senel commited on
Commit
07cf440
·
verified ·
1 Parent(s): 09feeb0

Update api.py

Browse files
Files changed (1) hide show
  1. api.py +22 -17
api.py CHANGED
@@ -1,4 +1,3 @@
1
- import cv2
2
  import os
3
  import numpy as np
4
  from io import BytesIO
@@ -7,6 +6,7 @@ from fastapi import FastAPI, File, UploadFile, HTTPException, Request
7
  import fitz
8
  import preproccess
9
  import detect_with_paddle
 
10
  from fastapi.templating import Jinja2Templates
11
  from fastapi.responses import HTMLResponse
12
 
@@ -19,6 +19,7 @@ templates = Jinja2Templates(directory="templates")
19
  async def read_item(request: Request):
20
  # index.html dosyasını render et
21
  return templates.TemplateResponse("home.html", {"request": request})
 
22
  @app.post("/uploadfile/")
23
  async def create_upload_file(file: UploadFile = File(...)):
24
  try:
@@ -30,21 +31,25 @@ async def create_upload_file(file: UploadFile = File(...)):
30
  raise HTTPException(status_code=500, detail=str(e))
31
 
32
  def perform_ocr(file_content, file_name):
33
- # Dosyayı bellekte geçici bir dosya olarak kullan
34
- with BytesIO(file_content) as buffer:
35
- if file_name.lower().endswith('.pdf'):
36
- # PDF dosyasını işle
37
- results = apply_ocr_on_pdf(buffer)
38
- elif file_name.lower().endswith(('.jpg', '.jpeg', '.png')):
39
- # Resmi doğrudan işle
40
- results = apply_ocr_on_image(buffer)
41
- else:
42
- raise HTTPException(status_code=400, detail="Unsupported file format")
 
43
 
44
  return results
45
 
46
- def apply_ocr_on_pdf(buffer):
47
- pdf_document = fitz.open("pdf", buffer.read())
 
 
 
48
  images = []
49
  for page_number in range(pdf_document.page_count):
50
  page = pdf_document[page_number]
@@ -66,10 +71,10 @@ def apply_ocr_on_pdf(buffer):
66
 
67
  return results
68
 
69
- def apply_ocr_on_image(buffer):
70
- image = cv2.imdecode(np.frombuffer(buffer.read(), np.uint8), cv2.IMREAD_COLOR)
 
 
71
  image = preproccess.preprocess_image(image)
72
  result = detect_with_paddle.detect_paddle_OCR(image)
73
  return result
74
-
75
- # Rotate the image around its center
 
 
1
  import os
2
  import numpy as np
3
  from io import BytesIO
 
6
  import fitz
7
  import preproccess
8
  import detect_with_paddle
9
+ import base64
10
  from fastapi.templating import Jinja2Templates
11
  from fastapi.responses import HTMLResponse
12
 
 
19
  async def read_item(request: Request):
20
  # index.html dosyasını render et
21
  return templates.TemplateResponse("home.html", {"request": request})
22
+
23
  @app.post("/uploadfile/")
24
  async def create_upload_file(file: UploadFile = File(...)):
25
  try:
 
31
  raise HTTPException(status_code=500, detail=str(e))
32
 
33
  def perform_ocr(file_content, file_name):
34
+ # Dosyayı base64'e kodla
35
+ base64_encoded = base64.b64encode(file_content).decode('utf-8')
36
+
37
+ if file_name.lower().endswith('.pdf'):
38
+ # PDF dosyasını işle
39
+ results = apply_ocr_on_pdf(base64_encoded)
40
+ elif file_name.lower().endswith(('.jpg', '.jpeg', '.png')):
41
+ # Resmi doğrudan işle
42
+ results = apply_ocr_on_image(base64_encoded)
43
+ else:
44
+ raise HTTPException(status_code=400, detail="Unsupported file format")
45
 
46
  return results
47
 
48
+ def apply_ocr_on_pdf(base64_encoded):
49
+ # Base64 kodlu veriyi geri çöz
50
+ pdf_bytes = base64.b64decode(base64_encoded)
51
+ pdf_document = fitz.open("pdf", pdf_bytes)
52
+
53
  images = []
54
  for page_number in range(pdf_document.page_count):
55
  page = pdf_document[page_number]
 
71
 
72
  return results
73
 
74
+ def apply_ocr_on_image(base64_encoded):
75
+ # Base64 kodlu veriyi geri çöz ve resmi işle
76
+ image_bytes = base64.b64decode(base64_encoded)
77
+ image = cv2.imdecode(np.frombuffer(image_bytes, np.uint8), cv2.IMREAD_COLOR)
78
  image = preproccess.preprocess_image(image)
79
  result = detect_with_paddle.detect_paddle_OCR(image)
80
  return result