feat: add choice_info field to CSV export with ICL truncation

- Add choice_info and label fields to InputOutputItem model
- Add choice_info column to CSV export containing full JSON data
- Truncate icl_example_responses to first example + "truncated" indicator
- Reduces CSV size from 130MB to 14MB for large datasets while preserving sample data

Author: PaulHax

align_browser/csv_exporter.py

@@ -67,6 +67,35 @@ def extract_justification(item: InputOutputItem) -> str:
     return action.get("justification", "")
 
 
+def extract_choice_info(item: InputOutputItem) -> str:
+    """Extract the choice_info as a JSON string, truncating ICL examples."""
+    if not item.choice_info:
+        return ""
+
+    import json
+
+    # Create a copy to avoid modifying the original
+    filtered_choice_info = {}
+
+    for key, value in item.choice_info.items():
+        if key == "icl_example_responses" and isinstance(value, dict):
+            # Keep only first example for each KDMA, truncate the rest
+            truncated_icl = {}
+            for kdma, examples in value.items():
+                if isinstance(examples, list) and len(examples) > 0:
+                    # Keep first example, replace rest with "truncated"
+                    truncated_icl[kdma] = [examples[0]]
+                    if len(examples) > 1:
+                        truncated_icl[kdma].append("truncated")
+                else:
+                    truncated_icl[kdma] = examples
+            filtered_choice_info[key] = truncated_icl
+        else:
+            filtered_choice_info[key] = value
+
+    return json.dumps(filtered_choice_info, separators=(",", ":"))
+
+
 def get_decision_time(
     timing_data: Optional[Dict[str, Any]], item_index: int
 ) -> Optional[float]:
@@ -156,6 +185,7 @@ def experiment_to_csv_rows(
             else "",
             "choice_text": extract_choice_text(item),
             "choice_kdma_association": extract_choice_kdma(item),
+            "choice_info": extract_choice_info(item),
             "justification": extract_justification(item),
             "decision_time_s": get_decision_time(timing_data, idx),
             "score": get_score(scores_data, idx),
@@ -188,6 +218,7 @@ def write_experiments_to_csv(
         "state_description",
         "choice_text",
         "choice_kdma_association",
+        "choice_info",
         "justification",
         "decision_time_s",
         "score",

align_browser/experiment_models.py

@@ -184,6 +184,8 @@ class InputOutputItem(BaseModel):
 
     input: InputData
     output: Optional[Dict[str, Any]] = None
+    choice_info: Optional[Dict[str, Any]] = None
+    label: Optional[List[Dict[str, Any]]] = None
     original_index: int  # Index in the original file
 
 

align_browser/test_link_cascade.py

@@ -1,55 +1,60 @@
 """Test that linked parameter cascading works properly"""
+
 from playwright.sync_api import Page
 
 
 def test_linked_scenario_scene_cascade(page: Page, real_data_test_server: str):
     """Test that when scenario and scene are linked, changing scenario cascades scene properly."""
     page.goto(real_data_test_server)
-    
+
     # Wait for page to load
     page.wait_for_selector(".comparison-table", timeout=10000)
     page.wait_for_function(
         "document.querySelectorAll('.table-scenario-select').length > 0", timeout=10000
     )
-    
+
     # Add a second column for testing
     add_column_btn = page.locator("#add-column-btn")
     if add_column_btn.is_visible():
         add_column_btn.click()
         page.wait_for_timeout(1000)
-    
+
     # Get initial scenario values from both columns
     scenario_selects = page.locator(".table-scenario-select")
     assert scenario_selects.count() >= 2, "Need at least 2 columns to test"
-    
+
     initial_scenario_col1 = scenario_selects.nth(0).input_value()
     initial_scenario_col2 = scenario_selects.nth(1).input_value()
-    
+
     # Get initial scene values - scene dropdowns use table-scenario-select class but in scene row
     scene_row = page.locator("tr.parameter-row[data-parameter='scene']")
     scene_selects = scene_row.locator(".table-scenario-select")
     initial_scene_col1 = scene_selects.nth(0).input_value()
     initial_scene_col2 = scene_selects.nth(1).input_value()
-    
+
     print("Initial states:")
     print(f"  Column 1: scenario={initial_scenario_col1}, scene={initial_scene_col1}")
     print(f"  Column 2: scenario={initial_scenario_col2}, scene={initial_scene_col2}")
-    
+
     # Link both scenario and scene parameters
     scenario_row = page.locator("tr.parameter-row[data-parameter='scenario']")
     scenario_link = scenario_row.locator(".link-toggle")
     scenario_link.click()
     page.wait_for_timeout(500)
-    
-    scene_row = page.locator("tr.parameter-row[data-parameter='scene']")  
+
+    scene_row = page.locator("tr.parameter-row[data-parameter='scene']")
     scene_link = scene_row.locator(".link-toggle")
     scene_link.click()
     page.wait_for_timeout(500)
-    
+
     # Verify both are linked - the row itself gets the linked class
-    assert "linked" in scenario_row.get_attribute("class"), "Scenario row should have 'linked' class"
-    assert "linked" in scene_row.get_attribute("class"), "Scene row should have 'linked' class"
-    
+    assert "linked" in scenario_row.get_attribute("class"), (
+        "Scenario row should have 'linked' class"
+    )
+    assert "linked" in scene_row.get_attribute("class"), (
+        "Scene row should have 'linked' class"
+    )
+
     # Change scenario in first column to a different value
     # Find a different scenario option
     scenario_select_col1 = scenario_selects.nth(0)
@@ -60,39 +65,46 @@ def test_linked_scenario_scene_cascade(page: Page, real_data_test_server: str):
         if value and value != initial_scenario_col1:
             new_scenario = value
             break
-    
+
     assert new_scenario is not None, "Could not find a different scenario to switch to"
-    
+
     print(f"\nChanging scenario from {initial_scenario_col1} to {new_scenario}")
     scenario_select_col1.select_option(new_scenario)
     page.wait_for_timeout(2000)  # Wait for cascading and data reload
-    
+
     # Check the results after changing scenario
     final_scenario_col1 = scenario_selects.nth(0).input_value()
     final_scenario_col2 = scenario_selects.nth(1).input_value()
     final_scene_col1 = scene_selects.nth(0).input_value()
     final_scene_col2 = scene_selects.nth(1).input_value()
-    
+
     print("\nFinal states:")
     print(f"  Column 1: scenario={final_scenario_col1}, scene={final_scene_col1}")
     print(f"  Column 2: scenario={final_scenario_col2}, scene={final_scene_col2}")
-    
+
     # Verify scenarios are synced (both should have the new scenario)
     assert final_scenario_col1 == new_scenario, "Column 1 should have the new scenario"
-    assert final_scenario_col2 == new_scenario, "Column 2 should also have the new scenario (linked)"
-    
+    assert final_scenario_col2 == new_scenario, (
+        "Column 2 should also have the new scenario (linked)"
+    )
+
     # Verify scenes are synced AND valid for the new scenario
-    assert final_scene_col1 == final_scene_col2, "Scenes should be synced across columns"
-    
+    assert final_scene_col1 == final_scene_col2, (
+        "Scenes should be synced across columns"
+    )
+
     # Check that we don't have "No data available" messages
     no_data_messages = page.locator(".no-data-message").all()
     for msg in no_data_messages:
         if msg.is_visible():
             parent_cell = msg.locator("..").first
-            print(f"WARNING: Found 'No data available' message in cell: {parent_cell.inner_text()}")
-    
+            print(
+                f"WARNING: Found 'No data available' message in cell: {parent_cell.inner_text()}"
+            )
+
     # The key test: scenes should have cascaded to valid values, not preserved invalid ones
-    assert len([msg for msg in no_data_messages if msg.is_visible()]) == 0, \
+    assert len([msg for msg in no_data_messages if msg.is_visible()]) == 0, (
         "Should not have any 'No data available' messages after linked cascade"
-    
-    print("\nTest passed! Linked parameters cascade properly.")
+    )
+
+    print("\nTest passed! Linked parameters cascade properly.")