Revision 76
Added by Aaron Marcuse-Kubitza about 13 years ago
| scripts/lib/Parser.py | ||
|---|---|---|
| 5 | 5 |
class SyntaxException(Exception): pass |
| 6 | 6 |
|
| 7 | 7 |
class Parser: |
| 8 |
def __init__(self, string):
|
|
| 9 |
self._str = string
|
|
| 8 |
def __init__(self, str_):
|
|
| 9 |
self._str = str_
|
|
| 10 | 10 |
self._pos = 0 |
| 11 | 11 |
|
| 12 |
def parse(self): |
|
| 13 |
tree = self._main() |
|
| 12 |
def end(self): |
|
| 14 | 13 |
if not self._pos == len(self._str): self._syntax_err('End of string')
|
| 15 |
return tree |
|
| 16 | 14 |
|
| 17 |
def _match_re(self, pattern, required=False): |
|
| 15 |
def str_(self, str_, required=False): |
|
| 16 |
end_pos = self._pos + len(str_) |
|
| 17 |
if self._str[self._pos:end_pos] == str_: |
|
| 18 |
self._pos = end_pos |
|
| 19 |
return True |
|
| 20 |
elif required: self._syntax_err(str_) |
|
| 21 |
else: return False |
|
| 22 |
|
|
| 23 |
def re(self, pattern, required=False): |
|
| 18 | 24 |
matcher = re.compile(pattern).match(self._str, self._pos) |
| 19 | 25 |
if matcher: |
| 20 | 26 |
self._pos = matcher.end(0) |
| ... | ... | |
| 22 | 28 |
elif required: self._syntax_err(pattern) |
| 23 | 29 |
else: return None |
| 24 | 30 |
|
| 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 | 31 |
def _syntax_err(self, token): |
| 34 | 32 |
raise SyntaxException(self.__class__.__name__+' syntax error: '+token |
| 35 | 33 |
+' expected in '+self._str[self._pos:]) |
| scripts/lib/xpath.py | ||
|---|---|---|
| 34 | 34 |
len(elem.attrs[0]) == 1: return elem.attrs[0] |
| 35 | 35 |
else: return None |
| 36 | 36 |
|
| 37 |
class XpathParser(Parser): |
|
| 38 |
def _main(self): |
|
| 39 |
self._match_str('/') # optional leading /
|
|
| 40 |
return self._path() |
|
| 37 |
def parse(str_): |
|
| 38 |
parser = Parser(str_) |
|
| 41 | 39 |
|
| 42 |
def _path(self):
|
|
| 40 |
def _path(): |
|
| 43 | 41 |
tree = [] |
| 44 | 42 |
trailing_slash = False |
| 45 | 43 |
while True: |
| 46 | 44 |
# Split path |
| 47 |
if self._match_str('{'):
|
|
| 45 |
if parser.str_('{'):
|
|
| 48 | 46 |
paths = [] |
| 49 | 47 |
while True: |
| 50 |
paths.append(tree + self._path())
|
|
| 51 |
if not self._match_str(','): break
|
|
| 52 |
self._match_str('}', required=True)
|
|
| 48 |
paths.append(tree + _path()) |
|
| 49 |
if not parser.str_(','): break
|
|
| 50 |
parser.str_('}', required=True)
|
|
| 53 | 51 |
tree = paths[0] # just use first subpath for now |
| 54 | 52 |
break # nothing allowed after split path |
| 55 | 53 |
|
| 56 |
elem = XpathElem(is_attr=self._match_str('@'),
|
|
| 57 |
name=self._match_re(r'[\w.*]+', required=True))
|
|
| 54 |
elem = XpathElem(is_attr=parser.str_('@'),
|
|
| 55 |
name=parser.re(r'[\w.*]+', required=True))
|
|
| 58 | 56 |
|
| 59 | 57 |
# Attrs |
| 60 |
if self._match_str('['):
|
|
| 58 |
if parser.str_('['):
|
|
| 61 | 59 |
elem.attrs = [] |
| 62 | 60 |
while True: |
| 63 |
path = self._path()
|
|
| 64 |
if self._match_str('='):
|
|
| 65 |
set_value(path, self._match_re(r'[\w.|]*'))
|
|
| 61 |
path = _path() |
|
| 62 |
if parser.str_('='):
|
|
| 63 |
set_value(path, parser.re(r'[\w.|]*'))
|
|
| 66 | 64 |
elem.attrs.append(path) |
| 67 |
if not self._match_str(','): break
|
|
| 68 |
self._match_str(']', required=True)
|
|
| 65 |
if not parser.str_(','): break
|
|
| 66 |
parser.str_(']', required=True)
|
|
| 69 | 67 |
|
| 70 |
elem.is_ptr = self._match_str('->')
|
|
| 68 |
elem.is_ptr = parser.str_('->')
|
|
| 71 | 69 |
tree.append(elem) |
| 72 | 70 |
|
| 73 | 71 |
# Lookahead assertion |
| 74 |
if self._match_str('('):
|
|
| 75 |
self._match_str('/', required=True) # next / is inside ()
|
|
| 76 |
path = self._path()
|
|
| 77 |
self._match_str(')', required=True)
|
|
| 72 |
if parser.str_('('):
|
|
| 73 |
parser.str_('/', required=True) # next / is inside ()
|
|
| 74 |
path = _path() |
|
| 75 |
parser.str_(')', required=True)
|
|
| 78 | 76 |
elem.attrs.append(path) |
| 79 | 77 |
tree += path |
| 80 | 78 |
|
| 81 |
if not self._match_str('/'): break
|
|
| 79 |
if not parser.str_('/'): break
|
|
| 82 | 80 |
|
| 83 | 81 |
# Expand * abbrs |
| 84 | 82 |
for i in reversed(xrange(len(tree))): |
| ... | ... | |
| 93 | 91 |
except IndexError: pass # no replacement elem |
| 94 | 92 |
|
| 95 | 93 |
return tree |
| 94 |
|
|
| 95 |
parser.str_('/') # optional leading /
|
|
| 96 |
path = _path() |
|
| 97 |
parser.end() |
|
| 98 |
return path |
|
| 96 | 99 |
|
| 97 |
def parse(string): return XpathParser(string).parse() |
|
| 98 |
|
|
| 99 | 100 |
instance_level = 1 |
| 100 | 101 |
|
| 101 | 102 |
def obj(path): |
Also available in: Unified diff
Refactored xpath.parse() to use a nested function instead of a class extending Parser