Published

Indonesia (ID) build pipeline

DataHub
DataHub
0

Postal Codes Dataset for Indonesia, ID including name of the city, town, or place, various administrative divisions and alternative city names.

Indonesia (ID) build pipeline

Self-contained — runs entirely from this repo, no fedex-agents-prototype dependency. Fixed 2026-08-19: every script previously computed ROOT = parents[3] and hardcoded countries/IDN/... subpaths, which only resolved correctly in that other repo's layout. All paths are now self-relative (ID_DIR = parent of this scripts/ folder), and data/, sources/, id_id/ all live as siblings under datasets/id/.

The buyer-facing README.md shipped in the delivery is a static, hand-authored document (real row counts and coverage stats baked in — not regenerated per run). It's committed here as delivery_README_template.md (next to this file, since data/ and id_id/ are gitignored/regenerable) and run_qa.py copies it into the delivery folder — update that template by hand if the shipped numbers ever change.

Run order (from this directory)

mkdir -p ../sources
curl -sL -o ../sources/wilayah.sql          https://raw.githubusercontent.com/cahyadsn/wilayah/master/db/wilayah.sql
curl -sL -o ../sources/wilayah_kodepos.json https://raw.githubusercontent.com/cahyadsn/wilayah_kodepos/master/json/wilayah_kodepos.json
curl -sL -o ../sources/ID.zip               https://download.geonames.org/export/zip/ID.zip
unzip -o -q ../sources/ID.zip ID.txt -d ../sources/

python3 build_admin.py
python3 build_postal.py
python3 build_deliverable.py
python3 run_qa.py
python3 publish_r2.py    # dry run by default; --apply to publish

See countries/IDN/README.md in fedex-agents-prototype (if available — not required to run this pipeline) for the original documented known-issues list: why the two admin sources disagree on ~14% of villages, and why unresolved rows are shipped with empty admin fields rather than guessed.

Sources

  • cahyadsn/wilayah + wilayah_kodepos (GitHub, MIT) — Indonesian administrative hierarchy and village-to-postcode mapping
  • GeoNames ID — CC BY 4.0

Verification (2026-08-19)

Fresh end-to-end run, MD5-identical to live R2. All sources fetched fresh into ../sources/. build_admin.py produced 38 provinsi / 514 kab-kota / 7,285 kecamatan / 83,762 desa — exact match to the counts documented in fedex-agents-prototype/CLAUDE.md's Admin Level Mapping Registry. build_deliverable.py produced 81,044 rows, cross-check agreement 69.2%/30.8% — exact match to the documented figures. Final id.csv MD5 f8535e27c7cced5181606a65e3233b8c, byte-for-byte identical to the live R2 object (same MD5, same 7,656,248 bytes). Frictionless valid. Publish dry-run: csv/datapackage/README byte-identical, zip differs by 2 bytes (zip container metadata/timestamp only, not content).