Skip to content

VSF Data Profiler v0.2.0-rc1

Pre-release
Pre-release

Choose a tag to compare

@Tan-Long Tan-Long released this 16 Jun 08:36

VSF Data Profiler v0.2.0-rc1 local release candidate.

Highlights:

  • Local-first DBML + multi-CSV profiling with DuckDB-backed scans and bounded pandas materialization.
  • Postgres and MySQL/MariaDB connector modes with chunked extracts, cleanup, and redaction.
  • Deterministic JSON artifacts, Markdown/HTML reports, dashboard chart specs, relationship graph, lineage graph, and dataset verdict.
  • Local web runner on 127.0.0.1 with upload/path modes, dashboard filters, and graph visualization.
  • Export package with offline index, manifest checksums, deterministic zip, and optional analysis_report.pdf.
  • Doctor, artifact audit, benchmark guardrails, and optional guarded LLM narrative.

Validation at release commit 542e374:

  • pytest: 76 passed, 3 skipped.
  • Ruff: passed.
  • node --check web/app.js: passed.
  • npm run test:e2e:dashboard: passed.
  • vsf-profiler doctor: passed.
  • make demo-small, make demo-full, make benchmark-small: passed.
  • make postgres-smoke and make mysql-smoke: clean-skipped without fixture URLs.

Deployment boundary:

  • Vercel is static preflight UI only.
  • Full profiling/dashboard jobs require the local Python backend via vsf-profiler web on 127.0.0.1.