refactor: DRY shared abstractions in TS features

[carlos-alm]

Summary

• Extract shared node-role classification logic from classifyNodeRolesFull and classifyNodeRolesIncremental (near-duplicates, both cog=27) into a single classifyNodeRole function
• Unify buildNodeDataflowResult and buildNativeDataflowResult (near-duplicate result builders) into a single parameterized buildDataflowResult function

Titan Audit Context

• Phase: abstraction
• Domain: features
• Commits: 2
• Depends on: none

Changes

• src/features/structure.ts -- extract shared classifyNodeRole, reducing duplication between Full and Incremental variants
• src/features/dataflow.ts -- unify duplicate dataflow result builders into parameterized function

Metrics Impact

• classifyNodeRolesFull: cog 27 -> 4, classifyNodeRolesIncremental: cog 27 -> 4
• Eliminates ~90 lines of duplicated logic

Test plan

• CI passes (lint + build + tests)
• Structure analysis produces same results
• Dataflow analysis produces same results for both native and WASM engines

[claude]

Claude finished @carlos-alm's task —— View job

---

Reviewing PR: refactor: DRY shared abstractions in TS features

Todo List:

• Read modified files to understand the refactoring approach
• Phase 1: Root cause analysis - determine what problem this solves
• Phase 2: Configuration change detection - check for any rule relaxation
• Phase 3: Test integrity - verify no test weakening
• Phase 4: Code quality assessment
• Phase 5: Backlog compliance evaluation
• Provide final structured assessment

Root Cause Analysis

Problem Being Solved: Technical debt from code duplication in two critical areas:

1. structure.ts - classifyNodeRolesFull and classifyNodeRolesIncremental had near-identical logic (both cog=27) for building role summaries and batch-updating roles
2. dataflow.ts - buildNodeDataflowResult and buildNativeDataflowResult contained ~45 lines of duplicated filter/assembly logic for the same result shape

Why This Approach: ✓ The refactoring extracts shared logic into parameterized helper functions while preserving the existing APIs. This is a proper abstraction that reduces cognitive complexity from 27→4 and eliminates duplicate code without changing behavior.

Risk Assessment: Low risk - pure refactoring with no API changes, no configuration modifications, and only internal helper function extraction.

Backlog Compliance

• Zero-dep: ✓ - No new runtime dependencies added
• Foundation-aligned: ✓ - Pure internal refactoring, no principle violations
• Problem-fit: 3/5 - Improves maintainability and reduces complexity for agent-facing features, but doesn't directly address AI token waste/hallucination
• Breaking: No - All existing APIs preserved, purely additive internal helpers
• Tier: 1 (zero-dep ✓ + foundation-aligned ✓)

Code Quality Assessment ✓

Abstraction Quality: Excellent - both extractions follow clean abstraction principles:

• buildRoleSummary() + batchUpdateRoles() in structure.ts:453,491 eliminate role classification duplication
• buildDataflowResult() in dataflow.ts:480 unifies SQL/native result building with proper parameterization

Type Safety: Strong - new interfaces RawDataflowEdges and parameterized functions maintain full type safety

Performance: Neutral/positive - eliminates code duplication without performance overhead

Maintainability: Major improvement - single source of truth for complex logic, cognitive complexity reduced from 27→4 per function

Final Recommendation

• Rating: ⭐⭐⭐⭐☆ (4/5)
• Action: APPROVE
• Reasoning: Excellent refactoring that meaningfully reduces technical debt in critical graph analysis components. Clean abstractions with no API breakage. The complexity reduction (cog 27→4) and ~90 lines of duplicate code elimination provide clear maintainability wins. Problem-fit is moderate (3/5) as this improves code quality but doesn't directly solve the core AI navigation problem - however, the quality improvements to these foundational components are valuable.

[greptile-apps]

Greptile Summary

This PR DRYs up two pairs of near-duplicate functions: buildRoleSummary and batchUpdateRoles are extracted from classifyNodeRolesFull/classifyNodeRolesIncremental in structure.ts, while buildDataflowResult (backed by the new RawDataflowEdges interface) replaces the twin result builders in dataflow.ts. All original behaviour is faithfully preserved — transaction semantics, closure-captured SQL parameters, and the pre-existing non-filtering of mutates/mutatedBy under noTests are all unchanged.

Confidence Score: 5/5

Safe to merge — a clean structural refactor with no behaviour changes.

Both extractions are faithful: closure-captured variables resolve correctly in the resetFn lambda, transaction boundaries are identical to the originals, all edge-field mappings match their pre-refactor counterparts, and the NormalizedSymbol import resolves to a proper exported type. No P0 or P1 findings.

No files require special attention.

Important Files Changed

Filename	Overview
src/features/structure.ts	Extracts buildRoleSummary and batchUpdateRoles helpers to DRY up classifyNodeRolesFull and classifyNodeRolesIncremental; logic is faithfully preserved, closure-captured variables (placeholders, allAffectedFiles) correctly resolve inside the resetFn lambda, and transaction semantics are unchanged.
src/features/dataflow.ts	Unifies buildNodeDataflowResult and buildNativeDataflowResult behind a shared buildDataflowResult(sym, edges, noTests) function with a typed RawDataflowEdges interface; all field mappings are correct, NormalizedSymbol import resolves, and the pre-existing behaviour of not filtering mutates/mutatedBy by isTestFile is preserved faithfully.

Flowchart

%%{init: {'theme': 'neutral'}}%%
flowchart TD
    subgraph structure.ts
        CNR[classifyNodeRoles] -->|full| CNRF[classifyNodeRolesFull]
        CNR -->|incremental| CNRI[classifyNodeRolesIncremental]
        CNRF --> BRS[buildRoleSummary]
        CNRI --> BRS
        CNRF --> BUR[batchUpdateRoles]
        CNRI --> BUR
        BUR -->|resetFn callback| RESET_FULL["reset: UPDATE nodes SET role = NULL"]
        BUR -->|resetFn callback| RESET_INC["reset: UPDATE nodes SET role = NULL WHERE file IN (...)"]
    end

    subgraph dataflow.ts
        DD[dataflowData] -->|native path| BNDR[buildNativeDataflowResult]
        DD -->|SQL path| BNDR2[buildNodeDataflowResult]
        BNDR --> BDR[buildDataflowResult]
        BNDR2 --> BDR
        BDR -->|noTests=true| FILTERED[filtered result]
        BDR -->|noTests=false| UNFILTERED[unfiltered result]
    end

Loading

_{Reviews (1): Last reviewed commit: "refactor: unify duplicate dataflow resul..." | Re-trigger Greptile}

[github-actions]

Codegraph Impact Analysis

13 functions changed → 9 callers affected across 5 files

• RawDataflowEdges.flowsTo in src/features/dataflow.ts:446 (0 transitive callers)
• RawDataflowEdges.flowsFrom in src/features/dataflow.ts:455 (0 transitive callers)
• RawDataflowEdges.returnConsumers in src/features/dataflow.ts:464 (0 transitive callers)
• RawDataflowEdges.returnedBy in src/features/dataflow.ts:471 (0 transitive callers)
• RawDataflowEdges.mutatesTargets in src/features/dataflow.ts:472 (0 transitive callers)
• RawDataflowEdges.mutatedBy in src/features/dataflow.ts:473 (0 transitive callers)
• buildDataflowResult in src/features/dataflow.ts:480 (5 transitive callers)
• buildNodeDataflowResult in src/features/dataflow.ts:509 (4 transitive callers)
• buildNativeDataflowResult in src/features/dataflow.ts:564 (4 transitive callers)
• buildRoleSummary in src/features/structure.ts:453 (3 transitive callers)
• batchUpdateRoles in src/features/structure.ts:491 (3 transitive callers)
• classifyNodeRolesFull in src/features/structure.ts:518 (1 transitive callers)
• classifyNodeRolesIncremental in src/features/structure.ts:617 (1 transitive callers)