linkedsoftwaredependencies / npm-extraction-server Goto Github PK

Stuff like "https://linkedsoftwaredependencies.orgengines/node" should be "https://linkedsoftwaredependencies.org/engines/node" I guess

Hi,

Is it possible to add the FOAF profile URL of the authors, contributors, and so on to the RDF?

KR
Pieter

When the server receives a path such as http://localhost:3000/bundles/npm/@solid/community-server/3.0.0/components/context.jsonld, it has to download the tarball and look in there to find that exact resource. It does this by taking the tail, components/context.jsonld, and looking for that path. The problem is that this doesn't work if there is no direct mapping between the URL tail and the file path. The package above, for example, has the following in it's package.json:

  "lsd:importPaths": {
    "https://linkedsoftwaredependencies.org/bundles/npm/@solid/community-server/^3.0.0/components/": "dist/components/",
    "https://linkedsoftwaredependencies.org/bundles/npm/@solid/community-server/^3.0.0/config/": "config/",
    "https://linkedsoftwaredependencies.org/bundles/npm/@solid/community-server/^3.0.0/dist/": "dist/",
    "https://linkedsoftwaredependencies.org/bundles/npm/@solid/community-server/^3.0.0/templates/config/": "templates/config/"
  }

So it should actually find dist/components/context.jsonld. This requires checking if the incoming URL starts with one of the import paths, or is at least equal by using semver rules on the version number, notice the difference between 3.0.0 and ^3.0.0 above. It then has to check that the trailing part of the URL matches the trailing part of one of the importPath URLs. And then it has to replace the overlapping part with the value of the map. This makes the operation quite a bit more complicated but is feasible to do.

The issue that I have, and why I'm tagging @rubensworks, is because there are 2 other possible ways to configure the package.json and I'm not sure what is expected there.

First case is having "lsd:module": true in your config. How is the server supposed to know how to map the URL to a file path in the tarball? I had a look at the example in the documentation, but that caused more confusion since I have no idea how the equivalence there is achieved: https://componentsjs.readthedocs.io/en/latest/getting_started/basics/exposing_components/#packagejson-discovery

The other case is when there are only lsd:contexts entries. What should be done in that case? Or should only the exact URLs there be supported and mapped to their values, while other paths should be ignored?

Hopefully this reduces the amount of errors

We should add a license to the dataset.