1 |
55
|
aaronmk
|
# XPath parsing
|
2 |
21
|
aaronmk
|
|
3 |
55
|
aaronmk
|
import copy
|
4 |
21
|
aaronmk
|
|
5 |
|
|
from Parser import Parser
|
6 |
|
|
|
7 |
|
|
class XpathElem:
|
8 |
|
|
def __init__(self, name, value=None, attrs=None, is_attr=False,
|
9 |
|
|
is_ptr=False):
|
10 |
|
|
if attrs == None: attrs = []
|
11 |
|
|
self.name = name
|
12 |
|
|
self.value = value
|
13 |
|
|
self.attrs = attrs
|
14 |
|
|
self.is_attr = is_attr
|
15 |
|
|
self.is_ptr = is_ptr
|
16 |
|
|
|
17 |
|
|
def __repr__(self):
|
18 |
|
|
str_ = ''
|
19 |
|
|
if self.is_attr: str_ += '@'
|
20 |
25
|
aaronmk
|
str_ += self.name
|
21 |
|
|
if self.attrs != []: str_ += repr(self.attrs)
|
22 |
|
|
if self.value != None: str_ += '='+repr(self.value)
|
23 |
21
|
aaronmk
|
if self.is_ptr: str_ += '->'
|
24 |
|
|
return str_
|
25 |
|
|
|
26 |
|
|
def __eq__(self, other): return self.__dict__ == other.__dict__
|
27 |
|
|
|
28 |
24
|
aaronmk
|
def value(path): return path[-1].value
|
29 |
|
|
|
30 |
22
|
aaronmk
|
def set_value(path, value): path[-1].value = value
|
31 |
|
|
|
32 |
32
|
aaronmk
|
def backward_id(elem):
|
33 |
59
|
aaronmk
|
if len(elem.attrs) >= 1 and value(elem.attrs[0]) == None and\
|
34 |
|
|
len(elem.attrs[0]) == 1: return elem.attrs[0]
|
35 |
32
|
aaronmk
|
else: return None
|
36 |
|
|
|
37 |
21
|
aaronmk
|
class XpathParser(Parser):
|
38 |
|
|
def _main(self):
|
39 |
36
|
aaronmk
|
self._match_str('/') # optional leading /
|
40 |
21
|
aaronmk
|
return self._path()
|
41 |
|
|
|
42 |
|
|
def _path(self):
|
43 |
|
|
tree = []
|
44 |
38
|
aaronmk
|
trailing_slash = False
|
45 |
21
|
aaronmk
|
while True:
|
46 |
38
|
aaronmk
|
# Split path
|
47 |
|
|
if self._match_str('{'):
|
48 |
|
|
paths = []
|
49 |
|
|
while True:
|
50 |
|
|
paths.append(tree + self._path())
|
51 |
|
|
if not self._match_str(','): break
|
52 |
|
|
self._match_str('}', required=True)
|
53 |
|
|
tree = paths[0] # just use first subpath for now
|
54 |
|
|
break # nothing allowed after split path
|
55 |
36
|
aaronmk
|
|
56 |
38
|
aaronmk
|
elem = XpathElem(is_attr=self._match_str('@'),
|
57 |
|
|
name=self._match_re(r'[\w.*]+', required=True))
|
58 |
|
|
|
59 |
36
|
aaronmk
|
# Attrs
|
60 |
21
|
aaronmk
|
if self._match_str('['):
|
61 |
36
|
aaronmk
|
elem.attrs = []
|
62 |
|
|
while True:
|
63 |
|
|
path = self._path()
|
64 |
38
|
aaronmk
|
if self._match_str('='):
|
65 |
|
|
set_value(path, self._match_re(r'[\w.|]*'))
|
66 |
36
|
aaronmk
|
elem.attrs.append(path)
|
67 |
|
|
if not self._match_str(','): break
|
68 |
21
|
aaronmk
|
self._match_str(']', required=True)
|
69 |
36
|
aaronmk
|
|
70 |
21
|
aaronmk
|
elem.is_ptr = self._match_str('->')
|
71 |
|
|
tree.append(elem)
|
72 |
36
|
aaronmk
|
|
73 |
|
|
# 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)
|
78 |
|
|
elem.attrs.append(path)
|
79 |
|
|
tree += path
|
80 |
|
|
|
81 |
21
|
aaronmk
|
if not self._match_str('/'): break
|
82 |
32
|
aaronmk
|
|
83 |
|
|
# Expand * abbrs
|
84 |
|
|
elem_idx = 0
|
85 |
|
|
for elem in tree:
|
86 |
|
|
id_ = backward_id(elem)
|
87 |
|
|
if id_ != None: elem = id_[0]; offset = -2
|
88 |
|
|
elif elem.is_ptr: offset = 2
|
89 |
|
|
else: offset = 1
|
90 |
|
|
before, abbr, after = elem.name.partition('*')
|
91 |
|
|
if abbr != '':
|
92 |
|
|
try: elem.name = before+tree[elem_idx+offset].name+after
|
93 |
|
|
except IndexError: pass # no replacement elem
|
94 |
|
|
elem_idx += 1
|
95 |
|
|
|
96 |
21
|
aaronmk
|
return tree
|
97 |
|
|
|
98 |
57
|
aaronmk
|
def parse(string): return XpathParser(string).parse()
|
99 |
|
|
|
100 |
26
|
aaronmk
|
instance_level = 1
|
101 |
22
|
aaronmk
|
|
102 |
26
|
aaronmk
|
def obj(path):
|
103 |
55
|
aaronmk
|
obj_path = copy.deepcopy(path[:instance_level+1])
|
104 |
26
|
aaronmk
|
obj_path[-1].is_ptr = False # prevent pointer w/o target
|
105 |
|
|
return obj_path
|
106 |
|
|
|
107 |
22
|
aaronmk
|
def set_id(path, id_, has_types=True):
|
108 |
26
|
aaronmk
|
if has_types: id_level = instance_level
|
109 |
21
|
aaronmk
|
else: id_level = 0
|
110 |
|
|
path[id_level].attrs.append([XpathElem('id', id_, is_attr=True)])
|
111 |
62
|
aaronmk
|
|
112 |
|
|
def is_id(path): return path[0].is_attr and path[0].name == 'id'
|
113 |
|
|
|
114 |
|
|
def is_instance(elem): return elem.attrs != [] and is_id(elem.attrs[0])
|