tests/test_parsers_only.py

import asyncio
import os
import sys

# Disable tokenizers parallelism
os.environ["TOKENIZERS_PARALLELISM"] = "false"

sys.path.append(os.path.join(os.path.dirname(__file__), '../server'))

from app.services.parsers.medical_parsers import MedQuADParser, HealthTapParser, PubMedParser, WhoParser

def create_mock_files():
    base = "server/datasets/raw"
    
    # PubMed
    pm_path = os.path.join(base, "pubmedqa", "mock.json")
    with open(pm_path, "w") as f:
        f.write('''[{"question": "Is X effective?", "ideal_answer": "Yes, X is effective.", "documents": ["doc1"]}]''')
        
    # WHO
    who_path = os.path.join(base, "who", "mock.json")
    with open(who_path, "w") as f:
        f.write('''{"topic": "Flu", "qa_pairs": [{"q": "What is flu?", "a": "Influenza."}]}''')

    return pm_path, who_path

def test_parsers():
    print("🏥 Testing EXTENDED Parsers...\n")
    pm_path, who_path = create_mock_files()
    
    # 1. PubMed
    pm = PubMedParser()
    res = pm.parse(pm_path)
    print(f"✅ PubMed: {res[0]}")
    assert res[0]['source'] == "PubMedQA"
    assert "effective" in res[0]['answer']
    
    # 2. WHO
    who = WhoParser()
    res = who.parse(who_path)
    print(f"✅ WHO: {res[0]}")
    assert res[0]['source'] == "WHO/CDC"
    assert res[0]['context'] == "Topic: Flu"
    
    print("\n🎉 ALL Parsers Verified!")

if __name__ == "__main__":
    test_parsers()