rollup-plugin-analyzer
Mad metrics for your rollup bundles, know all the things
TOC
Install
$ npm install --save-dev rollup-plugin-analyzer
Usage
Importing or Requiring
Import as ES Module
import analyze from 'rollup-plugin-analyzer'
Requiring as CJS
const analyze = require('rollup-plugin-analyzer')
Usage from rollup config
export default {
entry: 'module.js',
dest: 'index.js',
format: 'cjs',
plugins: [analyze()]
}
Usage from build script
rollup({
entry: 'main.js',
plugins: [analyze()]
}).then(...)
CI usage example
const limitBytes = 1e6
const onAnalysis = ({ bundleSize }) => {
if (bundleSize < limitBytes) return
console.log(`Bundle size exceeds ${limitBytes} bytes: ${bundleSize} bytes`)
return process.exit(1)
}
rollup({
entry: 'main.js',
plugins: [analyze({ onAnalysis, skipFormatted: true })]
}).then(...)
results
logged to console on rollup completion
-----------------------------
Rollup File Analysis
-----------------------------
bundle size: 2.809 KB
original size: 11.436 KB
code reduction: 75.44 %
module count: 5
█████████████████████████████████████████████░░░░░
file: /virtual-insanity.js
bundle space: 90.64 %
rendered size: 2.546 KB
original size: 2.57 KB
code reduction: 0.93 %
dependents: 1
- /jamiroquai.js
██░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░
file: /bundle-a.js
bundle space: 4.27 %
rendered size: 120 Bytes
original size: 309 Bytes
code reduction: 61.17 %
dependents: 0
█░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░
file: /jamiroquai.js
bundle space: 2.95 %
rendered size: 83 Bytes
original size: 169 Bytes
code reduction: 50.89 %
dependents: 1
- /the-alphabet-but-incomplete.js
...
results (with summaryOnly
enabled)
-----------------------------
Rollup File Analysis
-----------------------------
bundle size: 2.809 KB
original size: 11.436 KB
code reduction: 75.44 %
module count: 5
/virtual-insanity.js
█████████████████████████████████████████████░░░░░ 90.64 % (2.546 KB)
/bundle-a.js
██░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░ 4.27 % (120 Bytes)
/jamiroquai.js
█░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░ 2.95 % (83 Bytes)
/the-alphabet-but-incomplete.js
░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░ 1.17 % (33 Bytes)
/the-declaration-of-independence.js
░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░ 0.96 % (27 Bytes)
Options
- stdout - optional
- type: Boolean
- default:
false
- description: Print to stdout (console.log) instead of stderr (console.error)
- limit - optional
- type: Number
- default:
null
- description: Limit number of files to output analysis of, sorted by DESC size
- filter - optional
-
type: Array String Function - default:
null
- description: Filter module to output in analysis results.
- If a string is passed, checks if module name contains the string
- If array it checks the same for each string in the array
- If function, return a boolean to indicate if a module should be in analysis results
- This only suppresses the modules details, it does not affect the summary output
- If you would like it to filter from summary info as well set
filterSummary
totrue
- notes: Function receives
module
object specified below, should return boolean
-
- filterSummary - optional
- type: Boolean
- default:
false
- description: If
true
thefilter
andlimit
options will also remove any filtered out module data from the summary
- root - optional
- type: String
- default:
process.cwd()
- description: Application directory, used to display file paths relatively
- hideDeps - optional
- type: Boolean
- default:
false
- description: Don’t itemize dependents in the formatted output
- showExports - optional
- type: Boolean
- default:
false
- description: Show used and unused exports
- summaryOnly - optional
- type: Boolean
- default:
false
- description: Only output bundle summary and module usage bar graphs
- skipFormatted - optional
- type: Boolean
- default:
false
- description: Don’t output formatted string
- writeTo - optional
- type: Function
- default:
null
- description: Callback to be invoked with formatted string
- function will be invoked with:
- analysisString (String)
- transformModuleId - optional
- type: Function
- default:
null
- description: Modify module ids
- function will be invoked with:
- id (String) - path of module / rollup module id
- function should return:
- id (String) - desired final module id to display in analysis results
- example:
(id) => id.replace(/^\0(?:commonjs-proxy:)?/, '')
- onAnalysis - optional
- type: Function
- default:
null
- description: Callback to be invoked with analysis object
- function will be invoked with:
- analysisObject (Object)
- bundleSize (Number) - rendered bundle size in bytes
- bundleOrigSize (Number) - original bundle size in bytes
- bundleReduction (Number) - percentage of rendered bundle size reduction
- moduleCount (Number) - Count of all included modules
- modules (Array) - array of
module
analysis objects- module (Object)
- id (String) - path of module / rollup module id
- size (Number) - size of rendered module code in bytes
- origSize (Number) - size of module’s original code in bytes
- dependents (Array) - list of dependent module ids / paths
- percent (Number) - percentage of module size relative to entire bundle
- reduction (Number) - percentage of rendered size reduction
- renderedExports (Array) - list of used named exports
- removedExports (Array) - list of unused named exports
- module (Object)
- analysisObject (Object)
FAQ
Why is the reported size not the same as the file on disk?
This module is geared towards the details of the individual modules that make up the bundle and their relative impact to bundle size. That’s a detailed way of saying, it doesn’t really care about size on disk. There are other options which focus on size on disk as well as delivery size which can be used alongside this module (or in place of if your concern is not per module impact). In particular rollup-plugin-size-snapshot seems like a great option for that.
Getting a bit further into the details, rather than just intent, of why the reported size differs from that on disk. We get the module data from Rollup which reports it after chunk (module) resolution and tree-shaking, but before post-processing (such as minification and compression). We then add the sizes of each of those modules together, this is the bundle size
that we report.
That means it won’t account for post-processing from other plugins and also won’t account for post-processing by Rollup itself, which includes boilerplate / shims depending on what the output format is (CJS, ESM, iife, etc…).
Why am I seeing multiple analysis outputs emitted?
Rollup allows you to output to multiple files. If you are outputting to multiple files you will get a distinct analysis for each output file. Each analysis will contain data on the files imported by the respective target.
One way to manipulate the number of reports in this scenario is through the onAnalysis
callback option:
// Track iterations over output files
let analyzePluginIterations = 0;
export default {
input: 'myIsomorphicModule.js',
output: [
{
name: 'myIsomorphicModule',
format: 'cjs',
entryFileNames: '[name].cjs'
},
{
name: 'myIsomorphicModule',
format: 'es',
entryFileNames: '[name].mjs'
}
],
plugins: [
analyze({
onAnalysis: () => {
if (analyzePluginIterations > 0) {
throw ''; // We only want reports on the first output
}
analyzePluginIterations++;
}
})
]
}
License
MIT © Andrew Carpenter