Japan data.go.jp Open Data Catalog (CKAN API)

Search Japan's government open-data catalog (data.go.jp) across every ministry and municipality by keyword, publisher, category or format — then actually download a dataset's CSV rows as clean, structured JSON (Shift_JIS handled automatically).

This Actor is a thin, polite client for the CKAN API v3 (data.e-gov.go.jp/data/api/3/action/) behind the データカタログサイト (data.go.jp), the nationwide open-data catalog operated by the Digital Agency (デジタル庁) — the legacy data.go.jp host now redirects to the e-Gov data portal that serves it. It is not a scraper — it calls the CKAN JSON API, needs no API key, and returns each result as a flat record with a legal envelope (source / sourceUrl / license / retrievedAt) on every item.

Unlike a single-city catalog, this covers all central-government ministries and local governments in one search (~18,000 datasets). The catalogue as a whole is a directory, not a product. The value here is get_records: point it at a dataset and it fetches the actual CSV distribution and returns its rows — so you get the data, not just a link to it.

Unofficial tool. Independently built and maintained. Not affiliated with, endorsed by, or connected to the Japan Digital Agency or any publishing ministry/municipality. Catalogue content is provided by default under the 公共データ利用規約 第1.0版 (PDL1.0), which permits commercial reuse with attribution; the required 出典 statement 「出典：データカタログサイト（data.go.jp）」 is embedded in the license of every record.

---

Quick start — verified input

Copy, paste, run. This exact input is verified on the platform (SUCCEEDED, items > 0):

{
  "mode": "search_datasets",
  "query": "人口"
}

Running with no input at all also works (same defaults). To download a dataset's rows, take a datasetId from the search output and run {"mode": "get_records", "datasetId": "<that id>"} (only CSV/TSV resources are read — add "format": "CSV" to the search to find them).

No API key required

The data.go.jp CKAN API is public and unauthenticated. Just pick a mode and run. The Actor keeps a single connection, spaces requests ≥ 1.2 s apart, backs off exponentially on HTTP 429/5xx, and fails visibly on a persistent block — it never attempts rate-limit evasion.

License, per-dataset diversity & source-display obligation (出典記載義務)

The 利用規約 states (原文): 「本サイトで掲載・発信している情報……の著作権は、特記されていない限りデジタル庁に帰属し、権利表記の記載がない限り「公共データ利用規約（第1.0版）」（PDL1.0）が適用されています。」「本コンテンツを利用する際は出典を記載してください。」 — i.e. the default license is PDL1.0 (CC BY 4.0-compatible), commercial use permitted, with attribution.

• The credit 「出典：データカタログサイト（data.go.jp）」 is embedded verbatim in the license field of every record. If you display or republish the data, keep that credit visible.

• Per-dataset licenses vary and are your responsibility. The terms explicitly carve out content that is 「特記されている」 / carries its own 権利表記. This catalogue's CKAN license vocabulary includes not just open licenses (cc-by, cc-by-4.0, cc-by-sa, cc-zero, pdl-1.0, gjstu) but also non-commercial ones (cc-by-nc, cc-by-nc-sa, cc-by-nc-nd) and non-open ones (other-closed). At data.go.jp the real per-dataset signal usually lives on each resource, so the Actor surfaces:

  • licenseId — the dataset-level CKAN license id (often empty here).
  • resourceLicenseIds — the distinct license ids across the dataset's resources.
  • resources[].licenseId — each individual resource's own license id.
  • In get_records, every output row inherits its source resource's licenseId.

  You must confirm and comply with each dataset's own license and content — including whether it is non-commercial or non-open, and whether it contains personal data — before use.

Two modes

1. search_datasets — find datasets (CKAN package_search)

Give a query (free text) and/or an organization, tags, group, or format filter, and get back matching datasets: datasetId, title, publisher (ministry/municipality), update date, license ids, a catalogue deep link, and the full resources[] list (each resource's URL, format, size and license). At least one filter is required (a targeted search, not a full-catalogue dump).

2. get_records — download a dataset's CSV rows

Give a datasetId (from a search_datasets run) and the Actor resolves that dataset's CSV/TSV resources, downloads them, and returns one record per CSV row — the columns keyed by header, plus provenance (datasetId, resourceName, resourceUrl, licenseId, rowIndex). Or pass a resourceUrl directly. Japanese open-data CSVs are frequently Shift_JIS (confirmed live: an e-stat resource served charset=UTF-8 in its header yet was Shift_JIS on the wire); encoding: "auto" (default) sniffs UTF-8 (incl. BOM) then falls back to Shift_JIS, or force utf-8 / shift_jis.

Sample output

search_datasets (dataset item):

{
  "datasetId": "mhlw_20260624_2024",
  "title": "食中毒統計調査＿令和６年食中毒統計調査＿年次＿2024年",
  "organization": "厚生労働省",
  "organizationId": "org_1600",
  "publisher": "厚生労働省",
  "licenseId": null,
  "resourceLicenseIds": ["cc-by"],
  "isOpen": false,
  "lastModified": "2026-06-24T...",
  "datasetUrl": "https://data.e-gov.go.jp/data/dataset/mhlw_20260624_2024",
  "resources": [
    { "name": "第1表 ...", "format": "CSV", "url": "https://www.e-stat.go.jp/.../file-download?...", "licenseId": "cc-by" }
  ],
  "source": "データカタログサイト（data.go.jp）/ Japan Data Catalog site (data.go.jp, e-Govデータポータル, CKAN)",
  "sourceUrl": "https://www.data.go.jp/",
  "license": "出典：データカタログサイト（data.go.jp）… PDL1.0 … licenseId … Unofficial …",
  "retrievedAt": "2026-08-26T12:00:00.000Z"
}

get_records (dataset item — one CSV row):

{
  "datasetId": "mhlw_20260624_2024",
  "datasetTitle": "食中毒統計調査 ...",
  "resourceName": "第1表 ...",
  "resourceUrl": "https://www.e-stat.go.jp/.../file-download?...",
  "licenseId": "cc-by",
  "rowIndex": 1,
  "fields": { "都道府県名等": "全国", "...": "..." },
  "source": "データカタログサイト（data.go.jp）...",
  "sourceUrl": "https://www.data.go.jp/",
  "license": "出典：データカタログサイト（data.go.jp）… PDL1.0 … Unofficial …",
  "retrievedAt": "2026-08-26T12:00:00.000Z"
}

Input reference

Field	Mode	Description
mode	both	search_datasets (default) or get_records. Case-insensitive; if you pass only a datasetId/resourceUrl the Actor switches to get_records for you.
query	search	Free-text query (CKAN q), Japanese works best, e.g. 人口, 医療機関.
organization	search	CKAN organization name, e.g. org_1600 (厚生労働省).
tags	search	Restrict to a CKAN tag.
group	search	Category/group name (see the groups field of search results).
format	search	Resource format filter, e.g. CSV, XLSX, PDF, JSON.
datasetId	get_records	Dataset name/slug to read (from the search output's datasetId).
resourceUrl	get_records	Fetch a specific CSV resource URL directly.
encoding	get_records	auto (default), utf-8, or shift_jis (sjis, cp932 … accepted).
maxItems	both	Max records to output, 1–100000 (default 1000).
maxApiRequests	both	Hard per-run upstream request cap, 1–25 (default 10).
proxyConfiguration	both	Optional Apify proxy; default is a direct connection.

At least one of query / organization / tags / group / format is required in search_datasets mode; datasetId or resourceUrl is required in get_records mode. Values are validated before the first request; an invalid value fails the run immediately with a message that lists the valid values.

Common input mistakes

Mistake	Correct
{"mode": "search_datasets"} with no filter	add "query": "人口" (or organization / tags / group / format)
{"mode": "search_datasets", "datasetId": "mhlw_20260624_2024"}	"mode": "get_records" — datasetId is a get_records field
{"mode": "get_records", "query": "人口"}	get_records needs datasetId or resourceUrl, not a query
"datasetId": "https://data.go.jp/data/dataset/mhlw_20260624_2024"	just the slug: "mhlw_20260624_2024"
"organization": "厚生労働省"	the CKAN name: "org_1600" (find it in the organizationId field of a search)
"encoding": "euc-jp"	"auto", "utf-8" or "shift_jis"

Empty results?

A run that finds nothing completes with 0 items and a warning in the log (not a failure). Typical causes: an English query (the catalogue is Japanese — try 人口 instead of population), a tags/group/organization value that is not an exact CKAN name, or a get_records dataset whose resources are all XLSX/PDF (only CSV/TSV are read — search with "format": "CSV" or pass a resourceUrl to a CSV). Broaden the query or drop a filter and retry.

Notes & limits

• get_records reads CSV/TSV resources. Non-CSV distributions (XLSX, PDF, HTML, etc.) are skipped — pass a resourceUrl to a CSV, or read the other format yourself. Catalogue-wide, PDF/HTML/XLS dominate and CSV is a minority, so use the format: "CSV" search filter to find machine-readable datasets.
• Resource files are hosted on the publishers' own servers (ministry / e-Stat / municipal domains). A resource that responds with a redirect surfaces as an error (the Actor does not follow redirects to keep the request budget honest) — re-run with the final URL.
• search_datasets pages at up to 1000 datasets per request under maxItems.

Disclaimer

1. Unofficial — not affiliated with, endorsed by, or connected to the Japan Digital Agency or any publishing ministry / municipality. This tool uses the public data.go.jp CKAN API.
2. Public data, your responsibility — content is provided by default under PDL1.0 (公共データ利用規約 第1.0版). Keep the 出典 credit visible. Each dataset's own license (licenseId / resourceLicenseIds) — which may be non-commercial or non-open — and its content, including any personal data it may contain, are the user's responsibility to verify and comply with before use. Compliance with the source's terms, and applicable law (incl. GDPR / Japan's APPI where relevant), is your responsibility.
3. Polite by design — one connection, ≥ 1.2 s spacing, exponential backoff, a hard request budget, and a visible failure on a persistent block. No rate-limit evasion.
4. No warranty — provided "as is"; verify anything important against the source catalogue.

Search terms

japan government data catalog, data.go.jp api, japan open data search english, japan open data csv, CKAN japan, japanese government datasets, e-gov data portal, japan ministry open data, digital agency japan data, japan municipal open data, 政府 オープンデータ, data.go.jp csv download