ExpressionPredicate's handling of regex function seems to fail if regex does not match

[Eric Milles]

The implementation for the regex function differs significantly between the ARRAY and OBJECT branches. The array branch appears correct to me. I am testing the object branch and when my regex does not match, I am still getting values added and so JPath.evaluate is returning non-null.

[John Marsden]

I will check this in my next attempt at a query language. I have used the regex function quite a bit so I would be interested in seeing some examples.

In any case, I did not respond within a reasonable timeframe so I am assuming you have moved on to something that is actually working for you.

John

[Eric Milles]

I have made a number of fixes for issues reported in my local copy. I was unable to reach you for creation of pull requests so I needed to fix on my side. I could share them back.

[John Marsden]

Hello Eric, share away - happy to include anyting of value and applogies for the lack of reponse. My Bitbucket account was pointing to an email address I did not actually use :-S.

I am currently working on BSON support for an internal memory file store so I have spent a bit of time optimising the main model and the parsing.

I was planning on cleaning up JPath next. I was hoping a good standard had emerged. I did a bit of research over the last couple of days about how JSON is generally getting queried these days but there still seems to be no really good definition of a query language.

• JMESPath - This looks ok to me but I dont know how wide spread its adoption is http://jmespath.org/
• JSON Pointer - This one is at least a formal standard but its really basic and I dont like the replacement choices. https://tools.ietf.org/html/rfc6901
• Jaql - This looks like a totally different path and I would have to do a lot of leg work to support it. https://code.google.com/archive/p/jaql/

Out of interest, what are you using JSONiJ for?

John

[Eric Milles]

Here is what I have in terms of fixes or enhancements. I believe it was based off version 0.3.1. I can answer any questions related to the changes you have. Sorry I don't have more background to give; it has been over a year since I last did anything with this.

In terms of how it is used here (at Thomson Reuters), we have a JMS/MQ pathway for application events. These events have JSON payloads. At some point we have a dispatcher process setup that checks for certain characteristics in events and decides where it should be routed next. This makes heavy use of JSONiJ's JPath matching. Here is a sample from our rules XML:

<eventdispatcher>

<!-- source connection factory/pool and queue

(factoryOrPoolName:queueName) --> <source>EventDispatchQueueConnection:COBALT.EVENTS.ALL</source>

<!--
  Rule mappings (see

com.trgr.cobalt.infrastructure.eventdispatcher.Rule) @category: product or service identifier for traceability jpath: an XPath-like syntax for selecting events; see the JSONiJ site for more details destination(s): destination connection factory/pool and queue (factoryOrPoolName:queueName) -->

<rules>
    <rule category="BusinessIntelligence">

<jpath>/[?(regex(@.name,"MyBusinessIntelligence.Website.NewsPlusData"))]</jpath> <destination> RnDNewsPlusQueueConnectionPool:COBALT.RND.NEWSPLUS.EVENTS</destination> </rule> <rule category="BusinessIntelligence">

<jpath>/[?(regex(@.name,"MyBusinessIntelligence.Website.FullText"))] AND /properties[?(@.DocumentContentType="news")]</jpath> <destination> RnDNewsPlusQueueConnectionPool:COBALT.RND.NEWSPLUS.EVENTS</destination> </rule>

    <rule category="HistoryListener">

<jpath>/[?(regex(@.name,"Cobalt.Document.FullText.Display"))]</jpath>

<destination>HistoryQueueConnectionPool:COBALT.FOLDERING.HISTORY.EVENTS.SHARED</destination> </rule>

[John Marsden]

Hello Eric,

Here is what I have in terms of fixes or enhancements.

Where are the changes? I cant see any repositories on your user or pull requests. You could zip and email to me at [e-mail].

Thanks,

John