WIP

tests/pipeline/test_value_selector.py

@@ -203,15 +203,33 @@ class TestValueSelectorOcrField:
         assert len(result) == 1
         assert result[0].text == "94228110015950070"
 
-    def test_ignores_short_digit_tokens(self):
-        """Tokens with fewer than 5 digits are not OCR references."""
-        tokens = _tokens("OCR", "123")
+    def test_ignores_single_digit_tokens(self):
+        """Tokens with fewer than 2 digits are not OCR references."""
+        tokens = _tokens("OCR", "5")
 
         result = ValueSelector.select_value_tokens(tokens, "OCR")
 
         # Fallback: return all tokens since no valid OCR found
         assert len(result) == 2
 
+    def test_ocr_4_digit_token_selected(self):
+        """4-digit OCR token should be selected."""
+        tokens = _tokens("OCR", "3046")
+
+        result = ValueSelector.select_value_tokens(tokens, "OCR")
+
+        assert len(result) == 1
+        assert result[0].text == "3046"
+
+    def test_ocr_2_digit_token_selected(self):
+        """2-digit OCR token should be selected."""
+        tokens = _tokens("OCR", "42")
+
+        result = ValueSelector.select_value_tokens(tokens, "OCR")
+
+        assert len(result) == 1
+        assert result[0].text == "42"
+
 
 class TestValueSelectorInvoiceNumberField:
     """Tests for InvoiceNumber field value selection."""