mitai-jinkendo/backend/bls/attribute_labels.py

322 lines
14 KiB
Python
Raw Normal View History

"""Official BLS 4.0 component labels (MRI documentation) and placeholder repair.
``attr_key`` stays the identity. These names only fill empty or key-equal ``name_de``.
"""
from __future__ import annotations
from typing import Any
# Keys that feed nutrition_log kcal / protein_g / fat_g / carbs_g
MACRO_KEYS = ("ENERCC", "PROT625", "FAT", "CHO")
GROUP_META = (
("energy", "Energie"),
("macro", "Makronährstoffe"),
("carbs", "Kohlenhydrate"),
("fiber", "Ballaststoffe"),
("fatty_acids", "Fettsäuren"),
("amino_acids", "Aminosäuren"),
("vitamins", "Vitamine"),
("minerals", "Mineralstoffe"),
("organic_acids", "Organische Säuren"),
("polyols", "Zuckeralkohole"),
("other", "Sonstige"),
("extension", "Eigene Erweiterungen"),
("unknown", "Ohne Gruppe"),
)
GROUP_LABELS = {key: label for key, label in GROUP_META}
GROUP_RANK = {key: i for i, (key, _) in enumerate(GROUP_META)}
# Code → (name_de, unit, category) from BLS 4.0 Referenzhandbuch (Jan 2026)
_OFFICIAL: dict[str, tuple[str, str, str]] = {
"ENERCJ": ("Energie (Kilojoule)", "kJ", "energy"),
"ENERCC": ("Energie (Kilokalorien)", "kcal", "energy"),
"WATER": ("Wasser", "g", "macro"),
"PROT625": ("Protein (N×6,25)", "g", "macro"),
"FAT": ("Fett", "g", "macro"),
"CHO": ("Kohlenhydrate, verfügbar", "g", "macro"),
"FIBT": ("Ballaststoffe, gesamt", "g", "fiber"),
"ALC": ("Alkohol (Ethanol)", "g", "macro"),
"OA": ("Organische Säuren, gesamt", "g", "organic_acids"),
"ASH": ("Rohasche", "g", "macro"),
"POLYL": ("Zuckeralkohole, gesamt", "g", "polyols"),
"GLUS": ("Glucose", "g", "carbs"),
"FRUS": ("Fructose", "g", "carbs"),
"GALS": ("Galactose", "g", "carbs"),
"MNSAC": ("Monosaccharide, gesamt", "g", "carbs"),
"SUCS": ("Saccharose", "g", "carbs"),
"MALS": ("Maltose", "g", "carbs"),
"LACS": ("Lactose", "g", "carbs"),
"DISAC": ("Disaccharide, gesamt", "g", "carbs"),
"SUGAR": ("Zucker, gesamt", "g", "carbs"),
"OLSAC": ("Oligosaccharide, verfügbar", "g", "carbs"),
"STARCH": ("Stärke", "g", "carbs"),
"FIBLMW": ("Ballaststoffe, niedermolekular", "g", "fiber"),
"FIBHMW": ("Ballaststoffe, hochmolekular", "g", "fiber"),
"FIBINS": ("Ballaststoffe, wasserunlöslich", "g", "fiber"),
"FIBSOL": ("Ballaststoffe, wasserlöslich", "g", "fiber"),
"FIBHMWS": ("Ballaststoffe, hochmolekular, wasserlöslich", "g", "fiber"),
"FIBHMWI": ("Ballaststoffe, hochmolekular, wasserunlöslich", "g", "fiber"),
"FASAT": ("Fettsäuren, gesättigt, gesamt", "g", "fatty_acids"),
"F4:0": ("Buttersäure (C4:0)", "g", "fatty_acids"),
"F6:0": ("Capronsäure (C6:0)", "g", "fatty_acids"),
"F8:0": ("Caprylsäure (C8:0)", "g", "fatty_acids"),
"F10:0": ("Caprinsäure (C10:0)", "g", "fatty_acids"),
"F12:0": ("Laurinsäure (C12:0)", "g", "fatty_acids"),
"F14:0": ("Myristinsäure (C14:0)", "g", "fatty_acids"),
"F15:0": ("Pentadecylsäure (C15:0)", "g", "fatty_acids"),
"F16:0": ("Palmitinsäure (C16:0)", "g", "fatty_acids"),
"F17:0": ("Margarinsäure (C17:0)", "g", "fatty_acids"),
"F18:0": ("Stearinsäure (C18:0)", "g", "fatty_acids"),
"F20:0": ("Arachinsäure (C20:0)", "g", "fatty_acids"),
"F22:0": ("Behensäure (C22:0)", "g", "fatty_acids"),
"F24:0": ("Lignocerinsäure (C24:0)", "g", "fatty_acids"),
"FAMS": ("Fettsäuren, einfach ungesättigt, gesamt", "g", "fatty_acids"),
"F14:1CN5": ("Myristoleinsäure (C14:1 n-5 cis)", "g", "fatty_acids"),
"F16:1CN7": ("Palmitoleinsäure (C16:1 n-7 cis)", "g", "fatty_acids"),
"F18:1CN7": ("Vaccensäure (C18:1 n-7 cis)", "g", "fatty_acids"),
"F18:1CN9": ("Ölsäure (C18:1 n-9 cis)", "g", "fatty_acids"),
"F20:1CN9": ("Gondosäure (C20:1 n-9 cis)", "g", "fatty_acids"),
"F22:1CN9": ("Erucasäure (C22:1 n-9 cis)", "g", "fatty_acids"),
"FAPU": ("Fettsäuren, mehrfach ungesättigt, gesamt", "g", "fatty_acids"),
"FAPUN3": ("Omega-3-Fettsäuren, gesamt", "g", "fatty_acids"),
"FAPUN6": ("Omega-6-Fettsäuren, gesamt", "g", "fatty_acids"),
"F18:2CN6": ("Linolsäure (C18:2 n-6 cis)", "g", "fatty_acids"),
"F18:2C9T11": ("Konjugierte Linolsäure (C18:2 cis-9, trans-11)", "g", "fatty_acids"),
"F18:3CN3": ("Alpha-Linolensäure (C18:3 n-3)", "g", "fatty_acids"),
"F18:3CN6": ("Gamma-Linolensäure (C18:3 n-6)", "g", "fatty_acids"),
"F18:4CN3": ("Stearidonsäure (C18:4 n-3)", "g", "fatty_acids"),
"F20:2CN6": ("Eicosadiensäure (C20:2 n-6)", "g", "fatty_acids"),
"F20:3CN6": ("Dihomo-Gamma-Linolensäure (C20:3 n-6)", "g", "fatty_acids"),
"F20:4CN6": ("Arachidonsäure (C20:4 n-6)", "g", "fatty_acids"),
"F20:5CN3": ("Eicosapentaensäure EPA (C20:5 n-3)", "g", "fatty_acids"),
"F22:5CN3": ("Docosapentaensäure DPA (C22:5 n-3)", "g", "fatty_acids"),
"F22:6CN3": ("Docosahexaensäure DHA (C22:6 n-3)", "g", "fatty_acids"),
"FAX": ("Fettsäuren, sonstige", "g", "fatty_acids"),
"AAE9": ("Aminosäuren, unentbehrlich, gesamt", "g", "amino_acids"),
"ILE": ("Isoleucin", "g", "amino_acids"),
"LEU": ("Leucin", "g", "amino_acids"),
"LYS": ("Lysin", "g", "amino_acids"),
"MET": ("Methionin", "g", "amino_acids"),
"PHE": ("Phenylalanin", "g", "amino_acids"),
"THR": ("Threonin", "g", "amino_acids"),
"TRP": ("Tryptophan", "g", "amino_acids"),
"VAL": ("Valin", "g", "amino_acids"),
"HIS": ("Histidin", "g", "amino_acids"),
"ALA": ("Alanin", "g", "amino_acids"),
"ARG": ("Arginin", "g", "amino_acids"),
"ASP": ("Asparaginsäure inkl. Asparagin", "g", "amino_acids"),
"CYS": ("Cystein", "g", "amino_acids"),
"GLU": ("Glutaminsäure inkl. Glutamin", "g", "amino_acids"),
"GLY": ("Glycin", "g", "amino_acids"),
"PRO": ("Prolin", "g", "amino_acids"),
"SER": ("Serin", "g", "amino_acids"),
"TYR": ("Tyrosin", "g", "amino_acids"),
"VITA": ("Vitamin A, Retinol-Äquivalent (RE)", "µg", "vitamins"),
"VITAA": ("Vitamin A, Retinol-Aktivitätsäquivalent (RAE)", "µg", "vitamins"),
"RETOL": ("Retinol", "µg", "vitamins"),
"CARTB": ("Beta-Carotin", "µg", "vitamins"),
"CAROTPAXB": ("Carotinoide außer Beta-Carotin", "µg", "vitamins"),
"VITD": ("Vitamin D", "µg", "vitamins"),
"ERGCAL": ("Vitamin D2 (Ergocalciferol)", "µg", "vitamins"),
"CHOCAL": ("Vitamin D3 (Cholecalciferol)", "µg", "vitamins"),
"VITE": ("Vitamin E (Alpha-Tocopherol)", "mg", "vitamins"),
"TOCPHA": ("Alpha-Tocopherol", "mg", "vitamins"),
"TOCPHB": ("Beta-Tocopherol", "mg", "vitamins"),
"TOCPHG": ("Gamma-Tocopherol", "mg", "vitamins"),
"TOCPHD": ("Delta-Tocopherol", "mg", "vitamins"),
"TOCTRA": ("Alpha-Tocotrienol", "mg", "vitamins"),
"VITK": ("Vitamin K", "µg", "vitamins"),
"VITK1": ("Vitamin K1 (Phyllochinon)", "µg", "vitamins"),
"VITK2": ("Vitamin K2 (Menachinone)", "µg", "vitamins"),
"THIA": ("Vitamin B1 (Thiamin)", "mg", "vitamins"),
"RIBF": ("Vitamin B2 (Riboflavin)", "mg", "vitamins"),
"NIA": ("Niacin", "mg", "vitamins"),
"NIAEQ": ("Niacin-Äquivalent", "mg", "vitamins"),
"PANTAC": ("Pantothensäure", "mg", "vitamins"),
"VITB6": ("Vitamin B6", "µg", "vitamins"),
"BIOT": ("Biotin", "µg", "vitamins"),
"FOL": ("Folat-Äquivalent", "µg", "vitamins"),
"FOLFD": ("Folat", "µg", "vitamins"),
"FOLAC": ("Folsäure, synthetisch", "µg", "vitamins"),
"VITB12": ("Vitamin B12 (Cobalamine)", "µg", "vitamins"),
"VITC": ("Vitamin C", "mg", "vitamins"),
"NACL": ("Salz (Natriumchlorid)", "g", "minerals"),
"NA": ("Natrium", "mg", "minerals"),
"CLD": ("Chlorid", "mg", "minerals"),
"K": ("Kalium", "mg", "minerals"),
"CA": ("Calcium", "mg", "minerals"),
"MG": ("Magnesium", "mg", "minerals"),
"P": ("Phosphor", "mg", "minerals"),
"S": ("Schwefel", "mg", "minerals"),
"FE": ("Eisen", "mg", "minerals"),
"ZN": ("Zink", "mg", "minerals"),
"ID": ("Iodid", "µg", "minerals"),
"CU": ("Kupfer", "µg", "minerals"),
"MN": ("Mangan", "µg", "minerals"),
"FD": ("Fluorid", "µg", "minerals"),
"CR": ("Chrom", "µg", "minerals"),
"MO": ("Molybdän", "µg", "minerals"),
"ACEAC": ("Essigsäure", "g", "organic_acids"),
"CITAC": ("Zitronensäure", "g", "organic_acids"),
"LACAC": ("Milchsäure", "g", "organic_acids"),
"MALAC": ("Äpfelsäure", "g", "organic_acids"),
"TARAC": ("Weinsäure", "g", "organic_acids"),
"MANTL": ("Mannit", "g", "polyols"),
"SORTL": ("Sorbit", "g", "polyols"),
"XYLTL": ("Xylit", "g", "polyols"),
"CHORL": ("Cholesterin", "mg", "other"),
"NT": ("Stickstoff, gesamt", "g", "other"),
"EPA": ("Eicosapentaensäure (EPA)", "g", "fatty_acids"),
"DHA": ("Docosahexaensäure (DHA)", "g", "fatty_acids"),
"DPA": ("Docosapentaensäure (DPA)", "g", "fatty_acids"),
"ALA": ("Alpha-Linolensäure (ALA)", "g", "fatty_acids"),
"OMEGA3": ("Omega-3-Fettsäuren gesamt", "g", "fatty_acids"),
"OMEGA6": ("Omega-6-Fettsäuren gesamt", "g", "fatty_acids"),
}
def lookup_key(raw: str | None) -> str:
return str(raw or "").strip().upper()
def resolve_group(attr_key: str | None, category: str | None = None, origin: str | None = None) -> tuple[str, str]:
cat = (category or "").strip().lower()
if cat in GROUP_LABELS:
return cat, GROUP_LABELS[cat]
lab = official_label(attr_key)
if lab and lab["category"] in GROUP_LABELS:
return lab["category"], GROUP_LABELS[lab["category"]]
if (origin or "") == "extension":
return "extension", GROUP_LABELS["extension"]
return "unknown", GROUP_LABELS["unknown"]
def annotate_attribute(row: dict[str, Any]) -> dict[str, Any]:
key = row.get("attr_key")
group_key, group_label = resolve_group(key, row.get("category"), row.get("origin"))
row["is_macro"] = key in MACRO_KEYS
row["needs_label"] = label_is_placeholder(row.get("name_de"), key)
row["group_key"] = group_key
row["group_label"] = group_label
row["group_rank"] = GROUP_RANK[group_key]
return row
def sort_attributes(rows: list[dict[str, Any]]) -> list[dict[str, Any]]:
def key_of(row: dict[str, Any]):
return (
row.get("group_rank", 99),
0 if row.get("is_macro") else 1,
row.get("sort_order") if row.get("sort_order") is not None else 0,
(row.get("name_de") or row.get("attr_key") or "").casefold(),
row.get("attr_key") or "",
)
return sorted(rows, key=key_of)
def official_label(attr_key: str | None) -> dict[str, str] | None:
key = lookup_key(attr_key)
hit = _OFFICIAL.get(key)
if not hit:
hit = _OFFICIAL.get(key.replace("_", ":"))
if not hit:
return None
name_de, unit, category = hit
return {"name_de": name_de, "unit": unit, "category": category}
def label_is_placeholder(name_de: str | None, attr_key: str | None) -> bool:
name = (name_de or "").strip()
key = lookup_key(attr_key)
return (not name) or name.upper() == key
def apply_official_name(attr_key: str | None, name_de: str | None) -> str:
if not label_is_placeholder(name_de, attr_key):
return (name_de or "").strip()
lab = official_label(attr_key)
if lab:
return lab["name_de"]
return (name_de or "").strip() or lookup_key(attr_key)
def enrich_attributes_in_memory(attributes: list[dict[str, Any]]) -> list[dict[str, Any]]:
for a in attributes:
key = a.get("attr_key")
lab = official_label(key)
if label_is_placeholder(a.get("name_de"), key) and lab:
a["name_de"] = lab["name_de"]
a["unit"] = a.get("unit") or lab["unit"]
a["category"] = a.get("category") or lab["category"]
return attributes
def fill_placeholder_labels(cur) -> dict[str, int]:
cur.execute(
"""
SELECT id, attr_key, name_de, unit, category
FROM food_attributes
WHERE is_active = true
"""
)
updated = 0
for row in cur.fetchall():
rec = dict(row)
lab = official_label(rec.get("attr_key"))
if not lab:
continue
need_name = label_is_placeholder(rec.get("name_de"), rec.get("attr_key"))
need_cat = not (rec.get("category") or "").strip()
if not need_name and not need_cat:
continue
cur.execute(
"""
UPDATE food_attributes
SET name_de = %s,
unit = COALESCE(NULLIF(BTRIM(COALESCE(unit, '')), ''), %s),
category = COALESCE(NULLIF(BTRIM(COALESCE(category, '')), ''), %s),
updated_at = NOW()
WHERE id = %s
""",
(
lab["name_de"] if need_name else rec.get("name_de"),
lab["unit"],
lab["category"],
rec["id"],
),
)
updated += 1
return {"updated": updated}
def enrich_labels_from_food_headers(cur, headers: list[dict[str, Any]] | None) -> int:
updated = 0
for h in headers or []:
key = h.get("attr_key")
name = (h.get("name_de") or "").strip()
unit = (h.get("unit") or "").strip() or None
if not key or not name or label_is_placeholder(name, key):
lab = official_label(key)
if not lab:
continue
name = lab["name_de"]
unit = unit or lab["unit"]
cur.execute(
"""
UPDATE food_attributes
SET name_de = %s,
unit = COALESCE(%s, unit),
updated_at = NOW()
WHERE attr_key = %s
AND (
name_de IS NULL OR BTRIM(name_de) = '' OR name_de = attr_key
)
""",
(name, unit, key),
)
if cur.rowcount:
updated += cur.rowcount
return updated