cookbook/app/ingredients.py

# -*- coding: utf-8 -*-
"""
Helper module for ingredient parsing, table structuring, and quantity scaling.
"""
import re
from typing import List, Dict, Any, Optional

COMMON_UNITS = [
    "kg", "g", "mg", "l", "liter", "ml", "cl", "el", "tl", "esslöffel", "teelöffel",
    "pck.", "pck", "päckchen", "pkt.", "pkt", "dose", "dosen", "glas", "gläser",
    "prise", "prisen", "msp.", "msp", "messerspitze", "bund", "stk.", "stk",
    "stück", "stücke", "zehe", "zehen", "zweig", "zweige", "blatt", "blätter",
    "scheibe", "scheiben", "stange", "stangen", "kopf", "nest", "nester", "tropfen", "tr."
]

def parse_fraction(val_str: str) -> Optional[float]:
    val_str = val_str.strip()
    if "/" in val_str:
        parts = val_str.split("/")
        if len(parts) == 2:
            try:
                return float(parts[0]) / float(parts[1])
            except ValueError:
                pass
    try:
        return float(val_str.replace(",", "."))
    except ValueError:
        return None

def parse_ingredient_line(line: str) -> Dict[str, Any]:
    raw = line.strip()
    if not raw:
        return {"type": "empty", "raw": ""}
    
    # Check if section header (ends with colon or keywords)
    lower = raw.lower()
    if raw.endswith(":") or lower.startswith("für ") or lower in [
        "teig", "sauce", "soße", "füllung", "belag", "creme", "glasur", "marinade", "deko",
        "schichten & deko", "keksschicht", "heidelbeersoße", "zitronen-zimt-sirup", "zutaten",
        "einlage & toppings", "gemüse & toppings", "lemon curd", "mascarponecreme"
    ]:
        hdr_text = raw.rstrip(":")
        return {
            "type": "header",
            "is_header": True,
            "text": hdr_text,
            "name": hdr_text,
            "raw": raw
        }
    
    # Regex to match leading amount
    match = re.match(r"^(\d+\s+\d+/\d+|\d+/\d+|\d+(?:[.,]\d+)?)(?:\s*-\s*(\d+(?:[.,]\d+)?))?\s*(.*)$", raw)
    if match:
        amt_str = match.group(1).strip()
        range_end = match.group(2)
        rest = match.group(3).strip()
        
        if " " in amt_str and "/" in amt_str:
            whole, frac = amt_str.split()
            val = float(whole) + (parse_fraction(frac) or 0.0)
        elif "/" in amt_str:
            val = parse_fraction(amt_str) or 0.0
        else:
            val = parse_fraction(amt_str) or 0.0
            
        display_amt = f"{amt_str}-{range_end}" if range_end else amt_str
            
        rest_parts = rest.split(None, 1)
        if rest_parts:
            first_word_clean = rest_parts[0].lower().rstrip(".,")
            if first_word_clean in COMMON_UNITS or rest_parts[0].lower() in COMMON_UNITS:
                unit = rest_parts[0]
                ing_name = rest_parts[1] if len(rest_parts) > 1 else ""
                return {
                    "type": "ingredient",
                    "is_header": False,
                    "amount_val": val,
                    "amount": val,
                    "amount_str": display_amt,
                    "amount_formatted": display_amt,
                    "unit": unit,
                    "name": ing_name,
                    "raw": raw
                }
        
        return {
            "type": "ingredient",
            "is_header": False,
            "amount_val": val,
            "amount": val,
            "amount_str": display_amt,
            "amount_formatted": display_amt,
            "unit": "",
            "name": rest,
            "raw": raw
        }
        
    return {
        "type": "ingredient",
        "is_header": False,
        "amount_val": None,
        "amount": None,
        "amount_str": "",
        "amount_formatted": "",
        "unit": "",
        "name": raw,
        "raw": raw
    }

def parse_ingredients_block(text: str) -> List[Dict[str, Any]]:
    lines = [l.strip() for l in (text or "").split("\n") if l.strip()]
    return [parse_ingredient_line(l) for l in lines]


def format_quantity_py(num: Optional[float]) -> str:
    """Formatiert eine Zahl lesbar für Rezepte (z. B. 0.5 -> ½, 1.5 -> 1 ½, 2.0 -> 2)."""
    if num is None or num <= 0:
        return ""
    whole = int(num)
    frac = round((num - whole) * 100) / 100.0

    frac_str = ""
    if 0.20 <= frac <= 0.30:
        frac_str = "¼"
    elif 0.31 <= frac <= 0.40:
        frac_str = "⅓"
    elif 0.45 <= frac <= 0.55:
        frac_str = "½"
    elif 0.63 <= frac <= 0.70:
        frac_str = "⅔"
    elif 0.72 <= frac <= 0.80:
        frac_str = "¾"

    if frac_str:
        return f"{whole} {frac_str}".strip() if whole > 0 else frac_str

    if abs(num - round(num)) < 0.05:
        return str(int(round(num)))

    # Deutsche Kommaschreibweise
    val_str = f"{num:.1f}".rstrip("0").rstrip(".").replace(".", ",")
    return val_str


def calculate_portion_matrix(
    parsed_ingredients: List[Dict[str, Any]], 
    base_portions: int = 4, 
    targets: tuple = (2, 4, 6, 8)
) -> List[Dict[str, Any]]:
    """Berechnet für Waffeln / Pancakes etc. eine vorberechnete Tabelle für 2, 4, 6, 8 Portionen."""
    if not base_portions or base_portions <= 0:
        base_portions = 4

    matrix = []
    for item in parsed_ingredients:
        if item.get("is_header"):
            matrix.append({
                "is_header": True,
                "text": item.get("text") or item.get("name") or "",
                "amounts": {p: "" for p in targets}
            })
            continue

        row_amounts = {}
        for p in targets:
            amt_val = item.get("amount_val")
            unit = item.get("unit") or ""
            if amt_val is not None:
                scaled_val = amt_val * (p / float(base_portions))
                formatted_amt = format_quantity_py(scaled_val)
                row_amounts[p] = f"{formatted_amt} {unit}".strip()
            else:
                row_amounts[p] = "—"

        matrix.append({
            "is_header": False,
            "name": item.get("name") or item.get("raw") or "",
            "raw": item.get("raw", ""),
            "amounts": row_amounts
        })

    return matrix