app.py

import os
import base64
import io
import numpy as np
import cv2
from PIL import Image
from fastapi import FastAPI, Request
from fastapi.middleware.cors import CORSMiddleware
from fastapi.responses import JSONResponse
import easyocr

app = FastAPI()

app.add_middleware(
    CORSMiddleware,
    allow_origins=["*"],
    allow_credentials=True,
    allow_methods=["*"],
    allow_headers=["*"],
)

BASE_DIR = os.path.dirname(os.path.abspath(__file__))

print("Loading Case-Sensitive Custom Model V6 (Lazy Loaded as V2)...")
reader = easyocr.Reader(
    ['en'], 
    recog_network='cuims_model_v2',
    model_storage_directory=os.path.join(BASE_DIR, 'model'),
    user_network_directory=os.path.join(BASE_DIR, 'user_network')
)
print("Model Loaded!")

@app.post("/solve")
async def solve_captcha(request: Request):
    try:
        img = None
        content_type = request.headers.get('content-type', '')
        
        
        if 'application/json' in content_type:
            # Handles Chrome Extension Base64 payload
            data = await request.json()
            base64_data = data.get('image', '')
            if "," in base64_data:
                base64_data = base64_data.split(",")[1]
            img_bytes = base64.b64decode(base64_data)
            img = Image.open(io.BytesIO(img_bytes))
            
        elif 'multipart/form-data' in content_type:
            # Handles Flutter Raw File Upload payload
            form = await request.form()
            file = form.get('file')
            if file:
                contents = await file.read()
                img = Image.open(io.BytesIO(contents))
        else:
            return JSONResponse(status_code=400, content={"text": "", "error": "Invalid content type"})

        if not img:
            return JSONResponse(status_code=400, content={"text": "", "error": "No image provided"})

        
        img = img.convert("RGBA")
        white_bg = Image.new("RGBA", img.size, "WHITE")
        img_with_bg = Image.alpha_composite(white_bg, img)

        
        img_native = img_with_bg.resize((100, 32), resample=Image.Resampling.BICUBIC)
        img_cv_grey = np.array(img_native.convert("L"))
        
        
        _, binarized_img = cv2.threshold(img_cv_grey, 0, 255, cv2.THRESH_BINARY + cv2.THRESH_OTSU)
        
        
        results = reader.recognize(
            binarized_img,
            horizontal_list=[[0, 100, 0, 32]], 
            free_list=[],
            detail=0,
            decoder='beamsearch',
            beamWidth=10,
            allowlist='0123456789abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ'
        )
        
        text = "".join(results)
        cleaned = "".join(e for e in text if e.isalnum())[:4]
        
        print(f"AI Predicted: {cleaned}")
        
        
        return {"text": cleaned}
    
    except Exception as e:
        import traceback
        print(f"Detailed Error: {traceback.format_exc()}")
        return {"text": "", "error": str(e)}