CSV to JSON/RDF test cases

I've created several test cases [1] for a direct mapping to either JSON or RDF considering different types of input files that I think correlate well with our requirements. The tests are organized as a test suite, with a test manifest and input and result files, although it's entire the creation of Andy and myself and nothing currently generates this.

Some things to note:
* All fields are normalized to remove surrounding whitespace.
* Fields that look like numbers (or booleans) are represented as native types in the output.
* The RDF output includes extra column metadata that the JSON does not. The JSON could after some work on CSV-LD, but it's not clear that it's important for this use case (anyway, ripe for WG discussion).
* The direct mapping does not rely on any metadata, and assumes that the first row from the CSV is the header.

Subsequently, we can use some trial metadata definitions to provide alternate transformations of these files, and others to address skip lines, multi-line headers and more field micro-formats. Anyway, I think this is a good place to start having a WG discussion to ensure we agree on what the expected output of a direct transformation would be.

Gregg Kellogg
gregg@greggkellogg.net

[1] https://github.com/w3c/csvw/tree/gh-pages/tests

Received on Saturday, 29 March 2014 22:40:38 UTC