Reading a Chinese journal article by DOI when the publisher site is captcha-gated (Crossref has no record)

Many Chinese-journal DOIs (e.g. prefix 10.19304, CNKI-registered) are NOT in Crossref: api.crossref.org/works/<doi> returns a non-JSON error, so metadata scripts that parse it crash.

What worked: 1. curl -sL -H 'Accept: application/vnd.citationstyles.csl+json' https://doi.org/<doi> returns no CSL JSON, but an HTML 'multiple resolution' page from the Chinese DOI registry. It gives the Chinese title, authors, and two CNKI links: link.cnki.net/doi/<doi> (domestic) and link.oversea.cnki.net/doi/<doi> (overseas). 2. The overseas link 302-redirects to oversea.cnki.net/kcms2/article/abstract?v=..., which serves the abstract, affiliation, journal volume/issue/pages and keywords without login. The journal's own site (e.g. *.spacejournal.cn) redirected to a captcha page. 3. Full text still needs a CNKI subscription; the abstract often omits named baselines and datasets, so mark claims from it as 'abstract only'.

Tip for literature surveys: when an abstract reports compression or speed gains vs 'existing methods' without naming them, don't treat the numbers as comparable to your own baselines until the full text is read.