[SP-2993] feat: add support for recursive search #19
Conversation
matiasdaloia
force-pushed
the
feature/mdaloia/SP-2993-Folder-Hashing-Service-Support-recursive-search
branch
2 times, most recently
from
a26dae1 to
3619481
Compare
WalkthroughImplements recursive, node-driven scanning with per-node thresholding and deduplication; switches to distance-based scoring; expands domain models and mapper; updates Qdrant repository interface and implementation; adds a CSV import CLI and README updates; removes Docker packaging/scripts and replaces CI/build with a streamlined Makefile and build variables. Changes
Sequence Diagram(s)sequenceDiagram
autonumber
participant C as Client (gRPC/REST)
participant H as Handler
participant S as ScanService
participant R as Repository (Qdrant)
C->>H: HFHRequest (root, thresholds, rankThreshold...)
H->>S: ScanFolder(domain req)
S->>S: start scanNode(root)
loop traverse nodes
S->>R: SearchByHashes(dirHash,nameHash,contentHash,langExt,rankThreshold)
R-->>S: ComponentGroups (distance-based)
alt has match >= RecursiveThreshold
Note right of S #dff0d8: Early-stop this branch
S->>S: processComponentGroups(path, MinAcceptedScore)
else
S->>S: recurse into children
S->>S: processComponentGroups(path, MinAcceptedScore)
end
end
S->>S: deduplicateComponents (keep highest per component)
S-->>H: Aggregated ScanResults
H-->>C: HFHResponse (status, timing, results)
Estimated code review effort🎯 5 (Critical) | ⏱️ ~120 minutes Possibly related PRs
Poem
Pre-merge checks and finishing touches✅ Passed checks (3 passed)
✨ Finishing touches
🧪 Generate unit tests (beta)
📜 Recent review detailsConfiguration used: CodeRabbit UI Review profile: CHILL Plan: Pro 📒 Files selected for processing (2)
🚧 Files skipped from review as they are similar to previous changes (1)
🧰 Additional context used🪛 checkmake (0.2.2)Makefile[warning] 14-14: Missing required phony target "all" (minphony) [warning] 14-14: Missing required phony target "clean" (minphony) [warning] 14-14: Missing required phony target "test" (minphony) [warning] 49-49: Target "build" should be declared PHONY. (phonydeclared) [warning] 80-80: Target "ci" should be declared PHONY. (phonydeclared) ⏰ Context from checks skipped due to timeout of 90000ms. You can increase the timeout in your CodeRabbit configuration to a maximum of 15 minutes (900000ms). (1)
Comment |
![coderabbitai[bot]](https://avatars.githubusercontent.com/in/347564?s=60&v=4){kind=small}
There was a problem hiding this comment.
Actionable comments posted: 2
🧹 Nitpick comments (7)
Makefile (1)
48-48: OK; consider embedding version for parity with builds (optional).If helpful for local diagnostics, mirror ldflags used in build targets.
internal/domain/entities/scan_request.go (1)
7-10: Default not enforced; align code with comment or update comment.Field MinAcceptedScore mentions default 0.15, but no defaulting here. Apply default in mapper/service (see suggested mapper change) or adjust the comment.
internal/repository/scan_repository_qdrant_impl.go (4)
187-193: Replace fmt.Printf with structured logging; avoid stdout in repository code.Use ctxzap logger and pass ctx into helpers; gate at debug level.
- fmt.Printf("Stage 1 query returned %d results\n", len(stage1Results)) + s.Debugf("Stage 1 query returned %d results", len(stage1Results)) - fmt.Printf("Returning early with %d results (<=20)\n", len(stage1Results)) + s.Debugf("Returning early with %d results (<=20)", len(stage1Results)) - fmt.Printf("Stage 2 query (dirs filtering) returned %d results\n", len(stage2Results)) + s.Debugf("Stage 2 query (dirs filtering) returned %d results", len(stage2Results)) - fmt.Printf("Returning with %d results after stage 2 (<=20)\n", len(stage2Results)) + s.Debugf("Returning with %d results after stage 2 (<=20)", len(stage2Results))And refactor processSearchResults to receive ctx and use logger:
- return r.processSearchResults(searchResp) + return r.processSearchResults(ctx, searchResp)-func (r *ScanRepositoryQdrantImpl) processSearchResults(searchResp []*qdrant.ScoredPoint) ([]entities.ComponentGroup, error) { +func (r *ScanRepositoryQdrantImpl) processSearchResults(ctx context.Context, searchResp []*qdrant.ScoredPoint) ([]entities.ComponentGroup, error) { + s := ctxzap.Extract(ctx).Sugar() ... - fmt.Printf("processSearchResults: converted %d points to %d results\n", len(searchResp), len(allResults)) + s.Debugf("processSearchResults: converted %d points to %d results", len(searchResp), len(allResults)) ... - fmt.Printf("processSearchResults: grouped into %d component groups\n", len(purlGroups)) + s.Debugf("processSearchResults: grouped into %d component groups", len(purlGroups))Also applies to: 232-237, 407-413
227-248: Ensure order before thresholding; explicitly sort stage2Results by increasing distance.Assume Qdrant returns sorted, but make it explicit to avoid brittle early-break logic.
stage2Results, err := r.client.Query(ctx, stage2Query) if err != nil { return nil, fmt.Errorf("stage 2 search failed: %w", err) } + // Ensure results are ordered by ascending distance (lower is better) + sort.Slice(stage2Results, func(i, j int) bool { return stage2Results[i].Score < stage2Results[j].Score })
262-273: Stale comment: stage2IDs vs “stage3IDs”.The note says it “should be stage3IDs,” but code uses stage2IDs. Either compute stage3IDs or remove/fix the comment to prevent confusion.
568-578: Constant/comment mismatch in distance/score conversion.k = 0.065 doesn’t match “-ln(0.2)/30” (~0.0536). Align the constant or fix the comment; same in both functions.
-const k = 0.065 // -ln(0.2) / 30 +const k = 0.065 // Tuned decay; was previously documented as -ln(0.2)/30internal/service/scan_service_impl.go (1)
102-110: Consider clarifying the threshold check logic.The
shouldCheckThresholdexpression!(isRoot && len(node.Children) > 0)is logically correct but somewhat subtle. It means:
- Root nodes with children always explore children (skip threshold check)
- All other nodes check the threshold
Consider adding a brief inline comment explaining the boolean logic, or using a more explicit condition:
// Only check threshold for non-root nodes, or root nodes without children shouldCheckThreshold := !isRoot || len(node.Children) == 0This equivalent formulation may be clearer to readers.
📜 Review details
Configuration used: CodeRabbit UI
Review profile: CHILL
Plan: Pro
⛔ Files ignored due to path filters (1)
go.sumis excluded by!**/*.sum
📒 Files selected for processing (9)
Makefile(1 hunks)README.md(1 hunks)go.mod(1 hunks)internal/domain/entities/scan_request.go(1 hunks)internal/handler/scan_handler.go(0 hunks)internal/mapper/scan_mapper_impl.go(1 hunks)internal/repository/scan_repository.go(1 hunks)internal/repository/scan_repository_qdrant_impl.go(6 hunks)internal/service/scan_service_impl.go(2 hunks)
💤 Files with no reviewable changes (1)
- internal/handler/scan_handler.go
🧰 Additional context used
🧬 Code graph analysis (3)
internal/service/scan_service_impl.go (5)
internal/repository/scan_repository.go (1)
ScanRepository(16-28)internal/service/scan_service.go (1)
ScanService(9-12)internal/domain/entities/scan_request.go (2)
ScanRequest(4-17)FolderNode(20-33)internal/domain/entities/scan_response.go (2)
ScanResponse(4-7)ScanResult(10-15)internal/domain/entities/component.go (2)
ComponentGroup(4-11)Version(14-17)
internal/repository/scan_repository.go (2)
internal/domain/entities/scan_request.go (1)
LanguageExtensions(36-36)internal/domain/entities/component.go (1)
ComponentGroup(4-11)
internal/repository/scan_repository_qdrant_impl.go (2)
internal/domain/entities/scan_request.go (1)
LanguageExtensions(36-36)internal/domain/entities/component.go (1)
ComponentGroup(4-11)
🔇 Additional comments (9)
internal/repository/scan_repository_qdrant_impl.go (2)
551-552: OK on distance→score conversion for API output.Returning match scores [0..1] externally while ranking internally by distance is sound.
151-179: ClarifyScoreThresholdsemantics for distance metrics (L2/Hamming).
For distance-based scores,ScoreThresholdis a maximum acceptable distance—any result with distance > threshold is excluded. Ensure your thresholds (e.g. 30) match the maximum distances you intend to allow and adjust tighter (smaller) or looser (larger) as needed.README.md (1)
109-152: README flags match implementation. The -dir, -top-purls, and -overwrite flags in cmd/import/main.go are present and correctly documented.internal/repository/scan_repository.go (1)
18-18: All SearchByHashes call sites updatedService layer and repository implementation use the new signature without topK; no further changes needed.
go.mod (1)
13-13: Tidy and verifygitproxy.zycloud.tk/scanoss/papi v0.22.0
Rungo mod tidy, compile the code and run tests; confirm there are no breaking changes in thescanningv2API.internal/service/scan_service_impl.go (4)
13-20: LGTM! Clean field rename.The renaming of
scanRepotorepoimproves readability and aligns with common Go conventions for shorter field names.
47-83: LGTM! Effective filtering logic.The updated signature and filtering implementation correctly:
- Filters versions by
minAcceptedScore- Excludes groups with no qualifying versions
- Only creates results when relevant matches exist
This prevents low-quality results from cluttering the response.
127-137: LGTM! Clean helper function.The implementation correctly checks for any version meeting the threshold across all component groups. The use of
>=is appropriate for threshold semantics.
23-45: Confirm recursiveThreshold short-circuits subtree scanningCurrent logic stops descending into a node’s children once any component meets the
recursiveThreshold, which may skip deeper, higher-scoring matches. No docs or tests explicitly define this behavior—confirm it’s intentional.
… duplicated components
matiasdaloia
force-pushed
the
feature/mdaloia/SP-2993-Folder-Hashing-Service-Support-recursive-search
branch
from
e7a97d6 to
fd98961
Compare
There was a problem hiding this comment.
Actionable comments posted: 0
🧹 Nitpick comments (1)
scripts/README.md (1)
89-102: Add language identifier to fenced code block.The fenced code block lacks a language identifier. Add
textor an appropriate identifier to satisfy markdown linting.Apply this diff:
-``` +```text /usr/local/etc/scanoss/folder-hashing-api/ └── app-config-prod.json
📜 Review details
Configuration used: CodeRabbit UI
Review profile: CHILL
Plan: Pro
📒 Files selected for processing (12)
CHANGELOG.md(1 hunks)Dockerfile(0 hunks)README.md(3 hunks)cmd/import/main.go(2 hunks)docker-compose.yml(0 hunks)scripts/README.md(1 hunks)scripts/create-collection-snapshots.sh(0 hunks)scripts/create-docker-package.sh(0 hunks)scripts/docker-deploy.sh(0 hunks)scripts/import-collections.sh(0 hunks)scripts/import-single-collection.sh(0 hunks)scripts/setup-tls.sh(0 hunks)
💤 Files with no reviewable changes (8)
- scripts/create-docker-package.sh
- scripts/docker-deploy.sh
- scripts/setup-tls.sh
- scripts/create-collection-snapshots.sh
- scripts/import-single-collection.sh
- scripts/import-collections.sh
- docker-compose.yml
- Dockerfile
🧰 Additional context used
🪛 markdownlint-cli2 (0.18.1)
README.md
40-40: Bare URL used
(MD034, no-bare-urls)
scripts/README.md
89-89: Fenced code blocks should have a language specified
(MD040, fenced-code-language)
🔇 Additional comments (4)
cmd/import/main.go (1)
37-38: LGTM! Clean addition of configurable Qdrant connection parameters.The new CLI flags for
--qdrant-hostand--qdrant-portare well-integrated, with sensible defaults from existing constants and proper usage in the client configuration.Also applies to: 69-70
CHANGELOG.md (1)
9-9: Verify the release date.The release date
2025-10-08is set for October 2025, but this PR was created on 2025-08-06 (August 2025). Ensure this future date is intentional and not a typo.README.md (2)
8-32: LGTM! Clear prerequisites and quick start guide.The new Prerequisites and Quick Start sections provide a straightforward path for users to get up and running with the service. The step-by-step instructions are clear and include verification steps.
124-186: LGTM! Comprehensive import tool documentation.The new "Importing Data" section provides excellent documentation for the CSV import tool, including:
- Clear usage examples
- Command-line options table
- Explanation of how the tool works
- End-to-end workflow example
This will help users understand how to populate the vector database.
There was a problem hiding this comment.
Actionable comments posted: 5
Caution
Some comments are outside the diff and can’t be posted inline due to platform limitations.
⚠️ Outside diff range comments (2)
internal/config/config.go (1)
137-139: Reconsider enabling dynamic logging by default
SetDefaultsnow flipsLogging.DynamicLoggingfrom the old zero-value (false) totrue. With zap-logging-helper this spins up the dynamic logging HTTP listener on every boot, which is a behavioral regression: deployments that relied on the prior default will start a new server onlocalhost:60061and fail if that port is occupied (or expose runtime level toggling without opting in). Please keep the default disabled unless you also update the consuming code/docs to expect it.- c.Logging.DynamicLogging = true - c.Logging.DynamicPort = "localhost:60061" + c.Logging.DynamicLogging = false + c.Logging.DynamicPort = "localhost:60061"Based on learnings
internal/repository/scan_repository_qdrant_impl.go (1)
263-276: Fix incorrect variable reference in finalFilterPreferred.Line 269's comment (in Spanish) correctly notes that
finalFilterPreferredshould referencestage3IDs, notstage2IDs. The current code usesstage2IDswhich includes candidates from an earlier stage, potentially causing incorrect filtering in the final query.The fix requires defining
stage3IDsfrom the appropriate result set before using it. However, the current code structure doesn't create astage3result set—it goes directly fromstage2ResultstofinalFilter. You need to either:
- Create an explicit
stage3IDsslice from the filteredstage2Results(lines 243-249), or- Use a more descriptive variable name like
topStage2IDsorfilteredStage2IDsand update the comment accordinglyIf option 1 is intended:
// Stage 3: Final refinement using nested prefetch for the best results - stage2IDs := make([]uint64, 0, len(stage2Results)) + stage3IDs := make([]uint64, 0, len(stage2Results)) for _, point := range stage2Results { if point.Score > (stage2Results[0].Score*1.1)+5 { break } - stage2IDs = append(stage2IDs, point.Id.GetNum()) + stage3IDs = append(stage3IDs, point.Id.GetNum()) } finalFilter := &qdrant.Filter{ Must: []*qdrant.Condition{ { ConditionOneOf: &qdrant.Condition_HasId{ HasId: &qdrant.HasIdCondition{ - HasId: r.convertIDsToPointIDs(stage2IDs), + HasId: r.convertIDsToPointIDs(stage3IDs), }, }, }, }, } finalFilterPreferred := &qdrant.Filter{ Must: append( []*qdrant.Condition{ { ConditionOneOf: &qdrant.Condition_HasId{ HasId: &qdrant.HasIdCondition{ - HasId: r.convertIDsToPointIDs(stage2IDs), // Nota: debería ser stage3IDs, no stage2IDs + HasId: r.convertIDsToPointIDs(stage3IDs), }, }, }, }, exactRank1Conditions..., ), }
♻️ Duplicate comments (1)
internal/mapper/scan_mapper_impl.go (1)
25-32: Restore the documented MinAcceptedScore defaultMapping a nil/omitted
MinAcceptedScorestraight through keeps it at zero, which contradicts the struct comment inentities.ScanRequest(“default: 0.15”) and materially widens the result set by accepting every low-score match. Apply the domain default before returning the request (same concern raised earlier).- return &entities.ScanRequest{ + sr := &entities.ScanRequest{ RankThreshold: int(req.RankThreshold), RecursiveThreshold: req.RecursiveThreshold, MinAcceptedScore: req.MinAcceptedScore, Category: req.Category, QueryLimit: int(req.QueryLimit), Root: m.ChildrenToDomain(req.Root), - } + } + if sr.MinAcceptedScore == 0 { + sr.MinAcceptedScore = 0.15 + } + return sr +``` </blockquote></details> </blockquote></details> <details> <summary>🧹 Nitpick comments (3)</summary><blockquote> <details> <summary>internal/repository/scan_repository_qdrant_impl.go (2)</summary><blockquote> `398-398`: **Extract magic numbers to named constants.** The scoring and filtering logic uses several magic numbers (1.1, 3.0, 0.2, 2, 5) that make the thresholding algorithm harder to understand and maintain. Consider extracting these as named constants with explanatory comments. Add constants at the package level: ```go const ( // Scoring threshold multipliers and offsets scoreMultiplier = 1.1 // Allow 10% score variance scoreOffsetSmall = 2.0 // Offset for zero-score edge cases scoreOffsetMedium = 3.0 // Offset for garbage result filtering scoreOffsetLarge = 5.0 // Offset for stage filtering scoreSimilarityPercent = 0.2 // 20% similarity threshold )Then use them in the code:
- scoreThreshold := searchResp[0].Score*1.1 + 3.0 + scoreThreshold := searchResp[0].Score*scoreMultiplier + scoreOffsetMediumAlso applies to: 523-523, 279-279, 245-245
157-157: Clarify comment phrasing.The comments "limit to garbage results" at lines 157 and 168 are ambiguous. Consider rephrasing to "exclude garbage results" or "filter out low-quality results" for clarity.
Also applies to: 168-168
internal/service/scan_service_impl.go (1)
46-47: UseInfoffor formatted logging.
SugaredLogger.Infoignores%d, so this logs “Scan completed: found %d results ”. Switch toInfof(orInfowithfmt.Sprintf) to format correctly.
📜 Review details
Configuration used: CodeRabbit UI
Review profile: CHILL
Plan: Pro
📒 Files selected for processing (22)
.golangci.yml(1 hunks)Makefile(1 hunks)cmd/import/main.go(17 hunks)cmd/server/main.go(4 hunks)go.mod(1 hunks)internal/config/config.go(4 hunks)internal/domain/entities/component.go(2 hunks)internal/domain/entities/language.go(4 hunks)internal/domain/entities/metrics.go(1 hunks)internal/domain/entities/scan_request.go(3 hunks)internal/domain/entities/scan_response.go(1 hunks)internal/domain/entities/version.go(1 hunks)internal/handler/scan_handler.go(3 hunks)internal/mapper/scan_mapper.go(1 hunks)internal/mapper/scan_mapper_impl.go(1 hunks)internal/protocol/grpc/server.go(1 hunks)internal/protocol/rest/server.go(1 hunks)internal/repository/scan_repository.go(2 hunks)internal/repository/scan_repository_qdrant_impl.go(18 hunks)internal/service/scan_service.go(1 hunks)internal/service/scan_service_impl.go(2 hunks)internal/validation/validator.go(1 hunks)
✅ Files skipped from review due to trivial changes (1)
- internal/domain/entities/scan_response.go
🧰 Additional context used
🧬 Code graph analysis (6)
internal/domain/entities/language.go (1)
internal/domain/entities/scan_request.go (1)
LanguageExtensions(36-36)
internal/service/scan_service.go (2)
internal/domain/entities/scan_request.go (1)
ScanRequest(4-17)internal/domain/entities/scan_response.go (1)
ScanResponse(4-7)
internal/protocol/grpc/server.go (2)
internal/protocol/rest/server.go (1)
RunServer(35-63)internal/config/config.go (1)
Config(38-72)
internal/handler/scan_handler.go (2)
internal/service/scan_service.go (1)
ScanService(11-14)internal/mapper/scan_mapper.go (1)
ScanMapper(11-15)
internal/service/scan_service_impl.go (5)
internal/repository/scan_repository.go (1)
ScanRepository(19-31)internal/service/scan_service.go (1)
ScanService(11-14)internal/domain/entities/scan_request.go (2)
ScanRequest(4-17)FolderNode(20-33)internal/domain/entities/scan_response.go (2)
ScanResponse(4-7)ScanResult(10-15)internal/domain/entities/component.go (2)
ComponentGroup(5-12)Version(15-18)
internal/repository/scan_repository.go (2)
internal/domain/entities/scan_request.go (1)
LanguageExtensions(36-36)internal/domain/entities/component.go (1)
ComponentGroup(5-12)
🪛 checkmake (0.2.2)
Makefile
[warning] 1-1: Missing required phony target "all"
(minphony)
[warning] 76-76: Target "ci" should be declared PHONY.
(phonydeclared)
🔇 Additional comments (12)
internal/domain/entities/metrics.go (1)
15-15: LGTM: Comment capitalization matches function name.Good practice to capitalize the comment to match the exported function name.
internal/service/scan_service.go (1)
1-14: Docstring addition looks goodThanks for adding the package-level context; it helps clarify the service surface.
go.mod (1)
3-13: papi upgrade acknowledgedMoving to go1.23.x and bumping
github.com/scanoss/papito v0.22.0 keeps the module in sync with the latest helper features. Based on learningsinternal/protocol/grpc/server.go (1)
27-65: Scoped shutdown helper reads cleanerThe short declaration keeps
oltpShutdownlocal while preserving the teardown path—looks good.internal/domain/entities/version.go (1)
19-21: Clearer AppVersion commentNice clarification on how to inject the version via ldflags; matches the build tooling updates.
.golangci.yml (1)
1-235: LGTM! Comprehensive linting configuration.The expanded linting setup significantly improves code quality enforcement with appropriate timeouts, a well-chosen linter suite, sensible per-linter settings, and practical exclusion rules. This aligns well with the PR's broader refactoring and quality improvements.
internal/protocol/rest/server.go (1)
21-35: LGTM!The import reordering improves organization, and the nolint directive appropriately documents the context parameter position constraint imposed by the existing API design.
internal/validation/validator.go (1)
1-26: LGTM!The added package documentation and improved formatting enhance code clarity without altering behavior.
internal/handler/scan_handler.go (1)
1-104: LGTM!The enhanced validation flow (protobuf validation → domain validation) and structured error responses improve robustness and API contract enforcement. The implementation correctly handles validation failures with appropriate status codes.
cmd/server/main.go (1)
20-110: LGTM!The import consolidation improves readability, the nolint directive appropriately documents the early-exit pattern before defer setup, and the REST port check change (
!= ""instead oflen > 0) is more idiomatic for string emptiness checks in Go.Makefile (1)
11-11: Confirmedentities.AppVersionexists
AppVersionis declared ininternal/domain/entities/version.go(line 21); LDFLAGS embedding in the Makefile is valid.internal/mapper/scan_mapper.go (1)
1-15: ApproveChildrenToDomainimplementationFound implementation at internal/mapper/scan_mapper_impl.go:55; no further changes needed.
There was a problem hiding this comment.
Actionable comments posted: 1
Caution
Some comments are outside the diff and can’t be posted inline due to platform limitations.
⚠️ Outside diff range comments (1)
.github/workflows/release.yml (1)
24-45: Release artifacts no longer exist with the expected names.
make build_amd64/build_arm64now emittarget/folder-hashing-api-linux-*(noscanoss-prefix), yet the release step still uploadstarget/scanoss-folder-hashing-api-linux-*. On the next tagged release,action-gh-releasewill fail because those files are missing. Please align the file list with the new build output names.- target/scanoss-folder-hashing-api-linux-amd64 - target/scanoss-folder-hashing-api-linux-arm64 - target/scanoss-hfh-import-linux-amd64 - target/scanoss-hfh-import-linux-arm64 + target/folder-hashing-api-linux-amd64 + target/folder-hashing-api-linux-arm64 + target/hfh-import-linux-amd64 + target/hfh-import-linux-arm64
♻️ Duplicate comments (1)
Makefile (1)
1-1: Expand the.PHONYlist for the new targets.We now call targets such as
ci,build_amd64,build_import_amd64,test-coverage, etc., but they are not marked phony. Marking them prevents make from treating similarly named files as up-to-date and matches the last review’s guidance.-.PHONY: help test lint fmt build clean run +.PHONY: help test lint fmt build clean run ci version tidy build_amd64 build_arm64 build_import_amd64 build_import_arm64 test-coverage clean-testcache lint-fix vet package_amd64 package_arm64
📜 Review details
Configuration used: CodeRabbit UI
Review profile: CHILL
Plan: Pro
📒 Files selected for processing (3)
.github/workflows/go-ci.yml(1 hunks).github/workflows/release.yml(1 hunks)Makefile(1 hunks)
🧰 Additional context used
🪛 checkmake (0.2.2)
Makefile
[warning] 76-76: Target "ci" should be declared PHONY.
(phonydeclared)
[warning] 1-1: Missing required phony target "all"
(minphony)
matiasdaloia
deleted the
feature/mdaloia/SP-2993-Folder-Hashing-Service-Support-recursive-search
branch
Summary by CodeRabbit
New Features
Documentation
Refactor
Chores / Removed