source map support

[michaelficarra]

This adds source map support to escodegen through use of the Mozilla Source Map library.

[travisbot]

This pull request fails (merged 1312afc4 into ab43e95).

[travisbot]

This pull request fails (merged 1312afc4 into ab43e95).

[michaelficarra]

@travisbot: The comment tests are failing. That's due to the code marked FIXME.

[michaelficarra]

@travisbot: The comment tests are failing. That's due to the code marked FIXME.

[Constellation]

Great work!!
I'll see code :)

[Constellation]

Great work!!
I'll see code :)

[Constellation]

I've created pull request for your branch. michaelficarra#1

[Constellation]

I've created pull request for your branch. michaelficarra#1

[michaelficarra]

Waiting on mozilla/source-map#19

[michaelficarra]

Waiting on mozilla/source-map#19

[michaelficarra]

Only one failing test remains, and this is due to running SourceNode.prototype.replaceRight on a SourceNode that looks like

{ children: [ '//A\n', '', '' ],
  line: null,
  column: null,
  source: undefined }

Since the rightmost strings are empty, replacing \s+$ does nothing. I'm going to change the mozilla source map code to ignore empty strings, guaranteeing that the rightmost string is non-empty.

[michaelficarra]

Only one failing test remains, and this is due to running SourceNode.prototype.replaceRight on a SourceNode that looks like

{ children: [ '//A\n', '', '' ],
  line: null,
  column: null,
  source: undefined }

Since the rightmost strings are empty, replacing \s+$ does nothing. I'm going to change the mozilla source map code to ignore empty strings, guaranteeing that the rightmost string is non-empty.

[michaelficarra]

Still TODO: I need to find a more efficient way to test the generated string than toSourceNode(X).toString() (or avoid the tests).

[michaelficarra]

Still TODO: I need to find a more efficient way to test the generated string than toSourceNode(X).toString() (or avoid the tests).

[michaelficarra]

With mozilla/source-map#19 and now mozilla/source-map#20, all tests are passing.

[michaelficarra]

With mozilla/source-map#19 and now mozilla/source-map#20, all tests are passing.

[michaelficarra]

Alright, this LGTM. Usage with source map generation will have to wait until those mozilla/source-map pulls are merged, but it's working fine as-is. Let's 🚢 it.

Future work: optimise toSourceNode(X).toString() usage depending on how it is used.

[michaelficarra]

Alright, this LGTM. Usage with source map generation will have to wait until those mozilla/source-map pulls are merged, but it's working fine as-is. Let's 🚢 it.

Future work: optimise toSourceNode(X).toString() usage depending on how it is used.

[Constellation]

Alright, this LGTM. Usage with source map generation will have to wait until those mozilla/source-map pulls are merged, but it's working fine as-is. Let's it.

Yes, I think so too. Merging this. Thanks for your great contribution :)

Future work: optimise toSourceNode(X).toString() usage depending on how it is used.

NOTE: In my environment (node version 0.8.8), npm test result is following.

• master branch
  test: 312 tests. 0 failures. 72 ms
  api: 7 tests. 0 failures. 1 ms
  options: 468 tests. 0 failures. 92 ms
  comment: 18 tests. 0 failures. 13 ms
  compare: 6 tests. 0 failures. 11 ms
  identity: 11 tests. 0 failures. 11064 ms
• source-maps branch
  test: 312 tests. 0 failures. 77 ms
  api: 7 tests. 0 failures. 1 ms
  options: 468 tests. 0 failures. 103 ms
  comment: 18 tests. 0 failures. 17 ms
  compare: 6 tests. 0 failures. 14 ms
  identity: 11 tests. 0 failures. 13038 ms

[Constellation]

Alright, this LGTM. Usage with source map generation will have to wait until those mozilla/source-map pulls are merged, but it's working fine as-is. Let's it.

Yes, I think so too. Merging this. Thanks for your great contribution :)

Future work: optimise toSourceNode(X).toString() usage depending on how it is used.

NOTE: In my environment (node version 0.8.8), npm test result is following.

• master branch
  test: 312 tests. 0 failures. 72 ms
  api: 7 tests. 0 failures. 1 ms
  options: 468 tests. 0 failures. 92 ms
  comment: 18 tests. 0 failures. 13 ms
  compare: 6 tests. 0 failures. 11 ms
  identity: 11 tests. 0 failures. 11064 ms
• source-maps branch
  test: 312 tests. 0 failures. 77 ms
  api: 7 tests. 0 failures. 1 ms
  options: 468 tests. 0 failures. 103 ms
  comment: 18 tests. 0 failures. 17 ms
  compare: 6 tests. 0 failures. 14 ms
  identity: 11 tests. 0 failures. 13038 ms

[michaelficarra]

NOTE: In my environment (node version 0.8.8), npm test result is following.

Remember, that's using the SourceNodeMock constructor. I wouldn't expect that to be much slower. When actually generating source maps, we will see the biggest hit because we're flattening the entire tree structure into a string, usually just for a small portion of it.

[michaelficarra]

NOTE: In my environment (node version 0.8.8), npm test result is following.

Remember, that's using the SourceNodeMock constructor. I wouldn't expect that to be much slower. When actually generating source maps, we will see the biggest hit because we're flattening the entire tree structure into a string, usually just for a small portion of it.

[michaelficarra]

The mozilla/source-map pulls have been merged. Everything should be 100% operational.

[michaelficarra]

The mozilla/source-map pulls have been merged. Everything should be 100% operational.

[Constellation]

The mozilla/source-map pulls have been merged. Everything should be 100% operational.

Great. After source-map package is published, we can generate source-map easily.

[Constellation]

The mozilla/source-map pulls have been merged. Everything should be 100% operational.

Great. After source-map package is published, we can generate source-map easily.

[fitzgen]

I just pushed source-map@0.1.2 to npm.

[fitzgen]

I just pushed source-map@0.1.2 to npm.

[Constellation]

@fitzgen
Awesome!
Because this is big change, I'm also planning to publish this as version 0.0.7 soon.

[Constellation]

@fitzgen
Awesome!
Because this is big change, I'm also planning to publish this as version 0.0.7 soon.

[fitzgen]

I am very excited at the thought of many existing and new compilers which target JS using escodegen as a backend so that they can get source maps for free!

[fitzgen]

I am very excited at the thought of many existing and new compilers which target JS using escodegen as a backend so that they can get source maps for free!

[Constellation]

I've just published version 0.0.7, shipping source map option!
Thanks all :-)

[Constellation]

I've just published version 0.0.7, shipping source map option!
Thanks all :-)

[michaelficarra]

We should probably document the usage of the sourceMap option. Basically, if you provide a non-empty string value for options.sourceMap, escodegen will generate a source map, using the given string as the source map's source file name.

[michaelficarra]

We should probably document the usage of the sourceMap option. Basically, if you provide a non-empty string value for options.sourceMap, escodegen will generate a source map, using the given string as the source map's source file name.

[Constellation]

We should probably document the usage of the sourceMap option.

Yes. I've just updated API wiki page.
https://github.com/Constellation/escodegen/wiki/API
Please feel free to update wiki page.

[Constellation]

We should probably document the usage of the sourceMap option.

Yes. I've just updated API wiki page.
https://github.com/Constellation/escodegen/wiki/API
Please feel free to update wiki page.