cambodia-geo

Lightweight, zero-dependency Python library for Cambodia's administrative divisions: provinces (ខេត្ត), districts (ស្រុក/ខណ្ឌ), communes (ឃុំ/សង្កាត់) and villages (ភូមិ). Look things up by code, Khmer name or English name, in any mix.

Data source: data.mef.gov.kh (bundled dataset year: see cambodia_geo.DATA_YEAR).

pip install cambodia-geo

Python 3.9+. No runtime dependencies. Data is bundled and loaded lazily on first use.

Quick start

import cambodia_geo as kh

kh.provinces()                                 # all provinces
kh.districts("Siemreap")                       # districts of a province
kh.communes("17", "Angkor Thum")               # communes of a district
kh.villages("សៀមរាប", 1702, "ជប់តាត្រាវ")        # villages of a commune

p = kh.get_province("សៀមរាប")                   # Province(code='17', name_kh=..., name_en=...)
p.code, p.name_kh, p.name_en
p.to_dict()                                    # plain dict, ready for JSON

kh.get_district("Angkor Thum", province="Siemreap")
kh.get_commune("170201")
kh.search("angkor")                            # across all levels, best matches first

Records can be passed straight back in:

for province in kh.provinces():
    for district in kh.districts(province):
        ...

API

Also: DATA_YEAR (int) and __version__ (str).

Records are frozen dataclasses (hashable, typed, py.typed included):

Matching rules

• A query is a code ("17", 17, "017" and Khmer digits "១៧" are equal), a Khmer name or an English name.
• Names ignore case, spaces and - ' . , : "Siem Reap" == "siemreap".
• With a parent (province=, district=, ...) the first match inside it is returned.
• Without a parent, a name must be unambiguous; if several places share the name, None is returned instead of guessing. Codes are always safe to use alone.

Data notes

The bundled MEF files are shipped exactly as published. The 2025 dataset contains a few upstream quirks:

• Some village codes are shared by several villages of the same commune (e.g. 08120101, and the placeholder 0506140-). Listing functions return all of them; get_village(code) without a parent returns None for these codes instead of guessing, and a scoped get_village(name, commune=...) returns the first match.
• Commune 120913 (Phnom Penh, district 1209) has villages but is missing from the commune list, so villages("12", "1209", "120913") returns [].

Stability (no surprise breaking changes)

• Semantic Versioning. Only names in cambodia_geo.__all__ are public.
• Within a major version, functions, parameters and record fields are never removed or renamed; new options are keyword-only with defaults, new record fields are appended.
• Removals are announced with a DeprecationWarning and a changelog entry first.
• Data updates (once a year) ship as a MINOR release. Government reorganisations can change codes or names between years; pin the version (cambodia-geo~=1.0) if you store codes long-term, and check DATA_YEAR.

Data & licence

The code is MIT licensed. The bundled data comes from the Ministry of Economy and Finance (data.mef.gov.kh, terms of use) and remains subject to its terms.

ភាសាខ្មែរ (សង្ខេប)

បណ្ណាល័យ Python ស្រាល គ្មាន dependency សម្រាប់ស្វែងរក ខេត្ត ស្រុក ឃុំ ភូមិ នៃកម្ពុជា តាមលេខកូដ ឈ្មោះខ្មែរ ឬឈ្មោះអង់គ្លេស។ ដំឡើងដោយ pip install cambodia-geo ហើយប្រើ import cambodia_geo as kh។ ទិន្នន័យអាប់ដេតរៀងរាល់ឆ្នាំ (មើល kh.DATA_YEAR)។

Contributing

pip install -e ".[dev]" then pytest. See RELEASING.md.