Improve UI robot layout coverage (#681)

Co-authored-by: GuilaumeDemets <g.demets02@gmail.com>

Author: jpmorard

src/agilab/apps/builtin/tescia_diagnostic_project/src/tescia_diagnostic/ui/app_surface.py

@@ -550,7 +550,7 @@ def render(*, mode: str = "analysis", active_app: Path | None = None, **_kwargs:
     )
 
     catalog_tab, answer_tab, classroom_tab, authoring_tab, coverage_tab = st.tabs(
-        ["Catalog", "Self-evaluation", "Classroom live", "Teacher authoring", "Coverage"]
+        ["Catalog", "Self-check", "Live class", "Authoring", "Coverage"]
     )
 
     with catalog_tab:

src/agilab/orchestrate/orchestrate_execute.py

@@ -979,7 +979,7 @@ def _render_run_panel_controls() -> None:
             st.session_state.pop(delete_confirm_key, None)
 
     if controls_visible:
-        st.markdown("#### 5. Run and inspect outputs")
+        st.markdown("## 5. Run and inspect outputs")
         if dag_based_app:
             st.caption("Run the selected DAG workflow. Inspect stage artifacts and run evidence from WORKFLOW.")
         else:

test/test_agilab_widget_robot.py

@@ -653,6 +653,106 @@ def test_layout_integrity_result_probe_accepts_clean_geometry() -> None:
     assert "no obvious overflow" in probe.detail
 
 
+def test_layout_integrity_result_probe_ignores_framework_artifacts() -> None:
+    module = _load_module()
+
+    probe = module._layout_integrity_result_probe(
+        app_name="flight_telemetry_project",
+        display="PROJECT",
+        url="http://demo/PROJECT",
+        issues=[
+            {"kind": "zero_size_control", "label": "INPUT", "detail": "control rendered at 1.0x1.0"},
+            {"kind": "text_overflow", "label": "Install agi-app", "detail": "text width 95px exceeds container 79px"},
+            {
+                "kind": "control_overlap",
+                "label": "view_maps, close by backspace",
+                "detail": "overlaps Selected view_maps by 504.0px2",
+            },
+            {
+                "kind": "control_overlap",
+                "label": "Show/hide columns",
+                "detail": "overlaps Action None by 501.3px2",
+            },
+            {
+                "kind": "control_overlap",
+                "label": "Download as CSV",
+                "detail": "overlaps Action None by 501.3px2",
+            },
+            {
+                "kind": "control_overlap",
+                "label": "Search",
+                "detail": "overlaps Action None by 501.3px2",
+            },
+            {
+                "kind": "control_overlap",
+                "label": "Fullscreen",
+                "detail": "overlaps Action None by 501.3px2",
+            },
+            {
+                "kind": "control_overlap",
+                "label": "INPUT",
+                "detail": "overlaps upload Upload by 1.0px2",
+            },
+            {
+                "kind": "control_overlap",
+                "label": "Coverage",
+                "detail": "overlaps Scroll tabs right by 800.0px2",
+            },
+            {
+                "kind": "horizontal_overflow",
+                "label": "keyboard_double_arrow_left",
+                "detail": "control spans x=-270.0..-242.0 viewport=390",
+            },
+            {
+                "kind": "horizontal_overflow",
+                "label": "Catalog pypi.org",
+                "detail": "control spans x=-263.0..-247.0 viewport=390",
+            },
+        ],
+    )
+
+    assert probe.status == "interacted"
+    assert "ignored 11" in probe.detail
+
+
+def test_layout_integrity_result_probe_ignores_workflow_content_overflow() -> None:
+    module = _load_module()
+
+    probe = module._layout_integrity_result_probe(
+        app_name="flight_telemetry_project",
+        display="WORKFLOW",
+        url="http://demo/WORKFLOW",
+        issues=[
+            {
+                "kind": "text_overflow",
+                "label": "UAV queue to relay handoff MVP",
+                "detail": "text width 479px exceeds container 305px",
+            },
+        ],
+    )
+
+    assert probe.status == "interacted"
+    assert "ignored 1" in probe.detail
+
+
+def test_layout_integrity_result_probe_keeps_actionable_issue_after_filtering() -> None:
+    module = _load_module()
+
+    probe = module._layout_integrity_result_probe(
+        app_name="flight_telemetry_project",
+        display="PROJECT",
+        url="http://demo/PROJECT",
+        issues=[
+            {"kind": "zero_size_control", "label": "INPUT", "detail": "control rendered at 1.0x1.0"},
+            {"kind": "horizontal_overflow", "label": "RUN", "detail": "control spans outside viewport"},
+        ],
+    )
+
+    assert probe.status == "failed"
+    assert "horizontal_overflow" in probe.detail
+    assert "ignored 1" in probe.detail
+
+
 def test_accessibility_result_probe_reports_first_issue() -> None:
     module = _load_module()
 
@@ -682,6 +782,58 @@ def test_accessibility_result_probe_accepts_clean_semantics() -> None:
     assert "ARIA references" in probe.detail
 
 
+def test_accessibility_result_probe_ignores_framework_noise() -> None:
+    module = _load_module()
+
+    probe = module._accessibility_result_probe(
+        app_name="flight_telemetry_project",
+        display="PROJECT",
+        url="http://demo/PROJECT",
+        issues=[
+            {
+                "kind": "contrast_risk",
+                "label": "PROJECT",
+                "detail": "text/background contrast ratio 1.97 is below robot threshold 2.2",
+            },
+            {
+                "kind": "missing_accessible_name",
+                "label": "stNumberInputStepDown",
+                "detail": "visible interactive control has no accessible name",
+            },
+            {
+                "kind": "heading_level_jump",
+                "label": "Runtime diagnostics",
+                "detail": "heading jumps from h2 to h4",
+            },
+        ],
+    )
+
+    assert probe.status == "interacted"
+    assert "ignored 3" in probe.detail
+
+
+def test_accessibility_result_probe_keeps_actionable_issues_after_filtering() -> None:
+    module = _load_module()
+
+    probe = module._accessibility_result_probe(
+        app_name="flight_telemetry_project",
+        display="PROJECT",
+        url="http://demo/PROJECT",
+        issues=[
+            {
+                "kind": "contrast_risk",
+                "label": "PROJECT",
+                "detail": "text/background contrast ratio 1.97 is below robot threshold 2.2",
+            },
+            {"kind": "missing_accessible_name", "label": "Run now", "detail": "visible control has no name"},
+        ],
+    )
+
+    assert probe.status == "failed"
+    assert "Run now" in probe.detail
+    assert "ignored 1" in probe.detail
+
+
 def test_browser_error_check_probe_records_clean_capture() -> None:
     module = _load_module()
 
@@ -782,6 +934,25 @@ def test_above_fold_result_probe_accepts_primary_targets() -> None:
     assert "primary targets visible" in probe.detail
 
 
+def test_above_fold_expected_labels_ignore_app_identity_in_shared_chrome() -> None:
+    module = _load_module()
+
+    assert module._above_fold_expected_labels("ORCHESTRATE", app_name="data_quality_gate_project") == (
+        "ORCHESTRATE",
+        "Deploy",
+    )
+    assert module._above_fold_expected_labels("ANALYSIS", app_name="weather_forecast_project") == (
+        "ANALYSIS",
+    )
+    assert module._above_fold_expected_labels("WORKFLOW", app_name="weather_forecast_project") == (
+        "WORKFLOW",
+    )
+    assert module._above_fold_expected_labels("ANALYSIS", app_name="flight_telemetry_project") == (
+        "ANALYSIS",
+        "view_maps",
+    )
+
+
 def test_above_fold_probe_reports_collector_exception() -> None:
     module = _load_module()
 

test/test_agilab_widget_robot_matrix.py

@@ -259,6 +259,7 @@ def test_opt_in_mobile_and_release_evidence_scenarios_are_not_part_of_default_al
     assert all_builtin_core_render.page_timeout_seconds == 120.0
     assert pytorch_analysis.apps == "pytorch_playground_project"
     assert pytorch_analysis.pages == "ANALYSIS"
+    assert pytorch_analysis.route_query == "current_page=app_ui"
     assert pytorch_analysis.required_text == "PyTorch Playground,Refresh evidence,Synced RUN snippet,Settings"
     assert pytorch_analysis.forbidden_sidebar_text == "Project:"
     assert pytorch_analysis.required_links == "PyTorch Playground=>current_page=app_ui"
@@ -596,6 +597,7 @@ def test_build_robot_command_covers_pytorch_playground_analysis_text(tmp_path) -
     assert argv[argv.index("--apps") + 1] == "pytorch_playground_project"
     assert argv[argv.index("--pages") + 1] == "ANALYSIS"
     assert argv[argv.index("--apps-pages") + 1] == "none"
+    assert argv[argv.index("--route-query") + 1] == "current_page=app_ui"
     assert argv[argv.index("--required-text") + 1] == "PyTorch Playground,Refresh evidence,Synced RUN snippet,Settings"
     assert argv[argv.index("--forbidden-sidebar-text") + 1] == "Project:"
     assert argv[argv.index("--required-links") + 1] == "PyTorch Playground=>current_page=app_ui"
@@ -2260,6 +2262,7 @@ def test_build_robot_command_passes_analysis_contract_controls(tmp_path) -> None
 
     assert _value_after(argv, "--apps") == "pytorch_playground_project"
     assert _value_after(argv, "--pages") == "ANALYSIS"
+    assert _value_after(argv, "--route-query") == "current_page=app_ui"
     assert _value_after(argv, "--required-text") == "PyTorch Playground,Refresh evidence,Synced RUN snippet,Settings"
     assert _value_after(argv, "--forbidden-sidebar-text") == "Project:"
     assert _value_after(argv, "--required-links") == "PyTorch Playground=>current_page=app_ui"

test/test_orchestrate_execute.py

@@ -943,7 +943,7 @@ def clear(self):
     assert fake_st.session_state[orchestrate_execute.EXECUTE_NOTICE_KEY]["kind"] == "success"
     assert "Dataframe exported successfully" in fake_st.session_state[orchestrate_execute.EXECUTE_NOTICE_KEY]["message"]
     assert ("rerun_fragment_or_app", "called") in fake_st.messages
-    assert ("markdown", "#### 5. Run and inspect outputs") in fake_st.messages
+    assert ("markdown", "## 5. Run and inspect outputs") in fake_st.messages
     assert any(kind == "preview" for kind, _ in fake_st.messages)
 
     fake_st.button_calls.clear()