Spaces:
Sleeping
Sleeping
Mise en forme
Browse files- quick_deploy_agent.py +110 -0
quick_deploy_agent.py
CHANGED
|
@@ -129,6 +129,116 @@ class OFFtoCOICOP(Tool):
|
|
| 129 |
|
| 130 |
return json.dumps({"candidates": c})
|
| 131 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 132 |
|
| 133 |
|
| 134 |
|
|
|
|
| 129 |
|
| 130 |
return json.dumps({"candidates": c})
|
| 131 |
|
| 132 |
+
# ---- OFFByEAN : robuste + sortie normalisée + step3_inputs ----
|
| 133 |
+
class OFFByEAN(Tool):
|
| 134 |
+
name = "openfoodfacts_product_by_ean"
|
| 135 |
+
description = "Open Food Facts /api/v0|v2/product/{ean} (name, brands, categories...)."
|
| 136 |
+
inputs = {"ean": {"type": "string", "description": "EAN à interroger sur l'API OFF."}}
|
| 137 |
+
output_type = "string"
|
| 138 |
+
requirements = ["requests"] # urllib3 vient via requests
|
| 139 |
+
|
| 140 |
+
def forward(self, ean: str) -> str:
|
| 141 |
+
import re, json, requests
|
| 142 |
+
from requests.adapters import HTTPAdapter
|
| 143 |
+
try:
|
| 144 |
+
from urllib3.util.retry import Retry
|
| 145 |
+
except Exception:
|
| 146 |
+
Retry = None # environnement minimal
|
| 147 |
+
|
| 148 |
+
def _to_list(x):
|
| 149 |
+
if x is None:
|
| 150 |
+
return []
|
| 151 |
+
if isinstance(x, list):
|
| 152 |
+
# stringifier proprement les éléments non-str
|
| 153 |
+
return [str(t).strip() for t in x if str(t).strip()]
|
| 154 |
+
if isinstance(x, str):
|
| 155 |
+
parts = [p.strip() for p in re.split(r"[,\|;]", x) if p.strip()]
|
| 156 |
+
return parts
|
| 157 |
+
return [str(x).strip()]
|
| 158 |
+
|
| 159 |
+
def _first(*vals):
|
| 160 |
+
for v in vals:
|
| 161 |
+
if isinstance(v, str) and v.strip():
|
| 162 |
+
return v.strip()
|
| 163 |
+
return ""
|
| 164 |
+
|
| 165 |
+
code = re.sub(r"\D", "", ean or "")
|
| 166 |
+
if not code:
|
| 167 |
+
return json.dumps({"ok": False, "status": 0, "code": "", "error": "EAN vide"})
|
| 168 |
+
|
| 169 |
+
sess = requests.Session()
|
| 170 |
+
sess.headers.update({
|
| 171 |
+
"User-Agent": "insee-coicop-agent/1.0",
|
| 172 |
+
"Accept": "application/json",
|
| 173 |
+
})
|
| 174 |
+
if Retry:
|
| 175 |
+
retry = Retry(
|
| 176 |
+
total=3, backoff_factor=0.5,
|
| 177 |
+
status_forcelist=[429, 500, 502, 503, 504],
|
| 178 |
+
allowed_methods=frozenset(["GET"]),
|
| 179 |
+
raise_on_status=False,
|
| 180 |
+
)
|
| 181 |
+
sess.mount("https://", HTTPAdapter(max_retries=retry))
|
| 182 |
+
|
| 183 |
+
urls = [
|
| 184 |
+
f"https://world.openfoodfacts.org/api/v0/product/{code}.json",
|
| 185 |
+
"https://world.openfoodfacts.org/api/v2/product/"
|
| 186 |
+
f"{code}?lc=fr&fields=code,product_name,product_name_fr,brands,"
|
| 187 |
+
"categories_tags,categories_tags_fr,ingredients_text,ingredients_text_fr,"
|
| 188 |
+
"stores,status,status_verbose",
|
| 189 |
+
f"https://world.openfoodfacts.net/api/v0/product/{code}.json",
|
| 190 |
+
]
|
| 191 |
+
|
| 192 |
+
last_err = None
|
| 193 |
+
for u in urls:
|
| 194 |
+
try:
|
| 195 |
+
r = sess.get(u, timeout=15)
|
| 196 |
+
if not r.ok:
|
| 197 |
+
last_err = f"HTTP {r.status_code}"
|
| 198 |
+
continue
|
| 199 |
+
data = r.json()
|
| 200 |
+
product = data.get("product")
|
| 201 |
+
status = data.get("status", 1 if product else 0)
|
| 202 |
+
if status == 1 or product:
|
| 203 |
+
p = product or {}
|
| 204 |
+
# Normalisation stricte des champs
|
| 205 |
+
product_name = _first(p.get("product_name_fr"), p.get("product_name"))
|
| 206 |
+
categories_tags = (
|
| 207 |
+
p.get("categories_tags_fr")
|
| 208 |
+
or p.get("categories_tags")
|
| 209 |
+
or p.get("categories")
|
| 210 |
+
)
|
| 211 |
+
categories_tags = _to_list(categories_tags)
|
| 212 |
+
ingredients_text = _first(p.get("ingredients_text_fr"), p.get("ingredients_text"))
|
| 213 |
+
brands = _first(p.get("brands"), None)
|
| 214 |
+
stores = _first(p.get("stores"), None)
|
| 215 |
+
|
| 216 |
+
out = {
|
| 217 |
+
"ok": True,
|
| 218 |
+
"status": status,
|
| 219 |
+
"status_verbose": data.get("status_verbose"),
|
| 220 |
+
"code": code,
|
| 221 |
+
"used_url": u,
|
| 222 |
+
# Champs lisibles directement par le LLM
|
| 223 |
+
"product_name": product_name,
|
| 224 |
+
"categories_tags": categories_tags, # toujours list[str]
|
| 225 |
+
"ingredients_text": ingredients_text,
|
| 226 |
+
"brands": brands, # string (OFF est souvent "brand1, brand2")
|
| 227 |
+
"brands_list": _to_list(brands), # list[str] pratique
|
| 228 |
+
"stores": stores,
|
| 229 |
+
"stores_list": _to_list(stores),
|
| 230 |
+
# 🔑 Bloc prêt pour l'étape 3
|
| 231 |
+
"step3_inputs": {
|
| 232 |
+
"product_name": product_name,
|
| 233 |
+
"categories_tags": categories_tags,
|
| 234 |
+
"ingredients_text": ingredients_text,
|
| 235 |
+
},
|
| 236 |
+
}
|
| 237 |
+
return json.dumps(out)
|
| 238 |
+
except Exception as e:
|
| 239 |
+
last_err = str(e)
|
| 240 |
+
|
| 241 |
+
return json.dumps({"ok": False, "status": 0, "code": code, "error": last_err or "not found"})
|
| 242 |
|
| 243 |
|
| 244 |
|