Spaces:
Sleeping
Sleeping
Can Günen
commited on
Commit
·
dee6fb6
1
Parent(s):
6a757a3
fixed path redirection
Browse files
app.py
CHANGED
|
@@ -1,34 +1,26 @@
|
|
| 1 |
-
# -*- coding: utf-8 -*-
|
| 2 |
-
"""
|
| 3 |
-
Created on Mon May 8 00:32:30 2023
|
| 4 |
-
|
| 5 |
-
@author: ahmet
|
| 6 |
-
"""
|
| 7 |
-
import pdfplumber
|
| 8 |
-
import gradio as gr
|
| 9 |
from transformers import AutoTokenizer, AutoModelForSeq2SeqLM
|
| 10 |
from pathlib import Path
|
|
|
|
|
|
|
|
|
|
|
|
|
| 11 |
|
| 12 |
def respond(pdf_file, upper_page=0):
|
| 13 |
pdf_file = Path(pdf_file.name)
|
| 14 |
-
|
| 15 |
-
all_text = ''
|
| 16 |
with pdfplumber.open(pdf_file) as pdf:
|
| 17 |
total_pages = len(pdf.pages)
|
| 18 |
for idx, pdf_page in enumerate(pdf.pages):
|
| 19 |
-
|
| 20 |
-
|
| 21 |
-
|
| 22 |
-
|
| 23 |
-
else:
|
| 24 |
-
break
|
| 25 |
-
|
| 26 |
tokenizer=AutoTokenizer.from_pretrained('Einmalumdiewelt/T5-Base_GNAD')
|
| 27 |
model=AutoModelForSeq2SeqLM.from_pretrained('Einmalumdiewelt/T5-Base_GNAD', return_dict=True)
|
| 28 |
-
inputs=tokenizer.encode("
|
| 29 |
output = model.generate(inputs, min_length=70, max_length=80)
|
| 30 |
summary=tokenizer.decode(output[0])
|
| 31 |
-
return summary
|
| 32 |
|
| 33 |
|
| 34 |
with gr.Blocks() as demo:
|
|
@@ -37,11 +29,14 @@ with gr.Blocks() as demo:
|
|
| 37 |
with gr.Row():
|
| 38 |
with gr.Column():
|
| 39 |
file_input = gr.File(label="PDF File", type="file")
|
| 40 |
-
page_input = gr.
|
| 41 |
-
|
| 42 |
-
|
|
|
|
|
|
|
|
|
|
| 43 |
|
| 44 |
-
summarize.click(fn=respond, inputs=[file_input, page_input], outputs=text_output)
|
| 45 |
-
|
| 46 |
|
|
|
|
|
|
|
| 47 |
demo.launch(debug=True)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
from transformers import AutoTokenizer, AutoModelForSeq2SeqLM
|
| 2 |
from pathlib import Path
|
| 3 |
+
import pdfplumber
|
| 4 |
+
import gradio as gr
|
| 5 |
+
|
| 6 |
+
|
| 7 |
|
| 8 |
def respond(pdf_file, upper_page=0):
|
| 9 |
pdf_file = Path(pdf_file.name)
|
| 10 |
+
all_text = ""
|
|
|
|
| 11 |
with pdfplumber.open(pdf_file) as pdf:
|
| 12 |
total_pages = len(pdf.pages)
|
| 13 |
for idx, pdf_page in enumerate(pdf.pages):
|
| 14 |
+
single_page_text = pdf_page.extract_text()
|
| 15 |
+
all_text = all_text + "\n" + single_page_text
|
| 16 |
+
#print(idx / total_pages)
|
| 17 |
+
|
|
|
|
|
|
|
|
|
|
| 18 |
tokenizer=AutoTokenizer.from_pretrained('Einmalumdiewelt/T5-Base_GNAD')
|
| 19 |
model=AutoModelForSeq2SeqLM.from_pretrained('Einmalumdiewelt/T5-Base_GNAD', return_dict=True)
|
| 20 |
+
inputs=tokenizer.encode("summarize: " +all_text, return_tensors='pt', max_length=512, truncation=True)
|
| 21 |
output = model.generate(inputs, min_length=70, max_length=80)
|
| 22 |
summary=tokenizer.decode(output[0])
|
| 23 |
+
return summary, all_text
|
| 24 |
|
| 25 |
|
| 26 |
with gr.Blocks() as demo:
|
|
|
|
| 29 |
with gr.Row():
|
| 30 |
with gr.Column():
|
| 31 |
file_input = gr.File(label="PDF File", type="file")
|
| 32 |
+
page_input = gr.Textbox(label="Page Limit")
|
| 33 |
+
summarize_button = gr.Button(label="Summarize")
|
| 34 |
+
with gr.Column():
|
| 35 |
+
summary_output = gr.Textbox(label="Summarized Text")
|
| 36 |
+
with gr.Column():
|
| 37 |
+
text_output =gr.Textbox(label="Extracted Text")
|
| 38 |
|
|
|
|
|
|
|
| 39 |
|
| 40 |
+
summarize_button.click(respond, inputs=[file_input, page_input], outputs=[summary_output, text_output])
|
| 41 |
+
|
| 42 |
demo.launch(debug=True)
|