From fc49ba60e66ccd8bde6abb5098f1134e26e599b4 Mon Sep 17 00:00:00 2001
From: tngo <tngo@datacom.jp>
Date: Tue, 29 Sep 2026 00:41:52 +0900
Subject: [PATCH] Cache only text reads that could name a product

The per-track OCR cache kept any non-empty read, so a first crop that read
only a size ("400g") was replayed on every later attempt. A size matches every
400g product's size line, so the track could never be named even after a clean
crop arrived: the cached read replaced it. Only a read carrying identity text
is cached now, so the next attempt gets a fresh OCR.

`carries_identity` moves to module scope in live.py; a function-local import
shadowed the name at the cache site and raised UnboundLocalError.
---
 recognization/code/live.py      | 10 +++++++---
 recognization/code/test_live.py | 23 +++++++++++++++++++++++
 2 files changed, 30 insertions(+), 3 deletions(-)

diff --git a/recognization/code/live.py b/recognization/code/live.py
index d4240ac..54e7421 100644
--- a/recognization/code/live.py
+++ b/recognization/code/live.py
@@ -39,6 +39,7 @@ from engine import (  # noqa: E402
     STATE_UNRECOGNIZED,
     RecognitionEngine,
 )
+from identity_lanes import carries_identity  # noqa: E402
 from tracking import SessionTracker  # noqa: E402
 
 EMPTY_SLOT_HISTORY = 5
@@ -1183,9 +1184,12 @@ class LiveSession:
             self._classify_times.append(finished_at - submitted_at)
 
             # The reading is cached on the track so later attempts skip OCR.
-            # An empty read is not cached: a crop that was too small or too
-            # blurry to read says nothing about the next one.
-            if text_read:
+            # Only a read that could name something is cached: a crop that was
+            # too small or too blurry says nothing about the next one, and a
+            # read of just a size ("400g") matches every 400g product's size
+            # line, so caching it would let one unreadable crop veto the
+            # product for the rest of the track's life.
+            if text_read and carries_identity(text_read):
                 track.text_rank = list(text_rank or [])
                 track.text_read = text_read
             elif track.text_read:
diff --git a/recognization/code/test_live.py b/recognization/code/test_live.py
index 6c6ef92..ecf721e 100644
--- a/recognization/code/test_live.py
+++ b/recognization/code/test_live.py
@@ -403,6 +403,29 @@ class LocalScaleTests(unittest.TestCase):
         self.assertEqual(engine.ocr_calls, 1, "OCR must run once per track")
         engine.shutdown()
 
+    def test_a_size_only_read_is_not_cached(self):
+        """An unreadable crop must not veto the product for the track's life.
+
+        The cache used to keep any non-empty read, so a first crop that read
+        only "400g" was replayed on every later attempt: the track could never
+        be named even after a clean crop arrived, because the cached read
+        replaced it. Only a read that carries identity text is cached now, so
+        the next attempt gets a fresh OCR.
+        """
+        session, engine, _, _ = make_session(engine_cls=LocalScaleEngine)
+        engine.lane_answers = [
+            ({}, [("sku_a", 0.9)], "400g"),
+            ({}, [("sku_a", 0.9)], "コクうま 65%"),
+        ] + [({}, [("sku_a", 0.9)], "コクうま 65%")] * 6
+        labelled = False
+        for frame_id in range(1, 14):
+            for track in process(session, frame_id)["tracks"]:
+                if track["identity"]["sku_id"] == "sku_a":
+                    labelled = True
+        self.assertTrue(labelled, "a later readable crop must still name the track")
+        self.assertGreater(engine.ocr_calls, 1, "the size-only read must not be cached")
+        engine.shutdown()
+
 
 class TagPassIntegrationTests(unittest.TestCase):
     """The tag pass must reach the live result without blocking frames."""