stdout-collator
v0.1.0
Published
Harvest and aggregate stdout so that it can be associated with test suites
Downloads
2
Readme
stdout-collator
Listen to all data written to process.stdout
and filter out sections that are wrapped in markers. The markers are written using the startGroup()
and endGroup()
functions.
why??
We are using this module to collect the log output produced during text execution, and to properly attribute the log messages normally sent to the console back to the test that produced them. You might be asking, why worry about text based markers? The reason we use this strategy is so we can observe the merged output of multiple sub processes. In our setup the markers are logged by one of the child processes (the test runner) but processed by the parent process.
example
In the scenario described above, this is in the parent process:
import { format } from 'util'
import { bindToStdout } from 'stdout-collator'
const parser = bindToStdout()
// Notice how the report consumers don't use console.log!
// Doing so would pollute the parser, instead use stderr,
// or parser.log.
parser.consumeReports({
groupStart(group) {
parser.log('group started', group.startMark)
},
groupEnd(group) {
parser.log('group ended', group.endMark)
}
})
And then, in the child process, at the beginning and end of each test, code similar to this is run:
import { startGroup, endGroup } from 'stdout-collator'
describe('root suite', () => {
beforeEach(() => {
startGroup({ testName: this.name })
})
afterEach(() => {
startGroup({ testName: this.name, success: this.success, time: this.ms })
})
})
api
bindToStdout(): Parser
Call this method to create a parser that is bound to stdout. You should only call this once, and there isn't currently any way to tear this method down. The created parser is returned
Parser
class
The Parser class is responsible for tracking and scanning all of the chunks written to stdout, finding the markers in the output, parsing the attributes of each mark, and emitting Group objects
parser.consumeReports({ ...handlers })
Register a consumer of the parser reports. Right now there are two reports; groupStart
and groupEnd
. Pass an object with with function properties matching these report names and they will be called with Group
objects at the correct time.
Group
class
The markers written to stdout indicate the beginning and end of output groups. Whenever a start mark is fully parsed a groupStart
will be reported. Once the end of a group is parsed a groupEnd
will be reported. Both of these reports provide the consumer the group instance, and only certain properties will be available at certain points:
group.parent : Group?
available at all times
The parent of this group, if there is one
group.startMark : { [key: string]: any }
available in the groupStart
and groupEnd
report handler
The attributes defined by the startMark
group.endMark : { [key: string]: any }
available in the groupEnd
report handler
The attributes defined by the endMark
group.output : string
available in the groupEnd
report handler
The text logged within this group