اطلس جامعه مدنی ایران — Atlas of Iranian Civil Society

A bilingual (Persian / English) SvelteKit website that maps Iranian civil-society organizations. RTL-first layout, pre-rendered from Markdown frontmatter files.

Live site: AtlasIran.org

git clone https://github.com/Atlasiran/Atlasiran.org.git
cd Atlasiran.org
npm install
npm run dev          # → http://localhost:5173

npm run build        # runs md_to_json.py then Vite build
npm run preview      # preview the production build locally

The source of truth is a CSV file in data/. When new data arrives:

1. python3 deploy/csv_to_json.py — reads the CSV and writes:

   • static/data/data.json — all 226+ orgs (civil + political)
   • static/data/political_parties.json — political orgs only (used by /parties/)

2. python3 deploy/update_org_pages.py — syncs data.json → src/content/org-pages/*.md

3. npm run build — runs deploy/md_to_json.py (frontmatter → data.json) then Vite

The following org_type values are political and appear at /parties/:

• حزب
• سازمان سیاسی
• شورا / کنگره / ائتلاف

All other types are civil society and appear at /groups/.

Each organization is a single .md file in src/content/org-pages/ with YAML frontmatter:

---
id: "123"
org_type: "سازمان غیرانتفاعی"   # e.g. حزب | سازمان سیاسی | سازمان غیرانتفاعی | رسانه | پروژه | ...
pageLink: "/op/slug"
logo: "logos/{id}.png"
name_fa: "نام فارسی"
name_en: "English Name"
name_short: ""
name_local: ""           # Kurdish / Balochi / Arabic etc.
location: ""
post_location: ""
contact: ""              # email or other contact info
phone: ""
about: ""
expertise: ""
history: ""
manifest: ""
coc: ""                  # code of conduct / مرام‌نامه
estimation_of_members: ""
political_orientation: ""
social_telegram: ""
social_instagram: ""
social_x: ""
social_facebook: ""
social_youtube: ""
social_bluesky: ""
social_linkedin: ""
social_tiktok: ""
internetAddress: ""
headerBg: ""
created_at: ""
updated_at: ""
---

After editing files manually, run npm run build to regenerate static/data/data.json.

OG images for org pages are pre-generated locally and committed to static/og/op/. They are not regenerated during CI — Cloudflare's Linux build environment produces distorted Persian text due to FreeType rendering differences vs macOS.

Regenerate after adding or updating org pages:

npm run gen:og
git add static/og/op/ && git commit -m "chore: regenerate OG images"

Requires Python 3 with Pillow, arabic-reshaper, python-bidi, PyYAML (the script installs them automatically via pip3).

The graph at /graph reads static/test.gexf. Regenerate it after data changes:

python3 deploy/gen_gexf.py

data/
  *.csv                       ← source-of-truth CSV from the Atlas database
src/
  routes/
    op/[page]/                ← individual org pages (civil + political)
    p/[page]/                 ← static content pages (about, contact, etc.)
    parties/                  ← political orgs listing
    groups/                   ← civil society listing (نهادها)
    graph/                    ← network graph view
    blog/                     ← blog posts
    api/                      ← JSON API endpoints
  lib/
    components/               ← shared Svelte components
  content/
    org-pages/                ← one .md file per organization
    pages/                    ← static content pages
    posts/                    ← blog posts
static/
  data/data.json              ← all orgs; regenerated at build time
  data/political_parties.json ← political orgs only; used by /parties/
  logos/                      ← org logos ({id}.png)
  og/op/                      ← pre-generated OG images
deploy/
  csv_to_json.py      ← CSV → data.json + political_parties.json
  update_org_pages.py ← data.json → src/content/org-pages/*.md
  md_to_json.py       ← frontmatter → data.json (run at build time)
  gen_gexf.py         ← data.json → static/test.gexf (network graph)
  gen_org_og_images.py← generates OG images
  update_db.py        ← Supabase sync script

AGPL 3