File size: 2,648 Bytes
6bbd3ca
 
 
 
 
dea0d8a
4bf804b
6bbd3ca
 
 
d3f8141
 
6bbd3ca
 
 
d3f8141
6bbd3ca
f8ec4b3
6bbd3ca
f8ec4b3
6bbd3ca
 
 
 
4bf804b
 
 
 
 
 
 
6bbd3ca
 
 
 
 
 
 
 
 
4bf804b
6bbd3ca
 
 
 
d3f8141
6bbd3ca
 
d3f8141
 
 
 
f8ec4b3
 
 
 
 
 
 
 
 
 
 
 
2181fee
f8ec4b3
 
d3f8141
f8ec4b3
2181fee
d3f8141
 
 
 
6bbd3ca
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
from io import BytesIO
from PIL import Image
from fastapi import FastAPI, File, UploadFile, Form
from fastapi.responses import JSONResponse
from transformers import pipeline
from pytesseract import pytesseract
import base64

app = FastAPI()

# Use a pipeline as a high-level helper
nlp_qa = pipeline("document-question-answering", model="impira/layoutlm-document-qa")

description = """
## Image-based Document QA
This API extracts text from an uploaded image using OCR and performs document question answering using a LayoutLM-based model.

### Endpoints:
- **POST /uploadfile/:** Upload an image file to extract text and answer provided questions.
- **POST /pdfUpload/:** Provide a file to extract text and answer provided questions.
"""

app = FastAPI(docs_url="/", description=description)

def get_image_content(contents):
    # Convert binary content to image
    image = Image.open(BytesIO(contents))
    # Perform OCR to extract text from the image
    text_content = pytesseract.image_to_string(image)
    return text_content

@app.post("/uploadfile/", description=description)
async def perform_document_qa(
    file: UploadFile = File(...),
    questions: str = Form(...),
):
    try:
        # Read the uploaded file
        contents = await file.read()

        text_content = get_image_content(contents)

        # Split the questions string into a list
        question_list = [q.strip() for q in questions.split(',')]

        # Perform document question answering for each question using LayoutLM-based model
        answers_dict = {}
        for question in question_list:
            result = nlp_qa(
                text_content,
                question
            )
            answers_dict[question] = result['answer']

        return answers_dict
    except Exception as e:
        return JSONResponse(content=f"Error processing file: {str(e)}", status_code=500)

@app.post("/pdfUpload/", description=description)
async def load_file(
    file: UploadFile = File(...),
    questions: str = Form(...),
):
    try:
        # Read the uploaded file as bytes
        contents = await file.read()

        # Perform document question answering for each question using LayoutLM-based model
        answers_dict = {}
        for question in questions.split(','):
            result = nlp_qa(
                contents.decode('utf-8'),  # Assuming the content is text, adjust as needed
                question.strip()
            )
            answers_dict[question] = result['answer']

        return answers_dict
    except Exception as e:
        return JSONResponse(content=f"Error processing file: {str(e)}", status_code=500)