hzoo / babel-plugin-dedupe-string-literals Goto Github PK

I'm targeting ES6 and would like to use this together with babili

Understandably this tool is slow for large files such as https://raw.githubusercontent.com/mathiasbynens/unicode-8.0.0/15228776965be9ef1af995403b7d47386607318c/bidi-classes/index.js (over 6 MB).

One idea is to use a Map rather than a plain object for the cache, which should result in faster lookups.

When using dedupe-string-literals with webpack and babel-loader, we don't want strings passed to require.ensure() or require() to be deduped. Having another option that we could pass into this plugin would us specify strings that we do not want deduped.

plugins: [
    ['dedupe-string-literals', {
        minimumStringLength: 2,
        ignore: ['polyfills/x', 'modules/chunk']
    }]
]

It's hardcoded for 2 cases

['a','a'] - ExpressionStatement, ArrayExpression in the body
var a = ['a','a'] - VariableDeclaration, VariableDeclarator, ArrayExpression.

Needs to traverse parents to some level where a VariableDeclaration is possible - basically the closest parent function scope?

It's only visiting arrays - doesn't have to be. but outside of the initial scope

Basically need to account for [['longstring', 'longstring'], ['longstring', 'longstring']], since currently it only dedupes per array.

https://github.com/hzoo/babel-plugin-dedupe-string-literals/blob/master/test/fixtures/example/expected.js as an example

cc @mathiasbynens - I think that's a blocker for the files you mentioned since I saw the nested arrays.