Skip to content

export-obsidian.md

Latest commit

 

History

History
executable file
·
131 lines (99 loc) · 4.45 KB

export-obsidian.md

File metadata and controls

executable file
·
131 lines (99 loc) · 4.45 KB

Export Obsidian Script

Overview

export-obsidian.mjs exports the Supabase database into an Obsidian-friendly folder structure. It pulls issues, pages, and OCR generation history, then writes Markdown notes that link to each other.

This script mirrors the .env loading pattern used in other root scripts.

What It Exports

  • Issues (from public.issues)
  • Pages (from public.pages)
  • OCR generations (from public.ocr_generations)

Each issue is grouped into a year based on publication_date, title, or volume (see "Year inference" below).

Output Structure

The export creates (or updates) these files under the target folder:

  • Overview.md
  • Years/<year>.md
  • Issues/<issue-slug>.md
  • Generations/<issue-slug>-page-###-gen-<id>.md
  • Dashboards/Authors.md
  • Dashboards/Tags.md
  • Dashboards/Recent Issues.md

Where:

  • <year> is a 4-digit year or unknown
  • <issue-slug> is a lowercase, dash-separated slug derived from the issue title
  • ### is zero-padded page number (e.g., 001)

Note Contents

Overview.md

  • Links to each year and issue.

Years/<year>.md

  • Links to all issues in that year.

Issues/<issue-slug>.md

  • Metadata: id, volume, publication_date, authors, tags, created_at, updated_at
  • Stats: total pages, pages with OCR text, pages with OCR generations
  • One line per page with an online image link and OCR generation links

Generations/<issue-slug>-page-###-gen-<id>.md

  • Metadata: generation id, page id, issue id, created_at, model, image_path
  • OCR prompt/output from ocr_generations

Dashboards/Authors.md

  • Groups issues by author, listing each author, the number of issues they appear on, and links to the related issue notes.
  • Uses the authors field already fetched from issues (no extra API calls or dependencies).

Dashboards/Tags.md

  • Groups issues by tag, showing each tag, how many issues use it, and links to those issue notes.
  • Uses the tags field already fetched from issues (no extra API calls or dependencies).

Dashboards/Recent Issues.md

  • Lists the most recently created issues (sorted by created_at) with links to their notes and any available authors and tags metadata inline.
  • Uses existing issue metadata only; no new dependencies are required.

Authors and tags are read directly from the issues rows. Arrays are used as-is; comma, semicolon, or newline-delimited strings are split into individual entries before grouping and linking.

Environment Variables

Required:

  • VITE_SUPABASE_URL
  • VITE_SUPABASE_ANON_KEY Optional:
  • VITE_IMAGE_BASE_URL (prefix for pages.image_path when linking images)

Optional:

  • OBSIDIAN_EXPORT_DIR (used if --out is not provided)

The script looks for .env in:

  1. ./.env (project root) if present
  2. ./kitanocr-web/.env otherwise

You can override with --env <path>.

Usage

node export-obsidian.mjs --out /path/to/ObsidianVault

Other CLI options:

  • --out, --folder, or --vault to set output folder
  • --env <path> to set the env file
  • -h or --help to print usage

Database Assumptions

Tables and columns used:

public.issues

  • id, title, volume, publication_date, authors, tags, created_at, updated_at
  • Dashboards assume authors and tags are available (arrays or delimited strings) so they can group and label issues without extra queries.

public.pages

  • id, issue_id, page_number, image_path, status, ocr_text, created_at, updated_at

public.ocr_generations

  • id, page_id, model, prompt, output, metadata, created_at

If columns are added/removed, update the select('*') fields or note templates.

Year Inference Logic

Year is determined in this order:

  1. publication_date (parsed as UTC year)
  2. title/volume containing a 4-digit year
  3. title/volume that looks like YYYYMM or YYYYMMDD (uses first 4)
  4. Fallback to unknown

To change this behavior, update inferYear() in export-obsidian.mjs.

Slugging Rules

Issue slugs:

  • Lowercase
  • Non-alphanumeric replaced with -
  • Multiple dashes collapsed

To change this behavior, update slugify() in export-obsidian.mjs.

Maintenance Notes

  • Export pagination is fixed at 1000 rows per request.
  • Obsidian notes are overwritten on each run.
  • OCR generation history is written to separate generation notes.

If exports become large, consider:

  • Adding filters (by year, issue, or page range)
  • Splitting OCR generations into separate notes
  • Using select() with only necessary columns to reduce payload size