Well I can hear some curious minds, well in the directory structure for example found in: https://sourceforge.net/projects/chronolabsapis/files/labs.coop/fonts/data%20packagings/ which will be periodically scanned by a cronjob to make up all the folder names of ‘tags’ which in the documentation will be refereed to as ‘nodes’ for example any keyword after the initial break down and uniqueness of the keyword, if it occurs more than once in a sub-branch of the tree is still consider on occurring once.
There will be call like http://fonts.labs.coop/v1/nodes.json or for a search on lesser not the complete list :- http://fonts.labs.coop/v1/node2-nodeb-nodec/nodes.json
Then there will be function say a random font selected from a batch of keyword provided in a style sheet: http://fonts.labs.coop/v1/nodeb-noded-nodee/random.css You will for example be able to get a populated list of all the fonts as:- http://fonts.labs.coop/v1/fonts.json or just with in terms http://fonts.labs.coop/v1/nodea-nodeb-nodef/fonts.json. each node is only lowercase and separated by a dash!
Each font will have when the list is populated by the cron cause in things like API if you can avoid using a database you do, solid storage is so much fast to access if you don’t need higher database functions. Anyway in the directory contains ZIP files with each individual zip having a unique fingerprint called an MD5 checksum which you can get of the api by for example a true type: http://fonts.labs.coop/v1/xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx.ttf or a woff of the same font: http://fonts.labs.coop/v1/xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx.woff which gets them directly from inside the zip and cue to the output buffer.
There is also going to be a http://email@example.com/upload.api for a form submission upload similar to OCR that periodically every half hour will sort and convert and store that font into the library section until the cron runs to index it and you get an email to say you checksum, nodes etc.