Improvements to Fast JSON Input

jadametz commented 9 years ago

Our company has struggled with many of the same issues that you describe in PDI-10344 and were looking for a solution to the JSON Input step when we came across your implementation. Our changes have been in an effort to both increase the stability of and improve your implementation for deployment to our Product Managers for their ETL work.

In order of the commits, below are roughly the major things that were implemented / changed:

6564f6f: fixes the nodevalue to not always blindly convert .toString(). This helps to return JSON blobs for addition parsing as JSON strings (still quoted), as well was avoid throwing NullPointerException when the value being parsed is null.
f5b0627 adds a checkbox to the Content tab allowing for the DEFAULT_PATH_LEAF_TO_NULL option within the JsonPath module to be enabled.
02cc740 updates the logo to match PDI 5.4's look -- don't worry, I opted to keep your lightning bolt idea :)
a9ba531 adds 4 test cases to FastJsonInput; testing the permutations of Ignore missing path and Default path leaf to null
I also felt at this point, the step is ready for 1.0.0 release and have moved it to the Input folder

Please reach out with any additional questions, as I'd be happy to discuss any changes that I made. Hope you enjoy the contribution!

etdube commented 9 years ago

Many thanks for your contribution. I'll merge your pull request right away and will do a quick review when I get a chance. As far as IP is concerned, are you okay with your changes going back into the Kettle code base if Pentaho ask if they can use it? The license implicitly allows this but I prefer to ask to clear any ambiguity.

jadametz commented 9 years ago

Absolutely. We'd love for Pentaho to take in these changes.

We'll be keeping up with development on our side as needed. I'll be sure to open pull requests over time so you've got the latest and greatest :+1:

etdube / pdi-fastjsoninput-plugin

Improvements to Fast JSON Input #1