caarlos0-graveyard / github-integrator Goto Github PK

View Code? Open in Web Editor NEW

3.0 1.0 1.0 618 KB

Binds GitHub's PullRequests comments to shell executions

License: MIT License

Java 99.90% Shell 0.10%

github-integrator's Introduction

github-integrator

Binds GitHub's PullRequests comments to shell executions.

Usage

You will have to create a .properties file following the example here;
You will have to create a .json file containing the executions configuration, like this;

Right now, the application passes the following parameters down to the script:

$1: the pull request origin repository owner login (e.g.: caarlos0);
$2: the pull request origin repository name (e.g.: github-integrator);
$3: the pull request origin branch (e.g.: feature/more-parameters);
$4: the pull request number;
$5 to $n: the group values from the regex expression.

Scripts should be fast. If your script is slow, it will hold the execution of other commands until it ends. This will probably be fixed soon.

While we don't have any stable release, you will need to compile the project using Maven:

mvn clean install

Then, you can execute the app like this:

INTEGRATOR_CONFIG=my.properties java -jar \
  target/github-integrator-1.0-SNAPSHOT-jar-with-dependencies.jar

Throughput

Contributing

Feel free to try to fix an existing issue or open new ones.

github-integrator's People

Contributors

Stargazers

Watchers

Forkers

diego-aslz

github-integrator's Issues

comment the result of the script

ability to define custom replies in the json file

bug while parsing github json

Exception in thread "main" java.io.IOException: Parse exception converting JSON to object
    at org.eclipse.egit.github.core.client.GitHubClient.parseJson(GitHubClient.java:424)
    at org.eclipse.egit.github.core.client.GitHubClient.parseJson(GitHubClient.java:403)
    at org.eclipse.egit.github.core.client.GitHubClient.parseError(GitHubClient.java:514)
    at org.eclipse.egit.github.core.client.GitHubClient.createException(GitHubClient.java:547)
    at org.eclipse.egit.github.core.client.GitHubClient.get(GitHubClient.java:740)
    at org.eclipse.egit.github.core.client.PageIterator.next(PageIterator.java:173)
    at org.eclipse.egit.github.core.service.GitHubService.getAll(GitHubService.java:151)
    at org.eclipse.egit.github.core.service.PullRequestService.getPullRequests(PullRequestService.java:152)
    at com.carlosbecker.MainIntegrator.work(MainIntegrator.java:52)
    at com.carlosbecker.MainIntegrator.work(MainIntegrator.java:48)
    at com.carlosbecker.App.runOnce(App.java:31)
    at com.carlosbecker.App.run(App.java:25)
    at com.carlosbecker.App.main(App.java:38)
Caused by: com.google.gson.JsonSyntaxException: java.lang.IllegalStateException: Expected BEGIN_OBJECT but was STRING at line 1 column 1 path $
    at com.google.gson.internal.bind.ReflectiveTypeAdapterFactory$Adapter.read(ReflectiveTypeAdapterFactory.java:187)
    at com.google.gson.Gson.fromJson(Gson.java:805)
    at com.google.gson.Gson.fromJson(Gson.java:770)
    at org.eclipse.egit.github.core.client.GitHubClient.parseJson(GitHubClient.java:422)
    ... 12 more
Caused by: java.lang.IllegalStateException: Expected BEGIN_OBJECT but was STRING at line 1 column 1 path $
    at com.google.gson.stream.JsonReader.beginObject(JsonReader.java:387)
    at com.google.gson.internal.bind.ReflectiveTypeAdapterFactory$Adapter.read(ReflectiveTypeAdapterFactory.java:176)
    ... 15 more

Happens only when the something is configured wrong in the executions.json (e.g.: the repository doesn't exist)

wrong reply message when using regex

If the regex actually contains a regular expression, the bot will comment the regex without it's parsed groups.

Example:

executions.json (partial):

"regex": "deploy this to (production|staging)"

Comment was:

deploy this to production please

Reply was:

Ok, working on 'deploy this to (production|staging)'...

Reply should have been:

Ok, working on 'deploy this to production'...

enforce politeness: "please" suffix should be default

add coveralls

parse regex and send groups it to the script

Use a file database (sqlite, for instance) to store processed comments

The current implementation (based on ask/reply count) is weak and hard to maintain.

It will probably be better to store those info in a sqlite db or something like that.,

improve documentation

disable travis-ci emails

reduce the amount of api calls by allowing multiple scripts per repo

Each repo can have one script/regex pair.

In a scenario when a single repo have multiple regex/script pairs, the bot will do more requests than needed to the github api (by getting again the open prs, the comments, et al).

This can be improved by allowing multiple script/regex pairs for repos.

multiple equal requests are not executed

do this
Ok, working on 'do this'
do this
nothing happens...