rollback passport_parser.py

utils/parsers/passport_parser.py

@@ -1,23 +1,42 @@
-import os
+import base64
-from mistralai import Mistral
+import io
-import config
+from tempfile import NamedTemporaryFile
-
+from PIL import Image, ImageEnhance
-api_key = config.MISTRAL_API_KEY
+import pytesseract
-client = Mistral(api_key=api_key)
+from passporteye import read_mrz
-
+import json
 
 def process_passport(passport_b64: str) -> str:
-    mistral_image_url = f"data:image/jpeg;base64,{passport_b64}"
+    """
+    Traite le passport :
+    - Décodage de l'image en base64.
+    - Application de l'OCR pour extraire le texte.
 
-    ocr_response = client.ocr.process(
+    :param passport_b64: Chaîne base64 représentant l'image du passport.
-        model="mistral-ocr-latest",
+    :return: Texte extrait de l'image.
-        document={
+    """
-            "type": "image_url",
+    image_bytes = base64.b64decode(passport_b64)
-            "image_url": mistral_image_url
+    with NamedTemporaryFile(mode="wb") as tmp_img:
+        tmp_img.write(image_bytes)
+        with open(tmp_img.name, "rb") as read_img:
+            mrz_obj = read_mrz(read_img)
+
+    image = Image.open(io.BytesIO(image_bytes))
+    enhancer = ImageEnhance.Contrast(image)
+    image = enhancer.enhance(2.0)  # 2.0 = double le contraste (1.0 = inchangé)
+    tesseract_text = pytesseract.image_to_string(image, lang='eng')
+    out_dict = {}
+    if not mrz_obj is None:
+        number_raw = str(mrz_obj.number)
+        # It's not called a 'Hack'athon for nothing...
+        number = number_raw.replace("B", "8")
+        out_dict = {
+            "country": mrz_obj.country,
+            "names": mrz_obj.names,
+            "number": number,
+            "surname": mrz_obj.surname,
+            "mrz": mrz_obj.aux["text"],
         }
-    )
+    out_dict["raw"] = tesseract_text
-
+    out = json.dumps(out_dict)
-    # Extraire le markdown de toutes les pages
+    return out
-    markdown_text = "\n\n".join(page.markdown for page in ocr_response.pages)
-
-    return markdown_text  # Tu peux aussi retourner juste le chemin si tu préfèresOPENAI_API_KEY