core/parsing.py

"""
Paso 4: Parsing y estructuración del texto del menú.
"""

import re
from typing import Dict, List


SECTION_PATTERNS = [
    r"(?i)(entrantes|starters|aperitivos)",
    r"(?i)(plato principal|principales|main courses?)",
    r"(?i)(postres|desserts)",
    r"(?i)(bebidas|drinks|beverages)",
]

PRICE_PATTERN = r"€\s*(\d+(?:[.,]\d{1,2})?)"


def parse_menu_structure(text: str) -> List[Dict]:
    """
    Parsea el texto del menú y extrae platos con precios.

    Returns:
        Lista de dicts con {name, price, section}.
    """
    dishes = []
    current_section = "General"

    for line in text.split("\n"):
        line = line.strip()
        if not line:
            continue

        # Detectar secciones
        for pattern in SECTION_PATTERNS:
            if re.search(pattern, line):
                current_section = line
                break

        # Detectar platos con precio
        price_match = re.search(PRICE_PATTERN, line)
        if price_match:
            dish_name = re.sub(PRICE_PATTERN, "", line).strip()
            if dish_name:
                dishes.append({
                    "name": dish_name,
                    "price": price_match.group(0),
                    "section": current_section,
                })

    return dishes