Published

Singapore (SG) build pipeline

DataHub
DataHub
0

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

Singapore (SG) build pipeline

Builds from a staged Universal-Schema (FedEx-pipeline) export, not a live fetch — the raw postal source was already scraped once and staged as sources/singapore_fedex_base.csv (121,513 rows, one per postal code, building-level, 0 duplicates). The original scraper still exists and is maintained separately at ~/Desktop/Research/fedex-agents-prototype/countries/SGP/scripts/scrape_postal.py if the base ever needs re-fetching from scratch — not duplicated here, this pipeline picks up from the staged export.

Column mapping (Universal Schema -> 13-column postal-codes schema):

SourceTargetNote
countryCodecountry_codeSG, not SGP — this repo's convention is alpha-2
postalCodepostal_code
additionalInfoplace_nameBUILDING name if present, else ROAD_NAME, else subzone as last resort
stateName / stateCodeadmin_name1 / admin_code1CDC District
countyadmin_name2URA Planning Area; no code source exists
cityNameadmin_name3URA Subzone; no code source exists
(none)alternative_city_namenothing available — left blank, not invented

latitude/longitude/accuracy are intentionally left blank by build_base.py — join_coordinates.py fills them from a separately staged geocode lookup (sources/singapore_coordinates.csv) in a second pass, the same two-stage shape as Canada's admin2 enrichment.

Run order

python3 build_base.py
python3 join_coordinates.py
python3 integrity_checker.py
python3 package.py
python3 publish_r2.py          # dry run by default; --apply to publish

Verification (2026-08-19)

Byte-identical. All 4 output files (csv, datapackage.json, README.md, zip) match what's live on R2 exactly. publish_r2.py (written this session — it didn't exist before) confirms a zero-diff plan on dry run.