Add application file

app.py

@@ -5,69 +5,57 @@ from huggingface_hub import hf_hub_download
 import pandas as pd
 import tempfile
 
-HF_REPO_ID = "stcoats/temp-duckdb-upload"
+# --- Config ---
+HF_REPO_ID = "stcoats/temp-duckdb-upload"  # Change if needed
 HF_FILENAME = "ycsep.duckdb"
 LOCAL_PATH = "./ycsep.duckdb"
 
+st.set_page_config(layout="wide")
 st.title("YCSEP Audio Dataset Viewer")
 
+# --- Download database if not present ---
 if not os.path.exists(LOCAL_PATH):
-    st.write("Downloading from HF Hub...")
-    downloaded_path = hf_hub_download(
+    st.write("Downloading ycsep.duckdb from HF Hub...")
+    hf_hub_download(
         repo_id=HF_REPO_ID,
         repo_type="dataset",
         filename=HF_FILENAME,
         local_dir="."
     )
-    st.success(f"Downloaded: {downloaded_path}")
-else:
-    st.write("DuckDB file already exists.")
-
-try:
-    con = duckdb.connect(LOCAL_PATH, read_only=True)
-    st.success("Connected to DuckDB.")
-except Exception as e:
-    st.error(f"Failed to connect: {e}")
-    st.stop()
-
-query = st.text_input("Search text or speaker (case-insensitive)")
-
-if query:
-    query_sql = f"""
-        SELECT speaker, text, audio FROM data
-        WHERE LOWER(text) LIKE '%{query.lower()}%'
-           OR LOWER(speaker) LIKE '%{query.lower()}%'
-        LIMIT 25
+    st.success("Download complete.")
+
+# --- Connect to DuckDB ---
+con = duckdb.connect(LOCAL_PATH, read_only=True)
+
+# --- Search input ---
+query = st.text_input("Search text or speaker (case-insensitive)", "")
+
+# --- Execute query ---
+if query.strip():
+    search_term = f"%{query.strip().lower()}%"
+    sql = """
+        SELECT speaker, text, audio
+        FROM data
+        WHERE LOWER(speaker) LIKE ? OR LOWER(text) LIKE ?
+        LIMIT 100
     """
+    df = con.execute(sql, [search_term, search_term]).df()
 else:
-    query_sql = "SELECT speaker, text, audio FROM data LIMIT 25"
-
-try:
-    df = con.execute(query_sql).df()
-    st.write(f"Showing {len(df)} results")
-
-    # Display text/speaker table
-    st.dataframe(df[['speaker', 'text']], use_container_width=True)
-
-    st.write("---")
-    st.markdown("### Audio Previews:")
-
-    for i, row in df.iterrows():
-        st.markdown(f"**Speaker:** {row['speaker']}")
-        st.markdown(f"{row['text']}")
-
-        # Write audio bytes to temp file
-        audio_bytes = row['audio']
-        if isinstance(audio_bytes, (bytes, bytearray)):
-            with tempfile.NamedTemporaryFile(delete=False, suffix=".mp3") as tmp_audio:
-                tmp_audio.write(audio_bytes)
-                tmp_audio_path = tmp_audio.name
-            st.audio(tmp_audio_path, format="audio/mp3")
-        else:
-            st.warning("Audio not available or in unexpected format.")
-
-        st.markdown("---")
-
-except Exception as e:
-    st.error(f"DuckDB query failed: {e}")
+    df = con.execute("SELECT speaker, text, audio FROM data LIMIT 100").df()
+
+# --- Show results ---
+st.markdown("### Search Results")
+for idx, row in df.iterrows():
+    col1, col2, col3 = st.columns([2, 5, 3])
+    col1.markdown(f"**{row['speaker']}**")
+    col2.markdown(row['text'])
+
+    audio_data = row['audio']
+    if isinstance(audio_data, (bytes, bytearray)):
+        with tempfile.NamedTemporaryFile(delete=False, suffix=".mp3") as tmp:
+            tmp.write(audio_data)
+            tmp.flush()
+            col3.audio(tmp.name, format="audio/mp3")
+    else:
+        col3.warning("Audio not available or invalid format.")