Update src/streamlit_app.py

src/streamlit_app.py

@@ -103,7 +103,7 @@ def filter_content(content_list, filters):
             continue
             
         # Genre filter
-        if filters['genres'] and not any(genre in content.get('listed_in', '') for genre in filters['genres']):
+        if filters['genres'] and not any(genre.strip() in content.get('listed_in', '') for genre in filters['genres']):
             continue
             
         # Release year filter
@@ -123,6 +123,14 @@ def filter_content(content_list, filters):
                     continue
             except (ValueError, IndexError):
                 continue
+
+        # Director filter (optional)
+        if filters['director'] and filters['director'].lower() not in content.get('director', '').lower():
+            continue
+
+        # Cast filter (optional)
+        if filters['cast'] and filters['cast'].lower() not in content.get('cast', '').lower():
+            continue
         
         filtered.append(content)
     
@@ -240,52 +248,74 @@ def main():
     
     # Sidebar filters
     st.sidebar.header("🔍 Filter Content")
-    
-    content_type = st.sidebar.selectbox(
-        "Content Type",
-        options=["All", "Movie", "TV Show"],
-        index=0
-    )
-    
-    selected_ratings = st.sidebar.multiselect(
-        "Ratings",
-        options=all_ratings,
-        default=[]
-    )
-    
-    selected_genres = st.sidebar.multiselect(
-        "Genres",
-        options=all_genres,
-        default=[]
-    )
-    
-    # Year range slider
-    years = [int(c.get('release_year', 0)) for c in all_content if c.get('release_year')]
-    min_year, max_year = min(years), max(years)
-    year_range = st.sidebar.slider(
-        "Release Year",
-        min_value=min_year,
-        max_value=max_year,
-        value=(min_year, max_year)
-    )
-    
-    # Duration range slider (for movies only)
-    movie_durations = [
-        int(c.get('duration', '0 min').split()[0])
-        for c in all_content
-        if c.get('type') == 'Movie' and 'min' in c.get('duration', '')
-    ]
-    if movie_durations:
-        min_duration, max_duration = min(movie_durations), max(movie_durations)
-        duration_range = st.sidebar.slider(
-            "Movie Duration (minutes)",
-            min_value=min_duration,
-            max_value=max_duration,
-            value=(min_duration, max_duration),
-            help="This filter only applies to movies"
+
+    with st.sidebar.form("filter_form"):
+        st.subheader("Required Filters")
+        
+        content_type = st.selectbox(
+            "Content Type",
+            options=["All", "Movie", "TV Show"],
+            index=0
         )
-    else:
-        duration_range = (0, 1000)  # Fallback values
+        
+        selected_ratings = st.multiselect(
+            "Ratings",
+            options=all_ratings,
+            default=[]
+        )
+        
+        selected_genres = st.multiselect(
+            "Genres",
+            options=all_genres,
+            default=[]
+        )
+        
+        # Year range slider
+        years = [int(c.get('release_year', 0)) for c in all_content if c.get('release_year')]
+        min_year, max_year = min(years), max(years)
+        year_range = st.slider(
+            "Release Year",
+            min_value=min_year,
+            max_value=max_year,
+            value=(min_year, max_year)
+        )
+        
+        # Duration range slider (for movies only)
+        movie_durations = [
+            int(c.get('duration', '0 min').split()[0])
+            for c in all_content
+            if c.get('type') == 'Movie' and 'min' in c.get('duration', '')
+        ]
+        if movie_durations:
+            min_duration, max_duration = min(movie_durations), max(movie_durations)
+            duration_range = st.slider(
+                "Movie Duration (minutes)",
+                min_value=min_duration,
+                max_value=max_duration,
+                value=(min_duration, max_duration),
+                help="This filter only applies to movies"
+            )
+        else:
+            duration_range = (0, 1000)  # Fallback values
+
+        st.subheader("Optional Filters")
+        
+        # Director filter
+        director_filter = st.text_input(
+            "Director Name",
+            placeholder="Enter director name...",
+            help="Filter by director name (case-insensitive, partial match)"
+        )
+
+        # Cast filter
+        cast_filter = st.text_input(
+            "Cast Member",
+            placeholder="Enter cast member name...",
+            help="Filter by cast member name (case-insensitive, partial match)"
+        )
+        
+        # Submit button
+        apply_filters = st.form_submit_button("🔍 Apply Filters", type="primary")
     
     # Create filter dictionary
     filters = {
@@ -293,22 +323,30 @@ def main():
         'ratings': selected_ratings,
         'genres': selected_genres,
         'year_range': year_range,
-        'duration_range': duration_range
+        'duration_range': duration_range,
+        'director': director_filter,
+        'cast': cast_filter
     }
     
-    # Apply filters
-    filtered_content = filter_content(all_content, filters)
-    
+    # Only apply filters when the button is clicked
+    if apply_filters:
+        filtered_content = filter_content(all_content, filters)
+        st.session_state.filtered_content = filtered_content
+    else:
+        # Initialize filtered content if not exists
+        if 'filtered_content' not in st.session_state:
+            st.session_state.filtered_content = all_content
+
     # Main content area
     col1, col2 = st.columns([2, 1])
     
     with col1:
         # Content listings
-        st.subheader(f"📋 Found {len(filtered_content)} Titles")
+        st.subheader(f"📋 Found {len(st.session_state.filtered_content)} Titles")
         
-        if filtered_content:
+        if st.session_state.filtered_content:
             # Show first 10 items
-            for i, content in enumerate(filtered_content[:10]):
+            for i, content in enumerate(st.session_state.filtered_content[:10]):
                 with st.expander(f"{content.get('title', 'N/A')} ({content.get('release_year', 'N/A')})"):
                     # Content details in columns
                     detail_col1, detail_col2 = st.columns(2)
@@ -322,7 +360,7 @@ def main():
                         st.write(f"**🎬 Genre:** {content.get('listed_in', 'N/A')}")
                         # Fix cast handling to prevent NoneType error
                         cast = content.get('cast')
-                        cast_display = cast[:100] + "..." if cast else "N/A"
+                        cast_display = cast[:100] + "..." if cast and len(cast) > 100 else cast if cast else "N/A"
                         st.write(f"**👥 Cast:** {cast_display}")
                         st.write(f"**📝 Director:** {content.get('director', 'N/A')}")
                     
@@ -330,8 +368,8 @@ def main():
                     st.write(f"**📖 Description:**")
                     st.write(content.get('description', 'N/A'))
             
-            if len(filtered_content) > 10:
-                st.info(f"Showing first 10 of {len(filtered_content)} titles. Adjust filters to narrow results.")
+            if len(st.session_state.filtered_content) > 10:
+                st.info(f"Showing first 10 of {len(st.session_state.filtered_content)} titles. Adjust filters to narrow results.")
         else:
             st.info("No content matches your current filters. Try adjusting the criteria.")
     
@@ -371,7 +409,7 @@ def main():
             if user_question:
                 with st.spinner("AI is analyzing the content..."):
                     # Create context from current filtered data
-                    context = create_content_context(filtered_content)
+                    context = create_content_context(st.session_state.filtered_content)
                     
                     try:
                         # Get AI response
@@ -383,7 +421,7 @@ def main():
                         # Show debug info
                         with st.expander("Debug Info"):
                             st.write(f"Model used: {model_choice}")
-                            st.write(f"Content items analyzed: {len(filtered_content)}")
+                            st.write(f"Content items analyzed: {len(st.session_state.filtered_content)}")
                             st.write(f"Context: {context[:150]}...")
                         
                     except Exception as e:
@@ -391,18 +429,18 @@ def main():
                         
                         # Fallback response with data analysis
                         st.info("**Fallback Analysis:**")
-                        if filtered_content:
-                            movies = sum(1 for c in filtered_content if c.get('type') == 'Movie')
-                            shows = sum(1 for c in filtered_content if c.get('type') == 'TV Show')
-                            st.write(f"• Found {len(filtered_content)} titles ({movies} movies, {shows} TV shows)")
+                        if st.session_state.filtered_content:
+                            movies = sum(1 for c in st.session_state.filtered_content if c.get('type') == 'Movie')
+                            shows = sum(1 for c in st.session_state.filtered_content if c.get('type') == 'TV Show')
+                            st.write(f"• Found {len(st.session_state.filtered_content)} titles ({movies} movies, {shows} TV shows)")
                             
                             # Show top genres
-                            genres = [g.strip() for c in filtered_content for g in c.get('listed_in', '').split(',')]
+                            genres = [g.strip() for c in st.session_state.filtered_content for g in c.get('listed_in', '').split(',')]
                             genre_counts = pd.Series(genres).value_counts()
                             st.write(f"• Top genres: {', '.join(genre_counts.head(3).index)}")
                             
                             # Show year range
-                            years = [int(c.get('release_year', 0)) for c in filtered_content if c.get('release_year')]
+                            years = [int(c.get('release_year', 0)) for c in st.session_state.filtered_content if c.get('release_year')]
                             if years:
                                 st.write(f"• Release years: {min(years)} - {max(years)}")
             else:
@@ -412,7 +450,7 @@ def main():
     st.markdown("---")
     if all_content:
         total_items = len(all_content)
-        filtered_items = len(filtered_content)
+        filtered_items = len(st.session_state.filtered_content)
         
         stat_col1, stat_col2, stat_col3, stat_col4 = st.columns(4)
         
@@ -423,11 +461,11 @@ def main():
             st.metric("Filtered Results", filtered_items)
         
         with stat_col3:
-            movies = sum(1 for c in filtered_content if c.get('type') == 'Movie')
+            movies = sum(1 for c in st.session_state.filtered_content if c.get('type') == 'Movie')
             st.metric("Movies", movies)
         
         with stat_col4:
-            shows = sum(1 for c in filtered_content if c.get('type') == 'TV Show')
+            shows = sum(1 for c in st.session_state.filtered_content if c.get('type') == 'TV Show')
             st.metric("TV Shows", shows)
 
 if __name__ == "__main__":