Update app.py

app.py

@@ -9,6 +9,7 @@ import ffmpeg
 import time
 import json
 import psutil
+from fuzzywuzzy import fuzz  # Added for fuzzy matching
 
 st.set_page_config(layout="wide")
 
@@ -589,14 +590,21 @@ def main():
         st.session_state['show_timeframe'] = st.checkbox("Show timeframe in transcript", value=st.session_state['show_timeframe'])
         st.markdown("### Search Subtitles")
         search_query = st.text_input("Search subtitles...", value=st.session_state['search_query'], key="search_input")
-        st.session_state['search_query'] = search_query.lower()
+        st.session_state['search_query'] = search_query.lower().strip()
 
-        # Primary Transcript
+        # Primary Transcript with Fuzzy Matching
         st.markdown(f"### {st.session_state['language']} Transcript")
         primary_matches = 0
         for text, start, end in st.session_state['primary_transcript']:
             display_text = text.lower()
-            if not search_query or search_query in display_text:
+            # Use fuzzy matching for partial and robust search
+            if not search_query:
+                match = True
+            elif len(search_query) < 3:  # Exact match for very short queries
+                match = search_query in display_text
+            else:  # Fuzzy matching for longer queries
+                match = fuzz.partial_ratio(search_query, display_text) >= 70
+            if match:
                 primary_matches += 1
                 label = f"[{format_time(start)} - {format_time(end)}] {text}" if st.session_state['show_timeframe'] else text
                 if st.button(label, key=f"primary_{start}"):
@@ -605,13 +613,20 @@ def main():
         if primary_matches == 0 and search_query:
             st.info("No matches found in primary transcript for the search query.")
 
-        # English Transcript
+        # English Transcript with Fuzzy Matching
         if st.session_state['english_transcript']:
             st.markdown("### English Translation")
             english_matches = 0
             for text, start, end in st.session_state['english_transcript']:
                 display_text = text.lower()
-                if not search_query or search_query in display_text:
+                # Use fuzzy matching for partial and robust search
+                if not search_query:
+                    match = True
+                elif len(search_query) < 3:  # Exact match for very short queries
+                    match = search_query in display_text
+                else:  # Fuzzy matching for longer queries
+                    match = fuzz.partial_ratio(search_query, display_text) >= 70
+                if match:
                     english_matches += 1
                     label = f"[{format_time(start)} - {format_time(end)}] {text}" if st.session_state['show_timeframe'] else text
                     if st.button(label, key=f"english_{start}"):