Project

General

Profile

Actions

Feature #26

closed
VJ

Epic #20: Client design adaptation

Adapt Guides filtering to the shared interaction

Feature #26: Adapt Guides filtering to the shared interaction

Added by Vince Jurić 25 days ago.

Status:
Done
Priority:
Normal
Assignee:
-
Start date:
Due date:
% Done:

0%

Estimated time:

Description

Imported from GitHub issue 268.
Source milestone: Redesign (milestone 3).
Source labels: enhancement, ready-for-agent.
Original state: closed.

What to build

Type: AFK

Deliver Guides sub-section filtering through the same toolbar, draft/apply sheet and empty-results interaction as Foods and Recipes. Replace immediate inline sub-section chips, retaining the release-defined guide sub-sections and existing content-access behaviour.

Use only the Design and Design tokens pages of https://www.figma.com/design/yisgB7iNKYtjNZdogARGIN/Bebin-prvi-meni. Exclude the reference-app page entirely. The food sheet is the visual source; recipe and guide sheets adapt its shared pattern while preserving their domain rules.

Acceptance criteria

  • Search remains visible and debounced; search combines with applied filter groups using AND semantics.
  • Opening Filter copies applied selections into a draft. Editing a draft updates its preview count without changing catalogue results.
  • Apply commits the draft, closes the sheet, and resets pagination and scroll position. Close, backdrop dismissal, Android Back and any supported drag dismissal discard pending changes.
  • Reset clears draft selections and requires Apply to commit. A separate clear-all action clears search, applied selections and any removable context filters; it is hidden when nothing is active.
  • The Filter button indicates the number of applied selections, excluding search; each selected category or non-default age/sub-section counts as one selection.
  • Counts include all matching public catalogue records, including locked previews, rather than only loaded result pages. Count predicates match list predicates, reuse the content cache, and never fetch protected details.
  • Count loading and failures have explicit states and retry; no partial or failed count is presented as an exact total. Stale list/count responses cannot overwrite newer selections, locale or content revisions.
  • Serbian Latin and English fixed copy is complete; content labels, sorting and search preserve locale fallback rules.
  • Back-navigation from details and access flows preserves list context. Section re-entry and active-tab reselection reset temporary state according to the domain rules; no cross-session persistence is introduced.
  • No-content and no-results states remain distinct; no-results provides a direct clear-all action.
  • Automated tests cover draft/apply/cancel/reset, matching/count consistency, stale requests and pagination where applicable. Client tests, lint and typecheck pass.
  • Guide sub-sections remain single-select; no selection means all guides. Use the canonical release-defined sub-section IDs and fixed localized labels rather than inventing tags or food groups.
  • Counts use the same search/sub-section predicate as guide results and include locked previews without fetching protected bodies.
  • Search-only, sub-section-only and combined no-results states offer clear-all recovery consistently.
  • Android ADB verification covers opening, selection, apply, reset, cancellation, long labels, empty results, locked previews and return navigation; record evidence.

Blocked by

Acceptance verification

All 15 criteria checked against PR GitHub 274 at e82ec48 and the recorded implementation, automated and native evidence. The PR remains open; this checklist records verified implementation rather than a merge or release.

  • Current acceptance report and original issue-specific device evidence.
  • Client: 466 tests in 63 files, typecheck and lint passed. Contracts: 57 tests in two files passed. Android build/install and ADB verification passed. These are the recorded checks for this commit; no new CI run is implied.
  • ADB verified localised ordering, exact counts, applied context, foreground scroll retention and access-screen return. Sign-in/Premium screens were entered using deep links while retaining the existing Premium account, then dismissed with Cancel/Android Back. OAuth provider-picker cancellation, purchase-sheet cancellation and a real guest account session are outside that native evidence; locked-preview destinations/badges also have automated and fixture coverage.
  • iOS acceptance remains tracked separately in GitHub 269.

No data to display

Actions

Also available in: PDF Atom