Japan Postal Code & Address Normalizer (郵便番号・住所)

Bulk postal code → address, address → postal code and address normalisation for Japan, built on Japan Post's official KEN_ALL data (124,000+ rows), downloaded fresh on every run (a bundled snapshot is the fallback if their site is unreachable).

• Handles 〒１００−０００１, 100 0001, full-width digits, kanji numerals (大通西二十丁目), ー used as a hyphen, 大字 prefixes, omitted prefectures, ヶ/ケ variants.
• Picks the right code when a town has several (e.g. 札幌市中央区大通西 1–19丁目 vs 20–28丁目) by reading the block number, and says so when it cannot: every address result carries a confidence.
• Deterministic, no scraping, no AI guessing, no API keys. Works through the Apify MCP server, so AI agents can validate addresses instead of inventing postal codes.

Input

• items: postal codes and/or addresses (up to 100,000 per run).
• mode: auto (7 digits = postal code, otherwise address), zip, address, normalize.

Output

Address → postal code:

{ "input": "北海道札幌市中央区大通西二十丁目", "prefecture": "北海道", "city": "札幌市中央区", "town": "大通西", "block": "20丁目",
  "normalized": "北海道札幌市中央区大通西20丁目", "zip": "064-0820", "confidence": "town", "found": true }

confidence: town (town and block range matched one code) · city (only the city-level general code is known) · ambiguous (several codes fit: zip is null, candidates lists them) · none.

Postal code → address returns every row with that code (some codes cover several towns): prefecture, city, town, notes, JIS city code, katakana readings.

Pricing

Pay per event: one result event per found item. Not-found items and input errors are free.

Measured accuracy and honest limits

• Self-consistency check (Japan Post's own town names + a dummy block number, 16,580 sampled towns): 99.92% exact code, 0.08% reported as ambiguous, 0 wrong. This tests the matching logic, not how people actually write addresses: real-world addresses with building names, typos, or old municipality names will do worse, and the confidence field tells you when.
• No geocoding (latitude/longitude) and no building-level matching. Large-office codes (個別事業所) are returned for postal code lookups but not guessed from addresses.
• Municipal mergers: the data reflects Japan Post's current list, so old names (e.g. pre-merger 町村) may return none.

Source: 日本郵便「郵便番号データダウンロード」(https://www.post.japanpost.jp/zipcode/download.html). Japan Post states it claims no copyright on this data and it may be freely distributed.