Update app.py
Browse files
app.py
CHANGED
|
@@ -39,7 +39,9 @@ class PesticideDataFetcher:
|
|
| 39 |
HEADERS = {
|
| 40 |
'User-Agent': 'Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/91.0.4472.124 Safari/537.36',
|
| 41 |
'Accept': 'application/json',
|
| 42 |
-
'Accept-Language': 'fr-FR,fr;q=0.9,en-US;q=0.8,en;q=0.7'
|
|
|
|
|
|
|
| 43 |
}
|
| 44 |
CACHE_DIR = "cache"
|
| 45 |
SUBSTANCE_CACHE_FILE = os.path.join(CACHE_DIR, "substance_cache.json")
|
|
@@ -62,10 +64,12 @@ class PesticideDataFetcher:
|
|
| 62 |
if use_cache:
|
| 63 |
self._load_caches()
|
| 64 |
|
| 65 |
-
# Préchargement des substances si le cache est vide
|
| 66 |
if not self._substance_cache:
|
| 67 |
self.preload_substance_names()
|
| 68 |
|
|
|
|
|
|
|
| 69 |
def _load_caches(self):
|
| 70 |
"""Charge les données de cache depuis les fichiers"""
|
| 71 |
try:
|
|
@@ -123,10 +127,10 @@ class PesticideDataFetcher:
|
|
| 123 |
logger.error(f"Erreur lors de la sauvegarde des caches: {e}")
|
| 124 |
|
| 125 |
@retry(stop=stop_after_attempt(3), wait=wait_exponential(multiplier=1, min=2, max=10))
|
| 126 |
-
def fetch_data(self, url: str) -> Dict[str, Any]:
|
| 127 |
-
"""Récupère les données depuis l'API avec mécanisme de retry"""
|
| 128 |
try:
|
| 129 |
-
response = self.session.get(url, timeout=15)
|
| 130 |
response.raise_for_status()
|
| 131 |
data = response.json()
|
| 132 |
logger.debug(f"Réponse API pour {url}: {str(data)[:200]}...")
|
|
@@ -138,46 +142,60 @@ class PesticideDataFetcher:
|
|
| 138 |
def preload_substance_names(self) -> None:
|
| 139 |
"""Précharge tous les noms de substances depuis l'API"""
|
| 140 |
logger.info("Préchargement des substances...")
|
| 141 |
-
url = f"{self.BASE_URL}/pesticide_residues
|
|
|
|
| 142 |
substances_loaded = 0
|
| 143 |
|
| 144 |
-
|
| 145 |
-
data = self.fetch_data(url)
|
| 146 |
-
if "
|
| 147 |
-
logger.
|
| 148 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 149 |
|
| 150 |
-
|
| 151 |
-
|
| 152 |
-
if substance_id and substance_id not in self._substance_cache:
|
| 153 |
-
self._substance_cache[substance_id] = SubstanceDetails(
|
| 154 |
-
name=substance.get("pesticideResidueName", "Inconnu"),
|
| 155 |
-
substance_id=substance_id,
|
| 156 |
-
status=substance.get("status"),
|
| 157 |
-
cas_number=substance.get("casNumber"),
|
| 158 |
-
ec_number=substance.get("ecNumber")
|
| 159 |
-
)
|
| 160 |
-
substances_loaded += 1
|
| 161 |
|
| 162 |
-
|
| 163 |
-
|
|
|
|
|
|
|
|
|
|
| 164 |
|
| 165 |
-
url = data.get("nextLink")
|
| 166 |
-
logger.info(f"Substances chargées jusqu'à présent: {substances_loaded}")
|
| 167 |
|
| 168 |
-
|
| 169 |
-
|
| 170 |
|
| 171 |
-
|
|
|
|
|
|
|
|
|
|
| 172 |
|
| 173 |
def _load_substance_details(self, substance_id: int) -> None:
|
| 174 |
"""Charge les détails supplémentaires d'une substance"""
|
| 175 |
-
url = f"{self.BASE_URL}/pesticide_residues/{substance_id}
|
|
|
|
| 176 |
try:
|
| 177 |
-
data = self.fetch_data(url)
|
| 178 |
|
| 179 |
substance = self._substance_cache.get(substance_id)
|
| 180 |
-
if substance:
|
| 181 |
substance.approval_date = data.get("approvalDate")
|
| 182 |
substance.expiry_date = data.get("expiryDate")
|
| 183 |
# Mise à jour du statut si non défini précédemment
|
|
@@ -193,8 +211,11 @@ class PesticideDataFetcher:
|
|
| 193 |
|
| 194 |
# Si la substance n'est pas dans le cache, essayer de la récupérer
|
| 195 |
try:
|
| 196 |
-
url = f"{self.BASE_URL}/pesticide_residues/{substance_id}
|
| 197 |
-
|
|
|
|
|
|
|
|
|
|
| 198 |
|
| 199 |
substance_name = data.get("pesticideResidueName", f"Substance {substance_id}")
|
| 200 |
|
|
@@ -223,23 +244,28 @@ class PesticideDataFetcher:
|
|
| 223 |
return list(self._product_cache.values())
|
| 224 |
|
| 225 |
logger.info("Récupération de la liste des produits...")
|
| 226 |
-
url = f"{self.BASE_URL}/pesticide_residues_products
|
|
|
|
| 227 |
products_loaded = 0
|
| 228 |
|
| 229 |
while url:
|
| 230 |
-
data = self.fetch_data(url)
|
| 231 |
if "error" in data:
|
| 232 |
logger.error(f"Erreur produits: {data.get('error', 'Aucune info')}")
|
| 233 |
break
|
| 234 |
|
| 235 |
-
|
| 236 |
-
|
| 237 |
-
|
| 238 |
-
|
| 239 |
-
|
|
|
|
| 240 |
|
| 241 |
url = data.get("nextLink")
|
| 242 |
logger.info(f"Produits récupérés jusqu'à présent: {products_loaded}")
|
|
|
|
|
|
|
|
|
|
| 243 |
|
| 244 |
if self.use_cache:
|
| 245 |
self._save_caches()
|
|
@@ -254,17 +280,21 @@ class PesticideDataFetcher:
|
|
| 254 |
return self._mrl_cache[product_id]
|
| 255 |
|
| 256 |
logger.info(f"Récupération des LMR pour le produit {product_id}...")
|
| 257 |
-
url = f"{self.BASE_URL}/pesticide_residues_products/{product_id}/mrls
|
|
|
|
| 258 |
|
| 259 |
mrls = []
|
| 260 |
while url:
|
| 261 |
-
data = self.fetch_data(url)
|
| 262 |
if "error" in data:
|
| 263 |
logger.error(f"Erreur lors de la récupération des LMR: {data.get('error', 'Aucune info')}")
|
| 264 |
break
|
| 265 |
-
|
| 266 |
-
|
| 267 |
url = data.get("nextLink")
|
|
|
|
|
|
|
|
|
|
| 268 |
|
| 269 |
# Mise à jour du cache
|
| 270 |
self._mrl_cache[product_id] = mrls
|
|
@@ -284,20 +314,31 @@ class PesticideDataFetcher:
|
|
| 284 |
]
|
| 285 |
return sorted(results, key=lambda x: x.name)
|
| 286 |
|
| 287 |
-
def get_substance_mrls(self, substance_id: int) -> List[Dict[str,Any]]:
|
| 288 |
-
"""Récupère tous les produits avec LMR pour une substance donnée"""
|
| 289 |
logger.info(f"Récupération des LMR pour la substance {substance_id}...")
|
| 290 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 291 |
|
| 292 |
all_mrls = []
|
| 293 |
while url:
|
| 294 |
-
data = self.fetch_data(url)
|
| 295 |
if "error" in data:
|
| 296 |
logger.error(f"Erreur lors de la récupération des LMR: {data.get('error', 'Aucune info')}")
|
| 297 |
break
|
| 298 |
|
| 299 |
-
|
|
|
|
| 300 |
url = data.get("nextLink")
|
|
|
|
|
|
|
|
|
|
|
|
|
| 301 |
|
| 302 |
logger.info(f"LMR récupérées pour la substance {substance_id}: {len(all_mrls)}")
|
| 303 |
return all_mrls
|
|
@@ -429,7 +470,7 @@ class PesticideApp:
|
|
| 429 |
|
| 430 |
df = pd.DataFrame(data) # Crée le DataFrame
|
| 431 |
logger.info(f"LMR récupérées pour {substance.name}: {len(df)} entrées")
|
| 432 |
-
return df
|
| 433 |
|
| 434 |
def create_histogram(self, df: pd.DataFrame) -> go.Figure:
|
| 435 |
"""Crée un histogramme des valeurs LMR"""
|
|
@@ -574,8 +615,8 @@ class PesticideApp:
|
|
| 574 |
|
| 575 |
substance_mrls_btn.click(
|
| 576 |
fn=lambda sid: (
|
| 577 |
-
self.get_substance_mrls(int(sid)),
|
| 578 |
-
self.create_histogram(self.get_substance_mrls(int(sid)))
|
| 579 |
),
|
| 580 |
inputs=[substance_select],
|
| 581 |
outputs=[substance_mrls, mrl_histogram]
|
|
|
|
| 39 |
HEADERS = {
|
| 40 |
'User-Agent': 'Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/91.0.4472.124 Safari/537.36',
|
| 41 |
'Accept': 'application/json',
|
| 42 |
+
'Accept-Language': 'fr-FR,fr;q=0.9,en-US;q=0.8,en;q=0.7',
|
| 43 |
+
"Content-Type": "application/json", # Added as per API docs
|
| 44 |
+
"Cache-Control": "no-cache" # Added as per API docs
|
| 45 |
}
|
| 46 |
CACHE_DIR = "cache"
|
| 47 |
SUBSTANCE_CACHE_FILE = os.path.join(CACHE_DIR, "substance_cache.json")
|
|
|
|
| 64 |
if use_cache:
|
| 65 |
self._load_caches()
|
| 66 |
|
| 67 |
+
# Préchargement des substances si le cache est vide et si le chargement réussi.
|
| 68 |
if not self._substance_cache:
|
| 69 |
self.preload_substance_names()
|
| 70 |
|
| 71 |
+
|
| 72 |
+
|
| 73 |
def _load_caches(self):
|
| 74 |
"""Charge les données de cache depuis les fichiers"""
|
| 75 |
try:
|
|
|
|
| 127 |
logger.error(f"Erreur lors de la sauvegarde des caches: {e}")
|
| 128 |
|
| 129 |
@retry(stop=stop_after_attempt(3), wait=wait_exponential(multiplier=1, min=2, max=10))
|
| 130 |
+
def fetch_data(self, url: str, params: Optional[Dict[str, Any]] = None) -> Dict[str, Any]:
|
| 131 |
+
"""Récupère les données depuis l'API avec mécanisme de retry et paramètres optionnels."""
|
| 132 |
try:
|
| 133 |
+
response = self.session.get(url, params=params, timeout=15)
|
| 134 |
response.raise_for_status()
|
| 135 |
data = response.json()
|
| 136 |
logger.debug(f"Réponse API pour {url}: {str(data)[:200]}...")
|
|
|
|
| 142 |
def preload_substance_names(self) -> None:
|
| 143 |
"""Précharge tous les noms de substances depuis l'API"""
|
| 144 |
logger.info("Préchargement des substances...")
|
| 145 |
+
url = f"{self.BASE_URL}/pesticide_residues" # Corrected endpoint
|
| 146 |
+
params = {"format": "json", "language": "FR", "api-version": "v1.0"} # Use params dict
|
| 147 |
substances_loaded = 0
|
| 148 |
|
| 149 |
+
try:
|
| 150 |
+
data = self.fetch_data(url, params=params) # Initial fetch outside the loop
|
| 151 |
+
if not data or "value" not in data: # Handle potential empty or malformed response
|
| 152 |
+
logger.warning("Initial fetch returned no data. Skipping preloading.")
|
| 153 |
+
return
|
| 154 |
+
|
| 155 |
+
while url:
|
| 156 |
+
if "error" in data:
|
| 157 |
+
logger.error(f"Erreur lors du préchargement des substances: {data.get('error', 'Erreur inconnue')}")
|
| 158 |
+
break
|
| 159 |
+
|
| 160 |
+
for substance in data.get("value", []):
|
| 161 |
+
substance_id = substance.get("pesticideResidueId")
|
| 162 |
+
if substance_id and substance_id not in self._substance_cache:
|
| 163 |
+
self._substance_cache[substance_id] = SubstanceDetails(
|
| 164 |
+
name=substance.get("pesticideResidueName", "Inconnu"),
|
| 165 |
+
substance_id=substance_id,
|
| 166 |
+
status=substance.get("status"),
|
| 167 |
+
cas_number=substance.get("casNumber"),
|
| 168 |
+
ec_number=substance.get("ecNumber")
|
| 169 |
+
)
|
| 170 |
+
substances_loaded += 1
|
| 171 |
|
| 172 |
+
# Chargement des détails supplémentaires pour chaque substance
|
| 173 |
+
self._load_substance_details(substance_id)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 174 |
|
| 175 |
+
url = data.get("nextLink")
|
| 176 |
+
logger.info(f"Substances chargées jusqu'à présent: {substances_loaded}")
|
| 177 |
+
if url: # Only fetch if there's a nextLink
|
| 178 |
+
# No need to pass params again for nextLink, the API handles it
|
| 179 |
+
data = self.fetch_data(url)
|
| 180 |
|
|
|
|
|
|
|
| 181 |
|
| 182 |
+
if self.use_cache:
|
| 183 |
+
self._save_caches()
|
| 184 |
|
| 185 |
+
logger.info(f"Préchargement terminé. Total des substances: {len(self._substance_cache)}")
|
| 186 |
+
except Exception as e:
|
| 187 |
+
logger.error(f"Erreur lors du préchargement des substances: {e}")
|
| 188 |
+
return # Exit gracefully
|
| 189 |
|
| 190 |
def _load_substance_details(self, substance_id: int) -> None:
|
| 191 |
"""Charge les détails supplémentaires d'une substance"""
|
| 192 |
+
url = f"{self.BASE_URL}/pesticide_residues/{substance_id}" # Corrected endpoint
|
| 193 |
+
params = {"format": "json", "language": "FR", "api-version": "v1.0"} # Use params dict
|
| 194 |
try:
|
| 195 |
+
data = self.fetch_data(url, params=params)
|
| 196 |
|
| 197 |
substance = self._substance_cache.get(substance_id)
|
| 198 |
+
if substance and data: #check also if data is valid
|
| 199 |
substance.approval_date = data.get("approvalDate")
|
| 200 |
substance.expiry_date = data.get("expiryDate")
|
| 201 |
# Mise à jour du statut si non défini précédemment
|
|
|
|
| 211 |
|
| 212 |
# Si la substance n'est pas dans le cache, essayer de la récupérer
|
| 213 |
try:
|
| 214 |
+
url = f"{self.BASE_URL}/pesticide_residues/{substance_id}" # Corrected endpoint
|
| 215 |
+
params = {"format": "json", "language": "FR", "api-version": "v1.0"} # Use params dict
|
| 216 |
+
data = self.fetch_data(url, params)
|
| 217 |
+
if not data:
|
| 218 |
+
return f"Substance inconnue ({substance_id})"
|
| 219 |
|
| 220 |
substance_name = data.get("pesticideResidueName", f"Substance {substance_id}")
|
| 221 |
|
|
|
|
| 244 |
return list(self._product_cache.values())
|
| 245 |
|
| 246 |
logger.info("Récupération de la liste des produits...")
|
| 247 |
+
url = f"{self.BASE_URL}/pesticide_residues_products" # Corrected endpoint
|
| 248 |
+
params = {"format": "json", "language": "FR", "api-version": "v1.0"} # Use params dict
|
| 249 |
products_loaded = 0
|
| 250 |
|
| 251 |
while url:
|
| 252 |
+
data = self.fetch_data(url, params=params)
|
| 253 |
if "error" in data:
|
| 254 |
logger.error(f"Erreur produits: {data.get('error', 'Aucune info')}")
|
| 255 |
break
|
| 256 |
|
| 257 |
+
if data:
|
| 258 |
+
for product in data.get("value", []):
|
| 259 |
+
product_id = product.get("productId")
|
| 260 |
+
if product_id:
|
| 261 |
+
self._product_cache[product_id] = product
|
| 262 |
+
products_loaded += 1
|
| 263 |
|
| 264 |
url = data.get("nextLink")
|
| 265 |
logger.info(f"Produits récupérés jusqu'à présent: {products_loaded}")
|
| 266 |
+
if url :
|
| 267 |
+
# No need to pass params again for nextLink
|
| 268 |
+
data = self.fetch_data(url)
|
| 269 |
|
| 270 |
if self.use_cache:
|
| 271 |
self._save_caches()
|
|
|
|
| 280 |
return self._mrl_cache[product_id]
|
| 281 |
|
| 282 |
logger.info(f"Récupération des LMR pour le produit {product_id}...")
|
| 283 |
+
url = f"{self.BASE_URL}/pesticide_residues_products/{product_id}/mrls" # Corrected endpoint
|
| 284 |
+
params = {"format": "json", "language": "FR", "api-version": "v1.0"} # Use params dict
|
| 285 |
|
| 286 |
mrls = []
|
| 287 |
while url:
|
| 288 |
+
data = self.fetch_data(url, params=params)
|
| 289 |
if "error" in data:
|
| 290 |
logger.error(f"Erreur lors de la récupération des LMR: {data.get('error', 'Aucune info')}")
|
| 291 |
break
|
| 292 |
+
if data:
|
| 293 |
+
mrls.extend(data.get("value", []))
|
| 294 |
url = data.get("nextLink")
|
| 295 |
+
if url:
|
| 296 |
+
# No need for params in nextLink
|
| 297 |
+
data = self.fetch_data(url)
|
| 298 |
|
| 299 |
# Mise à jour du cache
|
| 300 |
self._mrl_cache[product_id] = mrls
|
|
|
|
| 314 |
]
|
| 315 |
return sorted(results, key=lambda x: x.name)
|
| 316 |
|
| 317 |
+
def get_substance_mrls(self, substance_id: int) -> List[Dict[str, Any]]:
|
| 318 |
+
"""Récupère tous les produits avec LMR pour une substance donnée en utilisant le nouvel endpoint."""
|
| 319 |
logger.info(f"Récupération des LMR pour la substance {substance_id}...")
|
| 320 |
+
|
| 321 |
+
url = f"{self.BASE_URL}/pesticide_residues_mrls" # Correct endpoint
|
| 322 |
+
params = {
|
| 323 |
+
"format": "json",
|
| 324 |
+
"api-version": "v1.0",
|
| 325 |
+
"pesticide_residue_id": substance_id # Use the correct parameter name
|
| 326 |
+
}
|
| 327 |
|
| 328 |
all_mrls = []
|
| 329 |
while url:
|
| 330 |
+
data = self.fetch_data(url, params=params) # Pass params here
|
| 331 |
if "error" in data:
|
| 332 |
logger.error(f"Erreur lors de la récupération des LMR: {data.get('error', 'Aucune info')}")
|
| 333 |
break
|
| 334 |
|
| 335 |
+
if data:
|
| 336 |
+
all_mrls.extend(data.get("value", []))
|
| 337 |
url = data.get("nextLink")
|
| 338 |
+
if url:
|
| 339 |
+
# No need to pass params again for the nextLink.
|
| 340 |
+
data = self.fetch_data(url)
|
| 341 |
+
|
| 342 |
|
| 343 |
logger.info(f"LMR récupérées pour la substance {substance_id}: {len(all_mrls)}")
|
| 344 |
return all_mrls
|
|
|
|
| 470 |
|
| 471 |
df = pd.DataFrame(data) # Crée le DataFrame
|
| 472 |
logger.info(f"LMR récupérées pour {substance.name}: {len(df)} entrées")
|
| 473 |
+
return df
|
| 474 |
|
| 475 |
def create_histogram(self, df: pd.DataFrame) -> go.Figure:
|
| 476 |
"""Crée un histogramme des valeurs LMR"""
|
|
|
|
| 615 |
|
| 616 |
substance_mrls_btn.click(
|
| 617 |
fn=lambda sid: (
|
| 618 |
+
self.get_substance_mrls(int(sid)),
|
| 619 |
+
self.create_histogram(self.get_substance_mrls(int(sid)))
|
| 620 |
),
|
| 621 |
inputs=[substance_select],
|
| 622 |
outputs=[substance_mrls, mrl_histogram]
|