Update app.py

app.py

@@ -1,28 +1,65 @@
-import gradio as gr
 from transformers import AutoModelForQuestionAnswering, AutoTokenizer, pipeline
+from gensim.parsing.preprocessing import STOPWORDS
+import wikipedia
+import gradio as gr
+import nltk
+from nltk.tokenize import word_tokenize
+import re
+nltk.download('punkt')
 
 model_name = "deepset/roberta-base-squad2"
 model = AutoModelForQuestionAnswering.from_pretrained(model_name)
 tokenizer = AutoTokenizer.from_pretrained(model_name)
 
-def pred(que,cont):
-    nlp = pipeline('question-answering', model=model_name, tokenizer=model_name)
-    QA_input = {
-        'question': que,
-        'context': cont
-    }
-    res = nlp(QA_input)
-    print(res)
-    values = list(res.values())[3]
+def bullete(text,wikipedia_language="en"):
+  question_words = STOPWORDS.union(set(['likes','play','.',',','like',"don't",'?','use','choose','important','better','?']))
+  lower_text = text.lower()
+  lower_text = word_tokenize(lower_text)
+  new_text = [i for i in lower_text if i not in question_words]
+  new_txt = "".join(new_text)
+  if wikipedia_language:
+    wikipedia.set_lang(wikipedia_language)
 
-    return values
+  et_page = wikipedia.page(new_txt.replace(" ", ""))
+  title = et_page.title
+  content = et_page.content
+  page_url = et_page.url
+  linked_pages = et_page.links
 
+  text1 = content
+  final_out = re.sub(r'\=.+\=', '', text1)
+  result = list(filter(lambda x: x != '', final_out.split('\n\n')))
 
-interface = gr.Interface(fn=pred,
-                         inputs=["text", "text"],
-                         outputs="text",
-                         title='Bullet Point')
+  answer = []
+  for i in range(4):
+      if len(result[i]) > 500:
+          summary_point=result[i].split(".")[0:3]
+          answer.append(summary_point)
+  l = []
+  for i in range(len(answer)):
+      l.append("".join(answer[i]))
+  gen_output = []
+  for i in range(len(l)):
+      gen_output.append(l[i] + ".")
 
-interface.launch(inline=False)
+  listToStr = ' '.join([str(elem) for elem in gen_output])
+  listToStr = listToStr.replace("\n", "")
+  print(listToStr)
+  nlp = pipeline('question-answering', model=model_name, tokenizer=model_name)
+  QA_input = {
+      'question': text,
+      'context': listToStr
+  }
+  print(QA_input)
+  res = nlp(QA_input)
+  values = list(res.values())[3]
 
+  return values
+
+
+interface = gr.Interface(fn=bullete,
+                         inputs="text",
+                         outputs="text",
+                         title='Bullet Point')
 
+interface.launch(inline=False)