Detect RPM specfiles

[rakeshbalusa]

Currently scancode-toolkit can scan packages or any kind of code without targeting metadata like type of package (for example : RPM). For any given RPM the .spec has most of the metadata like architecture, source RPM, URL etc., and it is the most reliable metadata for that RPM. So, we need to have a separate scanning scenarios for RPMs. Everytime scancode enounters an RPM it should provide scan results using these scenarios.

[pombredanne]

@rakeshbalusa actually the .spec file is rather hard to parse short of running the tools to build an RPM proper. The more reliable source of metadata is the RPM headers in an .rpm file. The .spec file itself is rarely available anyway when you deal with built RPMs.

[rakeshbalusa]

@pombredanne I guess we have to use those headers info instead of .spec file in case of an RPM.

[balusarakesh]

@pombredanne
I think the line - https://github.com/nexB/scancode-toolkit/blob/develop/src/packagedcode/recognize.py#L57 should be if location.endswith(tuple(package.extensions)):. I mean we should look for extensions not metafiles, but this will be a problem for NPM packages as they have package.json as metafile which we must look for. I'm not sure how to solve this issue.

[pombredanne]

We are scanning RPMs alright and there is no plan for now to scan .spec files yet, though I am renaming this issue to do that