File discovery

information files can import other information files using the $ref operator. This is based on the jsonref standard, but we have expanded it to also work on YAML files and to allow several directories in which to search for the given filepaths. This is called the datapath and includes the files and urls specified in the ~/.obsinfo file.

The .obsinforc file is created in main/setupObsinfo.py and is read by Datapath objects through a call to ObsinfoConfiguration.datapath

I would also like to be able to search in the current file’s directory, but haven’t yet been able to figure out how, although this was probably the origina jsonRef default.

I think that the subfiles are read through the recursive JsonRef.replace_refs() classmethod, which is called by yamlref.load(), yamlref.loads() and yamlref.load_uri().

Searching for datapath in yamlref.py suggests that its methods are only called in the derived property JsonRef.full_uri (see below), which is called by the callback method. The callback method appears to be intrinsic to the “proxytypes” superclass, but I don’t understand it. According to python documents, it is called when its object is garbage collected???

One could possibly be inspired by the jsonschema is True case, because jsonschema always reads from the same directory. On the other hand, perhaps this directory is passed in the base_uri argument for jsonschema and not for the others?

@property
def full_uri(self):
    """
    This method/property returns the full uri to reference a ``$ref`` object.
    It's the heart of how a datapath is used to either access a local or remote (gitlab) file.
    All schema files are supposed to be local, part of the obsinfo distribution

    :returns: updated full uri
    :raises: ValueError
    """

    kwargs = self._ref_kwargs

    if kwargs['jsonschema']:
        return urlparse.urljoin(self.base_uri, self.__reference__["$ref"])
    else:
        dp = kwargs["datapath"]
        if not dp:
            msg = f'Error in datapath in full_uri, reference: {self.__reference__["$ref"]}'
            logger.error(msg)
            raise ValueError(msg)
        base_uri = Path(dp.build_datapath(self.__reference__["$ref"]))

        tupl = urlparse.urlsplit(str(base_uri))
        path = unquote(tupl.path)
        frag = tupl.fragment
        new_uri = Datapath.add_frag(path, frag)
        # define the uri depending on whether it is remote or not
        self.base_uri = new_uri if gitLabFile.isRemote(str(base_uri)) \
            else unquote(base_uri.as_uri())

        return(self.base_uri)

I don’t like that (the file discoveryfiles.py only contains the class Datapath, should rename to datapath.py)