undefined variable in _iter_from_tabulator function

[ndkv]

The value variable on line 325 in resource.py is undefined.

I encountered it while running the example code in the README.

import datapackage

dp = datapackage.DataPackage('http://data.okfn.org/data/core/gdp/datapackage.json')
brazil_gdp = [{'Year': int(row['Year']), 'Value': float(row['Value'])}
              for row in dp.resources[0].data if row['Country Code'] == 'BRA']

This throws a NameError for value in the _iter_from_tabulator function. Note: this error is not caught by Travis.

I'm not familiar with jsontableschema and can't quickly see how to fix this. Any help is greatly appreciated as I'd like to address #63.

[ndkv]

In addition, the example code will always fail because the dates in the referenced data package are formatted as yyyy which cannot be cast by jsontableschema.types.date.DateType as there is no cast_yyyy() function.

Adding one is easy but I'm not sure whether that's the desired solution.

[danfowler]

@ndkv Thanks for flagging! For your second issue, as of the current spec (http://specs.frictionlessdata.io/json-table-schema/#date), that datapackage.json should actually use this for the date:

"pattern": "fmt:%Y"

However, this is not particularly well liked and is currently under discussion for replacement: frictionlessdata/datapackage#260 (your thoughts welcome!)

We should update the gdp dataset to at least match the current spec, but it might break the old OpenSpending (https://openspending.org/) which relies on this formatting and might be pulling in that dataset.

[s-celles]

See also: datasets/gdp#5

[ndkv]

@danfowler Right, thanks for explaining the issue. I have to agree that fmt:%Y doesn't feel right.