Update app.py

app.py

@@ -1,48 +1,61 @@
 import streamlit as st
-import transformers
-from transformers import pipeline, set_seed
+from bs4 import BeautifulSoup
+from nltk.tokenize import word_tokenize
+from nltk.corpus import wordnet
 import random
-import time
-import textblob
-from textblob import TextBlob
-from flask import Flask, request
 
-app = Flask(__name__)
-
-@st.cache
-def run_model(input_text):
-    set_seed(42)
-    model = pipeline("text-generation", model="bert-base-cased", tokenizer="bert-base-cased")
-    generated_text = model(input_text, max_length=1024, num_return_sequences=1).generated_text[0]
-    return generated_text
-
-def main():
-    st.set_page_config(page_title="Paraphraser", page_icon=":guardsman:", layout="wide")
-    st.title("Paraphraser")
-
-    input_text = st.text_area("Input Text", "Type your text here")
-    options = st.selectbox("Paraphrase Options", ["Correct Grammar", "Remove Special Characters"])
-
-    if options == "Correct Grammar":
-        input_text = str(TextBlob(input_text).correct())
-
-    if options == "Remove Special Characters":
-        input_text = ''.join(e for e in input_text if e.isalnum() or e.isspace())
-
-    progress_bar = st.progress(0)
-
-    if st.button("Paraphrase"):
-        generated_text = run_model(input_text)
-        st.write("Generated Text:", generated_text)
-        st.write("Paraphrasing completed!")
-        st.write("")
-        st.write("Click the button below to copy the generated text to your clipboard:")
-        st.button("Copy to Clipboard", script="document.execCommand('copy');", key="copy")
-
-@app.route("/", methods=["GET", "POST"])
-def index():
-    main()
-    return st.get_html_str()
-
-if __name__ == "__main__":
-    app.run() #host="0.0.0.0",port=7860)
+def paraphrase_text(text, synonyms_num=5, random_synonym=True):
+    # Tokenize the text
+    tokens = word_tokenize(text)
+    # Create a list to hold the paraphrased words
+    paraphrased_tokens = []
+    for token in tokens:
+        # Check if the token is a word
+        if token.isalpha():
+            # Get the synonyms of the word
+            synonyms = []
+            for syn in wordnet.synsets(token):
+                for lemma in syn.lemmas():
+                    if lemma.name() != token:
+                        synonyms.append(lemma.name())
+            # If there are synonyms available, choose a random one
+            if synonyms:
+                if random_synonym:
+                    paraphrased_word = random.choice(synonyms)
+                else:
+                    paraphrased_word = ", ".join(synonyms[:synonyms_num])
+            # If no synonyms are available, use the original word
+            else:
+                paraphrased_word = token
+        # If the token is not a word, use it as-is
+        else:
+            paraphrased_word = token
+        # Add the paraphrased word to the list
+        paraphrased_tokens.append(paraphrased_word)
+    # Join the paraphrased tokens back into a string
+    paraphrased_text = ' '.join(paraphrased_tokens)
+    return paraphrased_text
+
+def paraphrase_html(html_text, synonyms_num, random_synonym):
+    # Parse the HTML using BeautifulSoup
+    soup = BeautifulSoup(html_text, 'html.parser')
+    # Find all the text nodes in the HTML
+    text_nodes = soup.find_all(text=True)
+    # Paraphrase the text nodes
+    for node in text_nodes:
+        node.replace_with(paraphrase_text(node.string, synonyms_num, random_synonym))
+    # Return the paraphrased HTML
+    paraphrased_html = str(soup)
+    return paraphrased_html
+
+st.set_page_config(page_title="HTML Paraphraser", page_icon=":pencil2:")
+st.title("HTML Paraphraser")
+
+synonyms_num = st.sidebar.slider("Number of Synonyms", min_value=1, max_value=10, value=5, step=1)
+random_synonym = st.sidebar.checkbox("Use Random Synonym", value=True)
+
+html_text = st.text_area("Enter HTML text to paraphrase", height=300)
+
+if st.button("Paraphrase"):
+    paraphrased_html = paraphrase_html(html_text, synonyms_num, random_synonym)
+    st.write(paraphrased_html, unsafe_allow_html=True)