Add search text matcher boundary by BigAkins · Pull Request #13 · BigAkins/AkFit-iOS

BigAkins · 2026-05-12T04:04:20Z

Summary

This cleanup adds a small pure helper boundary for AkFit search text behavior.

It moves current search normalization, alias handling, simple plural stemming, edit distance, match scoring, dessert/processed penalty behavior, and type-ahead suggestion matching into SearchTextMatcher.

What changed

Added

AkFit/AkFit/Search/SearchTextMatcher.swift
AkFit/AkFitTests/SearchTextMatcherTests.swift

Updated

SupabaseFoodSearchService now delegates normalization/filtering/scoring behavior to SearchTextMatcher
SearchView now delegates type-ahead matching to SearchTextMatcher

Why

Search behavior was becoming dense across SupabaseFoodSearchService and SearchView.

This PR creates a small testable boundary before future search/ranking work, reducing the chance that future feature work duplicates or changes search behavior accidentally.

Behavior

Search behavior is intended to be preserved exactly.

This PR does not:

redesign search
change ranking behavior
change UI
change Supabase schema
add RPCs
touch food seed data
change search routing

What is still not solved

This does not yet:

improve ranking quality
add new aliases
change fuzzy matching tolerance
fix seed-data gaps
add server-side type-ahead
solve broader restaurant/branded-food coverage

Those should be handled in future feature-specific branches.

Verification

Verified with the PR-safe AkFit test plan.

Result:

45 tests passed
0 failed

chatgpt-codex-connector

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 701307a35d

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

Open a pull request for review
Mark a draft as ready
Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".

chatgpt-codex-connector · 2026-05-12T04:06:24Z

+<!-- session:019e1a49-efbb-7911-936e-90e6dfd8b17e rollout:/Users/aki/.codex/sessions/2026/05/11/rollout-2026-05-11T22-45-13-019e1a49-efbb-7911-936e-90e6dfd8b17e.jsonl -->
+- User asked: Implement the next AkFit cleanup-only PR.


Drop committed local session log from repository

Remove this .memsearch memory file from the commit: it contains local machine paths and session metadata that are unrelated to app behavior, and committing it leaks developer-environment details while adding persistent noise to version control. These generated artifacts should be excluded (for example via ignore rules) rather than tracked.

Useful? React with 👍 / 👎.

Copilot

Pull request overview

This PR refactors shared search normalization/matching/scoring into a new pure helper (SearchTextMatcher) and updates both Supabase-backed search and SearchView type-ahead to delegate to it, with new unit tests to preserve behavior.

Changes:

Added SearchTextMatcher with normalization, aliasing, stemming, scoring, and suggestion-matching helpers.
Updated SupabaseFoodSearchService and SearchView to delegate search behavior to the helper.
Added SearchTextMatcherTests to lock in expected normalization and suggestion ranking behavior.

Reviewed changes

Copilot reviewed 6 out of 6 changed files in this pull request and generated 3 comments.

Show a summary per file

File	Description
AkFit/AkFit/Search/SearchTextMatcher.swift	Introduces centralized pure search-text normalization/matching/scoring/suggestions logic.
AkFit/AkFit/Search/SupabaseFoodSearchService.swift	Delegates query normalization, filtering, and scoring to `SearchTextMatcher`.
AkFit/AkFit/Views/Search/SearchView.swift	Delegates type-ahead suggestion matching to `SearchTextMatcher`.
AkFit/AkFitTests/SearchTextMatcherTests.swift	Adds unit tests for normalization, stemming, and suggestion ranking behavior.
.memsearch/memory/2026-05-11.md	Adds a local session log file (appears unrelated to app code).
.memsearch/.index.pid	Adds a transient PID artifact (appears unrelated to app code).

Comments suppressed due to low confidence (1)

.memsearch/.index.pid:2

.index.pid looks like a transient runtime artifact (PID) that will change between machines/runs and cause needless churn/merge conflicts. Suggest removing it from the PR and ignoring it (e.g. via .gitignore for .memsearch/).

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

+## Session 22:45
+
+### 22:45
+<!-- session:019e1a49-fa43-7912-b11c-9fd8c96a3034 rollout: -->
+- Codex: {"title":"Add search helper tests"}
+
+### 22:52
+<!-- session:019e1a49-efbb-7911-936e-90e6dfd8b17e rollout:/Users/aki/.codex/sessions/2026/05/11/rollout-2026-05-11T22-45-13-019e1a49-efbb-7911-936e-90e6dfd8b17e.jsonl -->
+- User asked: Implement the next AkFit cleanup-only PR.
+- Codex: Implemented the cleanup-only search boundary. No commit made.
+
+**Files Changed**
+- [SearchTextMatcher.swift](/Users/aki/Documents/AkTalkToEM/AkCodeForEm/Project%20Workspace/AkPerso/AkFit/AkFit/AkFit/Search/SearchTextMatcher.swift) new pure helper
+- [SupabaseFoodSearchService.swift](/Users/aki/Documents/AkTalkToEM/AkCodeForEm/Project%20Workspace/AkPerso/AkFit/AkFit/AkFit/Search/SupabaseFoodSearchService.swift) now delegates normalization/filtering/scoring
+- [SearchView.swift](/Users/aki/Documents/AkTalkToEM/AkCodeForEm/Project%20Workspace/AkPerso/AkFit/AkFit/AkFit/Views/Search/SearchView.swift) type-ahead matching now delegates to helper
+- [SearchTextMatcherTests.swift](/Users/aki/Documents/AkTalkToEM/AkCodeForEm/Project%20Workspace/AkPerso/AkFit/AkFit/AkFitTests/SearchTextMatcherTests.swif...


+    /// Returns true when every normalized query word matches the term either
+    /// directly or through the rough plural stem.
+    static func matchesAllQueryWords(term: String, normalizedQuery: String) -> Bool {
+        guard !normalizedQuery.isEmpty else { return false }
+        let words = normalizedQuery.split(separator: " ").map(String.init)
+        let stemmedWords = words.map { stemWord($0) }
+        let normalizedTerm = normalizeForSearch(term)
+        let stemmedTerm = stemmedForm(normalizedTerm)
+        return words.allSatisfy { normalizedTerm.contains($0) } ||
+               stemmedWords.allSatisfy { stemmedTerm.contains($0) }
+    }
+


                rows = candidates.filter { row in
-                    let n = Self.normalizeForSearch(row.foodName)
-                    let nStemmed = Self.stemmedForm(n)
-                    return words.allSatisfy { w in n.contains(w) } ||
-                           stemmedWords.allSatisfy { sw in nStemmed.contains(sw) }
+                    SearchTextMatcher.matchesAllQueryWords(
+                        term: row.foodName,
+                        normalizedQuery: normalized
+                    )
                }


BigAkins · 2026-05-12T05:27:47Z

Patched the review comments.

Changes:

removed .memsearch files from Git tracking
added .memsearch/ to .gitignore
added SearchTextMatcher.QueryMatch so query normalization, words, stems, and plain-food query state are computed once per query
updated Supabase search filtering and SearchView type-ahead matching to reuse the precomputed query match data
added tests comparing the precomputed helper path against the existing string-based helper path

Verification:

PR-safe AkFit test plan passed
47 tests passed, 0 failed
git diff --check passed
.memsearch/ is now ignored and no .memsearch files remain tracked

BigAkins · 2026-05-12T05:34:14Z

@copilot

Codex · 2026-05-12T05:35:10Z

@codex[agent]

No further changes needed on my side—latest commits bdbe855 (QueryMatch + reuse) and 6332426 (.memsearch ignored) cover the review items. Let me know if you want any follow-up tweaks.

BigAkins · 2026-05-12T05:35:18Z

@claude

claude · 2026-05-12T05:35:35Z

Claude finished @BigAkins's task in 2m 10s —— View job

Code Review

Gather context (CLAUDE.md, git diff, changed files)
Review SearchTextMatcher.swift
Review SupabaseFoodSearchService.swift
Review SearchView.swift
Review SearchTextMatcherTests.swift
Post final review

The refactor is well-executed. The QueryMatch struct directly addresses the Copilot feedback about per-row recomputation, SearchView shrinks from 55 lines of logic to 2, and the .memsearch/ artifacts are now properly ignored. A few things worth looking at before merging:

1. Implicit safety assumption in `matchScore` — `qStems[0]` without bounds check

SearchTextMatcher.swift:155-156

if qWords.count <= 1 {
    let qStem = qStems[0]   // ← unchecked access

If qWords.count == 0, this crashes. In practice it's unreachable today because Swift.String.hasPrefix("") is always true, so the n.hasPrefix(q) early-return on line 148 fires first when the query is empty. But that's an implicit invariant, not an explicit guard — it breaks silently if the control flow above ever changes. A simple fix:

if qWords.count == 1 {   // exact check instead of <= 1
    let qStem = qStems[0]

Or add a guard before the branch. Fix this →

2. `applyQueryAliases` rebuilds the tuple array on every call

SearchTextMatcher.swift:38-41

let aliases: [(from: String, to: String)] = [
    ("in and out", "in n out"),
    ("innout", "in n out"),
]

This allocates a new array on every normalization call. Move to private static let:

private static let queryAliases: [(from: String, to: String)] = [
    ("in and out", "in n out"),
    ("innout", "in n out"),
]

Fix this →

3. Redundant entry in `dessertKeywords`

SearchTextMatcher.swift:187

isDessertOrProcessed uses substring contains, so "milkshake" already triggers on "shake". The "milkshake" entry does no additional work. Not a bug, just noise.

4. Sort comparator still normalizes inside O(n log n) comparisons

SupabaseFoodSearchService.swift:70-75

return rows.map(FoodItem.init).sorted { a, b in
    ...
    if SearchTextMatcher.isDessertOrProcessed(SearchTextMatcher.normalizeForSearch(a.name)) { sa += 1 }
    if SearchTextMatcher.isDessertOrProcessed(SearchTextMatcher.normalizeForSearch(b.name)) { sb += 1 }

normalizeForSearch is called inside the comparator, which runs O(n log n) times. This was the same behavior before this PR (not a regression), but it's worth noting: if the candidate set ever grows, a pre-sort .map { ($0, SearchTextMatcher.normalizeForSearch($0.name)) } before the sort would avoid the repeated work.

5. Test coverage gaps

The test file does a good job covering normalization, stemming, alias handling, ranking, and the precomputed-vs-string parity. A few paths have no coverage:

Gap	Why it matters
`editDistance`	Directly tests the O(n*m) routine in isolation
Fuzzy fallback path (< 3 exact matches, query ≥ 3 chars, single-char typo)	The fallback branch in `suggestions` is never exercised by existing tests
`matchScore` for all score bands (0–4)	Verifies the scoring tiers explicitly rather than relying on end-to-end suggestion order

Overall

Solid PR. The QueryMatch struct is the right design, the two-overload pattern (String + QueryMatch) makes the parity tests clean, and the delegation in SearchView is a meaningful simplification. Items 1 and 2 above are quick fixes; the rest are minor quality notes that could reasonably go in a follow-up.
| Branch: Ak/search-quality-boundary

BigAkins added 2 commits May 11, 2026 22:58

Add search text matcher boundary

431cdbc

Add files

701307a

Copilot AI review requested due to automatic review settings May 12, 2026 04:04

Copilot started reviewing on behalf of BigAkins May 12, 2026 04:04 View session

chatgpt-codex-connector Bot reviewed May 12, 2026

View reviewed changes

Copilot AI reviewed May 12, 2026

View reviewed changes

BigAkins added 2 commits May 12, 2026 00:02

git ignore update

6332426

Patch search matcher cleanup

bdbe855

Claude started work on behalf of BigAkins May 12, 2026 05:33 View session

Copilot stopped work on behalf of BigAkins due to an error May 12, 2026 05:34
Claude Code process exited with code 1

Codex started work on behalf of BigAkins May 12, 2026 05:34 View session

Codex finished work on behalf of BigAkins May 12, 2026 05:35

Repository owner deleted a comment from chatgpt-codex-connector Bot May 12, 2026

BigAkins merged commit a44efe8 into main May 12, 2026
4 of 5 checks passed

BigAkins deleted the Ak/search-quality-boundary branch May 12, 2026 05:39

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Add search text matcher boundary#13

Add search text matcher boundary#13
BigAkins merged 4 commits into
mainfrom
Ak/search-quality-boundary

BigAkins commented May 12, 2026

Uh oh!

chatgpt-codex-connector Bot left a comment

Uh oh!

chatgpt-codex-connector Bot May 12, 2026

Uh oh!

Copilot AI left a comment

Uh oh!

BigAkins commented May 12, 2026

Uh oh!

BigAkins commented May 12, 2026

Uh oh!

Codex AI commented May 12, 2026

Uh oh!

BigAkins commented May 12, 2026

Uh oh!

claude Bot commented May 12, 2026 •

edited

Loading

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

		<!-- session:019e1a49-efbb-7911-936e-90e6dfd8b17e rollout:/Users/aki/.codex/sessions/2026/05/11/rollout-2026-05-11T22-45-13-019e1a49-efbb-7911-936e-90e6dfd8b17e.jsonl -->
		- User asked: Implement the next AkFit cleanup-only PR.

Conversation

BigAkins commented May 12, 2026

Summary

What changed

Added

Updated

Why

Behavior

What is still not solved

Verification

Uh oh!

chatgpt-codex-connector Bot left a comment

Choose a reason for hiding this comment

💡 Codex Review

Uh oh!

chatgpt-codex-connector Bot May 12, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

Pull request overview

Reviewed changes

Uh oh!

BigAkins commented May 12, 2026

Uh oh!

BigAkins commented May 12, 2026

Uh oh!

Codex AI commented May 12, 2026

Uh oh!

BigAkins commented May 12, 2026

Uh oh!

claude Bot commented May 12, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Code Review

1. Implicit safety assumption in matchScore — qStems[0] without bounds check

2. applyQueryAliases rebuilds the tuple array on every call

3. Redundant entry in dessertKeywords

4. Sort comparator still normalizes inside O(n log n) comparisons

5. Test coverage gaps

Overall

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

claude Bot commented May 12, 2026 •

edited

Loading

1. Implicit safety assumption in `matchScore` — `qStems[0]` without bounds check

2. `applyQueryAliases` rebuilds the tuple array on every call

3. Redundant entry in `dessertKeywords`