Feature Request: Add 国立国会図書館 (NDL) Search API as a Japanese metadata source

[yamanobe96]

What's Your Idea?

Add the 国立国会図書館 (National Diet Library / NDL) Search API as a metadata provider for Japanese books.

NDL provides a free, public API that supports both ISBN-based and keyword-based lookups:
https://ndlsearch.ndl.go.jp/help/api

Why Would This Be Helpful?

Amazon metadata is currently broken for Japanese users due to anti-scraping measures that return 503 errors regardless of cookie configuration (see #343). Google Books and Goodreads have very poor coverage of Japanese titles.

NDL is Japan's legal deposit library — all publishers are required by law to submit every publication they release. This means NDL's coverage of Japanese books is comprehensive and reliable.

Unlike Amazon, NDL's API is free, stable, and requires no authentication, making it a low-maintenance addition that would significantly improve the experience for Japanese users.

Anything Else? (Optional)

NDL API endpoint example (ISBN lookup):
https://ndlsearch.ndl.go.jp/api/opensearch?isbn=9784488725006

NDL API endpoint example (keyword search):
https://ndlsearch.ndl.go.jp/api/opensearch?title=銀河英雄伝説&creator=田中芳樹

The response is Dublin Core XML, which is straightforward to parse.

Note: NDL metadata quality is lower than Amazon — cover images may be missing or low quality, and descriptions are sometimes absent. However, basic fields (title, author, publisher, ISBN, publication date) are consistently available.

Want to Help Out?

Yes! I'd love to help implement this

Have You Considered Any Alternatives? (Optional)

Amazon (broken due to 503, see #343) and Google Books (poor Japanese coverage) are the only current options. OpenBD, another Japanese book API, shut down in 2023 and is no longer available.

Before Submitting

• I've searched existing issues and confirmed this feature hasn't been requested yet

[coderabbitai[bot]]

🔗 Related PRs

#85 - fix(metadata): fix metadata fetching relying for filename "first" instead of better fields like ISBN in Goodreads/Bookdrop [merged]

---

📝 Issue Planner

_{Check the box below or use the @coderabbitai plan command to generate an implementation plan and prompts that you can use with your favorite coding assistant.}

• Create Plan

---

🧪 Issue enrichment is currently in open beta.

You can configure auto-planning by selecting labels in the issue_enrichment configuration.

To disable automatic issue enrichment, add the following to your .coderabbit.yaml:

issue_enrichment:
  auto_enrich:
    enabled: false

💬 Have feedback or questions? Drop into our discord!

[yamanobe96]

I'm a Japanese user who relies on Grimmory to manage a large personal book collection. I don't have a programming background and cannot implement this myself, but I believe I can contribute by doing the research legwork — testing the API directly and documenting exactly what an implementer would need to know. I hope this saves whoever takes on this feature some time.

I've tested the NDL Search API hands-on from my own server and want to share what I found.

API Endpoints
ISBN lookup:

https://ndlsearch.ndl.go.jp/api/opensearch?isbn={isbn13}
Confirmed working example — Suzumiya Haruhi no Yuuutsu (ISBN 9784044292010) returns 1 result correctly.

Keyword search:

https://ndlsearch.ndl.go.jp/api/opensearch?title={title}&creator={author}&cnt={count}
The same title searched by keyword (title=涼宮ハルヒの憂鬱&creator=谷川流) returns 238 total results. The cnt parameter controls how many are returned per page.

No authentication required
Response encoding: UTF-8
Response format: RSS 2.0 with Dublin Core XML namespaces

metadata.xlsx

No cover image field exists in the NDL API response. This is a known limitation I mentioned in the original issue.

Data Cleaning Notes
Authors (dc:creator):

Birth/death years are often appended: 谷川, 流, 1970- → strip trailing , 1970-
Name separators vary across records: 田中/芳樹 (slash) and 谷川, 流 (comma-space) both appear
Multiple dc:creator elements should all be collected
ISBN (dc:identifier):

Hyphens are present: 4-04-429201-9 → strip hyphens before storing
Both ISBN-10 and ISBN-13 appear; distinguish by digit count after stripping hyphens (10 vs 13 digits)
Page count (dc:extent):

307p → extract 307
1冊(ページ付なし) → no page count available (common for manga)
Publisher (dc:publisher):

Multiple elements can appear: primary publisher + distributor
Example: 角川書店 and 角川グループパブリッシング (発売) — the distributor is marked with (発売) and can be filtered out
ISBN Lookup Behavior
ISBN lookup works but has coverage gaps:

Well-indexed books return 1 result cleanly (confirmed with 9784044292010)
Some ISBNs return 0 results even for legitimate books
Suggested fallback: when ISBN returns 0 results, retry with keyword search using title + creator
Result Filtering
Keyword searches return all media types, not just books. The element should be used to filter results:

Keep: 図書 (book / printed matter)
Exclude: 映像資料 (video), 録音資料 (audio), 地図 (map), etc.
The first result for 涼宮ハルヒの憂鬱 is the anime DVD, not the novel — so this filtering is important in practice.

NDL Detail Page URL Pattern
Each result includes a pointing to its NDL detail page, following this pattern:

https://ndlsearch.ndl.go.jp/books/R100000002-I{NDLBibID}
This could be stored as a source reference URL alongside the metadata.

I'm happy to run additional API tests or answer any questions if it would help move this forward.