Schemer

Schemer recursively reads a directory full of JSON docs and works out what the 'master' JSON schema would be to define them all. It does this by generating a JSON schema for each doc in turn and then using deltas between the schemas to produce a master. Once a master schema is produced it sanity checks it by doing a compatibility check against each JSON document previously read.

The master schema is then written to a docson directory and can be viewed in a browser.

Installation

git clone https://github.com/julianbrowne/schemer.git

Running

cd schemer

schemer example     # runs against the 'example' directory

Configuration

Some fields may have allowable values that are known about but do not exist in the source files. For example, if all JSON documents contain an email field then the schema will mark this as a mandatory string. If some contain an email field, but others do not, then the field will be marked as optional but still mandated as a string when it is present. To set the email field to allow nulls, add email to the fields_that_can_be_null list in config.yml. The setting should be the full path as seen in the JSON file but without the prefix properties or type found in the JSON schema file.

e.g. to allow the age field in { "name": "bob", "stats": { "age": 42 } } to be present but null in some JSON documents, the config file should look like this:

fields_that_can_be_null:
    - stats.age