Update app.py

app.py

@@ -0,0 +1,126 @@
+import streamlit
+import fitz           # PyMuPDF = read the contents of the pdf file
+import openai
+from fpdf import FPDF # library for generating pdf files
+import os
+import tempfile
+
+
+# function to extract pdf file
+def extract_text_from_pdf(pdf_file):
+
+  # save the uploaded file
+  temp_file = tempfile.NamedTemporaryFile(delete=False, suffix=".pdf")
+  temp_file.write(pdf_file.read())
+  temp_file.close() # close the file to ensure it is saved
+
+  # open the saved pdf file
+  doc = fitz.open(temp_file.name)
+  text = "" # extracted info saved here
+
+  # iterate 'doc' page wise iteration
+  for page_num in range(len(doc)):
+    page = doc.load_page(page_num)
+    text += page.get_text() # saving the text of each page into the 'text'
+
+  # delete the temp file after reading
+  os.remove(temp_file.name)
+
+
+
+# function to ensure if it is ends properly
+# it might happen that LLM text generation might end in between, at this point we require this function
+def ensure_full_stop(text):
+  text = text.strip()
+  if not text.endswith(('.', '!', '?')):
+    text += '.'
+  return text
+
+
+# function to summarize
+def summarize_text(api_key, text):
+  openai.api_key = api_key
+  response = openai.ChatCompletion.create(
+      model = "gpt-3.5-turbo",
+      messages = [
+          {"role": "system", "content": "You are a helpful assistant."},
+          {"role": "user", "content": f"Summarize the following text:\n\n{text}"}
+      ],
+      max_tokens = 2048,
+      temp = 0.5 # creativity, 0.5 is sweet spot
+  )
+
+  summary = response.choices[0].message['content'].strip()
+  return ensure_full_stop(summary)
+
+
+# understand the main gist of the pdf
+def predict_topic(api_key, text):
+    openai.api_key = api_key
+    response = openai.ChatCompletion.create(
+        model="gpt-3.5-turbo",  # Use "gpt-4" if you have access
+        messages=[{"role": "system", "content": "You are a helpful assistant."},
+                  {"role": "user", "content": f"What is the main topic of the following text?\n\n{text}"}],
+        max_tokens=500,
+        temperature=0.5
+    )
+    topic = response.choices[0].message['content'].strip()
+    return topic
+
+# Function to generate a PDF with summary and topic
+def create_pdf(summary, topic, original_file_name):
+    base_name = os.path.splitext(original_file_name)[0]  # Remove the .pdf extension
+    pdf_file_name = f"{base_name}_summary.pdf"  # Create the new filename
+
+    pdf = FPDF()
+    pdf.add_page()
+    pdf.set_font("Arial", size=12)
+
+    pdf.cell(200, 10, txt="Summary", ln=True, align='C')
+    pdf.multi_cell(0, 10, txt=summary)
+
+    pdf.cell(200, 10, txt="Predicted Main Topic", ln=True, align='C')
+    pdf.multi_cell(0, 10, txt=topic)
+
+    # Save the PDF to a file in memory
+    pdf_file_path = f"/tmp/{pdf_file_name}"
+    pdf.output(pdf_file_path)
+
+    return pdf_file_path
+
+st.title("Research paper summary")
+api_key = st.text_input("Enter your OpenAI API Key: ", type="password")
+
+# file uploading
+uploaded_file = st.file_uploader("Upload your research paper (PDF)", type=["pdf"])
+
+if uploaded_file is not None: 
+  text = extract_text_from_pdf(uploaded_file)
+
+  if len(text) > 1000: 
+    summary = summarize_text(api_key, text)
+    topic = predict_topic(api_key, text)
+
+    st.subheader("Summary")
+    st.write(summary)
+
+    st.subheader("Predicted Topic")
+    st.write(topic)
+
+    # Button to download results as a PDF
+    if st.button("Get the Summary PDF"):
+        pdf_path = create_pdf(summary, topic, uploaded_file.name)
+        st.download_button (
+            label="Download Summary PDF",
+            data=open(pdf_path, "rb").read(),
+            file_name=os.path.basename(pdf_path),
+            mime="application/pdf"
+        )
+  else:
+    st.warning("The document is too short for meaningful analysis.")
+  
+else:
+    st.info("Please upload a valid PDF file to proceed.")
+
+def get_text():
+  pass