feat(ui): Add Auto Rubric feature for automatic grading criteria gene… (#92)

* feat(ui): Add Auto Rubric feature for automatic grading criteria generation

Add new Auto Rubric feature that automatically generates evaluation rubrics
for LLM applications. Includes simple and iterative generation modes,
data upload, history management, and export functionality.

* fix: Rename test_panel.py to rubric_tester.py to avoid pytest collection

Pytest was treating test_panel.py as a test file due to the 'test_' prefix,
causing import errors in CI. Renamed to rubric_tester.py to fix this issue.

* fix(ui): stabilize widget state across language switch and persist language preference

- Use stable values (e.g., "_custom_", "python") instead of translated labels as selectbox options
- Prevents widget state loss when UI language changes
- Add localStorage persistence for language preference
- Add inject_language_loader() to restore language setting on page load
- Update navigation to use stable feature_ids for feature selector

Author: XiaoBoAI

ui/app.py

@@ -20,12 +20,17 @@
 from core.feature_registry import FeatureRegistry  # noqa: E402
 from core.navigation import Navigation  # noqa: E402
 from features.auto_arena import AutoArenaFeature  # noqa: E402
+from features.auto_rubric import AutoRubricFeature  # noqa: E402
 
 # Import feature modules
 from features.grader import GraderFeature  # noqa: E402
 from shared.components.common import render_footer  # noqa: E402
 from shared.components.logo import render_logo_and_title  # noqa: E402
-from shared.i18n import render_language_selector, t  # noqa: E402
+from shared.i18n import (  # noqa: E402
+    inject_language_loader,
+    render_language_selector,
+    t,
+)
 from shared.styles.theme import inject_css  # noqa: E402
 
 # pylint: enable=wrong-import-position
@@ -38,9 +43,7 @@
 # Add new features here as they are implemented
 FeatureRegistry.register(GraderFeature)
 FeatureRegistry.register(AutoArenaFeature)
-# Future features:
-# from features.autorubric import AutoRubricFeature
-# FeatureRegistry.register(AutoRubricFeature)
+FeatureRegistry.register(AutoRubricFeature)
 
 # ============================================================================
 # Page Configuration (must be first Streamlit command)
@@ -58,6 +61,9 @@ def main() -> None:
     # Inject custom CSS
     inject_css()
 
+    # Load language preference from browser localStorage
+    inject_language_loader()
+
     # ========================================================================
     # Sidebar Configuration
     # ========================================================================

ui/core/navigation.py

@@ -38,29 +38,31 @@ def render_feature_selector() -> str:
             st.warning("No features registered")
             return ""
 
-        # Build options - use display_label property for i18n support
-        # Note: get_all() returns classes, but display_label is a property that requires instances
+        # Build options - use stable feature_ids (not translated labels)
         feature_ids = [f.feature_id for f in features]
-        feature_labels = {f.feature_id: FeatureRegistry.get_instance(f.feature_id).display_label for f in features}
 
         # Get default feature id
         default_id = FeatureRegistry.get_default_feature_id()
 
-        # Use widget key directly for state management
+        # Widget key for selectbox
         widget_key = "_nav_feature_selector"
 
         # Initialize widget state if not exists
         if widget_key not in st.session_state:
             st.session_state[widget_key] = default_id
 
-        # Ensure current value is valid
+        # Ensure current value is valid (in case features changed)
         if st.session_state[widget_key] not in feature_ids:
             st.session_state[widget_key] = default_id
 
         # Get previous value for lifecycle hooks
         previous_id = st.session_state.get(CURRENT_FEATURE_KEY)
 
-        # Render selectbox (dropdown) for feature selection
+        # Build labels dynamically (these change with language, but values stay stable)
+        feature_labels = {f.feature_id: FeatureRegistry.get_instance(f.feature_id).display_label for f in features}
+
+        # Render selectbox with stable feature_ids as options
+        # No index parameter - let Streamlit manage state via key
         selected_id = st.selectbox(
             t("app.features"),
             options=feature_ids,

ui/features/__init__.py

@@ -7,6 +7,7 @@
 """
 
 from features.auto_arena import AutoArenaFeature
+from features.auto_rubric import AutoRubricFeature
 from features.grader import GraderFeature
 
-__all__ = ["GraderFeature", "AutoArenaFeature"]
+__all__ = ["GraderFeature", "AutoArenaFeature", "AutoRubricFeature"]

ui/features/auto_arena/components/config_panel.py

@@ -192,24 +192,30 @@ def _render_single_endpoint(
         )
 
     with col2:
+        # Use stable value for custom option to survive UI language switch
+        CUSTOM_VALUE = "_custom_"
+
         # Check if there's a custom model value already set (e.g., from preset)
         current_model = st.session_state.get(f"arena_ep_model_{endpoint_id}", "")
 
         # Build options list - include current model if it's custom
         model_options = list(DEFAULT_MODELS)
-        custom_label = t("model.custom")
-        if current_model and current_model not in model_options and current_model != custom_label:
+        if current_model and current_model not in model_options and current_model != CUSTOM_VALUE:
             model_options.insert(0, current_model)
-        model_options.append(custom_label)
+        model_options.append(CUSTOM_VALUE)
+
+        def format_model(x: str) -> str:
+            return t("model.custom") if x == CUSTOM_VALUE else x
 
         model_option = st.selectbox(
             t("model.select"),
             options=model_options,
+            format_func=format_model,
             key=f"arena_ep_model_{endpoint_id}",
         )
 
     # Row 1.5: Custom model input (only if "Custom..." selected)
-    if model_option == t("model.custom"):
+    if model_option == CUSTOM_VALUE:
         model = st.text_input(
             t("model.custom_input"),
             placeholder=t("model.custom_placeholder"),

ui/features/auto_arena/components/sidebar.py

@@ -32,12 +32,13 @@ def _apply_preset_sidebar_data() -> None:
 
     st.session_state["arena_judge_api_key"] = preset_data.get("judge_api_key", "")
 
-    # Judge model
+    # Judge model - use stable value for custom option
+    CUSTOM_VALUE = "_custom_"
     judge_model = preset_data.get("judge_model", "")
     if judge_model in DEFAULT_MODELS:
-        st.session_state["arena_judge_model"] = judge_model
+        st.session_state["arena_judge_model_value"] = judge_model
     else:
-        st.session_state["arena_judge_model"] = "Custom..."
+        st.session_state["arena_judge_model_value"] = CUSTOM_VALUE
         st.session_state["arena_judge_custom_model"] = judge_model
 
     # Evaluation settings
@@ -59,10 +60,13 @@ def _render_judge_settings(config: dict[str, Any]) -> None:
     """Render judge model settings section."""
     st.markdown(f'<div class="section-header">{t("arena.sidebar.judge_model")}</div>', unsafe_allow_html=True)
 
+    provider_options = list(DEFAULT_API_ENDPOINTS.keys())
+    if "arena_judge_provider" not in st.session_state:
+        st.session_state["arena_judge_provider"] = provider_options[0]
+
     endpoint_choice = st.selectbox(
         t("api.provider"),
-        options=list(DEFAULT_API_ENDPOINTS.keys()),
-        index=0,
+        options=provider_options,
         help=t("arena.sidebar.judge_provider_help"),
         key="arena_judge_provider",
     )
@@ -90,14 +94,25 @@ def _render_judge_settings(config: dict[str, Any]) -> None:
     else:
         st.warning(t("api.key_required"))
 
+    # Use stable value for custom option to survive UI language switch
+    CUSTOM_VALUE = "_custom_"
+    model_options = DEFAULT_MODELS + [CUSTOM_VALUE]
+
+    def format_model_option(x: str) -> str:
+        return t("model.custom") if x == CUSTOM_VALUE else x
+
+    # Initialize default value in session state if not exists
+    if "arena_judge_model_value" not in st.session_state:
+        st.session_state["arena_judge_model_value"] = DEFAULT_MODELS[0] if DEFAULT_MODELS else CUSTOM_VALUE
+
     model_option = st.selectbox(
         t("model.select"),
-        options=DEFAULT_MODELS + [t("model.custom")],
-        index=0,
-        key="arena_judge_model",
+        options=model_options,
+        format_func=format_model_option,
+        key="arena_judge_model_value",
     )
 
-    if model_option == t("model.custom"):
+    if model_option == CUSTOM_VALUE:
         model_name = st.text_input(
             t("model.custom_input"),
             placeholder=t("model.custom_placeholder"),

ui/features/auto_rubric/__init__.py

@@ -0,0 +1,15 @@
+# -*- coding: utf-8 -*-
+"""Auto Rubric feature module for OpenJudge Studio.
+
+This feature provides automatic rubric generation from task descriptions
+or labeled data, eliminating manual rubric design.
+
+Phase 1 implements:
+- Simple Rubric mode (zero-shot from task description)
+- Basic result display
+- Export functionality (Python/YAML/JSON)
+"""
+
+from features.auto_rubric.feature import AutoRubricFeature
+
+__all__ = ["AutoRubricFeature"]

ui/features/auto_rubric/components/__init__.py

@@ -0,0 +1,36 @@
+# -*- coding: utf-8 -*-
+"""UI components for Auto Rubric feature."""
+
+from features.auto_rubric.components.data_upload_panel import render_data_upload_panel
+from features.auto_rubric.components.history_panel import (
+    render_history_panel,
+    render_task_detail,
+)
+from features.auto_rubric.components.iterative_config_panel import (
+    render_iterative_config_panel,
+    validate_iterative_config,
+)
+from features.auto_rubric.components.result_panel import render_result_panel
+from features.auto_rubric.components.rubric_tester import (
+    render_test_panel,
+    render_test_section_compact,
+)
+from features.auto_rubric.components.sidebar import render_rubric_sidebar
+from features.auto_rubric.components.simple_config_panel import (
+    render_simple_config_panel,
+    validate_simple_config,
+)
+
+__all__ = [
+    "render_rubric_sidebar",
+    "render_simple_config_panel",
+    "validate_simple_config",
+    "render_iterative_config_panel",
+    "validate_iterative_config",
+    "render_data_upload_panel",
+    "render_result_panel",
+    "render_history_panel",
+    "render_task_detail",
+    "render_test_panel",
+    "render_test_section_compact",
+]