Update summariser.py

Author: heyronjmilton

Backend/summariser.py

@@ -5,46 +5,42 @@
 import os
 import asyncio
 
+progress = None # just for tracking progress
 
 def summary(text):
-
     # Load the summarization pipeline
     summarizer = pipeline("summarization")
-
     # Split the text into smaller chunks
     max_tokens_per_chunk = 1024  # Initial value
     max_words_in_summary = 2000000
-
     # Calculate the maximum number of chunks needed
     max_num_chunks = (max_words_in_summary // max_tokens_per_chunk) + 1
-
     # Split the text into chunks
     chunks = [text[i:i + max_tokens_per_chunk] for i in range(0, len(text), max_tokens_per_chunk)]
     # for the exceptions
     exceptions = "NULL"
-
+    global progress
+    progress = 0 
     # Generate summaries for each chunk
     summaries = []
     len_chunk=len(chunks)
     print("Note have been divided into chunks:"+str(len_chunk))
     for i, chunk in enumerate(chunks):
         # Reduce the chunk size dynamically if it exceeds the maximum sequence length
         while len(chunk) > max_tokens_per_chunk:
-            max_tokens_per_chunk -= 50
-        
+            max_tokens_per_chunk -= 50      
         try:
             summary = summarizer(chunk, max_length=200, min_length=100, do_sample=False)
             summaries.append(summary[0]['summary_text']+"\n\n")
             print(summary[0]['summary_text'])
             print("\n \n STATUS:"+str(i+1)+"/"+str(len_chunk))
-            print("\n \n COMPLETED:"+str((i+1)/len_chunk*100)+"%")
+            progress = (i+1)/len_chunk*100
+            print("\n \n COMPLETED:"+str(progress)+"%")
         except Exception as e:
             print(f"An error occurred while summarizing chunk {i}: {str(e)}")
             exceptions = "\n".join(f"An error occurred while summarizing chunk {i}: {str(e)}")
-
     # Combine the summaries into a single summary
     combined_summary = " ".join(summaries)
-
     # Print and return the combined summary
     print("Combined Summary:")
     print(combined_summary)
@@ -55,7 +51,6 @@ def summary(text):
 
 
 async def gen_summary(file):
-
     try:
         with open("dat.txt", "wb") as buffer:      # saving file
             shutil.copyfileobj(file.file, buffer)
@@ -75,8 +70,16 @@ async def gen_summary(file):
 @router_summariser.post("/get-summary")
 async def get_summary(file: UploadFile = File(...)):
     data = await gen_summary(file)
-
     return data
 
-
-
+@router_summariser.get("/summary-gen-progress") # route to track progress of summarization
+def get_summary_progress():
+    global progress
+    if progress is None :
+        return {"status" : "No summarisation process in progress" }
+    elif progress == 100 :
+        return {"status" : "Completed" , "value" : progress}
+    elif progress in range(0,101) :
+        return {"status" : progress}
+    else :
+        return {"invalid data detected"}