AdityaAjithKumar commited on
Commit
ae379dd
·
1 Parent(s): 22d960b

feat: add PDF support — convert first page to image before OCR

Browse files
Files changed (3) hide show
  1. app.py +18 -1
  2. requirements.txt +0 -0
  3. utils/pdf.py +16 -0
app.py CHANGED
@@ -4,13 +4,30 @@ from fastapi import FastAPI, UploadFile, File, HTTPException
4
  from ocr.gemini_service import parse_pan_with_fallback
5
  from ocr.gemini_face_extractor import detect_face
6
  from utils.crop import crop_face
 
7
 
8
  app = FastAPI()
9
 
 
 
10
 
11
  @app.post("/kyc/pan")
12
  async def process_pan(file: UploadFile = File(...)):
13
- image_bytes = await file.read()
 
 
 
 
 
 
 
 
 
 
 
 
 
 
14
 
15
  try:
16
  pan_result, face_result = await asyncio.gather(
 
4
  from ocr.gemini_service import parse_pan_with_fallback
5
  from ocr.gemini_face_extractor import detect_face
6
  from utils.crop import crop_face
7
+ from utils.pdf import pdf_to_image_bytes
8
 
9
  app = FastAPI()
10
 
11
+ SUPPORTED_TYPES = {"image/jpeg", "image/png", "image/webp", "application/pdf"}
12
+
13
 
14
  @app.post("/kyc/pan")
15
  async def process_pan(file: UploadFile = File(...)):
16
+ if file.content_type not in SUPPORTED_TYPES:
17
+ raise HTTPException(
18
+ status_code=415,
19
+ detail=f"Unsupported file type '{file.content_type}'. Upload a JPEG, PNG, WebP, or PDF.",
20
+ )
21
+
22
+ raw_bytes = await file.read()
23
+
24
+ if file.content_type == "application/pdf":
25
+ try:
26
+ image_bytes = pdf_to_image_bytes(raw_bytes)
27
+ except ValueError as e:
28
+ raise HTTPException(status_code=422, detail=str(e))
29
+ else:
30
+ image_bytes = raw_bytes
31
 
32
  try:
33
  pan_result, face_result = await asyncio.gather(
requirements.txt CHANGED
Binary files a/requirements.txt and b/requirements.txt differ
 
utils/pdf.py ADDED
@@ -0,0 +1,16 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ import io
2
+ import fitz # PyMuPDF
3
+
4
+
5
+ def pdf_to_image_bytes(pdf_bytes: bytes, dpi: int = 200) -> bytes:
6
+ """Render the first page of a PDF to JPEG bytes."""
7
+ doc = fitz.open(stream=pdf_bytes, filetype="pdf")
8
+ if doc.page_count == 0:
9
+ raise ValueError("The uploaded PDF has no pages.")
10
+
11
+ page = doc.load_page(0)
12
+ mat = fitz.Matrix(dpi / 72, dpi / 72)
13
+ pix = page.get_pixmap(matrix=mat, colorspace=fitz.csRGB)
14
+
15
+ buffer = io.BytesIO(pix.tobytes("jpeg"))
16
+ return buffer.getvalue()