Update

main.py

@@ -19,6 +19,169 @@ os.makedirs(DATA_DIR, exist_ok=True)
 
 _MASTER_INDEX_CACHE = {}
 
+# --- CATEGORY DEFINITIONS ---
+# Ordered by specificity. "Whole word" matching will apply to these keywords.
+CATEGORY_DEFINITIONS = {
+    "Pharmaceuticals & Biologics": [
+        "tablet", "tab", "capsule", "cap", "syrup", "suspension", "susp", "injection", "inj", "vial", "ampoule", "amp",
+        "drops", "gtt", "inhaler", "vaccine", "insulin", "dose", "drug", "medication", "ointment", "cream", "gel", 
+        "lotion", "suppository", "supp", "antibiotic", "antiviral", "analgesic", "anesthetic", "hormone", "steroid", 
+        "vitamin", "mineral", "supplement", "lozenge", "patch", "solution", "powder for suspension", "elixir", "serum",
+        "antitoxin"
+    ],
+    "Surgical Products": [
+        "scalpel", "forceps", "retractor", "clamp", "suture", "stapler", "surgical mesh", "hemostatic", "sealant", 
+        "surgical drape", "surgical gown", "laparoscopic", "robotic surgery", "electrosurgical", "surgical laser", 
+        "surgical blade", "trocar", "surgical clip", "surgical scissor", "needle holder"
+    ],
+    "Orthopedic & Spine": [
+        "orthopedic", "spine", "joint replacement", "trauma fixation", "bone plate", "bone screw", "intramedullary rod", 
+        "bone nail", "spinal implant", "spinal fusion", "bone graft", "orthopedic brace", "cast", "arthroscopy", 
+        "fixator", "prosthesis", "bone drill", "bone saw"
+    ],
+    "Cardiovascular Products": [
+        "cardiac stent", "pacemaker", "defibrillator", "icd", "heart valve", "vascular graft", "cardiac catheter", 
+        "guidewire", "cardiac balloon", "ablation", "coronary", "angioplasty", "introducer sheath"
+    ],
+    "Medical Imaging Equipment": [
+        "mri", "ct scanner", "x-ray", "ultrasound", "mammography", "fluoroscopy", "pet scanner", "c-arm", 
+        "medical imaging", "transducer", "x-ray film", "contrast media", "lead apron"
+    ],
+    "Diagnostic Products": [
+        "diagnostic", "test kit", "glucose test", "reagent", "immunoassay", "chemistry analyzer", "hematology", 
+        "microbiology", "culture media", "pregnancy test", "covid", "rapid test", "urinalysis", "penlight", 
+        "specula", "otoscope", "ophthalmoscope", "lancet", "glucometer strips", "test strip"
+    ],
+    "Patient Monitoring Equipment": [
+        "vital signs", "ecg", "ekg", "pulse oximeter", "blood pressure monitor", "sphygmomanometer", 
+        "medical thermometer", "capnography", "fetal monitor", "telemetry", "spo2 sensor", "bp cuff", "temperature probe"
+    ],
+    "Respiratory & Anesthesia": [
+        "ventilator", "anesthesia machine", "oxygen concentrator", "nebulizer", "cpap", "bipap", "respiratory", 
+        "endotracheal", "tracheostomy", "spirometer", "oxygen mask", "breathing circuit", "nasal cannula", 
+        "resuscitator", "laryngoscope"
+    ],
+    "Infusion & Vascular Access": [
+        "infusion pump", "syringe pump", "iv set", "iv catheter", "venous", "picc", "iv port", "dialysis catheter", 
+        "administration set", "extension set", "stopcock", "giving set", "saline", "dextrose", "ringer", 
+        "sodium chloride", "water for injection"
+    ],
+    "Wound Care & Tissue Management": [
+        "wound dressing", "bandage", "gauze", "medical tape", "plaster", "adhesive", "wound foam", "alginate", 
+        "hydrocolloid", "compression bandage", "ostomy", "skin substitute", "negative pressure"
+    ],
+    "Dialysis & Renal Care": [
+        "hemodialysis", "peritoneal", "dialyzer", "blood line", "fistula needle", "dialysis concentrate", "bicarbonate"
+    ],
+    "Ophthalmic Products": [
+        "intraocular", "intraocular lens", "phaco", "vitrectomy", "lasik", "contact lens", "viscoelastic", 
+        "ophthalmic solution", "eye drops"
+    ],
+    "Dental Products": [
+        "dental implant", "orthodontic", "dental bracket", "dental wire", "dental drill", "dental handpiece", 
+        "dental cement", "dental composite", "amalgam", "impression material", "teeth whitening", "dental chair"
+    ],
+    "Neurology & Neurosurgery": [
+        "neurostimulation", "spinal cord stimulator", "neuro coil", "flow diverter", "cranial", "shunt", 
+        "neuro electrode", "eeg", "emg"
+    ],
+    "Laboratory Equipment & Supplies": [
+        "microscope", "lab centrifuge", "incubator", "autoclave", "pipette", "glassware", "test tube", "petri dish", 
+        "flask", "beaker", "microscope slide", "cover glass", "fume hood", "biosafety cabinet"
+    ],
+    "Personal Protective Equipment (PPE)": [
+        "ppe", "n95", "face shield", "safety eyewear", "goggles", "protective apron", "shoe cover", "head cover", 
+        "coverall", "isolation gown", "hazmat", "surgical mask"
+    ],
+    "Sterilization & Disinfection": [
+        "sterilization", "disinfectant", "antiseptic", "povidone", "iodine", "chlorhexidine", "alcohol swab", 
+        "hand sanitizer", "medical soap", "enzymatic cleaner", "detergent", "washer disinfector", "sterilizer", 
+        "sterilization indicator"
+    ],
+    "Hospital Furniture & Equipment": [
+        "hospital bed", "examination table", "stretcher", "medical trolley", "medical cart", "medical cabinet", 
+        "bedside locker", "overbed table", "iv pole", "wheelchair"
+    ],
+    "Rehabilitation & Physical Therapy": [
+        "rehabilitation", "physiotherapy", "walker", "walking cane", "crutch", "exercise band", "traction", 
+        "electrotherapy", "massage table", "orthosis"
+    ],
+    "Home Healthcare Products": [
+        "home care", "blood glucose meter", "hearing aid", "mobility aid", "bathroom safety", "commode"
+    ],
+    "Emergency & Trauma Care": [
+        "emergency kit", "trauma kit", "first aid", "aed", "defibrillator", "manual resuscitator", "suction unit", 
+        "immobilizer", "cervical collar", "splint", "tourniquet", "crash cart"
+    ],
+    "Maternal & Neonatal Care": [
+        "maternal", "neonatal", "infant incubator", "infant warmer", "phototherapy", "breast pump", "obstetric", 
+        "birthing bed", "fetal doppler", "umbilical"
+    ],
+    "Urology Products": [
+        "urology", "foley catheter", "urine bag", "urinary drainage", "ureteral stent", "stone basket"
+    ],
+    "Gastroenterology & Endoscopy": [
+        "endoscope", "gastroscope", "colonoscope", "biopsy forceps", "polypectomy snare", "gastric balloon", "ercp"
+    ],
+    "Oncology Products": [
+        "oncology", "chemotherapy", "radiotherapy", "brachytherapy", "port-a-cath", "cancer diagnostic"
+    ],
+    "Pain Management": [
+        "pain management", "pca pump", "epidural", "nerve block", "tens unit"
+    ],
+    "Sleep Medicine": [
+        "sleep apnea", "cpap mask", "bipap mask", "sleep tubing", "polysomnography"
+    ],
+    "Telemedicine & Digital Health": [
+        "telemedicine", "telehealth", "remote monitor", "medical software", "health app"
+    ],
+    "Blood Management": [
+        "blood bag", "blood transfusion", "blood bank", "blood warmer", "apheresis"
+    ],
+    "Mortuary & Pathology": [
+        "mortuary", "autopsy", "body bag", "morgue fridge", "dissection table", "microtome", "tissue processor"
+    ],
+    "Environmental Control": [
+        "medical gas", "medical vacuum", "medical air plant", "gas manifold", "gas outlet", "gas alarm"
+    ],
+    "Mobility & Accessibility": [
+        "patient lift", "patient hoist", "wheelchair ramp", "stair lift", "transfer board"
+    ],
+    "Bariatric Products": [
+        "bariatric bed", "bariatric wheelchair", "heavy duty scale"
+    ],
+    "Medical Textiles": [
+        "hospital linen", "bed sheet", "pillow case", "medical blanket", "towel", "privacy curtain", "medical uniform", 
+        "scrub suit", "lab coat"
+    ],
+    "Infection Control Products": [
+        "waste bin", "sharps container", "biohazard bag", "spill kit", "air purifier"
+    ],
+    "Medical Gases & Cryogenics": [
+        "gas cylinder", "oxygen regulator", "flowmeter", "liquid oxygen", "nitrogen tank"
+    ],
+    "Nutrition & Feeding": [
+        "enteral feeding", "clinical nutrition", "nasogastric tube", "feeding pump", "feeding set", "peg tube"
+    ],
+    "Specimen Collection & Transport": [
+        "specimen container", "sample collection", "transport media", "transport swab", "urine container", 
+        "stool container", "cool box", "transport bag"
+    ],
+    "Medical Software & IT": [
+        "emr", "ehr", "pacs", "ris", "lis", "his", "hospital information system"
+    ],
+    "Aesthetics & Dermatology": [
+        "dermatology", "aesthetic laser", "ipl", "dermal filler", "botulinum", "botox", "chemical peel", 
+        "microdermabrasion"
+    ],
+    # Catch-all for basic items not caught above
+    "Medical Supplies & Consumables": [
+        "syringe", "needle", "glove", "examination glove", "disposable", "consumable", "cotton wool", "alcohol prep", 
+        "urinal", "bedpan", "underpad", "tongue depressor", "applicator", "lubricant jelly", "cannula" 
+        # Note: Cannula is here as fallback if not specific nasal/iv
+    ]
+}
+
 def load_master_index():
     global _MASTER_INDEX_CACHE
     if _MASTER_INDEX_CACHE: return _MASTER_INDEX_CACHE
@@ -47,45 +210,23 @@ def is_garbage_row(row_text: str) -> bool:
 def determine_item_type(description: str, form: str) -> str:
     """
     Determines the category of the item based on its description and form/unit.
-    Categories: Pharmaceuticals, Medical Supplies, Medical Equipment.
+    Uses regex for whole-word matching to prevent substring errors (e.g. 'fusion' in 'infusion').
     """
     text = (description + " " + form).lower()
     
-    # Priority 1: Medical Supplies (Consumables)
-    # Check these first to handle cases like "Insulin Syringe" (Supply) vs "Insulin" (Pharma)
-    supplies_keywords = [
-        'syringe', 'needle', 'cannula', 'catheter', 'glove', 'mask', 'gauze', 
-        'bandage', 'dressing', 'cotton', 'swab', 'lancet', 'strip', 'test kit', 
-        'blade', 'suture', 'plaster', 'gown', 'sheet', 'bag', 'alcohol', 
-        'disinfectant', 'sanitizer', 'tongue depressor', 'specula', 'paper',
-        'wipes', 'apron', 'cap', 'shoe cover', 'tape'
-    ]
-    if any(k in text for k in supplies_keywords):
-        return 'Medical Supplies'
-
-    # Priority 2: Medical Equipment (Devices/Durable)
-    equipment_keywords = [
-        'thermometer', 'sphygmomanometer', 'stethoscope', 'oximeter', 
-        'glucometer', 'nebulizer', 'otoscope', 'penlight', 'monitor', 
-        'scale', 'microscope', 'centrifuge', 'refrigerator', 'cool box',
-        'freezer', 'lamp', 'bed', 'chair', 'pump', 'bp machine', 'device'
-    ]
-    if any(k in text for k in equipment_keywords):
-        return 'Medical Equipment'
-
-    # Priority 3: Pharmaceuticals (Medicines/Drugs)
-    pharma_keywords = [
-        'tablet', 'capsule', 'cap', 'tab', 'syrup', 'suspension', 'susp', 
-        'injection', 'inj', 'ampoule', 'amp', 'vial', 'cream', 'ointment', 
-        'gel', 'suppository', 'supp', 'drops', 'inhaler', 'vaccine', 'sera', 
-        'insulin', 'medicine', 'drug', 'mg', 'ml', 'mcg', 'iu', 'dose',
-        'solution', 'infusion', 'spray', 'lozenge'
-    ]
-    if any(k in text for k in pharma_keywords):
-        return 'Pharmaceuticals'
-
-    # Fallback
-    return 'Medical Supplies'
+    for category, keywords in CATEGORY_DEFINITIONS.items():
+        for k in keywords:
+            # \b matches word boundaries. 
+            # This ensures "fusion" matches "spinal fusion" but NOT "infusion".
+            # It ensures "coat" matches "lab coat" but NOT "coated".
+            # We escape the keyword to handle any special characters safely.
+            pattern = r'\b' + re.escape(k) + r'\b'
+            
+            if re.search(pattern, text):
+                return category
+    
+    # Fallback default
+    return 'Medical Supplies & Consumables'
 
 async def delete_file_safety_net(file_path: str, delay: int = 600):
     await asyncio.sleep(delay)
@@ -215,20 +356,18 @@ async def match_all(req: MatchRequest):
         
         matches = []
         for v in vendors:
-            cats = [c.lower() for c in v.get('primary_categories', [])]
-            # Simple matching logic - can be expanded to use item['type'] if needed
-            if 'pharmaceuticals' in cats or 'medical devices' in cats or 'medical supplies' in cats:
-                 matches.append({
-                    'vendor_id': v.get('vendor_id'),
-                    'name': v.get('legal_name'),
-                    'country': (v.get('countries_served') or ['Unknown'])[0],
-                    'landedCost': v.get('landedCost', 10),
-                    'deliveryDays': v.get('deliveryDays', 5),
-                    'availableQty': v.get('availableQty', 1000),
-                    'qualityScore': v.get('confidence_score', 80) / 10.0,
-                    'reliabilityScore': 5,
-                    'score': 9.5
-                })
+            # Simplified matching logic for demonstration
+            matches.append({
+                'vendor_id': v.get('vendor_id'),
+                'name': v.get('legal_name'),
+                'country': (v.get('countries_served') or ['Unknown'])[0],
+                'landedCost': v.get('landedCost', 10),
+                'deliveryDays': v.get('deliveryDays', 5),
+                'availableQty': v.get('availableQty', 1000),
+                'qualityScore': v.get('confidence_score', 80) / 10.0,
+                'reliabilityScore': 5,
+                'score': 9.5
+            })
 
         results.append({
             "medicine": name,