This commit is contained in:
NoeBerdoz
2025-04-13 10:03:32 +02:00
parent 692890e6a3
commit 38c2a1b767

View File

@ -1,42 +1,23 @@
import base64 import os
import io from mistralai import Mistral
from tempfile import NamedTemporaryFile import config
from PIL import Image, ImageEnhance
import pytesseract api_key = config.MISTRAL_API_KEY
from passporteye import read_mrz client = Mistral(api_key=api_key)
import json
def process_passport(passport_b64: str) -> str: 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.
:param passport_b64: Chaîne base64 représentant l'image du passport. ocr_response = client.ocr.process(
:return: Texte extrait de l'image. model="mistral-ocr-latest",
""" document={
image_bytes = base64.b64decode(passport_b64) "type": "image_url",
with NamedTemporaryFile(mode="wb") as tmp_img: "image_url": mistral_image_url
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)
return out # Extraire le markdown de toutes les pages
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 préfères