ftdoi.org allows you to give a DOI and get back full text URLs for PDF/XML/etc. if
The goal is to allow browser flow (aka:
ftdoi.org/doi) like one often does with
https://oadoi.org, AND for programmatic flow in any programming language of choice (aka:
ftdoi.org/api/...). Right now, only the programmatic flow is supported, but you can of course use the api in browser if you like.
Figuring out URLs for full text version of articles is a huge PITA.
There's https://doi.org for resolving DOIs to their full URLs on the web, and there's https://oadoi.org/ for the same but to OA versions.
However, there's no good tool for figuring out links to versions for text mining: typically either pdf or xml.
There is Crossref's TDM (text and data mining) bit, where publishers voluntarily deposit full text links into Crossref's metadata. However, very few publishers do this; some that do don't deposit correct metadata; some deposit but don't update when they change their URL structure (and publishers change URL stucture A LOT). Checkout my brief analysis.
This API builds on work at https://github.com/ropenscilabs/pubpatterns - which is simply rules for building URLs. The idea here is that we can crowd source the best knowledgee on correct URL patterns to full text versions of articles on a per publisher basis, either by crossref member number, or by DOI prefix. If there's publishers that aren't included, send a PR.
source code for the API is at https://github.com/ropenscilabs/pubpatternsapi