jvalue / open-data-service-v1 Goto Github PK

For example http://faui2o2f.cs.fau.de:8080/open-data-service/foo returns

Invalid query, see /api for available queries.

It would be nicer to have a response along the lines of

{
   status: 400,
   message: "Invalid query, see /api for available queries."
}

ODS currently always returns 200 OK for each request, even though some are invalid and the response contains error messages.

No filter should be able to modify the callers data.

For example if there are two filters A --> B, then A should create a copy of the data before forwarding it to B.

QA filter throws exception (1) when pegelonline source is experiencing an internal error (2)

(1)

java.lang.NullPointerException at org.jvalue.ods.qa.improvement.CombineSourceFilter.filter(CombineSourceFilter.java:74)
at org.jvalue.ods.qa.improvement.CombineSourceFilter.filter(CombineSourceFilter.java:48)
at org.jvalue.ods.filter.FilterChain.filter(FilterChain.java:49)
at org.jvalue.ods.filter.FilterChain.filter(FilterChain.java:50)
at org.jvalue.ods.filter.FilterChainManager.startFilterChains(FilterChainManager.java:63)
at org.jvalue.ods.grabber.DataGrabberMain.updateData(DataGrabberMain.java:64)
at org.jvalue.ods.server.ContainerRestletApp.run(ContainerRestletApp.java:118)
at java.lang.Thread.run(Thread.java:744)

(2)
http://pegelonline.wsv.de/webservices/rest-api/v2/stations.json

currently (18.07.14 12:52) returns

{
   status: 500,
   message: "Internal server error. Please retry."
}

The content of

/admin/logs

keeps growing and is not limited by any size restrictions.

Instead of one log with all entries offer some pagination mechanism for fetching only the most recent entries.

Github commits should trigger jenkins build. They used to, but aren't doing it anymore.

Api could / should offer two version counts

• current build (automatically increased?)
• release version (incremented manually on each larger release)

Source schema definitions (how data is available in the original source and how it will be stored in the DB) are currently created by manually by constructing large nested map and list objects.

When wanting to create these schemata dynamically at runtime, having a builder object that takes care on the internal referencing makes constructing this definition at lot easier.

--> Create a builder class for constructing schema, eliminating the need for manual linking

Clients should again be able to extract parts of all kinds of documents.

with $1234 being the ID of the document, water/coordinate two of its fields:
GET /ods/$1234/water
GET /ods/$1234/water/longname
GET /ods/$1234/coordinate
GET /ods/$1234/coordinate/longitude
... as deep as they want (except lists)

The ugly, former previously used code can be found here:
https://github.com/jvalue/open-data-service/blob/76299da9d021fe1775118c008343ec66da9fb1be/src/main/java/org/jvalue/ods/server/restlet/DefaultRestlet.java

Allow simple key-value-queries

f.e.: ods/stations?timeseries.measurement.level=10

This currently happens via

ContainerRestletApp.run()
--> DataGrabberMain.initialize()
--> ConfigurationManager.configureAll()
--> DbAccessor.deleteDatabase()

... and they are

coordinate: {
longitude: 8.826504838107983,
latitude: 53.161504277715444
}

Each FilterChain should run in its own thread, using a custom update interval (pull or push triggered)..