Update app.py

app.py

@@ -227,6 +227,67 @@ _session_rag = SessionRAG()
 # NEW: session-scoped data registry
 _data_registry = DataRegistry()
 
+def _assess_scenario_completeness(scenario_text: str, data_registry: DataRegistry, mapping: MappingResult) -> bool:
+    """Intelligently assess if scenario has enough info to proceed directly to analysis."""
+    if not scenario_text or not data_registry.names():
+        return False
+    
+    scenario_lower = scenario_text.lower()
+    
+    # Check for explicit instructions/tasks
+    has_explicit_tasks = any(phrase in scenario_lower for phrase in [
+        'identify', 'analyze', 'calculate', 'determine', 'compare', 'assess', 'rank', 'list',
+        'your tasks', 'deliverables', 'requirements', 'you should', 'you need to',
+        'find', 'show', 'report', 'evaluate', 'examine', 'investigate'
+    ])
+    
+    # Check for data descriptions that match uploaded files
+    mentions_data_files = any(phrase in scenario_lower for phrase in [
+        '.csv', 'dataset', 'data file', 'database', 'records', 'columns', 'spreadsheet', 'table'
+    ])
+    
+    # Check if scenario describes what the data contains
+    describes_data_structure = any(phrase in scenario_lower for phrase in [
+        'column', 'field', 'contains', 'includes', 'reports', 'each record', 'data shows', 'file has'
+    ])
+    
+    # NEW: Check if files were uploaded (implicit data context)
+    has_uploaded_files = len(data_registry.names()) > 0
+    
+    # NEW: Check for general analysis requests that imply using uploaded data
+    implies_data_analysis = any(phrase in scenario_lower for phrase in [
+        'this data', 'the data', 'analyze', 'analysis', 'insights', 'patterns', 'trends'
+    ])
+    
+    # Check mapping success rate
+    total_concepts = len(mapping.resolved) + len(mapping.ambiguous) + len(mapping.missing)
+    if total_concepts == 0:
+        return False
+    
+    mapping_success_rate = len(mapping.resolved) / total_concepts
+    has_good_mappings = mapping_success_rate >= 0.5  # At least half of concepts mapped
+    
+    # Check if critical ambiguities exist (more than 3 unresolved concepts)
+    critical_ambiguities = len(mapping.ambiguous) + len(mapping.missing) > 3
+    
+    # Enhanced decision logic: proceed if scenario is instructional AND either:
+    # 1. Explicitly describes data/files, OR
+    # 2. Files are uploaded and scenario implies analysis of "the data"
+    data_context_clear = (
+        mentions_data_files or 
+        describes_data_structure or 
+        (has_uploaded_files and implies_data_analysis)
+    )
+    
+    can_proceed = (
+        has_explicit_tasks and 
+        data_context_clear and 
+        has_good_mappings and 
+        not critical_ambiguities
+    )
+    
+    return can_proceed
+
 # ---------- Core chat logic (generic scenario handling) ----------
 def clarityops_reply(user_msg, history, tz, uploaded_files_paths, awaiting_answers=False):
     try:
@@ -296,16 +357,26 @@ def clarityops_reply(user_msg, history, tz, uploaded_files_paths, awaiting_answe
         mapping = map_concepts(safe_in, _data_registry)
 
         if not awaiting_answers:
-            # PHASE 1: ask for missing/ambiguous information
-            phase1 = build_phase1_questions(scenario_text=safe_in, registry=_data_registry, mapping=mapping)
-            phase1 = _sanitize_text(phase1)
-            log_event("assistant_reply", None, {
-                **hash_summary("prompt", safe_in if not PERSIST_CONTENT else ""),
-                **hash_summary("reply", phase1 if not PERSIST_CONTENT else ""),
-                "mode": "scenario_phase1",
-                "awaiting_next_phase": True
-            })
-            return history + [(user_msg, phase1)], True
+            # Intelligent scenario assessment: can we proceed directly to analysis?
+            can_proceed = _assess_scenario_completeness(safe_in, _data_registry, mapping)
+            
+            if can_proceed:
+                awaiting_answers = True  # Skip directly to Phase 2
+            else:
+                # PHASE 1: ask for missing/ambiguous information only when truly needed
+                phase1 = build_phase1_questions(scenario_text=safe_in, registry=_data_registry, mapping=mapping)
+                if phase1.strip() == "**Data Analysis Ready**: Your data appears well-structured. Please provide any additional context about your analysis goals.":
+                    # If only generic message, skip to analysis
+                    awaiting_answers = True
+                else:
+                    phase1 = _sanitize_text(phase1)
+                    log_event("assistant_reply", None, {
+                        **hash_summary("prompt", safe_in if not PERSIST_CONTENT else ""),
+                        **hash_summary("reply", phase1 if not PERSIST_CONTENT else ""),
+                        "mode": "scenario_phase1",
+                        "awaiting_next_phase": True
+                    })
+                    return history + [(user_msg, phase1)], True
 
         # PHASE 2: compute data analysis and generate structured response
         data_findings_md, missing_keys = build_data_findings_markdown(_data_registry, mapping)
@@ -550,4 +621,3 @@ if __name__ == "__main__":
 
 
 
-