Python 2.3.3 Grammar Terence Parr and Loring Craymer February 2004 Converted to ANTLR v3 November 2005 by Terence Parr. This grammar was derived automatically from the Python 2.3.3 parser grammar to get a syntactically correct ANTLR grammar for Python. Then Terence hand tweaked it to be semantically correct; i.e., removed lookahead issues etc... It is LL(1) except for the (sometimes optional) trailing commas and semi-colons. It needs two symbols of lookahead in this case. Starting with Loring's preliminary lexer for Python, I modified it to do my version of the whole nasty INDENT/DEDENT issue just so I could understand the problem better. This grammar requires PythonTokenStream.java to work. Also I used some rules from the semi-formal grammar on the web for Python (automatically translated to ANTLR format by an ANTLR grammar, naturally ). The lexical rules for python are particularly nasty and it took me a long time to get it 'right'; i.e., think about it in the proper way. Resist changing the lexer unless you've used ANTLR a lot. ;) I (Terence) tested this by running it on the jython-2.1/Lib directory of 40k lines of Python. REQUIRES ANTLR v3 Updated the original parser for Python 2.5 features. The parser has been altered to produce an AST - the AST work started from tne newcompiler grammar from Jim Baker. The current parsing and compiling strategy looks like this: Python source->Python.g->AST (org/python/parser/ast/*)->CodeCompiler(ASM)->.class