Update app.py
Browse files
app.py
CHANGED
@@ -1,24 +1,16 @@
|
|
1 |
import fitz
|
2 |
import gradio as gr
|
3 |
-
import requests
|
4 |
-
import io
|
5 |
import re
|
6 |
-
import os
|
7 |
-
from PIL import Image
|
8 |
from transformers import pipeline
|
9 |
|
10 |
summarizer = pipeline("summarization", model="facebook/bart-large-cnn")
|
11 |
qa_model = pipeline("question-answering", model="deepset/bert-large-uncased-whole-word-masking-squad2")
|
12 |
|
13 |
-
os.environ["HUGGINGFACE_HUB_TOKEN"] = "ctp-hw"
|
14 |
-
my_key = os.environ["HUGGINGFACE_HUB_TOKEN"]
|
15 |
-
|
16 |
def extract_text_from_pdf(pdf_file):
|
17 |
with fitz.open(pdf_file) as pdf:
|
18 |
text = ""
|
19 |
for page in pdf:
|
20 |
text += page.get_text("text")
|
21 |
-
|
22 |
text = re.sub(r'\s+', ' ', text).strip()
|
23 |
return text
|
24 |
|
@@ -30,39 +22,23 @@ def summarize(text):
|
|
30 |
summary += summarizer(chunk, max_length=150, min_length=50, do_sample=False)[0]['summary_text'] + " "
|
31 |
else:
|
32 |
summary = summarizer(text, max_length=150, min_length=50, do_sample=False)[0]['summary_text']
|
33 |
-
|
34 |
return summary
|
35 |
|
36 |
def answer_question(text, question):
|
37 |
response = qa_model(question=question, context=text)
|
38 |
-
answer = response['answer']
|
39 |
return answer
|
40 |
|
41 |
-
def query(payload):
|
42 |
-
API_URL = "https://api-inference.huggingface.co/models/Shakker-Labs/FLUX.1-dev-LoRA-AntiBlur"
|
43 |
-
headers = {"Authorization": f"Bearer {my_key}"}
|
44 |
-
response = requests.post(API_URL, headers=headers, json=payload)
|
45 |
-
return response.content
|
46 |
-
|
47 |
def summarize_and_qa(pdf_file, question):
|
48 |
text = extract_text_from_pdf(pdf_file)
|
49 |
summary = summarize(text)
|
50 |
answer = answer_question(text, question)
|
51 |
-
|
52 |
-
if image_bytes:
|
53 |
-
try:
|
54 |
-
image = Image.open(io.BytesIO(image_bytes))
|
55 |
-
except Exception as e:
|
56 |
-
return summary, answer, None
|
57 |
-
else:
|
58 |
-
image = None
|
59 |
-
|
60 |
-
return summary, answer, image
|
61 |
|
62 |
gr.Interface(
|
63 |
fn=summarize_and_qa,
|
64 |
inputs=["file", "text"],
|
65 |
-
outputs=["textbox", "textbox"
|
66 |
title="Understand your PDF Better",
|
67 |
-
description="Upload a PDF to get a summary. You can ask any question
|
68 |
-
).launch(debug=True, share=True)
|
|
|
1 |
import fitz
|
2 |
import gradio as gr
|
|
|
|
|
3 |
import re
|
|
|
|
|
4 |
from transformers import pipeline
|
5 |
|
6 |
summarizer = pipeline("summarization", model="facebook/bart-large-cnn")
|
7 |
qa_model = pipeline("question-answering", model="deepset/bert-large-uncased-whole-word-masking-squad2")
|
8 |
|
|
|
|
|
|
|
9 |
def extract_text_from_pdf(pdf_file):
|
10 |
with fitz.open(pdf_file) as pdf:
|
11 |
text = ""
|
12 |
for page in pdf:
|
13 |
text += page.get_text("text")
|
|
|
14 |
text = re.sub(r'\s+', ' ', text).strip()
|
15 |
return text
|
16 |
|
|
|
22 |
summary += summarizer(chunk, max_length=150, min_length=50, do_sample=False)[0]['summary_text'] + " "
|
23 |
else:
|
24 |
summary = summarizer(text, max_length=150, min_length=50, do_sample=False)[0]['summary_text']
|
|
|
25 |
return summary
|
26 |
|
27 |
def answer_question(text, question):
|
28 |
response = qa_model(question=question, context=text)
|
29 |
+
answer = response['answer']
|
30 |
return answer
|
31 |
|
|
|
|
|
|
|
|
|
|
|
|
|
32 |
def summarize_and_qa(pdf_file, question):
|
33 |
text = extract_text_from_pdf(pdf_file)
|
34 |
summary = summarize(text)
|
35 |
answer = answer_question(text, question)
|
36 |
+
return summary, answer
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
37 |
|
38 |
gr.Interface(
|
39 |
fn=summarize_and_qa,
|
40 |
inputs=["file", "text"],
|
41 |
+
outputs=["textbox", "textbox"],
|
42 |
title="Understand your PDF Better",
|
43 |
+
description="Upload a PDF to get a summary. You can ask any question regarding the content of the PDF."
|
44 |
+
).launch(debug=True, share=True)
|