Skip to content


Subversion checkout URL

You can clone with
Download ZIP
Parser for JavaScript, in Haskell
Haskell Yacc Logos Other

Fetching latest commit…

Cannot retrieve the latest commit at this time

Failed to load latest commit information.


Parser for JavaScript

Based (loosely) on language-python

How to build


cabal clean && cabal configure && cabal build


cabal clean && cabal configure -fbuildtests && cabal build

Running the tests


To debug the grammar

happy -iparse.txt -g -a  -d src/Language/JavaScript/Parser/Grammar5.y

This generates src/Language/JavaScript/Parser/Grammar.hs, delete this
when done with the debug version

UTF8/Unicode version

Alex 3.0 now supports unicode natively, and has been included as a
dependency in the cabal file.

Note: The generation of the lexical analyser has been separated out,
      to remove the install-time dependency on Alex. If any changes
      need to be made to the lexer, the Lexer.x source lies in
      src-dev, and the script will invoke Alex with the
      appropriate directories.

      At the moment (0.5.6), the lexer can only be generated with Alex


0.5.7 Remove the hs-source-dirs from test suite to prevent compilation
      issues (@nomeata) 

      Introduce parseFileUtf8 to explicitly use utf8 for parsing a
      file, and update tests to use it where needed. Closes #21

0.5.6 Remove constraint on Alex 3.0.1, it is only required to make
      changes to the lexer. Closes #19

0.5.5 Updated version ranges for GHC 7.6.1, courtesy of @mietek 
      Note: requires alex 3.0.1 for development (not install)

0.5.4 Fixed bug where lexer switches to regexp mode after a ']'
      char, reported by @aszlig

0.5.3 Merge pull requests from @aszlig to add octal support, and to
      allow leading zeros in exponents

0.5.2 Merged pull request from @markwright added some missing test
      files to the cabal file.
      Added and Lexer.x to the cabal file.

0.5.1 Export CommentAnnotation(..). Simplify AST by getting rid of
      JSFunctionBody, JSSourceElements, JSStatementBlock, JSStatementList.
      They are replaced by JSBlock or a simple list.
      Also fix lexer mode in presence of whitespace.
      Changed way tests are invoked, to allow Travis integration.

0.5.0 Rework AST to allow full round-trip parsing and output of
      JavaScript. Breaks AST compatibility with prior versions

0.4.10 Moved Lexer.x into a separate directory, and made a script to
       call alex to generate Lexer.hs. This means alex is not required
       at install time

0.4.9 Make alex and happy versions more explicit. Expose the AlexSpan
      data type, so parse error positions can be reported.

0.4.8 Close issue by
      correcting the lexical analyser for DecimalLiteral

0.4.7 Continue ECMASCRIPT update, remove incorrect future reserved words for
      "code", "mode", "of", "one", "or" and "strict", put in by accident/stupidity.

0.4.6 Update to ECMASCRIPT 5.1 by allowing continuations in string literals
      Include build dependencies on alex >= 3.0 and happy.

0.4.5 Update cabal file and docs for Alex 3.0, giving out of the box unicode support

0.4.4 Allow unicode characters in comments

0.4.3 correct lexer for decimalToken, only leading digit needs to be non-zero.

0.4.2 Expose SrcSpan

0.4.1 Expose AST.Node as well.

0.4.0 Updated JSNode to include location information, and made the parse more true to the original by leaving blocks as such, not replacing with ';'.

0.3.0 ECMA-262 edition 5 compliance.  Required adding JSFunctionExpression and JSPropertyAccessor to the AST.

0.2.2 Heading toward compliance with edition 5. Trailing commas allowed in array literals too.

0.2.1 Allow trailing comma in object literal

0.2.0 ECMAScript 3 allows function expressions to have names, AST.JSFunctionExpression now reflects this

0.1.0 Simplified AST by removing JSElement and JSElementList components

0.0.3 Support for unicode in source. At the moment it only supports
UTF8 encoding, does not recognise byte order marks or UTF-16/UTF-32

0.0.2 Multiline comments were processed in greedy form, now end of
comment recognised properly. Thanks to Tony Morris for reporting this.

0.0.1 Initial release


Something went wrong with that request. Please try again.