cerxes/babel
3
0
Fork 0
Code Issues Pull Requests Releases Wiki Activity
443 Commits 5 Branches 418 Tags
Commit Graph

271 Commits

Author SHA1 Message Date
Ingvar Stepanyan
ff60ee4fdb [estree] Align export ... representations. 2015-03-10 10:37:00 +02:00
Ingvar Stepanyan
722bc65e79 [estree] Align Import specifiers. 2015-03-10 02:27:24 +02:00
Marijn Haverbeke
12558821c5 Use an object argument in the TokenType constuctor 2015-03-06 10:37:35 +01:00
Marijn Haverbeke
4e0a7fac71 Linting changes 2015-03-06 10:36:17 +01:00
Marijn Haverbeke
8459481e65 Represent a tokenizer as an instance of the parser 
This completely changes the interface, and removes most of the complexity
in the old tokenizer interface (jump-to-position was removed, since it is
all kinds of unreliable given the new tokenizer context system).
 2015-03-05 17:30:48 +01:00
Marijn Haverbeke
0df2affdfe Move loose parser to object style 
Drop dependency on tokenizer interface
 2015-03-05 17:22:06 +01:00
Marijn Haverbeke
3365478645 Add a plugin mechanism to allow JSX parser to be a module 
And export a few more things
 2015-03-05 16:00:34 +01:00
Marijn Haverbeke
06f3b3c224 Attach context update algorithm to token types 
For, theoretically, easier extendability
 2015-03-05 15:59:12 +01:00
Marijn Haverbeke
6bd1013f2c Use uniform object type for node types 
Speeds things up by about 9% on io.js 1.3
 2015-03-05 15:59:12 +01:00
Marijn Haverbeke
141905f9fd Consume whitespace before, rather than after, reading a token 
Simplifies several things
 2015-03-05 15:59:12 +01:00
Marijn Haverbeke
bc48c02a18 Move the parser state into an object 
Makes almost everything in acorn.js a method of this object.

Performance is not significantly affected on a modern V8. Makes the
code reentrant and allows us to more easily expose more methods
externally in the future.

On the other hand, the proliferation of `this.` is unfortunate.
 2015-03-05 15:59:12 +01:00
Marijn Haverbeke
33307e789a Update tools/generate-identifier-regex.js 
So that it can also spit out the astral maps.
 2015-03-05 15:55:38 +01:00
Marijn Haverbeke
d76ea4b3be Make the tokenizer aware of multi-byte characters in ES6 mode 
Add a data structure to recognize astral identifier chars. Parse whole
code points when looking for identifiers.

Issue #215
 2015-03-05 15:55:38 +01:00
Marijn Haverbeke
891d5d07dd Allow braced \u escapes in identifiers 
We still can't properly recognize code points as ES6-style
identifier chars.

Issue #214
 2015-03-04 15:37:26 +01:00
Marijn Haverbeke
59dc29f3f0 In ES6-mode, don't treat keywords with escaped characters as regular identifiers 
See https://mathiasbynens.be/notes/javascript-identifiers-es6
 2015-03-04 15:29:15 +01:00
Marijn Haverbeke
cbd8aa8f06 Bump version number post-0.12 2015-03-02 11:31:27 +01:00
Marijn Haverbeke
99f1989f72 Mark release 0.12 2015-03-02 11:31:11 +01:00
Ingvar Stepanyan
7264bc0178 Fix found V8 deopt. 
Increases perf from 230K to 360K lines/sec on std bench on my machine.
 2015-02-05 00:02:21 +02:00
Ingvar Stepanyan
c034550215 Disallow line break between arrow function params and =>. 2015-02-04 12:57:18 +02:00
Ingvar Stepanyan
6eb177582d Fix export default declarations. 
Closes #184.
 2015-01-27 16:21:41 +02:00
Ingvar Stepanyan
6660a21b79 Disallow MemberExpression in bindings; clarify function namings. 2015-01-27 15:23:09 +02:00
Ingvar Stepanyan
3d5964ceed Fix locations for AssignmentPatterns. 2015-01-27 15:10:25 +02:00
Ingvar Stepanyan
5d96bbd781 Simplify & fix rest argument validity checks. 2015-01-24 13:38:14 +02:00
Ingvar Stepanyan
cdd444eff1 Speed-up reading words, strings and templates. 
Now identifiers and strings even with escaped chars are read
in optimized way by reading entire chunks delimited by escape
chars (and not bailing to deopt mode on first one).
 2015-01-24 12:42:19 +02:00
Ingvar Stepanyan
65d09eac6e Implement shorthand property assignment in ambiguous contexts. 
Issue #181.
 2015-01-24 12:40:24 +02:00
Ingvar Stepanyan
f0569147e6 Avoid extra call and arg in parseExpression for single-expression case. 2015-01-23 21:48:33 +02:00
Ingvar Stepanyan
41ad304955 Introduce helpers for contextual keywords. 2015-01-23 21:36:16 +02:00
Ingvar Stepanyan
917de714b1 Switch from Function.{rest,defaults} to AssignmentPattern and RestElement. 
Closes #182.
 2015-01-23 15:29:37 +02:00
Ingvar Stepanyan
38609ae26d Closes #205. 2015-01-23 14:14:02 +02:00
Ingvar Stepanyan
e6f1e1027e Fix dividing function expression. 
Fixes #204.
 2015-01-23 13:17:03 +02:00
Ingvar Stepanyan
39d752dc6b Make top-level equal to block statement context. 
Fixes #203.
 2015-01-23 11:48:39 +02:00
Ingvar Stepanyan
555c7898c8 Remove Token::{startLoc,endLoc} so they don't appear in next release. 2015-01-22 18:21:38 +02:00
Ingvar Stepanyan
caa5da6ce1 Improve spread element parsing (fix allowed contexts and error locations). 2015-01-21 23:51:40 +02:00
Marijn Haverbeke
94b5efcd3e Disallow declaration statements in block-less context 
Closes #202
 2015-01-20 12:55:21 +01:00
Marijn Haverbeke
a1d2561cfa Restore patch 9f7cb552648829796ab4d6836d08246e95f16b9d to original shape 
And make loose parser handle stray class semicolons

Issue #190
 2015-01-20 12:02:30 +01:00
Sebastian McKenzie
9f7cb55264 Add stray semicolons as class elements 2015-01-20 11:56:33 +01:00
Ingvar Stepanyan
6dee98d1b9 Adapt ES6 template handling to new tokenizer. 
Avoid need for:
* extra `templates` array in favor of new `tokContext`;
* special location handling for first & last template elements;
* separate `_templateContinued` token in favor of same `_template`.

Adds:
* token types for backQuote and dollarBraceL instead of skipping them
so they can be handled (i.e. highlighted differently).
 2015-01-17 22:21:22 +01:00
Ingvar Stepanyan
f6c45ac59f Re-read only number or string after "use strict". 
Fixes double-entering same tokContext for various parentheses.
 2015-01-14 23:10:10 +02:00
Ingvar Stepanyan
e7beee177d Remove deprecated ComprehensionBlock.of property. 
Comprehensions were moved to ES7 anyway, so there is no sense
in keeping intermediate no-more-supported syntax.
 2015-01-14 12:35:00 +02:00
Ingvar Stepanyan
7e85da74cb shouldSkipSpace is no more needed in finishToken. 2015-01-14 12:31:59 +02:00
Ingvar Stepanyan
ad9411d2ae Made tokenize() compliant with ES6 iterables for easier processing. 2015-01-14 12:29:20 +02:00
Marijn Haverbeke
0f55a53a7d [loose parser] Fetch token before comment when tokenizer raises unterminated comment error 
Closes #197
 2015-01-13 22:18:55 +01:00
Ingvar Stepanyan
d1f95ece42 Revert "Disallow parentheses in lvalue except as in computed keys or default values." 
This reverts commit 85087f2a0951f60ca035ca6d785e4bab6d96d8f7.
Fixes #193.
 2015-01-12 20:31:38 +02:00
Ingvar Stepanyan
bc2e01aa03 Allow static as method name in class (fixes #192). 
Add uncommitted tests for #191.
 2015-01-09 15:31:01 +02:00
Ingvar Stepanyan
3d9048a692 Allow pattern in catch param (fixes #191). 2015-01-09 02:50:32 +02:00
Ingvar Stepanyan
85087f2a09 Disallow parentheses in lvalue except as in computed keys or default values. 2015-01-08 23:00:22 +02:00
Ingvar Stepanyan
4d4a76588c Parse assignment patterns in-place in certain contexts. 
* Parsing assignables without extra transform step when possible (speed-up).
* Added support for shorthand defaults in such certain contexts (issue #181).
 2015-01-08 20:19:48 +02:00
Marijn Haverbeke
0897901f1f Slight cleanup of '/' disambiguation 
Issue #189
 2015-01-06 11:03:50 +01:00
Marijn Haverbeke
b6b085ac8e Make tokenizer independent of parser (with regards to '/' disambiguation) 
Issue #189
 2015-01-06 10:58:52 +01:00
Marijn Haverbeke
7c6a8b7c76 Use parseExprSubscripts instead of parseExpression when parsing superclass 
Issue #187
 2015-01-04 22:30:50 +01:00