Update app.py

app.py

@@ -58,12 +58,12 @@ def get_connection():
 
 con = get_connection()
 
-# --- Cached Metadata for Filters ---
+# --- Metadata for Filters ---
 @st.cache_data(ttl=600)
 def get_filter_options():
-    # Fast counts via DuckDB
-    entities = con.sql("SELECT entity, COUNT(*) as c FROM sentiment_analysis GROUP BY 1 ORDER BY c DESC").df()
-    domains = con.sql("SELECT domain, COUNT(*) as c FROM sentiment_analysis GROUP BY 1 ORDER BY c DESC").df()
+    # Trim and count for cleaner lookups
+    entities = con.sql("SELECT trim(entity) as entity, COUNT(*) as c FROM sentiment_analysis GROUP BY 1 ORDER BY c DESC").df()
+    domains = con.sql("SELECT trim(domain) as domain, COUNT(*) as c FROM sentiment_analysis GROUP BY 1 ORDER BY c DESC").df()
     return entities, domains
 
 df_entities, df_domains = get_filter_options()
@@ -72,7 +72,7 @@ df_entities, df_domains = get_filter_options()
 st.title(T[LANG]["title"])
 
 with st.sidebar:
-    # Create lookup dictionaries to avoid filtering dataframes in the lambda (prevents crashes)
+    # Dictionary lookup prevents crashes on format_func
     entity_lookup = dict(zip(df_entities['entity'], df_entities['c']))
     domain_lookup = dict(zip(df_domains['domain'], df_domains['c']))
 
@@ -95,25 +95,45 @@ with st.sidebar:
     timeframes = {"All": 9999, "Last 7 Days": 7, "Last 30 Days": 30, "Last Year": 365}
     time_choice = st.selectbox("Timeframe", list(timeframes.keys()))
 
-# --- Query Building ---
+# --- Logic: Stop if no selection ---
 if not selected_entities:
     st.info(T[LANG]["select_entities_prompt"])
     st.stop()
 
-# Dynamic SQL construction
+# --- Dynamic Granularity Logic (Targeting ~30 points) ---
+days = timeframes[time_choice]
+
+# Calculate optimal bucket size to get ~30 data points
+if days <= 30:
+    bucket = "1 day"
+elif days <= 210:  # ~30 weeks
+    bucket = "1 week"
+elif days <= 900:  # ~30 months
+    bucket = "1 month"
+else:
+    bucket = "1 year"
+
+# --- Query Building ---
 where_clause = f"WHERE entity IN ({str(selected_entities)[1:-1]})"
 if selected_domains:
     where_clause += f" AND domain IN ({str(selected_domains)[1:-1]})"
 if time_choice != "All":
-    where_clause += f" AND created_at >= (epoch(now()) - {timeframes[time_choice] * 86400})"
+    where_clause += f" AND created_at >= (epoch(now()) - {days * 86400})"
 
+# --- CRITICAL FIX: Explicit Grouping for Smooth Lines ---
 if all_scores:
+    # Unpivot case: Group by Date, Entity, and Score Type
+    # If grouping by domain is active, add domain to the group key
+    group_keys = "1, 2, 3"
+    select_keys = "time_bucket(interval '{bucket}', created_at) as date, entity, score_name as score_type"
+    
+    if group_by_domain:
+        select_keys += ", domain"
+        group_keys += ", 4"
+
     sql_query = f"""
     SELECT 
-        time_bucket(interval '1 day', created_at) as date,
-        entity,
-        domain,
-        score_name as score_type,
+        {select_keys},
         AVG(score_value) as score
     FROM (
         UNPIVOT sentiment_analysis 
@@ -121,18 +141,26 @@ if all_scores:
         INTO NAME score_name VALUE score_value
     )
     {where_clause}
-    GROUP BY ALL
+    GROUP BY {group_keys}
+    ORDER BY 1 ASC
     """
 else:
+    # Standard case: Group by Date and Entity
+    group_keys = "1, 2"
+    select_keys = "time_bucket(interval '{bucket}', created_at) as date, entity"
+
+    if group_by_domain:
+        select_keys += ", domain"
+        group_keys += ", 3"
+
     sql_query = f"""
     SELECT 
-        time_bucket(interval '1 day', created_at) as date,
-        entity,
-        domain,
+        {select_keys},
         AVG({score_type}) as score
     FROM sentiment_analysis
     {where_clause}
-    GROUP BY ALL
+    GROUP BY {group_keys}
+    ORDER BY 1 ASC
     """
 
 # --- Execution & Plotting ---
@@ -145,32 +173,34 @@ except Exception as e:
 if filtered_df.empty:
     st.warning(T[LANG]["no_data"])
 else:
-    # Build dynamic labels for the legend
-    color_col = "entity"
+    # Build legend labels
     if group_by_domain:
         filtered_df["label"] = filtered_df["entity"] + " | " + filtered_df["domain"]
-        color_col = "label"
+    else:
+        filtered_df["label"] = filtered_df["entity"]
     
     if all_scores:
-        current_label = filtered_df["label"] if group_by_domain else filtered_df["entity"]
-        filtered_df["label"] = current_label + " | " + filtered_df["score_type"]
-        color_col = "label"
+        filtered_df["label"] = filtered_df["label"] + " | " + filtered_df["score_type"]
 
+    # Plot
     fig = px.line(
-        filtered_df.sort_values("date"),
-        x="date", y="score", color=color_col,
-        title=T[LANG]["scores_over_time"] if all_scores else T[LANG]["avg_over_time"].format(score_type.replace('_', ' ').title()),
-        labels={"score": "Sentiment Score", "date": "Date", "label": "Entity/Source"}
+        filtered_df,
+        x="date", 
+        y="score", 
+        color="label",
+        title=f"{T[LANG]['scores_over_time']} (Avg by {bucket})",
+        labels={"score": "Score", "date": "Date", "label": "Legend"},
+        markers=True  # Adds dots to points for clarity
     )
 
-    # Styling
+    # Layout styling
     fig.update_layout(
         yaxis=dict(range=[-10, 10], gridcolor="lightgrey"),
         plot_bgcolor="white",
         legend=dict(orientation="h", y=-0.2, x=0.5, xanchor="center")
     )
     
-    # Static horizontal reference lines
+    # Reference lines
     for val in [-5, 0, 5]:
         fig.add_hline(y=val, line_width=2 if val==0 else 1, line_dash="dash", line_color="black")