refactor: optimize dependency tree and remove unused packages (#175)

[kitelev]

Summary

Completed comprehensive dependency audit and optimization to address Issue #175 (Reduce node_modules size).

Changes

Optimizations Applied

1. Deduplication (npm dedupe)

   • Removed 3 duplicate packages
   • Added 2 packages for optimization
   • Net result: -1 package

2. Unused Package Removal

   • ❌ Removed eslint-plugin-import (not used in eslint.config.mjs)
   • ❌ Removed eslint-plugin-prettier (not used in eslint.config.mjs)
   • Net result: -3 packages

Results

Metric	Before	After	Change
Total size	264 MB	263 MB	-1 MB (-0.4%)
Package count	904	900	-4 packages
Dependency health	Multiple duplicates	Clean tree	✅ Improved

Analysis Performed

1. Size Analysis: Identified largest dependencies (eslint-plugin-obsidianmd: 34 MB, typescript: 23 MB)
2. Depcheck: Verified truly unused dependencies (filtered false positives)
3. Code Verification: Searched codebase to confirm actual usage patterns
4. Transitive Dependencies: Analyzed indirect dependencies (moment via obsidian)

False Positives (Verified as REQUIRED)

Depcheck incorrectly flagged these as unused:

• ✅ builtin-modules - used in esbuild.config.mjs:3,138
• ✅ preact/@preact/compat - used in esbuild.config.mjs:147-149 (React alias)
• ✅ ts-jest - used in jest.ui.config.js:15
• ✅ jest-environment-obsidian - used in jest.ui.config.js:9,12
• ✅ jest-environment-jsdom - used in jest.config.js:3
• ✅ @typescript-eslint/* - used in eslint.config.mjs

Constraints Preventing Further Reduction

1. Large Required Dependencies:

   • eslint-plugin-obsidianmd (34 MB) - Obsidian-specific linting
   • typescript (23 MB) - build toolchain
   • @babel (14 MB) - Jest ES module transformation
   • playwright-core (8.5 MB) - E2E testing

2. Transitive Dependencies:

   • moment (5.1 MB) - required by Obsidian package (cannot remove)

3. Essential Tooling:

   • Cannot remove testing infrastructure
   • Cannot remove build tools
   • Cannot remove linting/formatting

Documentation

Created comprehensive DEPENDENCY_OPTIMIZATION_REPORT.md with:

• Detailed analysis methodology
• Findings and rationale for each decision
• Recommendations for future optimizations
• pnpm migration evaluation (40-50% estimated savings)

Testing

All quality gates passed:

✅ Unit tests (803 tests) - PASSED
✅ UI tests - PASSED  
✅ Component tests (218 tests) - PASSED
✅ E2E tests (Docker) - PASSED
✅ Build successful - bundle: 118.6kb
✅ TypeScript compilation - PASSED
✅ Zero vulnerabilities (npm audit)

Why Target (<200 MB) Not Achieved

Initial issue estimated 367 MB, but actual size was already 264 MB (optimized by previous work).

The 1 MB savings represents maximum optimization within current npm + toolchain constraints.

To reach <200 MB would require:

• pnpm migration (40-50% savings) - separate initiative recommended
• or fundamental tooling changes (remove Babel, switch test frameworks)
• or removal of Obsidian plugin dev workflow (not feasible)

Recommendations

Short Term

• ✅ Monitor bundle size trends (already tracked in build-meta.json)
• ✅ Audit transitive dependencies quarterly

Medium Term

• 📋 Evaluate pnpm migration (Issue TBD - separate PR)
• 📋 Investigate Docker layer caching for CI speedup

Long Term

• 📋 Monitor Obsidian API updates (may reduce moment dependency)
• 📋 Track TypeScript/ESLint/Jest major version updates

Value Delivered

While the 1 MB savings is modest, the primary value is:

1. ✅ Improved dependency hygiene - removed unused packages
2. ✅ Cleaner dependency tree - deduplicated packages
3. ✅ Documented dependency rationale - clear understanding of requirements
4. ✅ Baseline established - future optimizations measurable against this

Closes

• Closes chore: Reduce node_modules size and optimize dependency tree #175

Breaking Changes

None

Additional Notes

This PR represents pragmatic optimization within current constraints. Further size reduction requires infrastructure changes (pnpm migration) warranting separate evaluation.