Schema file - #58

I may have overthought this a bit. This pull request lets you actually change the field names of system fields like title, notes, organization in the dataset yaml files, and the rest of the site still works. That's pretty cool, but I'm not sure that functionality is necessary; if the driver is being able to produce a data.json file, we can just map the fields. An alternative, simpler approach would be to use a core schema that is as flat as possible (the only depth would be the resources). Any interoperability would involve mapping the system fields to whatever name they need to be. This would reduce the complexity of this pull request (including that clever recursive function). Looking forward to feedback.

Finally getting a chance to review this! I think there is definitely a need to handle mapping from schema to schema so not to have to swap out all the jekyll templates when switching over. Think this is an awesome hack, but I fear the invention of a new recursive syntax with [ ] and the field to field remapping in the _data/schemas/default.yml is not easy to intuit. What if instead of assigning a value from the remapping into the template using logic to find/replace we instead use site_data/schema/default.yml to reassign what variable is to be rendered by jekyll...

Basically using a variable and jekyll's ability to embed a variable within a variable (ex. {{ page.{{ site.data.schema.default.title}} }} to override what attribute/value gets rendered by jekyll without having to change the template itself, ex.

for default in /_data/schema/default.yml:

title: title
notes: notes
maintainer: maintainer
maintainer_email: maintainer_email
organization: organization
category: category
resources: resources
  - url: url
    name: name
    format: format
    description: description

for data.json v1.1 schema /_data/schema/default.yml:

title: title
notes: description
maintainer: contactPoint.fn
maintainer_email: contactPoint.hasEmail
organization: publisher.name
category: theme
resources: distribution
  - url: downloadURL
    name: title
    format: format
    description: description

Alternative approach to editing schema as json to remove the need for creative logic/hacks of dataset entries... Using a JSON schema to control the hierarchy, ordering, description of the fields could significantly simplify the core jekyll site and reduce the workflow to:

1. GET raw yml for page from github API
2. CONVERT auto convert yml2json, dump into text area where the form is set to listen/watch as a JS event to pre-populate the form with the current pages data.
3. UPDATE info through the form UI
4. POST auto convert json2yaml output using github API to update

datajson-editor

using data.json but could control which schema is used and offer the dataset.md ckan model/schema currently being used by JKAN

DEMO

Related to issues #34 and #33

Closing in favor of #60