Spaces:
Sleeping
Sleeping
Update app.py
Browse files
app.py
CHANGED
@@ -1,25 +1,26 @@
|
|
1 |
-
import
|
2 |
import torch
|
3 |
from transformers import BertTokenizer, BertModel
|
4 |
-
from fastapi import FastAPI, HTTPException
|
5 |
-
from pydantic import BaseModel
|
6 |
import pdfplumber
|
7 |
|
8 |
-
|
|
|
|
|
|
|
9 |
|
10 |
-
|
11 |
-
|
12 |
-
|
13 |
-
|
14 |
-
|
15 |
-
|
16 |
-
|
17 |
-
tokenizer = BertTokenizer.from_pretrained(model_name)
|
18 |
-
model = BertModel.from_pretrained(model_name)
|
19 |
|
|
|
|
|
20 |
# Preprocess the input text
|
21 |
inputs = tokenizer.encode_plus(
|
22 |
-
|
23 |
add_special_tokens=True,
|
24 |
max_length=512,
|
25 |
return_attention_mask=True,
|
@@ -32,38 +33,22 @@ async def classify_text(request: TextClassificationRequest):
|
|
32 |
# Extract the features
|
33 |
features = outputs.last_hidden_state[:, 0, :]
|
34 |
|
35 |
-
|
36 |
-
return {"features": features.tolist()}
|
37 |
|
38 |
-
#
|
39 |
-
|
40 |
-
|
41 |
-
|
42 |
-
|
43 |
-
|
44 |
-
return text
|
45 |
|
46 |
-
|
47 |
-
def classify_pdf(pdf_file):
|
48 |
# Extract text from the uploaded PDF
|
49 |
extracted_text = extract_text_from_pdf(pdf_file)
|
50 |
-
|
51 |
-
|
52 |
-
request = TextClassificationRequest(text=extracted_text)
|
53 |
-
|
54 |
-
# Simulate calling the FastAPI endpoint
|
55 |
-
output = classify_text(request)
|
56 |
-
|
57 |
-
return output
|
58 |
-
|
59 |
-
# Create a Gradio interface
|
60 |
-
interface = gr.Interface(
|
61 |
-
fn=classify_pdf,
|
62 |
-
inputs="file", # Expecting PDF file input
|
63 |
-
outputs="json", # Outputs a JSON dictionary
|
64 |
-
title="PDF Text Classification",
|
65 |
-
description="Upload a PDF file to classify its text using BERT"
|
66 |
-
)
|
67 |
|
68 |
-
#
|
69 |
-
|
|
|
|
|
|
1 |
+
import streamlit as st
|
2 |
import torch
|
3 |
from transformers import BertTokenizer, BertModel
|
|
|
|
|
4 |
import pdfplumber
|
5 |
|
6 |
+
# Load the pre-trained BERT model and tokenizer outside the function for efficiency
|
7 |
+
model_name = "bert-base-uncased"
|
8 |
+
tokenizer = BertTokenizer.from_pretrained(model_name)
|
9 |
+
model = BertModel.from_pretrained(model_name)
|
10 |
|
11 |
+
# Define a function to extract text from a PDF
|
12 |
+
def extract_text_from_pdf(pdf_file):
|
13 |
+
with pdfplumber.open(pdf_file) as pdf:
|
14 |
+
text = ""
|
15 |
+
for page in pdf.pages:
|
16 |
+
text += page.extract_text()
|
17 |
+
return text
|
|
|
|
|
18 |
|
19 |
+
# Define a function to classify the extracted text
|
20 |
+
def classify_text(text):
|
21 |
# Preprocess the input text
|
22 |
inputs = tokenizer.encode_plus(
|
23 |
+
text,
|
24 |
add_special_tokens=True,
|
25 |
max_length=512,
|
26 |
return_attention_mask=True,
|
|
|
33 |
# Extract the features
|
34 |
features = outputs.last_hidden_state[:, 0, :]
|
35 |
|
36 |
+
return features.tolist()
|
|
|
37 |
|
38 |
+
# Streamlit app setup
|
39 |
+
st.title("PDF Text Classification")
|
40 |
+
st.write("Upload a PDF file to classify its text using BERT")
|
41 |
+
|
42 |
+
# File uploader for PDFs
|
43 |
+
pdf_file = st.file_uploader("Choose a PDF file", type="pdf")
|
|
|
44 |
|
45 |
+
if pdf_file is not None:
|
|
|
46 |
# Extract text from the uploaded PDF
|
47 |
extracted_text = extract_text_from_pdf(pdf_file)
|
48 |
+
st.write("Extracted Text:")
|
49 |
+
st.write(extracted_text)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
50 |
|
51 |
+
# Classify the extracted text
|
52 |
+
if st.button("Classify"):
|
53 |
+
features = classify_text(extracted_text)
|
54 |
+
st.json({"features": features}) # Display the features in JSON format
|