This is my senior project for running sentiment analysis on commit messages for Jira projects. This package includes the code that I wrote for obtaining the data from Jira to run the analysis.
There were a few APIs that I tried out for querying Jira, but JRJC (Jira REST Java Client) was the one that worked best for me. It is a bit wonky and weird to use, but I was able to get it work well enough.
All the magic happens in JRJC.java and SSWrapper.java. Here's a walkthrough of the code:
The setup() method sets up the Jira client and creates a new CSV file to dump the data. Notice that the createWithBasicHttpAuthentication() call requires your Jira username and password. If you don't have a Jira account, signing up is free and easy.
I used CSVPrinter to print the data to a csv file. This example outputs the data to a file called sample.csv.
main() will search Jira with a JQL (Jira Query Language) search. The code right now asks for the first 1000 issues for project HADOOP.
runSearch() will take each issue returned by the search, use the issueClient to obtain more information about it, and then write the info to the CSV. The code is pretty self-explanatory -- I had to write a bit of logic to format the data so that it would be easier to read in the CSV, but that's about it. The csvPrinter.flush() call will write to the CSV after each individual issue has been analyzed.
runSearch() makes two calls to ssWrapper, a helper class I made for calling the sentiment analysis library. In the SSWrapper constructor, SentiStrength has to be initialized and you'll have to provide the path to where the SentiStrength data files are located. The process method will process the inputted text and prepare it for scoring. The compute call will actually compute the scores. Look here for more information about the sentiment analysis library. That page includes a word document that explains how to use the Java library.
This code isn't perfect, but it got the job done for me. The most annoying part, by far, was how inconsistent JRJC would be for me. Sometimes I would be able to query for 1000 issues and it would finish within 7-8 minutes, yet other times it would take 20+ minutes and not even finish 200. There isn't much information for this library online and the error messages were quite cryptic, so I thought it was easier just to deal with the annoyances rather than look into fully resolving the issues. It's possible that the service doesn't like receiving such a large number of requests at once, causing it to error out. Anyway, if further investigation is needed, there are a lot of messages being logged that can be looked into.