cand = '1Z' + fixShip + fixDigits;
+    if (!UPS_STRICT.test(cand)) return null;
+    if (!/[A-Z]/.test(cand.slice(2, 8))) return null; // real shipper numbers contain a letter
+    return cand;
+  }
+  function looseUpsIn(text, found, seen, conf) {
+    // Loose: anchor + 16 chars from an OCR-confusion-tolerant alphabet,
+    // spaces/dashes allowed. repairUpsTail decides; strictness lives there.
+    var m, re = /\b(?:1Z|12|IZ)[\s-]?((?:[0-9A-Z)\]}(\[{|!][\s-]?){16})\b/gi;
+    while ((m = re.exec(text)) !== null) {
+      var tail = m[1].replace(/[\s-]/g, '').toUpperCase();
+      if (tail.length !== 16) continue;
+      var num = repairUpsTail(tail);
+      if (num && !seen[num]) {
+        seen[num] = true;
+        found.push({ number: num, carrier: 'ups', confidence: conf, ocr_fixed: true });
+      }
+    }
+  }
+
   function extractTrackingNumbers(text) {
     var found = [];
     var seen = {};
@@ -95,6 +134,7 @@
     anchorZones(text).forEach(function (z) {
       runPatterns(z.text, found, seen, z.carrier);
       spacedUpsIn(z.text, found, seen, 'high');
+      looseUpsIn(z.text, found, seen, 'high');
     });
     // 1) Whole-text strict patterns
     runPatterns(text, found, seen, false);
@@ -194,6 +234,7 @@
     // a) UPS number split by spaces/dashes (whole-text fallback; the anchor
     // pass already tried the hot zones at high confidence)
     spacedUpsIn(text, found, seen, 'medium');
+    looseUpsIn(text, found, seen, 'medium');
     // b) long digit-like runs with OCR confusions -> fix -> re-match digit formats
     var runRe = /[0-9OILSBoilszb]{10,24}/g, run;
     while ((run = runRe.exec(text)) !== null) {