Spaces:
Sleeping
Sleeping
Create app.py
Browse files
app.py
ADDED
@@ -0,0 +1,70 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
import gradio as gr
|
2 |
+
from sentence_transformers import SentenceTransformer
|
3 |
+
import fitz # PyMuPDF
|
4 |
+
import numpy as np
|
5 |
+
from bokeh.plotting import figure, output_file, save
|
6 |
+
from bokeh.io import export_png
|
7 |
+
from bokeh.embed import file_html
|
8 |
+
from bokeh.resources import CDN
|
9 |
+
import tempfile
|
10 |
+
import os
|
11 |
+
|
12 |
+
# Load your model
|
13 |
+
model = SentenceTransformer('all-MiniLM-L6-v2')
|
14 |
+
|
15 |
+
def process_pdf(pdf_path):
|
16 |
+
# Open the PDF
|
17 |
+
doc = fitz.open(pdf_path)
|
18 |
+
texts = []
|
19 |
+
for page in doc:
|
20 |
+
texts.append(page.get_text())
|
21 |
+
return " ".join(texts)
|
22 |
+
|
23 |
+
def create_embeddings(text):
|
24 |
+
# Split the text into sentences/chunks and generate embeddings
|
25 |
+
# This is a placeholder for your actual text splitting and embedding code
|
26 |
+
sentences = text.split(".") # Simplistic split, consider using a better sentence splitter
|
27 |
+
embeddings = model.encode(sentences)
|
28 |
+
return embeddings, sentences
|
29 |
+
|
30 |
+
def generate_plot(query, pdf_file):
|
31 |
+
# Process the PDF and create embeddings
|
32 |
+
text = process_pdf(pdf_file)
|
33 |
+
embeddings, sentences = create_embeddings(text)
|
34 |
+
|
35 |
+
# Here, you'll integrate the UMAP and Bokeh visualization code you have,
|
36 |
+
# and then save the Bokeh plot to a file.
|
37 |
+
# For simplicity, let's assume it's saved to 'plot.html'
|
38 |
+
|
39 |
+
output_file("plot.html")
|
40 |
+
# Your Bokeh plot creation code here...
|
41 |
+
save(p) # Assuming 'p' is your Bokeh figure
|
42 |
+
|
43 |
+
# Alternatively, you can save as PNG
|
44 |
+
# export_png(p, filename="plot.png")
|
45 |
+
|
46 |
+
# Return the path to the saved file
|
47 |
+
return "plot.html" # or "plot.png"
|
48 |
+
|
49 |
+
def gradio_interface(pdf_file, query):
|
50 |
+
plot_path = generate_plot(query, pdf_file.name)
|
51 |
+
|
52 |
+
# If returning HTML file
|
53 |
+
with open(plot_path, "r") as f:
|
54 |
+
html_content = f.read()
|
55 |
+
return html_content
|
56 |
+
|
57 |
+
# If returning an image
|
58 |
+
# return plot_path
|
59 |
+
|
60 |
+
# Set up the Gradio app
|
61 |
+
iface = gr.Interface(
|
62 |
+
fn=gradio_interface,
|
63 |
+
inputs=[gr.inputs.File(label="Upload PDF"), gr.inputs.Textbox(label="Query")],
|
64 |
+
outputs=gr.outputs.HTML(label="Visualization"), # Use gr.outputs.Image for image output
|
65 |
+
title="PDF Content Visualizer",
|
66 |
+
description="Upload a PDF and enter a query to visualize the content."
|
67 |
+
)
|
68 |
+
|
69 |
+
# Run the app
|
70 |
+
iface.launch()
|