Implements lexical analyzers as iterators yielding tokens.
Text file tokenizers that support multiple start states and lexical tracking implemented as standard Python iterators.
The class RegexTokenizer implements a tokenizer using the re module to recognize tokens in the input stream. Tokens and actions are defined by rules. The tokenizer calls user action functions associated with each rule. In most cases, the user action function can simply be a @classmethod constructor of a user-provided token class.
Each rule is specified as a tuple. The first element of the tuple is a regular expression that will be compiled by re and used to match a token. The second element of the tuple is a user-provided callable that will be passed the recognized text, along with the current lexical position.
In this example, the user class Token implements constructors as @classmethod functions, and these serve as the callables in each lexical rule.
Rules are specified in the class variable spec, which is a list of rules.
import tokenizertools as tt class MyTokenizer(tt.RegexTokenizer): spec = [ (r'[a-zA-Z][a-zA-Z0-9_]*',Token.type_ident), # idents and keywords (r'[0-9]+\.[0-9]+',Token.type_float), # floats (r'[0-9]+', Token.type_int), # ints (r'\s*',None), # ignore white space ]
Nothing else needs to be defined. All methods are inherited. Instantiate a lexer and commence parsing. The specification rules are compiled and cached on creation of the first instance.:
tokenizer = MyTokenizer() with open('foo.bar') as f: token_stream = Lookahead(tokenizer.lex(f, f.name)) compiled_stuff = my_parser.parse(token_stream)
1.0 Time to put some mileage on it.
Clean ups, PEP8-ification. Improved exception handling for bad begin() state.
0.1a Initial alpha.