#!/usr/bin/env python3
import sys
sys.path.insert(0, 'c:/MONIKA')
from salience_os_seed.proto_lm.vocab import Vocabulary

v = Vocabulary()
print(f"Vocab size: {v.size()}")
print(f"Has 'Hello': {'Hello' in v.token_to_id}")
print(f"Has ' I': {' I' in v.token_to_id}")
print(f"Has ' am': {' am' in v.token_to_id}")
print(f"Has ' Monika': {' Monika' in v.token_to_id}")

# Test encoding
text = "Hello I am Monika"
encoded = v.encode(text)
print(f"\nEncoded: {encoded}")