Protractor Flake

Rerun potentially flakey protractor tests before failing.

npm i protractor-flake
 
# or globally for easier cli usage 
npm i -g protractor-flake

Usage

Via the CLI:

npm i -g protractor-flake
 
# protractor-flake <protractor-flake-options> -- <options to be passed to protractor> 
protractor-flake --parser standard  --max-attempts=3 -- path/to/protractor.conf.js

See src/parse-options.js for the full list of command line options.

Protractor flake expects protractor to be on $PATH by default, but you can use the --protractor-path argument to point to the protractor executable.

Or programmatically:

var protractorFlake = require('protractor-flake');
 
// Default Options
protractorFlake({
  parser: 'standard'
}, function (status, output) {
  proces.exit(status)
})
 
// Full Options
protractorFlake({
  protractorPath: '/path/to/protractor',
  maxAttempts: 3,
  parser: 'standard',
  // expects node to be in path
  // set this to wherever the node bin is located
  nodeBin: 'node',
  // set color to one of the colors available at 'chalk' - https://github.com/chalk/ansi-styles#colors
  color: 'magenta',
  protractorArgs: []
}, function (status, output) {
  process.exit(status);
});

Parsers

Protractor flake defaults to using the standard parser, which will typically pick up failures run from non-sharded/multi-capability test runs using Jasmine 1 + 2 and Mocha.

There are a few other ways that you can customize your parsing:

overriding this with the parser option, specifying one of the built in parsers.
providing a path to a module (e.g. /my/module.js or ./module.js) that exports a parser
a parser (if used programatically)

Parsers should be defined as an object with a parse method (and optionally a name property):

module.exports = {
  parse (protractorTestOutput) {
    let failedSpecs = new Set()
    // ... analyze protractor test output
    // ... and add to specFiles
    failedSpecs.add('path/to/failed/specfile')
 
    // specFiles to be re-run by protractor-flake
    // if an empty array is returned, all specs will be re-run
    return [...failedSpecs]
  }
}

Parser documentation

Mocha (TODO)
Jasmine (TODO)
cucumber

Caveats

This has not yet been tested with Protractor + Mocha. It should function similarly. Please update with an issue or PR if this is not the case.

Tests will not re-run properly (all tests will run each time) if you use a custom reporter that does not log stacktraces for failed tests. For example, if you are using jasmine-spec-reporter with Jasmine 2.0, make sure to set displayStacktrace: 'specs' or displayStacktrace: 'all'.

Contributors

See CONTRIBUTING.md