Skip to content

Green-Guides Sources Status

Upstream BMP manual references:

  • 2026: https://dep.nj.gov/stormwater/stormwater-management/bmp-manual/nj-stormwater-best-management-practices-manual-2026/
  • 2023: https://dep.nj.gov/stormwater/stormwater-management/bmp-manual/nj-stormwater-best-management-practices-manual-2023/

Current Canon

For this repository, the authoritative maintained source corpus lives under:

  • docs/Sources

That means:

  • docs/Sources/*/Markdown/*.md is the working source-of-truth text corpus
  • docs/Sources/BMP_*/*.pdf and docs/Sources/NJAC_*/*.pdf are the paired binary source documents
  • docs/Sources/*/raw/ is the repo-local OCR evidence backend
  • docs/Sources/citations/* is the canonical citations and registry layer

External shares, backups, or synced copies may exist, but they are not the live authority for this repo unless they are explicitly re-imported into docs/Sources.

Current Layout

Managed source families:

  • BMP_2023
  • BMP_2026
  • NJAC_2023
  • NJAC_2026
  • citations

Within each BMP or NJAC family:

  • PDFs live at the family root
  • extracted markdown lives under Markdown/
  • raw OCR evidence lives under raw/

Examples:

  • BMP_2023/2023_BMP_1_Impacts_of_Development_on_Runoff.pdf
  • BMP_2023/Markdown/2023_BMP_1_Impacts_of_Development_on_Runoff.md
  • BMP_2023/raw/2023_BMP_1_Impacts_of_Development_on_Runoff/manifest.json

Portable Path Contract

Markdown frontmatter uses file-relative source_pdf paths:

source_pdf: ../2023_BMP_1_Impacts_of_Development_on_Runoff.pdf

Registry and generated citations artifacts use Sources-root-relative paths:

  • BMP_2023/2023_BMP_1_Impacts_of_Development_on_Runoff.pdf
  • BMP_2023/Markdown/2023_BMP_1_Impacts_of_Development_on_Runoff.md
  • BMP_2023/raw/2023_BMP_1_Impacts_of_Development_on_Runoff/manifest.json

This is intentional:

  • markdown frontmatter stays self-referential from the markdown file
  • registry and citations paths stay stable relative to the published Sources/ root
  • no managed contract should emit obsolete *_OCR or *_EXTRACTED family-directory names

Current Generated Artifacts

Authoritative maintained files under Sources/citations/:

  • source_document_registry.json
  • bmp_crosswalk.json
  • bmp_crosswalk.md
  • njac_section_index.json
  • raw_index.json
  • report_claims.json
  • report_claims.md
  • frontmatter_template.md
  • README.md

Current status:

  • registry paths are normalized to the portable family layout
  • citation artifacts are maintained only under docs/Sources/citations
  • the former docs/assets/tools/citations mirror has been retired

Citations Contract

Sources/citations/ is the only canonical citations location in this repo.

Use it for:

  • source identity and path joins
  • raw-evidence indexing
  • BMP crosswalk maintenance
  • report claim tracing
  • frontmatter contract reference

Do not treat assets/tools/citations as authored storage. That mirror was removed so citation maintenance stays single-source.

Raw Evidence Notes

Sources/*/raw/ contains the repo-local OCR evidence bundles used to support QA and citation maintenance.

Important rules:

  • manifest.json is the preferred metadata entrypoint
  • response.json is the deep evidence payload
  • pages/ contains decoded page images when available
  • raw evidence should override stale prose summaries when conflicts appear

Source Identity Rules

  • BMP and NJAC source documents should retain stable registry-style src-* identities
  • PDFs and Markdown twins should share the same source identity
  • broad browse tags remain semantic and structural, not machine-path-derived
  • path fields in registries and manifests should remain relative to the Sources/ root

Deprecated Assumptions

These are obsolete and should not be reused:

  • F:\Paperless-lib
  • \\TRUENAS\media\Paperless-lib
  • \\TRUENAS\knowledge\Curated\Paperless-lib\Engineering\Sources as the live repo authority
  • assets/tools/citations as a maintained citations mirror
  • Sources/2023_BMP_OCR/...
  • Sources/2026_BMP_OCR/...
  • Sources/2023_BMP_EXTRACTED/...
  • Sources/2026_BMP_EXTRACTED/...
  • Sources/2023_NJAC_OCR/...
  • Sources/2026_NJAC_OCR/...
  • Sources/2023_NJAC_EXTRACTED/...
  • Sources/2026_NJAC_EXTRACTED/...

Practical Rule

If you need to regenerate or validate the corpus:

  1. update files directly under docs/Sources
  2. rebuild or refresh the citations artifacts under docs/Sources/citations
  3. validate path consistency against the Sources/-root-relative contract
  4. run the normal site build so published site/Sources/... output stays in parity

As of 2026-04-09, this note reflects the live repo-local source-corpus contract.