add description extraction

services/extractor.py

@@ -7,7 +7,7 @@ from langchain_core.output_parsers import PydanticOutputParser
 from langchain_openai.chat_models import ChatOpenAI
 from pydantic import BaseModel
 
-from utils.parsers import process_profile, process_passport, process_account
+from utils.parsers import process_profile, process_passport, process_account, process_description
 from validation.from_account import FromAccount
 from validation.from_passport import FromPassport
 from validation.from_profile import FromProfile
@@ -23,9 +23,10 @@ def extract_description(client_data: dict[str, Any]) -> FromDescription:
         "Passport text:\n{processed_text}"
     )
 
+
     result = __run_extraction_chain(
         raw_file_data=passport_data,
-        file_processor=lambda passport_data: passport_data,
+        file_processor=process_description,
         pydantic_model=FromDescription,
         prompt_template=prompt_template,
     )

utils/parsers/__init__.py

@@ -1,9 +1,11 @@
 from .passport_parser import process_passport
 from .profile_parser import process_profile
 from .account_parser import process_account
+from .description_parser import process_description
 
 __all__ = [
     "process_passport",
     "process_profile",
-    "process_account"
+    "process_account",
+    "process_description",
 ]

utils/parsers/description_parser.py

@@ -0,0 +1,15 @@
+import base64
+
+def process_description(desc_b64: str) -> str:
+    """
+            Décode une chaîne base64 et retourne le texte UTF-8 correspondant.
+
+            :param base64_text: Donnée encodée en base64.
+            :return: Chaîne de caractères décodée.
+            """
+    try:
+        decoded_bytes = base64.b64decode(desc_b64)
+        decoded_text = decoded_bytes.decode('utf-8')
+        return decoded_text
+    except Exception as e:
+        return f"[Erreur de décodage] {str(e)}"