diff options
author | Daniel Friesel <daniel.friesel@uos.de> | 2019-12-12 17:34:01 +0100 |
---|---|---|
committer | Daniel Friesel <daniel.friesel@uos.de> | 2019-12-12 17:34:01 +0100 |
commit | de382bd34807a32af948a017164f0fb91cb2e6cb (patch) | |
tree | c4b0f968637c2e4ebf120373cc42a27cfea3bc47 /lib | |
parent | ff6fc70456354b23663bee037c5e737a2b437ca8 (diff) |
lexer and parser for basic timed word strings
Diffstat (limited to 'lib')
-rw-r--r-- | lib/lex.py | 62 |
1 files changed, 62 insertions, 0 deletions
diff --git a/lib/lex.py b/lib/lex.py new file mode 100644 index 0000000..c0323fa --- /dev/null +++ b/lib/lex.py @@ -0,0 +1,62 @@ +from sly import Lexer, Parser + + +class TimedWordLexer(Lexer): + tokens = {LPAREN, RPAREN, IDENTIFIER, NUMBER, ARGSEP, FUNCTIONSEP} + ignore = ' \t' + + LPAREN = r'\(' + RPAREN = r'\)' + IDENTIFIER = r'[a-zA-Z_][a-zA-Z0-9_]*' + NUMBER = r'[0-9e.]+' + ARGSEP = r',' + FUNCTIONSEP = r';' + + +class TimedWordParser(Parser): + tokens = TimedWordLexer.tokens + + @_('timedSymbol FUNCTIONSEP timedWord') + def timedWord(self, p): + ret = [p.timedSymbol] + ret.extend(p.timedWord) + return ret + + @_('timedSymbol FUNCTIONSEP', 'timedSymbol') + def timedWord(self, p): + return [p.timedSymbol] + + @_('IDENTIFIER', 'IDENTIFIER LPAREN RPAREN') + def timedSymbol(self, p): + return (p.IDENTIFIER,) + + @_('IDENTIFIER LPAREN args RPAREN') + def timedSymbol(self, p): + return (p.IDENTIFIER, *p.args) + + @_('arg ARGSEP args') + def args(self, p): + ret = [p.arg] + ret.extend(p.args) + return ret + + @_('arg') + def args(self, p): + return [p.arg] + + @_('NUMBER') + def arg(self, p): + return [float(p.NUMBER)] + + @_('IDENTIFIER') + def arg(self, p): + return [p.IDENTIFIER] + + +if __name__ == '__main__': + data = 'init(); sleep(12345); foo(_, 7);' + lexer = TimedWordLexer() + parser = TimedWordParser() + for tok in lexer.tokenize(data): + print('type={}, value={}'.format(tok.type, tok.value)) + print(parser.parse(lexer.tokenize(data))) |