source-map-resolve

Resolve the source map and/or sources for a generated file.


Keywords
source map, sourcemap, source, map, sourceMappingURL, resolve, resolver, locate, locator, find, finder
License
MIT
Install
npm install source-map-resolve@0.6.0

Documentation

Deprecated

Is npm bugging you about this module being deprecated? You probably depend on it via an old version of micromatch:

some-package > some-other-package > micromatch > nanomatch > snapdragon > source-map-resolve

In the above case, you could try:

  • To make some-other-package update to the latest version of micromatch, which doesn’t depend on nanomatch anymore.
  • To make some-package not depend on some-other-package.
  • Check that you use the latest version of some-package. Maybe they have already updated the deprecated stuff away!
  • Maybe your dependency chain is even longer? See if the bad stuff can be cut off at some other point.

Or maybe some package you use depends on nanomatch? If so, try to make that package switch to micromatch instead.

So why is source-map-resolve deprecated? Well, source-map-resolve tries to work with source maps completely in general – support every feature and be completely language agnostic. That is a cool goal, but has a couple of flaws:

  • Finding source map comments. This is done via regex to support any language, but in reality you need to know the language to know what a comment is, as opposed to something that looks like a comment inside a string. In reality, only JS and CSS use source maps, but even then you have further problems. The source map specification is vague on exactly how to find a source map comment, and browsers differ. So what does it even mean to find the source map comment of a file? You could try to reverse engineer how browsers do it, but what to do when they disagree?
  • URLs. Source maps link to source files via URLs, not file paths. But source-map-resolve works with file paths. How should it translate URLs to file paths? source-map-resolve makes assumptions and has some issues on Windows.
  • The “general” approach. The source map spec is, well, pretty under specified, so it’s difficult to decide how source-map-resolve should work at all.

If you still find this package useful and would like to use it:

  • First think through if you really need it. Why are you “resolving source maps” (whatever that even means)? Is there some simpler way to solve your problem? Maybe in your case you don’t have to support everything because your tool only deals with JS and deals with source maps that look a certain way?
  • Copy good parts of the code.
  • Fork it. If you do, please let me know so I can link to your fork. Also note that there is a branch called next with more modern code, from another contributor (that I shamefully never released).

Overview

Resolve the source map and/or sources for a generated file.

var sourceMapResolve = require("source-map-resolve")
var sourceMap        = require("source-map")

var code = [
  "!function(){...}();",
  "/*# sourceMappingURL=foo.js.map */"
].join("\n")

sourceMapResolve.resolveSourceMap(code, "/js/foo.js", fs.readFile, function(error, result) {
  if (error) {
    return notifyFailure(error)
  }
  result
  // {
  //   map: {file: "foo.js", mappings: "...", sources: ["/coffee/foo.coffee"], names: []},
  //   url: "/js/foo.js.map",
  //   sourcesRelativeTo: "/js/foo.js.map",
  //   sourceMappingURL: "foo.js.map"
  // }

  sourceMapResolve.resolveSources(result.map, result.sourcesRelativeTo, fs.readFile, function(error, result) {
    if (error) {
      return notifyFailure(error)
    }
    result
    // {
    //   sourcesResolved: ["/coffee/foo.coffee"],
    //   sourcesContent: ["<contents of /coffee/foo.coffee>"]
    // }
  })
})

sourceMapResolve.resolve(code, "/js/foo.js", fs.readFile, function(error, result) {
  if (error) {
    return notifyFailure(error)
  }
  result
  // {
  //   map: {file: "foo.js", mappings: "...", sources: ["/coffee/foo.coffee"], names: []},
  //   url: "/js/foo.js.map",
  //   sourcesRelativeTo: "/js/foo.js.map",
  //   sourceMappingURL: "foo.js.map",
  //   sourcesResolved: ["/coffee/foo.coffee"],
  //   sourcesContent: ["<contents of /coffee/foo.coffee>"]
  // }
  result.map.sourcesContent = result.sourcesContent
  var map = new sourceMap.sourceMapConsumer(result.map)
  map.sourceContentFor("/coffee/foo.coffee")
  // "<contents of /coffee/foo.coffee>"
})

Installation

npm install source-map-resolve

Usage

sourceMapResolve.resolveSourceMap(code, codeUrl, read, callback)

  • code is a string of code that may or may not contain a sourceMappingURL comment. Such a comment is used to resolve the source map.
  • codeUrl is the url to the file containing code. If the sourceMappingURL is relative, it is resolved against codeUrl.
  • read(url, callback) is a function that reads url and responds using callback(error, content). In Node.js you might want to use fs.readFile, while in the browser you might want to use an asynchronus XMLHttpRequest.
  • callback(error, result) is a function that is invoked with either an error or null and the result.

The result is an object with the following properties:

  • map: The source map for code, as an object (not a string).
  • url: The url to the source map. If the source map came from a data uri, this property is null, since then there is no url to it.
  • sourcesRelativeTo: The url that the sources of the source map are relative to. Since the sources are relative to the source map, and the url to the source map is provided as the url property, this property might seem superfluos. However, remember that the url property can be null if the source map came from a data uri. If so, the sources are relative to the file containing the data uri—codeUrl. This property will be identical to the url property or codeUrl, whichever is appropriate. This way you can conveniently resolve the sources without having to think about where the source map came from.
  • sourceMappingURL: The url of the sourceMappingURL comment in code.

If code contains no sourceMappingURL, the result is null.

sourceMapResolve.resolveSources(map, mapUrl, read, [options], callback)

  • map is a source map, as an object (not a string).
  • mapUrl is the url to the file containing map. Relative sources in the source map, if any, are resolved against mapUrl.
  • read(url, callback) is a function that reads url and responds using callback(error, content). In Node.js you might want to use fs.readFile, while in the browser you might want to use an asynchronus XMLHttpRequest.
  • options is an optional object with any of the following properties:
    • sourceRoot: Override the sourceRoot property of the source map, which might only be relevant when resolving sources in the browser. This lets you bypass it when using the module outside of a browser, if needed. Pass a string to replace the sourceRoot property with, or false to ignore it. Defaults to undefined.
  • callback(error, result) is a function that is invoked with either an error or null and the result.

The result is an object with the following properties:

  • sourcesResolved: The same as map.sources, except all the sources are fully resolved.
  • sourcesContent: An array with the contents of all sources in map.sources, in the same order as map.sources. If getting the contents of a source fails, an error object is put into the array instead.

sourceMapResolve.resolve(code, codeUrl, read, [options], callback)

The arguments are identical to sourceMapResolve.resolveSourceMap, except that you may also provide the same options as in sourceMapResolve.resolveSources.

This is a convenience method that first resolves the source map and then its sources. You could also do this by first calling sourceMapResolve.resolveSourceMap and then sourceMapResolve.resolveSources.

The result is identical to sourceMapResolve.resolveSourceMap, with the properties from sourceMapResolve.resolveSources merged into it.

There is one extra feature available, though. If code is null, codeUrl is treated as a url to the source map instead of to code, and will be read. This is handy if you sometimes get the source map url from the SourceMap: <url> header (see the Notes section). In this case, the sourceMappingURL property of the result is null.

sourceMapResolve.*Sync()

There are also sync versions of the three previous functions. They are identical to the async versions, except:

  • They expect a sync reading function. In Node.js you might want to use fs.readFileSync, while in the browser you might want to use a synchronus XMLHttpRequest.
  • They throw errors and return the result instead of using a callback.

sourceMapResolve.resolveSourcesSync also accepts null as the read parameter. The result is the same as when passing a function as the read parameter, except that the sourcesContent property of the result will be an empty array. In other words, the sources aren’t read. You only get the sourcesResolved property. (This only supported in the synchronus version, since there is no point doing it asynchronusly.)

sourceMapResolve.parseMapToJSON(string, [data])

The spec says that if a source map (as a string) starts with )]}', it should be stripped off. This is to prevent XSSI attacks. This function does that and returns the result of JSON.parseing what’s left.

If this function throws error, error.sourceMapData === data.

Errors

All errors passed to callbacks or thrown by this module have a sourceMapData property that contain as much as possible of the intended result of the function up until the error occurred.

Note that while the map property of result objects always is an object, error.sourceMapData.map will be a string if parsing that string fails.

Note

This module resolves the source map for a given generated file by looking for a sourceMappingURL comment. The spec defines yet a way to provide the URL to the source map: By sending the SourceMap: <url> header along with the generated file. Since this module doesn’t retrive the generated code for you (instead you give the generated code to the module), it’s up to you to look for such a header when you retrieve the file (should the need arise).

License

MIT.