fixes #1874 adding a substring extraction function, tests, and documentation

[clintropolis]

SQL style substring extraction function, which takes an index and optionally a length, returning a substring of the dimension value starting from the index of size length, or the remainder of the string if it is shorter than length or length is not supplied. If the starting index is greater than the length of the dimension value, null is returned.

{ "type" : "substring", "index" : 3, "length" : 4 }

{ "type" : "substring", "index" : 3 }

[rasahner]

I was looking at the CACHE_TYPE_ID values and saw that TimeFormatExtractionFn.java and LookupExtractionFn.java both define it to 5.

[pjain1]

Is this check required ? At this point length will be greater than 0 as we will return null if the length is 0 at line 62 Strings.isNullOrEmpty(dimValue)

[pjain1]

sorry I confused length with len - ignore the comment

[fjy]

@clintropolis can we fix merge conflicts? I think this is almost ready to go

[clintropolis]

Yep, will do this evening, along with other suggestions.

[drcrallen]

io.druid.server.namespace.cache.NamespaceExtractionCacheManagerExecutorsTest
testConcurrentAddDelete(io.druid.server.namespace.cache.NamespaceExtractionCacheManagerExecutorsTest)  Time elapsed: 600.018 sec  <<< ERROR!
java.lang.Exception: test timed out after 600000 milliseconds
    at sun.misc.Unsafe.park(Native Method)
    at java.util.concurrent.locks.LockSupport.park(LockSupport.java:175)
    at java.util.concurrent.FutureTask.awaitDone(FutureTask.java:429)
    at java.util.concurrent.FutureTask.get(FutureTask.java:191)
    at 

Bouncing for travis

[b-slim]

thought @drcrallen want to have at lease one UT for getCacheKey ?

[drcrallen]

Yes please :)

[fjy]

@clintropolis can we finish this off?

[drcrallen]

BrokerServerViewTest.testSingleServerAddedRemovedSegment:112->setupZNodeForServer:383 » NodeExists

[clintropolis]

@fjy I think it should be good to go now

[xvrl]

this should be true when index == 0 to allow optimizations for this particular case.

[himanshug]

is length == 0 allowed? then it is expected to be "non-negative" not "positive"

[himanshug]

actually, from the code below it appears that length being 0 is not allowed.

[clintropolis]

I probably should not allow a length of 0, since it just returns null across the board. I can change the check.

[xvrl]

@himanshug @clintropolis "positive" means >= 0 "strictly positive" means > 0. If we don't allow 0, it should be say strictly positive to be clear.

[clintropolis]

sure, will change message since I don't think having a length of 0 is very useful.

[fjy]

👍 @himanshug @drcrallen i think this is ready to go

[himanshug]

👍