BUG: Handle Array().concat() version of webpackJsonp

[ryan-roemer]

From @ryan-codingintrigue in FormidableLabs/webpack-dashboard#182 (comment)

@ryan-roemer I've been getting this error too since 1.0.0-7. I have a pretty big SPA, so it's been hard to get a repro of the issue, but I think I've narrowed it down to something to do with my bundle splitting.

My minimal example outputs a few bundles like this:

// bundle.0.js
webpackJsonp([0],[ ...
// bundle.1.js
webpackJsonp([1],{ ...
// etc.

However, my large SPA's bundles look like this:

// bundle.0.js
webpackJsonp([0],Array(171).concat([
// bundle.1.js
webpackJsonp([1],Array(171).concat([

I haven't been able to identify why, but it looks like this is causing the inspectpack parser/index.js#179 to be hit, making the parser assume throw that there "No code sections found"

The issue looks to be in getJsonpModules, where we've found a webpackJsonp, but expect second arg to be [] instead of Array(NUM).concat([]): https://github.com/FormidableLabs/inspectpack/blob/master/lib/parser/find-module-expression.js#L17-L22

[ryan-codingintrigue]

Given that the parameter index in webpackJsonp is still correct, would it be easier to add the check into extractModules? This seems to work for me:

if (t.isObjectExpression(node)) {
  // ...
}
if (t.isCallExpression(node) && node.arguments.length) {
  const potentialArrayNode = node.arguments[0];
  if (t.isArrayExpression(potentialArrayNode)) {
    return extractFromArrayExpression(potentialArrayNode, rawCode, opts);
  }
}

[ryan-roemer]

@ryan-codingintrigue -- Cool! I'll wrap that into the branch I'm working on with regression tests...