fix(swift): resolve seven reported iOS defects across weather, packs, trips, and catalog search - #2697
Merged
Merged
Conversation
- #2693 Weather: omit the search-state row when empty. The row rendered unconditionally, reserving inset-grouped height and padding, which showed as a large gap between the search field and Saved Locations. - #2690 Pack Details: read weight totals from `currentPack` rather than the captured `pack`. The latter is a snapshot from push time, so the summary kept showing weights for items the live pack no longer had. - #2692 Pack Item Details: make Similar Gear cards tappable, presenting CatalogItemDetailView via the same contentShape/onTapGesture/sheet pattern CatalogItemRow already uses. The cards previously had no action at all. - #2691 Packs list: drop the List selection binding on compact iOS, where navigation is a NavigationLink push. Tracking selection there left the tapped row rendered gray after popping back. Split-view keeps selection. - #2661 Pack Item Details: re-read the item from the view model instead of rendering the construction-time snapshot, so the weight shown right after Save reflects the edit. Edit re-open now prefills from the live item too. - #2660 Trip edit: encode `packId` unconditionally on update, as explicit null when "None" is chosen. Synthesised Encodable omits nil keys, and the route keys off `if ('packId' in data)`, so unassigning never persisted.
Searching "Tent" surfaced leggings, tote bags and repair kits above actual
tents. Two independent causes:
1. Unanchored `ilike '%q%'` matched the letters anywhere, including inside
unrelated words — "consistent", "content", "intent", "patented", "latent"
are common in gear marketing copy, and `description` was matched with the
same weight as `name`. Free-text fields (name, description, categories)
now require a POSIX word-boundary match, so incidental hits inside longer
words no longer qualify. Brand and model keep substring matching, where
typing a prefix is the normal way to search.
2. Nothing ranked by the query at all — ordering was purely by how often an
item had been added to packs, so a popular incidental match outranked an
exact-name tent. Matches are now tiered by where they hit (exact name >
name prefix > name word > category > brand/model > description-only),
with popularity kept as a tiebreaker within a tier.
The query is interpolated into a regex operator, so metacharacters in `q`
are escaped — otherwise a stray "(" would error the statement.
An explicit caller-supplied `sort` still overrides relevance, and the
`desc(id)` pagination tiebreaker is preserved on every branch.
Verified against Postgres with fixtures drawn from the issue: the leggings
and tote-bag decoys drop out of the result set entirely, and the remaining
matches order Tent > Tent Footprint > Copper Spur UL2 Tent > Fabric Repair
Kit (description-only mention, last).
Contributor
WalkthroughThe PR updates Swift pack, weather, and trip request behavior. It also changes catalog search matching and relevance ordering, with regression tests for search boundaries, ranking, brand/model matching, and literal regex characters. ChangesSwift application updates
Catalog search updates
Estimated code review effort: 3 (Moderate) | ~25 minutes Possibly related PRs
Suggested labels: Suggested reviewers: 🚥 Pre-merge checks | ✅ 5✅ Passed checks (5 passed)
✨ Finishing Touches 💡 1🛠️ Fix failing CI checks 💡
📝 Generate docstrings
🧪 Generate unit tests (beta)
Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out. Comment |
Contributor
Coverage Report for packages/units (./packages/units)
File CoverageNo changed files found. |
Contributor
Coverage Report for packages/utils (./packages/utils)
File CoverageNo changed files found. |
Contributor
Coverage Report for packages/overpass (./packages/overpass)
File CoverageNo changed files found. |
Contributor
Coverage Report for packages/analytics (./packages/analytics)
File CoverageNo changed files found. |
Contributor
Coverage Report for packages/mcp (./packages/mcp)
File CoverageNo changed files found. |
Contributor
Coverage Report for apps/expo (./apps/expo)
File CoverageNo changed files found. |
Contributor
Coverage Report for packages/api (./packages/api)
File CoverageNo changed files found. |
![coderabbitai[bot]](https://avatars.githubusercontent.com/in/347564?s=60&v=4){kind=small}
Contributor
There was a problem hiding this comment.
Actionable comments posted: 4
🤖 Prompt for all review comments with AI agents
Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.
Inline comments:
In `@apps/swift/Sources/PackRat/Features/Packs/PackItemDetailView.swift`:
- Around line 315-323: Replace the onTapGesture interaction in the similar-item
card’s body with a Button that sets showingDetail to true, preserve the card
content and accessibility configuration, and apply .buttonStyle(.plain) to
retain the current visual appearance.
In `@apps/swift/Sources/PackRat/Models/Trip.swift`:
- Around line 41-57: Update Trip and TripService.updateTrip to represent packId
with distinct omitted, assigned, and explicitly cleared states instead of a
plain String?. Adjust encode(to:) to omit packId for the omitted state while
encoding a value or null for the other states, and add serialization tests
covering all three cases.
In `@packages/api/src/services/catalogService.ts`:
- Around line 134-152: Update the search flow around q and term so
whitespace-only input is ignored: trim q before the conditional branch, and only
construct the regex, ILIKE predicates, and searchCondition when the trimmed term
is non-empty. Preserve the no-query result behavior for empty or whitespace-only
terms.
In `@packages/api/test/catalog.test.ts`:
- Around line 166-189: Extend the “still matches brand and model by substring”
test with a fixture whose model, not brand, contains the query substring, and
assert that it is returned. Update the regex-metacharacter test to query an
encoded word-ending value such as “t.ent”, seed both a literal “T.ent” item and
a “Tent” item, and assert the literal item matches while the non-literal item
does not, while preserving the successful response assertion.
🪄 Autofix
Fix all unresolved CodeRabbit comments on this PR:
- Push a commit to this branch (recommended)
- Create a new PR with the fixes
ℹ️ Review info
⚙️ Run configuration
Configuration used: Path: .coderabbit.yaml
Review profile: ASSERTIVE
Plan: Pro Plus
Run ID: 30df1b57-f34f-4ff4-a96e-e6fa5b10f359
📒 Files selected for processing (7)
apps/swift/Sources/PackRat/Features/Packs/PackDetailView.swiftapps/swift/Sources/PackRat/Features/Packs/PackItemDetailView.swiftapps/swift/Sources/PackRat/Features/Packs/PacksListView.swiftapps/swift/Sources/PackRat/Features/Weather/WeatherView.swiftapps/swift/Sources/PackRat/Models/Trip.swiftpackages/api/src/services/catalogService.tspackages/api/test/catalog.test.ts
Comment on lines
315
to
+323
| var body: some View { | ||
| cardContent | ||
| .contentShape(Rectangle()) | ||
| .onTapGesture { showingDetail = true } | ||
| .accessibilityIdentifier("pack_item_similar_card_\(item.id)") | ||
| .accessibilityAddTraits(.isButton) | ||
| .sheet(isPresented: $showingDetail) { | ||
| CatalogItemDetailView(item: item, packsViewModel: packsViewModel) | ||
| } |
Contributor
There was a problem hiding this comment.
🎯 Functional Correctness | 🟠 Major | ⚡ Quick win
🧩 Analysis chain
🏁 Script executed:
#!/bin/bash
set -eu
file="apps/swift/Sources/PackRat/Features/Packs/PackItemDetailView.swift"
printf '%s\n' '--- target and file references ---'
rg -n "SimilarItemCard|onTapGesture|accessibilityAddTraits|CatalogItemDetailView|platform|macOS|iOS" \
apps/swift/Sources apps/swift/Package.swift apps/swift/*.xcodeproj 2>/dev/null || true
printf '%s\n' '--- relevant source ---'
sed -n '270,340p' "$file"
printf '%s\n' '--- package/project platform declarations ---'
rg -n -i "macos|ios|deploymentTarget|supportedplatform|swift-tools-version" \
apps/swift/Package.swift apps/swift/*.pbxproj apps/swift/*.xcodeproj/project.pbxproj 2>/dev/null || true
printf '%s\n' '--- related interaction patterns ---'
rg -n -g '*.swift' "Button\s*\{|\.onTapGesture|accessibilityAddTraits\(\.isButton\)|\.buttonStyle\(\.plain\)" apps/swift/Sources apps/swift/Tests 2>/dev/null | head -200Repository: PackRat-AI/PackRat
Length of output: 50374
🏁 Script executed:
#!/bin/bash
set -eu
python3 - <<'PY'
from pathlib import Path
import re
root = Path("apps/swift")
for path in root.rglob("*.swift"):
text = path.read_text(errors="replace")
if "SimilarItemCard" not in text:
continue
print(f"FILE {path}")
for match in re.finditer(r"SimilarItemCard", text):
start = max(0, text.rfind("\n", 0, match.start()-250))
end = text.find("\n", match.end()+400)
end = len(text) if end < 0 else end
print(text[start+1:end+1])
PYRepository: PackRat-AI/PackRat
Length of output: 1641
🌐 Web query:
Apple SwiftUI documentation onTapGesture Button keyboard activation accessibility macOS iOS
💡 Result:
In SwiftUI, using a Button is the recommended practice for interactive elements because it provides essential accessibility and keyboard activation capabilities out of the box [1][2][3][4]. Using.onTapGesture on non-button views (like Text or Image) is generally considered a significant accessibility anti-pattern because these views lack the necessary traits and behaviors required for assistive technologies and keyboard navigation [2][4][5]. Key differences include: 1. Keyboard Activation and Focus: Native SwiftUI Button elements are keyboard-accessible by default [6]. On macOS and platforms supporting Full Keyboard Access, a Button automatically receives focus and can be activated using standard keyboard inputs (such as the Space or Enter/Return keys) [6][7]. Conversely, views using.onTapGesture do not automatically receive keyboard focus and cannot be activated via standard keyboard navigation [6][2]. 2. Accessibility Traits: A Button is automatically identified by VoiceOver as a "button," informing users that the element is interactive [8][2][5]. Views with.onTapGesture do not carry this trait by default; to assistive technologies, they appear as static content [2][5]. If you must use.onTapGesture for a custom interactive view, you are required to manually add the accessibility trait using.accessibilityAddTraits(.isButton) and provide an appropriate accessibility label [6][8][2]. 3. User Experience: Button provides automatic visual feedback (such as a pressed state) that helps users understand when an interaction has been recognized [9][4]. It also handles platform-specific interaction models—like clicks on macOS or taps on iOS—consistently [9][7]. Apple explicitly advises that if you are creating a control functionally equivalent to a button, you should use the Button component, utilizing ButtonStyle to achieve custom designs rather than applying gestures to arbitrary views [9]. [6][8][1][2][9][4][7][5]
Citations:
- 1: https://forums.swift.org/t/text-with-ontapgesture-instead-of-button/58629
- 2: https://swiftcrafted.dev/article/swiftui-accessibility-complete-guide-voiceover-dynamic-type-inclusive-design
- 3: https://david.y4ng.fr/tap-gesture-vs-button/
- 4: https://www.swiftwithvincent.com/blog/be-careful-when-using-ontapgesture
- 5: https://blog.lars-richter.dev/multiple-buttons-in-a-swiftui-list-element/
- 6: https://github.com/pasqualevittoriosi/swift-accessibility-skill/blob/main/swift-accessibility-skill/references/motor-input.md
- 7: https://apple-docs.everest.mt/docs/swiftui/button/
- 8: https://realm.github.io/SwiftLint/accessibility_trait_for_button.html
- 9: https://stackoverflow.com/questions/61604277/difference-between-buttons-and-ontapgesture
Use Button for the similar-item card.
onTapGesture does not provide native keyboard activation or button control behavior. Replace it with a Button and retain .buttonStyle(.plain).
🤖 Prompt for AI Agents
Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.
In `@apps/swift/Sources/PackRat/Features/Packs/PackItemDetailView.swift` around
lines 315 - 323, Replace the onTapGesture interaction in the similar-item card’s
body with a Button that sets showingDetail to true, preserve the card content
and accessibility configuration, and apply .buttonStyle(.plain) to retain the
current visual appearance.
Comment on lines
+41
to
+57
| /// `packId` is encoded unconditionally — as an explicit `null` when the user | ||
| /// picks "None" — because the update route distinguishes the two cases with | ||
| /// `if ('packId' in data)`. Synthesised `Encodable` omits nil keys entirely, | ||
| /// which the server read as "leave unchanged", so unassigning a pack never | ||
| /// saved. Every other field keeps omit-when-nil so a partial update does not | ||
| /// clobber fields the form did not touch. | ||
| func encode(to encoder: any Encoder) throws { | ||
| var container = encoder.container(keyedBy: CodingKeys.self) | ||
| try container.encodeIfPresent(name, forKey: .name) | ||
| try container.encodeIfPresent(description, forKey: .description) | ||
| try container.encodeIfPresent(location, forKey: .location) | ||
| try container.encodeIfPresent(startDate, forKey: .startDate) | ||
| try container.encodeIfPresent(endDate, forKey: .endDate) | ||
| try container.encodeIfPresent(notes, forKey: .notes) | ||
| try container.encode(packId, forKey: .packId) | ||
| try container.encode(localUpdatedAt, forKey: .localUpdatedAt) | ||
| } |
Contributor
There was a problem hiding this comment.
🗄️ Data Integrity & Integration | 🟠 Major | 🏗️ Heavy lift
🧩 Analysis chain
🏁 Script executed:
#!/usr/bin/env bash
set -euo pipefail
rg -n -C 6 'UpdateTripRequest|\.updateTrip\s*\(' \
apps/swift/Sources/PackRat --glob '*.swift'Repository: PackRat-AI/PackRat
Length of output: 6578
🏁 Script executed:
#!/usr/bin/env bash
set -euo pipefail
printf '%s\n' '--- Trip.swift ---'
sed -n '1,90p' apps/swift/Sources/PackRat/Models/Trip.swift
printf '%s\n' '--- TripService.swift ---'
sed -n '35,75p' apps/swift/Sources/PackRat/Services/TripService.swift
printf '%s\n' '--- TripFormView.swift ---'
sed -n '175,215p' apps/swift/Sources/PackRat/Features/Trips/TripFormView.swift
printf '%s\n' '--- TripsViewModel.swift ---'
sed -n '205,255p' apps/swift/Sources/PackRat/Features/Trips/TripsViewModel.swift
printf '%s\n' '--- OutboxService.swift ---'
sed -n '315,355p' apps/swift/Sources/PackRat/Services/OutboxService.swift
printf '%s\n' '--- API route references ---'
rg -n -C 8 "packId|UpdateTrip|updateTrip" packages/api/src/routes/trips/index.tsRepository: PackRat-AI/PackRat
Length of output: 12084
🏁 Script executed:
#!/usr/bin/env bash
set -euo pipefail
printf '%s\n' '--- all updateTrip call sites ---'
rg -n -C 5 '\bupdateTrip\s*\(' apps/swift --glob '*.swift'
printf '%s\n' '--- form pack selection state and initialization ---'
rg -n -C 8 'selectedPackId|existingTrip|TripFormView' \
apps/swift/Sources/PackRat/Features/Trips/TripFormView.swift
printf '%s\n' '--- view-model update method ---'
rg -n -A 95 -B 8 'func updateTrip\s*\(' \
apps/swift/Sources/PackRat/Features/Trips/TripsViewModel.swift
printf '%s\n' '--- mutation payload definition and uses ---'
rg -n -C 8 'struct TripMutationPayload|TripMutationPayload' \
apps/swift/Sources/PackRat --glob '*.swift'
printf '%s\n' '--- update schema ---'
rg -n -C 12 'UpdateTripBodySchema' packages --glob '*.{ts,tsx}'Repository: PackRat-AI/PackRat
Length of output: 28711
🏁 Script executed:
#!/usr/bin/env bash
set -euo pipefail
python3 - <<'PY'
from pathlib import Path
trip = Path("apps/swift/Sources/PackRat/Models/Trip.swift").read_text()
service = Path("apps/swift/Sources/PackRat/Services/TripService.swift").read_text()
form = Path("apps/swift/Sources/PackRat/Features/Trips/TripFormView.swift").read_text()
route = Path("packages/api/src/routes/trips/index.ts").read_text()
assert "let packId: String?" in trip
assert "try container.encode(packId, forKey: .packId)" in trip
assert "packId: String? = nil" in service
assert "packId: selectedPackId" in form
assert "if ('packId' in data) updateData.packId = data.packId ?? null;" in route
# Model the two relevant API states produced by this request type.
def encoded_pack_id(pack_id):
return {"packId": pack_id} # encode(Optional.none) writes JSON null
for value, expected in [("pack-1", "set"), (None, "explicit-clear")]:
payload = encoded_pack_id(value)
route_action = "update" if "packId" in payload else "leave-unchanged"
if value is None:
assert route_action == "update"
assert expected == "explicit-clear"
# There is no third value in String? that can represent omission.
assert encoded_pack_id(None) == {"packId": None}
print("UpdateTripRequest has no representable omitted packId state.")
print("nil serializes as packId:null, and the API treats that as unassignment.")
PY
printf '%s\n' '--- relevant tests ---'
rg -n -i 'UpdateTripRequest|packId.*null|unassign|TripService.*update' \
apps/swift --glob '*Test*.swift' --glob '*.swift' || trueRepository: PackRat-AI/PackRat
Length of output: 865
Preserve omitted and explicit-clear states for packId.
TripService.updateTrip defaults packId to nil. The custom encoder then sends "packId": null, and the API treats that value as unassignment. A String? cannot represent both omitted and explicit “None” states. Use a three-state representation and add serialization tests for omitted, set, and clear.
🤖 Prompt for AI Agents
Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.
In `@apps/swift/Sources/PackRat/Models/Trip.swift` around lines 41 - 57, Update
Trip and TripService.updateTrip to represent packId with distinct omitted,
assigned, and explicitly cleared states instead of a plain String?. Adjust
encode(to:) to omit packId for the omitted state while encoding a value or null
for the other states, and add serialization tests covering all three cases.
Comment on lines
134
to
152
| if (q) { | ||
| const term = q.trim(); | ||
| // Escape regex metacharacters: `q` is user input and reaches a POSIX | ||
| // regex operator below, where an unescaped `(` or `*` would either error | ||
| // or silently change the match semantics. | ||
| const escaped = term.replace(/[.*+?^${}()|[\]\\]/g, '\\$&'); | ||
| // `\m` / `\M` are POSIX word-boundary escapes (start/end of word). | ||
| const wordBoundary = `\\m${escaped}\\M`; | ||
| const nameWord = sql`${catalogItems.name} ~* ${wordBoundary}`; | ||
| const descriptionWord = sql`COALESCE(${catalogItems.description}, '') ~* ${wordBoundary}`; | ||
| const categoriesWord = sql`${catalogItems.categories}::text ~* ${wordBoundary}`; | ||
|
|
||
| const searchCondition = or( | ||
| ilike(catalogItems.name, `%${q}%`), | ||
| ilike(catalogItems.description, `%${q}%`), | ||
| ilike(catalogItems.brand, `%${q}%`), | ||
| ilike(catalogItems.model, `%${q}%`), | ||
| ilike(sql`${catalogItems.categories}::text`, `%${q}%`), | ||
| nameWord, | ||
| descriptionWord, | ||
| categoriesWord, | ||
| ilike(catalogItems.brand, `%${term}%`), | ||
| ilike(catalogItems.model, `%${term}%`), | ||
| ); |
Contributor
There was a problem hiding this comment.
🎯 Functional Correctness | 🟡 Minor | ⚡ Quick win
Ignore whitespace-only search terms.
Line 134 enters the search branch for q=' '. Line 135 then creates an empty term. The brand and model predicates become ILIKE '%%', so this request applies an unintended filter and relevance ordering instead of using the no-query result set.
Trim before the branch. Only build search conditions when the trimmed term is non-empty.
Proposed fix
- if (q) {
- const term = q.trim();
+ const term = q?.trim();
+ if (term) {📝 Committable suggestion
‼️ IMPORTANT
Carefully review the code before committing. Ensure that it accurately replaces the highlighted code, contains no missing lines, and has no issues with indentation. Thoroughly test & benchmark the code to ensure it meets the requirements.
Suggested change
| if (q) { | |
| const term = q.trim(); | |
| // Escape regex metacharacters: `q` is user input and reaches a POSIX | |
| // regex operator below, where an unescaped `(` or `*` would either error | |
| // or silently change the match semantics. | |
| const escaped = term.replace(/[.*+?^${}()|[\]\\]/g, '\\$&'); | |
| // `\m` / `\M` are POSIX word-boundary escapes (start/end of word). | |
| const wordBoundary = `\\m${escaped}\\M`; | |
| const nameWord = sql`${catalogItems.name} ~* ${wordBoundary}`; | |
| const descriptionWord = sql`COALESCE(${catalogItems.description}, '') ~* ${wordBoundary}`; | |
| const categoriesWord = sql`${catalogItems.categories}::text ~* ${wordBoundary}`; | |
| const searchCondition = or( | |
| ilike(catalogItems.name, `%${q}%`), | |
| ilike(catalogItems.description, `%${q}%`), | |
| ilike(catalogItems.brand, `%${q}%`), | |
| ilike(catalogItems.model, `%${q}%`), | |
| ilike(sql`${catalogItems.categories}::text`, `%${q}%`), | |
| nameWord, | |
| descriptionWord, | |
| categoriesWord, | |
| ilike(catalogItems.brand, `%${term}%`), | |
| ilike(catalogItems.model, `%${term}%`), | |
| ); | |
| const term = q?.trim(); | |
| if (term) { | |
| // Escape regex metacharacters: `q` is user input and reaches a POSIX | |
| // regex operator below, where an unescaped `(` or `*` would either error | |
| // or silently change the match semantics. | |
| const escaped = term.replace(/[.*+?^${}()|[\]\\]/g, '\\$&'); | |
| // `\m` / `\M` are POSIX word-boundary escapes (start/end of word). | |
| const wordBoundary = `\\m${escaped}\\M`; | |
| const nameWord = sql`${catalogItems.name} ~* ${wordBoundary}`; | |
| const descriptionWord = sql`COALESCE(${catalogItems.description}, '') ~* ${wordBoundary}`; | |
| const categoriesWord = sql`${catalogItems.categories}::text ~* ${wordBoundary}`; | |
| const searchCondition = or( | |
| nameWord, | |
| descriptionWord, | |
| categoriesWord, | |
| ilike(catalogItems.brand, `%${term}%`), | |
| ilike(catalogItems.model, `%${term}%`), | |
| ); |
🤖 Prompt for AI Agents
Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.
In `@packages/api/src/services/catalogService.ts` around lines 134 - 152, Update
the search flow around q and term so whitespace-only input is ignored: trim q
before the conditional branch, and only construct the regex, ILIKE predicates,
and searchCondition when the trimmed term is non-empty. Preserve the no-query
result behavior for empty or whitespace-only terms.
Comment on lines
+166
to
+189
| it('still matches brand and model by substring', async () => { | ||
| const item = await seedCatalogItem({ | ||
| name: 'Ultralight Shelter', | ||
| brand: 'Hilleberg', | ||
| description: 'No matching word in here.', | ||
| categories: ['Shelter'], | ||
| }); | ||
|
|
||
| const res = await apiWithAuth('/catalog?q=Hille&limit=100'); | ||
|
|
||
| expect(res.status).toBe(200); | ||
| const data = await expectJsonResponse(res, ['items']); | ||
| const ids = data.items.map((catalogItem: { id: number }) => catalogItem.id); | ||
| expect(ids).toContain(item.id); | ||
| }); | ||
|
|
||
| it('treats regex metacharacters in the query as literal text', async () => { | ||
| const res = await apiWithAuth(`/catalog?q=${encodeURIComponent('tent(')}&limit=10`); | ||
|
|
||
| // An unescaped "(" would be an invalid POSIX regex and error the query. | ||
| expect(res.status).toBe(200); | ||
| await expectJsonResponse(res); | ||
| }); | ||
| }); |
Contributor
There was a problem hiding this comment.
📐 Maintainability & Code Quality | 🟡 Minor | ⚡ Quick win
Test the model branch and literal matching behavior.
Line 166 describes brand and model coverage, but the fixture only matches brand. Add a fixture where only model contains the query substring.
Lines 182-189 only verify that the request succeeds. Keep this error-regression test, but add a literal-match assertion. Use a metacharacter inside a word-ending query, such as t.ent, and assert that a literal T.ent item matches while a Tent item does not.
As per path instructions, tests must “test observable behaviour.”
🤖 Prompt for AI Agents
Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.
In `@packages/api/test/catalog.test.ts` around lines 166 - 189, Extend the “still
matches brand and model by substring” test with a fixture whose model, not
brand, contains the query substring, and assert that it is returned. Update the
regex-metacharacter test to query an encoded word-ending value such as “t.ent”,
seed both a literal “T.ent” item and a “Tent” item, and assert the literal item
matches while the non-literal item does not, while preserving the successful
response assertion.
Source: Path instructions
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Description
Fixes seven reported defects in the Swift app. Six are client-side; one (#2662) is a server-side search-ranking bug that surfaces in the app's Catalog screen.
Closes #2693
Closes #2692
Closes #2691
Closes #2690
Closes #2662
Closes #2661
Closes #2660
Listunconditionally, so it still reserved inset-grouped row height and padding while emptyweightSummaryread the capturedlet pack(a snapshot from push time) while the item list read the livecurrentPackcurrentPack, like the rest of the view already didSimilarItemCardwas a plainVStack— no button, link, or gestureCatalogItemDetailView, reusing thecontentShape/onTapGesture/sheetpatternCatalogItemRowalready usesNavigationLinkand theListhad aselection:binding, so the tapped row stayed selected after popping backnilselection on compact; split-view keeps selection for its detail panePackItemsnapshot it was constructed withEncodableomits nil keys, so nopackIdwas sent; the route keys offif ('packId' in data)and read the absence as "leave unchanged"packIdunconditionally, as an explicitnullwhen "None" is chosenOn #2662
Searching "Tent" returned leggings, tote bags and repair kits above actual tents, for two independent reasons:
ilike '%q%'matched the letters anywhere, including inside unrelated words. "consistent", "content", "intent", "patented" and "latent" are everywhere in gear marketing copy, anddescriptionwas weighted the same asname. Free-text fields (name, description, categories) now require a POSIX word-boundary match. Brand and model keep substring matching, since typing a prefix is the normal way to search those.qis interpolated into a regex operator, so metacharacters are escaped — otherwise a stray(errors the statement. An explicit caller-suppliedsortstill overrides relevance, and thedesc(id)pagination tiebreaker is preserved on every branch.Verified directly against Postgres using fixtures drawn from the issue:
Both decoys drop out of the result set entirely, and the genuine description-only match (repair kit) correctly sorts last.
Type of change
Area(s) affected
apps/expo)packages/api)apps/landing)apps/guides).github/)apps/swift) — not in the template listTesting
packages/api/test/catalog.test.tscovering word-boundary matching, name-over-description ranking, exact-over-partial ranking, brand substring, and regex-metacharacter safetycurlor Postman)What was actually run:
xcodebuildBUILD SUCCEEDED for bothPackRat-iOSandPackRat-macOS(macOS matters here: the iOS Mobile – Pack Card Remains Highlighted After Returning to Packs Screen #2691 fix routes throughisCompact, hardcodedfalsethere)bunx tsc --noEmit— exit 0bunx biome check— clean across 44 filesscripts/lint/no-unprojected-fat-table-queries.ts— clean, no new violationsbun test:swift:scripts— 76/76 passImportant
The five new API tests have never executed, and the reviewer should know why. The
packages/apiintegration harness is broken ondevelopment, independently of this branch: migration0033_social_feed_tables.sqlcreatescomment_likeswithIF NOT EXISTS, then0037_big_archangel.sqlrecreates it without the guard, andtest/vitest.global-setup.tsapplies every.sqlfile blindly. Startup therefore fails withrelation "comment_likes" already existson a clean volume. I confirmed this against pristinedbc84ca0ebefore concluding it wasn't mine.I did not fix it: those files are Drizzle-generated and
CLAUDE.mdforbids hand-editing migrations, so the repair needs a regeneration decision outside this PR's scope. That is why #2662 was validated by running the SQL directly against Postgres instead. The new tests will run once the harness is repaired — worth tracking separately.Screenshots / recordings
None. The six client-side fixes were verified by root-cause analysis and compilation rather than by running the app; each is a small, targeted change at an identified defect. The reproduction steps in each issue are the fastest way to confirm on-device.
Pre-merge checklist
bun format && bun lintpasses with no errorsbun check-typespasses with no errorsfeat:,fix:,chore:, etc.)Summary by CodeRabbit
New Features
Bug Fixes
Tests