Division -- 8/8 covered, including Mymensingh Division (corrected from GeoNames' stale 'Dhaka Division' label on the 77 rows GeoNames' own internal coding already separates, see ATTRIBUTION.txt)
Administrative Name 1
admin_code1
string
ISO 3166-2:BD code
Administrative Code 1
admin_name2
string
District -- 64 distinct, 100% populated
Administrative Name 2
admin_code2
string
GeoNames' own internal district code (not an independently verified official code)
Administrative Code 2
admin_name3
string
Thana/Upazila -- 483 distinct, 100% populated
Administrative Name 3
admin_code3
string
Not populated -- no code accompanies this tier in the source
No per-country caveat for Bangladesh in GeoNames' export/zip/readme.txt —
the general CC BY 4.0 licence applies without restriction.
Verified directly against the raw file: 1,349 rows, all 1,349 distinct
postal codes (no duplicates on the code itself). Unlike South Africa's
postal export, this one already carries three real admin tiers on every
single row: admin_name1 (Division, e.g. "Dhaka Division"), admin_name2
(District, e.g. "Dhaka"), and admin_name3 (Thana/Upazila, e.g. "Paltan") —
100% populated on all three, no join or enrichment needed.
Official source — checked, no bulk export found
Bangladesh Post Office's own site (bangladeshpost.gov.bd) was checked via
search; no bulk postal-code download, API, or open-data catalogue entry was
found. Bangladesh Bureau of Statistics was also checked — no postal-code
dataset located (BBS publishes census/administrative geography, not postal
codes).
Third-party alternatives — checked, none better than GeoNames
Search surfaced several community mirrors (postalbin.com, bgd.postcodebase.com)
and PDF scans of the printed postal directory (Scribd uploads of the
official directory, government district-portal PDFs). None offers a
machine-readable bulk file with a stated open licence — they are either
copies of the same printed directory with no licence grant, or read-only
lookup sites. None was pursued further given GeoNames already provides
equal or better structured coverage under a clean, unambiguous licence.
B ($49). Single source normalised in place — real work (ISO 3166-2:BD
crosswalk for admin_code1, cleaning GeoNames' internal admin1/admin2 codes
into real identifiers if any exist) but no second authoritative source to
consolidate. Not A.
Verdict
Clean and strong for its size — smallest of this batch's three
(1,349 rows) but the only one of the three with full three-tier admin
coverage already present in the postal export itself, no gazetteer join
required. Not yet built; this is the research/decision record only.
Build — 2026-08-30 (local only, not published)
Built via datasets/bd/scripts/build_postal_codes_product.py. Found and
fixed a real defect not caught in the research pass: 77 rows carried
admin_name1="Dhaka Division" but a DIFFERENT internal GeoNames code ("H")
than the other 298 real Dhaka rows ("81") — those 77 rows' districts
(Jamalpur, Mymensingh, Netrakona, Sherpur) are exactly Mymensingh Division,
split from Dhaka in 2015. GeoNames' own coding already separated them
correctly; only the name string was stale. Corrected to "Mymensingh
Division" — result is 8/8 real divisions, 64 districts (matches
Bangladesh's real district count exactly), 483 thanas, all populated on
100% of rows. admin_code1 derived to verified ISO 3166-2:BD codes
(A–H, confirmed against Wikipedia).
Frictionless validation: valid: True.
Status: built and validated locally only. Nothing published to R2,
nothing committed to git. datasets/bd/README.md and datapackage.yml
updated in place (tier B, premium_source_key: bd/bd.zip, real
sources/licenses, replacing the "Multiple sources" placeholder).
API access
Access dataset files directly from scripts, code, or AI agents.
Each file has a stable URL (r-link) that you can use directly in scripts, apps, or AI agents. These URLs are permanent and safe to hardcode.