Normalize wizard button roles

[ebelo]

Summary

• Normalize wizard action-row helpers around explicit primary, secondary, and destructive roles.
• Keep secondary actions on native Qt styling instead of qfit outline/grey chrome.
• Style Clear local database… as a separated destructive action, never as primary/green.
• Document the shared helper contract in the design system.

Closes #793.

Validation

• python3 -m pytest tests/test_wizard_action_row.py tests/test_sync_page.py tests/test_map_page.py tests/test_analysis_page.py tests/test_atlas_page.py tests/test_connection_page.py -q
• python3 -m pytest tests/ -x -q --tb=short
• git diff --check

Screenshot validation

Before:

After:

[greptile-apps]

Greptile Summary

This PR normalises wizard button roles by refactoring _button_stylesheet from a boolean primary flag to a string role ("primary", "secondary", "destructive"), adding a style_destructive_action_button helper with red/danger-tone chrome, and reclassifying clear_button in SyncPageContent accordingly. Secondary buttons intentionally receive an empty stylesheet so native Qt rendering takes over; only primary and destructive actions carry qfit-specific chrome.

Confidence Score: 4/5

Safe to merge; all changes are well-tested and the logic is straightforward.

Only P2 style observations (silent fallthrough for unknown roles, secondary buttons receiving non-stylesheet chrome); no P0 or P1 issues found.

ui/dockwidget/action_row.py — the _button_stylesheet fallthrough and _apply_button_chrome unconditional side effects are worth a second look.

Important Files Changed

Filename	Overview
ui/dockwidget/action_row.py	Refactors _button_stylesheet from a bool primary flag to a string role; adds style_destructive_action_button and danger-tone hover chrome; secondary role intentionally returns empty stylesheet for native Qt appearance.
ui/dockwidget/sync_page.py	Swaps clear_button from style_secondary_action_button to style_destructive_action_button; imports and layout unchanged.
tests/test_wizard_action_row.py	Updates secondary-button test to assert empty styleSheet(); adds new test_destructive_action_button_gets_danger_role_and_chrome covering #c01c28 presence and #589632 absence.
tests/test_sync_page.py	Replaces secondaryAction/"secondary" assertions with destructiveAction/"destructive" for clear_button; adds stylesheet inequality and green-colour exclusion guards.
docs/design-system.md	Documents the three shared wizard-button helpers and the intentional native-Qt treatment for secondary buttons.

Flowchart

%%{init: {'theme': 'neutral'}}%%
flowchart TD
    A[style_primary_action_button] -->|role=primary| C[_apply_button_chrome]
    B[style_secondary_action_button] -->|role=secondary| C
    D[style_destructive_action_button] -->|role=destructive| C

    C --> E[setToolButtonStyle ToolButtonTextBesideIcon]
    C --> F[_allow_button_shrink]
    C --> G[setCursor PointingHandCursor]
    C --> H[setStyleSheet _button_stylesheet role]

    H -->|role == primary| I[Green accent font-weight 700]
    H -->|role == destructive| J[Transparent bg Red text + danger hover font-weight 700]
    H -->|role == secondary| K[Empty string native Qt styling]

Loading

_{Reviews (1): Last reviewed commit: "Normalize wizard button roles" | Re-trigger Greptile}

[greptile-apps]

Silent fallthrough for unknown roles

_button_stylesheet returns "" for any string that is neither "primary" nor "destructive" — including the "secondary" role by design, but also any future typo or misuse. Since this is a private function only reachable through the three public style_* helpers the risk is low today, but an explicit elif role == "secondary": return "" (or a final else: raise ValueError(...)) would make the contract self-documenting and catch mistakes at the call site instead of silently producing an unstyled button.

[greptile-apps]

setToolButtonStyle and cursor applied to secondary buttons

_apply_button_chrome unconditionally calls setToolButtonStyle(Qt.ToolButtonTextBesideIcon) and setCursor(Qt.PointingHandCursor) for all roles, including "secondary". The PR description says secondary buttons should keep native Qt styling, but these two widget-level properties are not stylesheet-based and won't be reset by the empty setStyleSheet(""). Secondary buttons will still end up with ToolButtonTextBesideIcon and a pointer cursor even though they aren't receiving qfit chrome. If fully native secondary buttons are the goal, consider gating those two calls on role != "secondary".

[ebelo]

Addressed the first Greptile P2 in 39e92c4: _button_stylesheet now explicitly handles secondary and raises ValueError for unknown roles, with regression coverage for a typoed role.

For the second P2: I kept setToolButtonStyle(Qt.ToolButtonTextBesideIcon) and the pointing-hand cursor on secondary buttons intentionally. The goal is native visual styling for secondary actions (empty stylesheet, no qfit outline/green chrome), while preserving readable text on QToolButton and qfit's existing interaction convention that clickable action controls use a pointing cursor. The screenshot validation confirms the secondary buttons render with neutral/native visual styling while the destructive action is distinct and non-green.

[sonarqubecloud]

Quality Gate passed

Issues
0 New issues
0 Accepted issues

Measures
0 Security Hotspots
100.0% Coverage on New Code
0.0% Duplication on New Code

See analysis details on SonarQube Cloud