Every .expand request downloads 130K file from schema.org

[orglnte]

I suggest to add some caching because if the library is called many times it's painfully slow

[orglnte]

cache = {}
def caching_document_loader(url):
    loader = jsonld.requests_document_loader()
    if url in cache:
        return cache[url]
    resp = loader(url)
    cache[url] = resp
    return resp

jsonld.set_document_loader(caching_document_loader)

[anatoly-scherbakov]

Filed #253 proposing a disk-based caching DocumentLoader (subclass of the ABC from #250).

[orglnte]

@anatoly-scherbakov Do you check "on startup" if there is a newer document? Otherwise the other side of the medal will show up, which is stale cached docs

[anatoly-scherbakov]

@orglnte this depends on the use case.

• If we are on an air gapped, highly secure system where we value security much more than agility and only allow a whitelist of contexts ⇒ then we use FrozenDocumentLoader Add FrozenDocumentLoader for offline / vetted JSON-LD contexts #250
• If we just want to reduce the number of HTTP requests done from the same machine, we should respect HTTP caching directives and do Implement disk-based cache for remote documents #253 .

I believe we can create a modular system of document loaders fitting different use cases.