1 |
11
|
aaronmk
|
# A general recursive descent parser
|
2 |
|
|
|
3 |
|
|
import re
|
4 |
|
|
|
5 |
|
|
class SyntaxException(Exception): pass
|
6 |
|
|
|
7 |
|
|
class Parser:
|
8 |
|
|
def __init__(self, string):
|
9 |
|
|
self._str = string
|
10 |
|
|
self._pos = 0
|
11 |
|
|
|
12 |
|
|
def parse(self):
|
13 |
|
|
tree = self._main()
|
14 |
|
|
if not self._pos == len(self._str): self._syntax_err('End of string')
|
15 |
|
|
return tree
|
16 |
|
|
|
17 |
|
|
def _match_re(self, pattern, required=False):
|
18 |
|
|
matcher = re.compile(pattern).match(self._str, self._pos)
|
19 |
|
|
if matcher:
|
20 |
|
|
self._pos = matcher.end(0)
|
21 |
|
|
return matcher.group(0)
|
22 |
|
|
elif required: self._syntax_err(pattern)
|
23 |
38
|
aaronmk
|
else: return None
|
24 |
11
|
aaronmk
|
|
25 |
|
|
def _match_str(self, string, required=False):
|
26 |
|
|
end_pos = self._pos + len(string)
|
27 |
|
|
if self._str[self._pos:end_pos] == string:
|
28 |
|
|
self._pos = end_pos
|
29 |
|
|
return True
|
30 |
|
|
elif required: self._syntax_err(string)
|
31 |
|
|
else: return False
|
32 |
|
|
|
33 |
|
|
def _syntax_err(self, token):
|
34 |
|
|
raise SyntaxException(token+' expected in '+self._str[self._pos:])
|