timeliner
v3.0.0
Published
Automated Browser Timeline Analyser
Downloads
22
Readme
timeliner
Automated Browser Timeline Analyser
Uses chromedriver to load a webpage a number of times and aggregates performance metrics from Chrome's devtools timelines.
Installation
Install from npm. Use -g
flag to install CLI.
npm install timeliner [-g]
Usage
From the command line:
$ timeliner https://cubelets.digital/ --count 3 --sleep=15000
# outputs
┌──────────────────┬───────────────────┬────────┬────────┐
│ metric │ mean │ min │ max │
├──────────────────┼───────────────────┼────────┼────────┤
│ firstPaint │ 1.973 (+/-0.314) │ 1.532 │ 2.234 │
├──────────────────┼───────────────────┼────────┼────────┤
│ domcontentloaded │ 2.083 (+/-0.313) │ 1.643 │ 2.342 │
├──────────────────┼───────────────────┼────────┼────────┤
│ load │ 2.568 (+/-0.264) │ 2.194 │ 2.761 │
├──────────────────┼───────────────────┼────────┼────────┤
│ lastPaint │ 10.451 (+/-0.314) │ 10.010 │ 10.714 │
└──────────────────┴───────────────────┴────────┴────────┘
Side by side comparison:
You can run timeliner against two websites in parallel which will return a comparison of metrics, with an indicator of whether the difference is statistically significant (p<0.05).
$ timeliner https://facebook.com https://twitter.com --count=1
# outputs
┌──────────────────┬──────────────────────┬─────────────────────┬──────────┐
│ metric │ https://facebook.com │ https://twitter.com │ p < 0.05 │
├──────────────────┼──────────────────────┼─────────────────────┼──────────┤
│ firstPaint │ 2.189 (±0.000) │ 2.724 (±0.000) │ - │
├──────────────────┼──────────────────────┼─────────────────────┼──────────┤
│ domcontentloaded │ 2.223 (±0.000) │ 2.876 (±0.000) │ - │
├──────────────────┼──────────────────────┼─────────────────────┼──────────┤
│ load │ 2.432 (±0.000) │ 3.124 (±0.000) │ - │
├──────────────────┼──────────────────────┼─────────────────────┼──────────┤
│ lastPaint │ 2.434 (±0.000) │ 3.139 (±0.000) │ - │
└──────────────────┴──────────────────────┴─────────────────────┴──────────┘
Note: you may need to increase the count option (i.e. sample size) in order to see statistical significance.
Scrolling performance:
# analyse scrolling performance on a long webpage
$ timeliner http://buzzfeed.com --reporter fps --sleep 5000
From code:
const timeliner = require('timeliner');
timeliner({ url: 'http://google.com' })
.then(timeliner.reporters.basic)
.then(timeliner.reporters.table)
.then(result => console.log(result));
The reporter step can be omitted to provide the raw timeline logs to analyse as you require.
Options
All options can be passed as flags in the command line, or as arguments in code unless otherwise specified.
url
Required - set the page url to be loaded
count
set the number of times to load the page before aggregating results - Default 5
reporter
CLI only - set the reporter to be used to output results - supported values: table
(default), chart:<metric>
, basic
, fps
, json
If the chart:<metric>
reporter is used, then <metric>
must be a metric name present in the result set. e.g. chart:load
will output a chart showing the value of the load metric for each run.
Multiple reporters can also be specified by separating with a comma. e.g. chart:load,table
will output load time charts, and a table of all metrics.
progress
if set then a progress bar will be output to the console showing test execution progress
scroll
if set, injects a script into the page which binds a vertical scroll to window.requestAnimationFrame
making the page scroll continuously. If a numerical value is provided then the page will scroll by that amount per frame - Default false
sleep
set how long (in ms) after the page completes loading to continue recording metrics - Default 0
driver
sets the url of the webdriver remote server to use - Default http://localhost:9515
(note: default webdriver is started automatically)
inject
Function
- allows the definition of a custom step in the webdriver promise chain. Function is passed the webdriver browser object as a parameter and should return a promise. See example below.
Custom metrics
Using console.timeStamp
You can fire custom events by calling console.timeStamp
from anywhere within your code with a label that matches timeliner.*
. This will then report the first occurence of that event with a metric name of the wildcard portion of the timestamp label.
These can either be fired directly by the site under test, or injected as part of the test run using the inject
option.
Example - inject some custom javascript into your page to trigger a custom event after 1 second.
const timeliner = require('timeliner');
timeliner({
url: 'http://example.com',
inject: (browser) => {
return browser.execute(`setTimeout(() => console.timeStamp('timeliner.custom-metric'), 1000);`);
}
})
.then(timeliner.reporters.basic)
.then((result) => {
// result includes data for `custom-metric` event
// result = { ... , 'custom-metric': { ... } }
});
In Code
You can pass an optional function to the basic
reporter as a second argument which can execute custom metrics and output them in a form compatible with the table
reporter.
The function should take a single set of logs as an argument and return an object keyed by metric names and with values corresponding to the value of each metric.
Example:
const timeliner = require('timeliner');
function customMetrics (logs) {
// value = do some big map-reduce on the logs
return {
'my-metric': value
}
}
timeliner({ url: 'http://google.com' })
.then(logs => timeliner.reporters.basic(logs, customMetrics))
.then(timeliner.reporters.table)
.then(result => console.log(result));
See [./examples/image-count.js](a full worked example of using custom metrics in code).