Update app.py

app.py

@@ -8,6 +8,7 @@ import gradio as gr
 from gradio.mix import Parallel
 from transformers import pipeline
 import numpy as np
+import math
 
 # Defining a function to read in the text file
 def read_in_text(url):
@@ -15,6 +16,21 @@ def read_in_text(url):
     article = file.read()
     return article
     
+def clean_text(url):
+  text = url
+  text = text.encode("ascii", errors="ignore").decode(
+          "ascii"
+    )  # remove non-ascii, Chinese characters
+    
+  text = re.sub(r"\n", " ", text)
+  text = re.sub(r"\n\n", " ", text)
+  text = re.sub(r"\t", " ", text)
+  text = text.strip(" ")
+  text = re.sub(
+        " +", " ", text
+    ).strip()  # get rid of multiple spaces and replace with a single
+  return text
+    
 #initailizing the model pipeline   
 from transformers import BartTokenizer, BartForConditionalGeneration
 
@@ -24,19 +40,57 @@ tokenizer = BartTokenizer.from_pretrained("sshleifer/distilbart-cnn-12-6")
 #Defining a function to get the summary of the article
 def final_summary(file):
   #reading in the text and tokenizing it into sentence
-  text = read_in_text(file.name)
+  text = clean_text(file)
   chunks = sent_tokenize(text)
   output = []
+  sentences_remaining = len(chunks)
 
   #looping through the sentences in a batch of 10 and summarizing them
-  for i in range(0,len(chunks), 10):
-    sentence = ' '.join(chunks[i:i+10])
-    inputs = tokenizer(sentence, max_length=1024, return_tensors="pt")
-    summary_ids = model.generate(inputs["input_ids"])
-    summary = tokenizer.batch_decode(summary_ids, skip_special_tokens=True, clean_up_tokenization_spaces=False)[0]  
-    output.append(summary)
+  i=0
+  while (sentences_remaining > 0):
+
+    chunks_remaining = math.ceil(sentences_remaining / 10.0)
+    next_chunk_size = math.ceil(sentences_remaining / chunks_remaining)
+    sentence = ' '.join(chunks[i:i+(next_chunk_size)])
+
+    i += next_chunk_size
+    sentences_remaining -= next_chunk_size
+        
+    inputs = tokenizer(sentence, return_tensors="pt", padding='longest')
+    if (len(inputs['input_ids'][0])) < 150:
+      output.append(sentence)
+
+    elif (len(inputs['input_ids'][0])) > 1024:
+      sent = sent_tokenize(sentence)
+      length_sent = len(sent)
+
+      j=0
+      sent_remaining = math.ceil(length_sent / 2)
+      while length_sent >0:
+        
+        #next_sent_size = math.ceil(length_sent / sent_remaining)
+        halved_sentence = ' '.join(sent[j:j+(sent_remaining)])
+        inputs = tokenizer(halved_sentence, return_tensors="pt")
+        summary_ids = model.generate(inputs["input_ids"])
+        j += sent_remaining
+        length_sent -= sent_remaining
+
+        if (len(summary_ids[0])) < (len(inputs['input_ids'][0])):
+          summary = tokenizer.batch_decode(summary_ids, skip_special_tokens=True, clean_up_tokenization_spaces=False)[0]  
+          output.append(summary)
+        else:
+          continue
+
+    else:
+      summary_ids = model.generate(inputs["input_ids"])
+
+      if (len(summary_ids[0])) < (len(inputs['input_ids'][0])):
+        summary = tokenizer.batch_decode(summary_ids, skip_special_tokens=True, clean_up_tokenization_spaces=False)[0]  
+        output.append(summary)
+      else:
+        continue
   
-  #joining all the summary output together
+#joining all the summary output together
   summary = ' '.join(output)
   lines1 = sent_tokenize(summary)
   for i in range(len(lines1)):
@@ -46,7 +100,7 @@ def final_summary(file):
   return summ_bullet1  
   
   #creating an interface for the headline generator using gradio
-demo = gr.Interface(final_summary, inputs=[gr.inputs.File(label="Drop your .txt file here", optional=False)],
+demo = gr.Interface(final_summary, inputs=[gr.inputs.Textbox(label="Drop your article here", optional=False)],
                                           title = "ARTICLE SUMMARIZER",
                                           outputs=[gr.outputs.Textbox(label="Summary")],
                                           theme= "darkhuggingface")