Commit Test SRT

app.py

@@ -1,12 +1,15 @@
 import streamlit as st
+from flask import Flask, request, jsonify
 from transformers import pipeline
-from heapq import nlargest
 from keybert import KeyBERT
 
+# Initialize Flask app
+app = Flask(__name__)
+
 # Function to extract text from SRT-formatted text
 def extract_text_from_srt_text(srt_text):
-    lines = srt_text.strip().split("\n\n")  # Split by empty lines to separate subtitles
-    texts = [subtitle.split("\n")[2] for subtitle in lines if subtitle.strip()]  # Extract text from the third line of each subtitle
+    lines = srt_text.strip().split("\n\n")
+    texts = [subtitle.split("\n")[2] for subtitle in lines if subtitle.strip()]
     return " ".join(texts)
 
 # Function to generate summary from text
@@ -20,22 +23,11 @@ def generate_summary(text, summary_length):
 def extract_top_topics(text, n_top_topics):
     model = KeyBERT('distilbert-base-nli-mean-tokens')
     keywords = model.extract_keywords(text, keyphrase_ngram_range=(1, 3), stop_words='english', use_maxsum=True, nr_candidates=20, top_n=n_top_topics)
-    return keywords
+    return [topic for topic, _ in keywords]
 
 # Streamlit app
 st.title("SRT Summarization")
 
-# Logo image URL
-logo_url = "https://encrypted-tbn0.gstatic.com/images?q=tbn:ANd9GcQ6uQl0omK_PHXBbyaCHdmh3VjCo_Yvgwavmcs5XRF9Rkjx5FpflxyO4yfux6d2ojKsCOA&usqp=CAU"  # Replace with your logo image URL
-
-# Center the logo
-st.markdown(
-    f'<div style="display: flex; justify-content: center;">'
-    f'<img src="{logo_url}" style="width: 364px;">'
-    f'</div>',
-    unsafe_allow_html=True
-)
-
 # Text area for user to input SRT-formatted text
 srt_text_input = st.text_area("Paste SRT-formatted text here:")
 
@@ -43,20 +35,32 @@ srt_text_input = st.text_area("Paste SRT-formatted text here:")
 if st.button("Summarize"):
     # Check if text area is not empty
     if srt_text_input.strip():
-        # Show loading spinner while processing
-        with st.spinner("Summarizing..."):
-            # Extract text from SRT-formatted text
-            text_to_summarize = extract_text_from_srt_text(srt_text_input)
-            # Generate summary
-            summary = generate_summary(text_to_summarize, 150)  # You can adjust the summary length as needed
-            # Display summary
-            st.subheader("Summary:")
-            st.write(summary)
-            # Extract top 4 topics
-            top_topics = extract_top_topics(text_to_summarize, 4)
-            # Display top 4 topics
-            st.subheader("Top 4 Keywords:")
-            for topic, _ in top_topics:
-                st.write(f"- {topic}")
+        # Extract text from SRT-formatted text
+        text_to_summarize = extract_text_from_srt_text(srt_text_input)
+        # Generate summary
+        summary = generate_summary(text_to_summarize, 150)
+        # Extract top 4 topics
+        top_topics = extract_top_topics(text_to_summarize, 4)
+        # Display summary and top 4 topics
+        st.subheader("Summary:")
+        st.write(summary)
+        st.subheader("Top 4 Keywords:")
+        for topic in top_topics:
+            st.write(f"- {topic}")
     else:
         st.warning("Please enter some SRT-formatted text.")
+
+# Define endpoint for REST API
+@app.route("/summarize", methods=["POST"])
+def summarize():
+    data = request.json
+    if "srt_text" not in data:
+        return jsonify({"error": "Missing 'srt_text' parameter"}), 400
+    srt_text = data["srt_text"]
+    text_to_summarize = extract_text_from_srt_text(srt_text)
+    summary = generate_summary(text_to_summarize, 150)
+    top_topics = extract_top_topics(text_to_summarize, 4)
+    return jsonify({"summary": summary, "top_topics": top_topics})
+
+if __name__ == "__main__":
+    app.run()

requirements.txt

@@ -2,4 +2,5 @@ torch
 transformers
 streamlit
 gradio
-keybert
+keybert
+flask