Rajut commited on
Commit
3c40c63
1 Parent(s): ee2694e

Update app.py

Browse files
Files changed (1) hide show
  1. app.py +12 -1
app.py CHANGED
@@ -2,6 +2,7 @@ import streamlit as st
2
  import fitz
3
  from transformers import pipeline, MBart50TokenizerFast, MBartForConditionalGeneration
4
  from multiprocessing import Pool, cpu_count
 
5
 
6
  # Load summarization pipeline
7
  summarizer = pipeline("summarization", model="Falconsai/text_summarization")
@@ -62,10 +63,17 @@ def translate_summary(summary, lang):
62
 
63
  return " ".join(translated_chunks)
64
 
 
 
65
  # Function to read PDF and summarize and translate chunk by chunk
66
  def summarize_and_translate_pdf(uploaded_file, lang):
 
 
 
 
 
67
  try:
68
- doc = fitz.open(stream=uploaded_file.read(), filetype="pdf")
69
  except FileNotFoundError:
70
  st.error("File not found. Please make sure the file path is correct.")
71
  return []
@@ -82,6 +90,9 @@ def summarize_and_translate_pdf(uploaded_file, lang):
82
  with Pool(cpu_count()) as pool:
83
  translated_chunks = pool.starmap(summarize_and_translate_chunk, [(chunk, lang) for chunk in chunks])
84
 
 
 
 
85
  return translated_chunks
86
 
87
 
 
2
  import fitz
3
  from transformers import pipeline, MBart50TokenizerFast, MBartForConditionalGeneration
4
  from multiprocessing import Pool, cpu_count
5
+ import tempfile
6
 
7
  # Load summarization pipeline
8
  summarizer = pipeline("summarization", model="Falconsai/text_summarization")
 
63
 
64
  return " ".join(translated_chunks)
65
 
66
+
67
+
68
  # Function to read PDF and summarize and translate chunk by chunk
69
  def summarize_and_translate_pdf(uploaded_file, lang):
70
+ # Save uploaded PDF to a temporary file
71
+ with tempfile.NamedTemporaryFile(delete=False) as temp_file:
72
+ temp_file.write(uploaded_file.read())
73
+ temp_file_path = temp_file.name
74
+
75
  try:
76
+ doc = fitz.open(temp_file_path)
77
  except FileNotFoundError:
78
  st.error("File not found. Please make sure the file path is correct.")
79
  return []
 
90
  with Pool(cpu_count()) as pool:
91
  translated_chunks = pool.starmap(summarize_and_translate_chunk, [(chunk, lang) for chunk in chunks])
92
 
93
+ # Delete temporary file
94
+ os.unlink(temp_file_path)
95
+
96
  return translated_chunks
97
 
98