Add application file

app.py

@@ -42,13 +42,15 @@ query = st.text_input("Search text (case-insensitive)", "").strip()
 
 # Build query
 if query:
-    sql = """
+    search_terms = query.lower().split()
+    conditions = " AND ".join(["LOWER(text) LIKE ?" for _ in search_terms])
+    sql = f"""
         SELECT id, channel, video_id, speaker, start_time, end_time, upload_date, text, pos_tags, audio
         FROM data
-        WHERE LOWER(text) LIKE LOWER(?)
+        WHERE {conditions}
         LIMIT 100
     """
-    df = con.execute(sql, [f"%{query}%"]).df()
+    df = con.execute(sql, [f"%{term}%" for term in search_terms]).df()
 else:
     df = con.execute("""
         SELECT id, channel, video_id, speaker, start_time, end_time, upload_date, text, pos_tags, audio
@@ -61,7 +63,7 @@ st.markdown(f"### Showing {len(df)} results")
 if len(df) == 0:
     st.warning("No matches found.")
 else:
-    def render_audio_cell(audio_bytes):
+    def render_audio_html(audio_bytes):
         try:
             if isinstance(audio_bytes, (bytes, bytearray, memoryview)):
                 data = bytes(audio_bytes)
@@ -72,22 +74,25 @@ else:
             with tempfile.NamedTemporaryFile(delete=False, suffix=".mp3") as tmp:
                 tmp.write(data)
                 tmp.flush()
-                return f'<audio controls style="height:20px;"> <source src="file://{tmp.name}" type="audio/mpeg"></audio>'
+                return f'<audio controls style="height:20px;width:100%;"><source src="file://{tmp.name}" type="audio/mpeg"></audio>'
         except Exception:
             return ""
 
-    df["Audio"] = df["audio"].apply(render_audio_cell)
-    df_display = df.drop(columns=["audio"])  # Drop binary column before display
+    df["Audio"] = df["audio"].apply(render_audio_html)
+    df = df.drop(columns=["audio"])
+    df = df[["id", "channel", "video_id", "speaker", "start_time", "end_time", "upload_date", "text", "pos_tags", "Audio"]]
 
-    # Reorder columns for display
-    df_display = df_display[["id", "channel", "video_id", "speaker", "start_time", "end_time", "upload_date", "text", "pos_tags", "Audio"]]
+    from st_aggrid import AgGrid, GridOptionsBuilder
 
-    st.markdown("### Results Table (Sortable with Audio Column)")
-    st.markdown("(Scroll right to view audio controls)")
-    st.dataframe(df_display, use_container_width=True)
+    st.markdown("### Results Table (Sortable with Embedded Audio)")
 
-    # Optionally, render inline HTML audio
-    # for i, row in df_display.iterrows():
-    #     st.markdown(f"**{row['speaker']} | {row['text']}**")
-    #     st.markdown(row["Audio"], unsafe_allow_html=True)
+    gb = GridOptionsBuilder.from_dataframe(df.drop(columns=["Audio"]))
+    gb.configure_default_column(resizable=True, sortable=True, filter=True)
+    grid_options = gb.build()
+
+    AgGrid(df.drop(columns=["Audio"]), gridOptions=grid_options, fit_columns_on_grid_load=True)
+
+    st.markdown("### Audio Controls in Table")
+    for i in range(len(df)):
+        st.markdown(df.loc[i, "Audio"], unsafe_allow_html=True)