small fixes

app.py

@@ -29,6 +29,11 @@ run_id_options = []
 run_id_dropdown = None
 
 
+# Last source runs for retrieving full baseline_reason on selection
+last_source_runs = []
+# Maximum length for baseline_reason display
+TRUNCATE_REASON_LEN = 50
+
 # Get all questions
 
 # Initialize data in a single async function
@@ -143,7 +148,8 @@ async def update_sources_list_async(question_option, source_finder_name, run_id,
         # Get source runs data
         if finder_id_int:
             source_runs, stats = await get_unified_sources(conn, question_id, run_id_int, baseline_ranker_id_int)
-            # Create DataFrame for display
+            global last_source_runs
+            last_source_runs = source_runs
             df = pd.DataFrame(source_runs)
 
         if not source_runs:
@@ -268,7 +274,7 @@ async def main():
             with gr.Column(scale=3):
                 results_table = gr.DataFrame(
                     headers=['id', 'tractate', 'folio', 'in_baseline', 'baseline_rank', 'in_source_run',
-                             'source_run_rank', 'source_reason', 'metadata'],
+                             'source_run_rank', 'source_reason', 'baseline_reason'],
                     interactive=False
                 )
             with gr.Column(scale=1):

data_access.py

@@ -248,7 +248,7 @@ async def get_unified_sources(conn: asyncpg.Connection, question_id: int, source
                  """
     source_runs = await conn.fetch(query_runs, question_id, source_finder_run_id)
     # Get sources from baseline_sources
-    baseline_query = query_runs.replace("source_rank", "baseline_rank")
+    baseline_query = query_runs.replace("source_rank", "baseline_rank").replace("source_reason", "baseline_reason")
     baseline_sources = await conn.fetch(baseline_query, question_id, ranker_id)
     stats_df = await calculate_baseline_vs_source_stats_for_question(conn, baseline_sources, source_runs)
     # Convert to dictionaries for easier lookup
@@ -274,7 +274,7 @@ async def get_unified_sources(conn: asyncpg.Connection, question_id: int, source
             "in_source_run": "Yes" if in_source_run else "No",
             "source_run_rank": source_runs_dict.get(sugya_id, {}).get("source_rank", "N/A"),
             "source_reason": source_runs_dict.get(sugya_id, {}).get("reason", "N/A"),
-            "metadata": source_runs_dict.get(sugya_id, {}).get("metadata", "")
+            "baseline_reason": baseline_dict.get(sugya_id, {}).get("baseline_reason", "N/A"),
         }
         unified_results.append(result)