Pre-flight
Problem
Live ISBN/provider fallback testing showed that translated or noisy edition titles can make canonical OpenLibrary matching harder than it needs to be. For example, German Dune ISBNs can appear as Der Wüstenplanet or Dune - Der Wüstenplanet: Roman, while the desired canonical work is OpenLibrary OL893415W / Dune by Frank Herbert.
Hardcover can provide a useful optional bridge for this class of cases: ISBN edition records may carry the localized edition title while their linked Hardcover book carries the canonical title and author. In sampled Dune records, Hardcover mapped German edition titles back to canonical book title Dune. The alternative_titles field existed but was empty, so edition title plus canonical book title appears to be the useful datapoint.
Proposed solution
Add an optional Hardcover-backed canonicalization hint for ISBN lookups when a Hardcover token is configured:
- Query Hardcover editions by normalized ISBN and capture both the edition
title and linked book title/author.
- Treat the edition title and canonical Hardcover book title as title aliases or canonicalization hints during aggregator ISBN fallback.
- Use those hints to form additional OpenLibrary title+author canonical searches before accepting a noisy DNB/Google/duplicate-OpenLibrary result.
- Keep this strictly optional: if Hardcover is unconfigured, unavailable, or returns no result, the existing OpenLibrary/DNB/Google path should still work.
- Do not rely solely on Hardcover
alternative_titles; parse it when present, but use edition title and canonical book title as the primary signal.
Alternatives considered
- Rely only on OpenLibrary title+author search variants. This is still necessary, but Hardcover can provide a stronger external hint for ISBN-specific translated editions when configured.
- Make Hardcover required for multilingual canonicalization. This would regress no-token setups and DNB-only workflows.
- Use hard-coded translation aliases such as
Der Wüstenplanet → Dune. That would not scale and would be difficult to maintain.
Additional context
This is a follow-up enhancement, not the primary fix for OpenLibrary canonical matching. The OpenLibrary-only canonical path still needs to work because live canonical tests instantiate the aggregator with only OpenLibrary. Hardcover should be an optional extra datapoint for production ISBN fallback when credentials are available.
Pre-flight
Problem
Live ISBN/provider fallback testing showed that translated or noisy edition titles can make canonical OpenLibrary matching harder than it needs to be. For example, German Dune ISBNs can appear as
Der WüstenplanetorDune - Der Wüstenplanet: Roman, while the desired canonical work is OpenLibraryOL893415W/Duneby Frank Herbert.Hardcover can provide a useful optional bridge for this class of cases: ISBN edition records may carry the localized edition title while their linked Hardcover book carries the canonical title and author. In sampled Dune records, Hardcover mapped German edition titles back to canonical book title
Dune. Thealternative_titlesfield existed but was empty, so edition title plus canonical book title appears to be the useful datapoint.Proposed solution
Add an optional Hardcover-backed canonicalization hint for ISBN lookups when a Hardcover token is configured:
titleand linked booktitle/author.alternative_titles; parse it when present, but use edition title and canonical book title as the primary signal.Alternatives considered
Der Wüstenplanet→Dune. That would not scale and would be difficult to maintain.Additional context
This is a follow-up enhancement, not the primary fix for OpenLibrary canonical matching. The OpenLibrary-only canonical path still needs to work because live canonical tests instantiate the aggregator with only OpenLibrary. Hardcover should be an optional extra datapoint for production ISBN fallback when credentials are available.