Geographic source and license
Current postal geography comes from country-level GeoNames postal code files under CC BY 4.0. The import manifest records source URL, SHA-256 checksum, source-row count, import date and processing notes for each country.
Japanese place names are matched to Japan Post native and romanized postal datasets, preserving the original postcodes, coordinates and time zones. Unverified dependent locality names are omitted; source dates and checksums are recorded separately.
Address order is not inferred from GeoNames. It is documented separately using USPS, Japan Post, Deutsche Post, PTT or UPU postal guidance, with the relevant standards link shown in each country tool’s format section.
How regions and locality records are selected
The import pipeline first preserves complete first-level administrative coverage, then selects up to eight representative locality records per area based on postal coverage. Administrative area, locality, dependent locality, postcode, coordinates and derived IANA time zone remain on one record and are never randomized independently.
Which fields are synthetic
Street, building, unit, person, company, employment, TEST-marked document and non-account financial fields are created by the local generator. Names, companies, departments and job titles use localized Faker definitions, with project-owned German/Turkish job definitions and Turkish company patterns where native Faker data is missing. Job titles retain the selected department. Email uses the reserved example.com domain and phone values use Faker country templates without full numbering-plan or reachability verification. A complete record is never presented as a real person, account or deliverable address.
Seeds and reproducibility
The same country, region, locality, seed and dataset version produce the same result. A dataset update can change output even when the seed stays constant, so defect reports and automated cases should store both seed and dataset version.
Quality checks and known limits
Every published country requires fixed-seed sample tests, complete first-level-area checks, locality-to-postcode checks and native-language and single-line/multiline formatting tests. Coverage is representative rather than a complete national address database, and coordinates describe the postal record—not an exact synthetic building.
Updates and corrections
Dataset versions appear in the generator and country coverage sections. Every re-import records a new date, checksum and processing version. To report a missing area, postal relationship or format issue, include country, page URL, dataset version and supporting source.