Using the Magda API Externally
If you’re trying to write something that uses the Magda API externally, there are two main apis that you’re likely to be using - Registry and Search. Both of these have API documentation at
/api/v0/apidocs/index.html on the Magda instance that you’re using.
Using the Registry API
The registry api is where most metadata within Magda is stored - datasets, distributions (files), organisations etc, and these are ingested into the search as they change.
When using the registry API, it’s important to remember that the registry works a bit like a JSON document store - nearly all data is stored as a
record that can be retrieved through the
Getting the right kind of records
So how do you know which
records are datasets, which are distributions and so on? Each record has a set of
aspects, which are JSON documents that match a certain schema. So if you want to get all the datasets, you’d use
Getting records linked to other records
Records can also be linked to other records - for instance, a dataset is usually linked to a number of distributions (with aspect
dcat-distribution-strings). You can get the whole thing with the
dereference parameter. E.g. to get distributions for a dataset, you can use
Simple pagination with start and limit parameters is very slow in the Registry API. If you want to crawl through all the data available, please use the
nextPageToken returned with each page, passing it into the
pageToken parameter to get the next page.
You can see all the aspects and what’s in them in https://github.com/magda-io/magda/tree/master/magda-registry-aspects. You can also see all the aspects that a record has at
Using the Search API
Using the search api is pretty straightforward - you can search for datasets, organisations or regions as per the API documentation. Be aware that (at least for now), the schema for datasets and distributions in the search API doesn’t match the schema for
dcat-distribution-strings in the Registry API, although it’s fairly similar.
Also note that paging through to get all the records in the search doesn’t work past 10,000 - to get all the data, refer to “Paging through” above.