Pranavpai0309/TextExtraction_Payment_Screenshot

1from fastapi import FastAPI, File, UploadFile , HTTPException2from pydantic import BaseModel3import pytesseract4from PIL import Image5import io6import re7 8app = FastAPI()9 10class ExtractionResult(BaseModel):11    upi_handle: str | None = None12    upi_transaction_id: str | None = None13    google_transaction_id: str | None = None14 15image_types = ["image/jpeg", "image/png", "image/jpg"]16 17pytesseract.pytesseract.tesseract_cmd = "/usr/bin/tesseract"18 19def parse_transaction_ids(input_text: str):20    norm = input_text.replace('\r', '\n')21    lower = norm.lower()22 23    upi_txn = None24    m = re.search(r'upi transaction id[:\s]*([0-9]{5,20})', lower, flags=re.IGNORECASE)25    26    if m:27        upi_txn = m.group(1)28        29    else:30        m2 = re.search(r'upi[^ \n]{0,40}([0-9]{6,20})', lower, flags=re.IGNORECASE)31        32        if m2:33            upi_txn = m2.group(1)34 35    google_txn = None36    37    mg = re.search(r'google transaction id[:\s]*([^\s,;]+)', norm, flags=re.IGNORECASE)38    39    if mg:40        google_txn = mg.group(1).strip()41 42    upi_handle = None43    handles = re.findall(r'([^\s,;]+@[^\s,;]+)', norm)44    45    if handles:46        best = None47        for h in handles:48            pos = norm.find(h)49            window = norm[max(0, pos-40): pos+len(h)+40].lower()50            51            if 'upi' in window or 'pay' in window or 'gpay' in window or 'google' in window or 'paytm' in window:52                best = h53                break54            55        upi_handle = best or handles[0]56 57    return {58        "upi_handle": upi_handle,59        "upi_transaction_id": upi_txn,60        "google_transaction_id": google_txn61    }62    63@app.get("/")64def root():65    return {"Message" : "FastAPI is running smoothly ..."}66 67 68@app.post("/text-extraction", tags=["Text Extraction "], response_model=ExtractionResult)69async def upload_and_extract(file: UploadFile = File(...)):70    """71 72        Args:73            Input image file (jpg,jpeg and png)74 75        Raises:76            HTTPException: 400 - if not of image format ..77 78        Returns:79        80           extracted text in the format :81           {82                "upi_handle": upi_handle,83                "upi_transaction_id": upi_txn,84                "google_transaction_id": google_txn85            }86    """87    88    if file.content_type not in image_types:89        raise HTTPException(status_code=400,detail="Only images of format jpg,jpeg,png are allowed. Try uploading a valid image ...")90    91    image_contents = await file.read()92    93    image = Image.open(io.BytesIO(image_contents)).convert("RGB")94 95    extracted_text = pytesseract.image_to_string(image)96 97    parsed = parse_transaction_ids(extracted_text)98 99    return ExtractionResult(100        upi_handle=parsed["upi_handle"],101        upi_transaction_id=parsed["upi_transaction_id"],102        google_transaction_id=parsed["google_transaction_id"]103    )