Update engine/parser_rules.py

engine/parser_rules.py

@@ -2,7 +2,7 @@
 # ------------------------------------------------------------
 # Rule-based core parser for microbiology descriptions.
 #
-# Stage 11F (Option A ranges + fixes):
+# Stage 11F (Option A ranges + fixes) + Stage 11H additions:
 # - Always store Growth Temperature as "low//high"
 #   • single: 37 → "37//37"
 #   • two temps in text: min//max (e.g. "4 °C but not 45 °C" → "4//45")
@@ -12,6 +12,9 @@
 # - "aerobically" / "anaerobically" → Aerobic / Anaerobic
 # - NaCl tolerance phrases improved
 # - Colony morphology from "colonies dry, white and irregular on nutrient agar"
+# - NEW (11H):
+#   • "Gelatinase positive/negative" → Gelatin Hydrolysis Positive/Negative
+#   • "<sugar> fermenter" → <Sugar> Fermentation = Positive
 # ------------------------------------------------------------
 
 from __future__ import annotations
@@ -210,6 +213,7 @@ def _parse_core_bool_tests(text_lc: str, parsed: Dict[str, str]) -> None:
       - Nitrate reduction text
       - H2S production / non-production
       - DNase universal coverage
+      - NEW (11H): explicit gelatinase → Gelatin Hydrolysis mapping
     """
     for field, keywords in CORE_BOOL_FIELDS.items():
         for kw in keywords:
@@ -316,6 +320,14 @@ def _parse_core_bool_tests(text_lc: str, parsed: Dict[str, str]) -> None:
     if re.search(r"\bnon[- ]?dnase[- ]?producing\b", text_lc):
         _set_if_stronger(parsed, "DNase", "Negative")
 
+    # --- NEW: Gelatinase → Gelatin Hydrolysis ---
+    # Explicit mapping just in case generic patterns miss it
+    if re.search(r"\bgelatinase\s*(positive|pos|\+)\b", text_lc):
+        _set_if_stronger(parsed, "Gelatin Hydrolysis", "Positive")
+
+    if re.search(r"\bgelatinase\s*(negative|neg|\-)\b", text_lc):
+        _set_if_stronger(parsed, "Gelatin Hydrolysis", "Negative")
+
 
 # ------------------------------------------------------------
 # Motility / Capsule / Spores
@@ -551,6 +563,7 @@ def _parse_sugars(text_lc: str, parsed: Dict[str, str]) -> None:
       - "ferments glucose, mannitol and sucrose but not lactose"
       - "does not ferment lactose"
       - "non-lactose fermenter"
+      - "<sugar> fermenter" (positive)
       - global non-fermenter phrases
     """
 
@@ -565,6 +578,14 @@ def _parse_sugars(text_lc: str, parsed: Dict[str, str]) -> None:
             if val:
                 _set_if_stronger(parsed, field, val)
 
+    # 0b) NEW: "<sugar> fermenter" → Positive (unless "non-<sugar> fermenter")
+    for sugar_key, field in SUGAR_FIELDS.items():
+        # positive: "lactose fermenter"
+        if re.search(rf"\b{sugar_key}\s+fermenter\b", text_lc) and not re.search(
+            rf"\bnon[- ]{sugar_key}\s+fermenter\b", text_lc
+        ):
+            _set_if_stronger(parsed, field, "Positive")
+
     # 1) "ferments X, Y and Z but not A, B"
     ferments_pattern = re.compile(r"ferments\s+([a-z0-9 ,;/&\-]+)")
     for m in ferments_pattern.finditer(text_lc):