| 1 |
21
|
aaronmk
|
# XPath-based XML tree manipulation
|
| 2 |
|
|
|
| 3 |
22
|
aaronmk
|
from copy import deepcopy
|
| 4 |
21
|
aaronmk
|
from xml.dom import Node
|
| 5 |
|
|
|
| 6 |
|
|
from Parser import Parser
|
| 7 |
|
|
import xml_util
|
| 8 |
|
|
|
| 9 |
|
|
class XpathElem:
|
| 10 |
|
|
def __init__(self, name, value=None, attrs=None, is_attr=False,
|
| 11 |
|
|
is_ptr=False):
|
| 12 |
|
|
if attrs == None: attrs = []
|
| 13 |
|
|
self.name = name
|
| 14 |
|
|
self.value = value
|
| 15 |
|
|
self.attrs = attrs
|
| 16 |
|
|
self.is_attr = is_attr
|
| 17 |
|
|
self.is_ptr = is_ptr
|
| 18 |
|
|
|
| 19 |
|
|
def __repr__(self):
|
| 20 |
|
|
str_ = ''
|
| 21 |
|
|
if self.is_attr: str_ += '@'
|
| 22 |
25
|
aaronmk
|
str_ += self.name
|
| 23 |
|
|
if self.attrs != []: str_ += repr(self.attrs)
|
| 24 |
|
|
if self.value != None: str_ += '='+repr(self.value)
|
| 25 |
21
|
aaronmk
|
if self.is_ptr: str_ += '->'
|
| 26 |
|
|
return str_
|
| 27 |
|
|
|
| 28 |
|
|
def __eq__(self, other): return self.__dict__ == other.__dict__
|
| 29 |
|
|
|
| 30 |
24
|
aaronmk
|
def value(path): return path[-1].value
|
| 31 |
|
|
|
| 32 |
22
|
aaronmk
|
def set_value(path, value): path[-1].value = value
|
| 33 |
|
|
|
| 34 |
32
|
aaronmk
|
def backward_id(elem):
|
| 35 |
|
|
if len(elem.attrs) >= 1 and value(elem.attrs[0]) == None:
|
| 36 |
|
|
return elem.attrs[0]
|
| 37 |
|
|
else: return None
|
| 38 |
|
|
|
| 39 |
21
|
aaronmk
|
class XpathParser(Parser):
|
| 40 |
|
|
def _main(self):
|
| 41 |
|
|
self._match_str('/', required=True)
|
| 42 |
|
|
return self._path()
|
| 43 |
|
|
|
| 44 |
|
|
def _path(self):
|
| 45 |
|
|
tree = []
|
| 46 |
25
|
aaronmk
|
fork_idx = None
|
| 47 |
|
|
elem_idx = 0
|
| 48 |
21
|
aaronmk
|
while True:
|
| 49 |
|
|
elem = XpathElem(is_attr=self._match_str('@'), name=self._fields())
|
| 50 |
|
|
if self._match_str('['):
|
| 51 |
|
|
elem.attrs = self._attrs()
|
| 52 |
|
|
self._match_str(']', required=True)
|
| 53 |
|
|
elem.is_ptr = self._match_str('->')
|
| 54 |
25
|
aaronmk
|
if not elem.is_ptr and self._match_str('!'): fork_idx = elem_idx
|
| 55 |
21
|
aaronmk
|
tree.append(elem)
|
| 56 |
|
|
if not self._match_str('/'): break
|
| 57 |
25
|
aaronmk
|
elem_idx += 1
|
| 58 |
32
|
aaronmk
|
|
| 59 |
|
|
# Expand * abbrs
|
| 60 |
|
|
elem_idx = 0
|
| 61 |
|
|
for elem in tree:
|
| 62 |
|
|
id_ = backward_id(elem)
|
| 63 |
|
|
if id_ != None: elem = id_[0]; offset = -2
|
| 64 |
|
|
elif elem.is_ptr: offset = 2
|
| 65 |
|
|
else: offset = 1
|
| 66 |
|
|
before, abbr, after = elem.name.partition('*')
|
| 67 |
|
|
if abbr != '':
|
| 68 |
|
|
try: elem.name = before+tree[elem_idx+offset].name+after
|
| 69 |
|
|
except IndexError: pass # no replacement elem
|
| 70 |
|
|
elem_idx += 1
|
| 71 |
|
|
|
| 72 |
25
|
aaronmk
|
# Add lookahead assertion for rest of path
|
| 73 |
|
|
if fork_idx != None: tree[fork_idx].attrs.append(tree[fork_idx+1:])
|
| 74 |
32
|
aaronmk
|
|
| 75 |
21
|
aaronmk
|
return tree
|
| 76 |
|
|
|
| 77 |
|
|
def _fields(self):
|
| 78 |
|
|
if self._match_str('{'):
|
| 79 |
|
|
tree = []
|
| 80 |
|
|
while True:
|
| 81 |
|
|
tree.append(self._field())
|
| 82 |
|
|
if not self._match_str(','): break
|
| 83 |
|
|
self._match_str('}', required=True)
|
| 84 |
|
|
tree = tuple(tree)
|
| 85 |
|
|
tree = tree[0] # just use first field for now
|
| 86 |
|
|
else: tree = self._field()
|
| 87 |
|
|
return tree
|
| 88 |
|
|
|
| 89 |
|
|
def _attrs(self):
|
| 90 |
|
|
tree = []
|
| 91 |
|
|
while True:
|
| 92 |
|
|
path = self._path()
|
| 93 |
22
|
aaronmk
|
if self._match_str('='): set_value(path, self._value())
|
| 94 |
21
|
aaronmk
|
tree.append(path)
|
| 95 |
|
|
if not self._match_str(','): break
|
| 96 |
|
|
return tree
|
| 97 |
|
|
|
| 98 |
|
|
def _field(self):
|
| 99 |
|
|
return self._name()
|
| 100 |
|
|
|
| 101 |
32
|
aaronmk
|
def _name(self): return self._match_re(r'[\w.*]+', required=True)
|
| 102 |
21
|
aaronmk
|
|
| 103 |
|
|
def _value(self): return self._match_re(r'[\w.|]+', required=True)
|
| 104 |
|
|
|
| 105 |
26
|
aaronmk
|
instance_level = 1
|
| 106 |
22
|
aaronmk
|
|
| 107 |
26
|
aaronmk
|
def obj(path):
|
| 108 |
|
|
obj_path = deepcopy(path[:instance_level+1])
|
| 109 |
|
|
obj_path[-1].is_ptr = False # prevent pointer w/o target
|
| 110 |
|
|
return obj_path
|
| 111 |
|
|
|
| 112 |
22
|
aaronmk
|
def set_id(path, id_, has_types=True):
|
| 113 |
26
|
aaronmk
|
if has_types: id_level = instance_level
|
| 114 |
21
|
aaronmk
|
else: id_level = 0
|
| 115 |
|
|
path[id_level].attrs.append([XpathElem('id', id_, is_attr=True)])
|
| 116 |
|
|
|
| 117 |
22
|
aaronmk
|
def get(doc, path, create=False, last_only=None, parent=None):
|
| 118 |
32
|
aaronmk
|
# Warning: The last_only optimization may put data that should be together
|
| 119 |
|
|
# into separate nodes
|
| 120 |
21
|
aaronmk
|
if parent == None: parent = doc.documentElement
|
| 121 |
22
|
aaronmk
|
if last_only == None: last_only = create
|
| 122 |
|
|
elem_idx = 0
|
| 123 |
21
|
aaronmk
|
for elem in path:
|
| 124 |
22
|
aaronmk
|
# Find possible matches
|
| 125 |
|
|
children = []
|
| 126 |
|
|
if elem.is_attr:
|
| 127 |
|
|
child = parent.getAttributeNode(elem.name)
|
| 128 |
|
|
if child != None: children = [child]
|
| 129 |
|
|
elif elem.name == '.': children = [parent]
|
| 130 |
|
|
else: children = xml_util.by_tag_name(parent, elem.name, last_only)
|
| 131 |
|
|
|
| 132 |
|
|
# Check each match
|
| 133 |
21
|
aaronmk
|
node = None
|
| 134 |
22
|
aaronmk
|
for child in children:
|
| 135 |
|
|
is_match = elem.value == None or xml_util.value(child) == elem.value
|
| 136 |
|
|
for attr in elem.attrs:
|
| 137 |
|
|
if not is_match: break
|
| 138 |
|
|
is_match = get(doc, attr, False, last_only, child) != None
|
| 139 |
|
|
if is_match: node = child; break
|
| 140 |
|
|
|
| 141 |
|
|
# Create node
|
| 142 |
21
|
aaronmk
|
if node == None:
|
| 143 |
|
|
if not create: return None
|
| 144 |
|
|
if elem.is_attr:
|
| 145 |
|
|
parent.setAttribute(elem.name, '')
|
| 146 |
|
|
node = parent.getAttributeNode(elem.name)
|
| 147 |
|
|
else: node = parent.appendChild(doc.createElement(elem.name))
|
| 148 |
22
|
aaronmk
|
if elem.value != None: xml_util.set_value(doc, node, elem.value)
|
| 149 |
|
|
for attr in elem.attrs: get(doc, attr, create, last_only, node)
|
| 150 |
|
|
|
| 151 |
|
|
# Follow pointer
|
| 152 |
21
|
aaronmk
|
if elem.is_ptr:
|
| 153 |
25
|
aaronmk
|
path = deepcopy(path[elem_idx+1:]) # rest of path
|
| 154 |
32
|
aaronmk
|
id_elem = backward_id(path[instance_level])
|
| 155 |
|
|
if id_elem != None:
|
| 156 |
22
|
aaronmk
|
# backward (child-to-parent) pointer with target ID attr
|
| 157 |
32
|
aaronmk
|
set_value(id_elem, xml_util.get_id(node))
|
| 158 |
22
|
aaronmk
|
else: # forward (parent-to-child) pointer
|
| 159 |
|
|
id_ = xml_util.value(node)
|
| 160 |
26
|
aaronmk
|
obj_path = obj(path) # target object
|
| 161 |
25
|
aaronmk
|
if id_ == None or get(doc, obj_path, False, True) == None:
|
| 162 |
|
|
# no target or target attrs don't match
|
| 163 |
22
|
aaronmk
|
if not create: return None
|
| 164 |
25
|
aaronmk
|
|
| 165 |
22
|
aaronmk
|
# Use last target object's ID + 1
|
| 166 |
26
|
aaronmk
|
obj_path[-1].attrs = [] # just get by tag name
|
| 167 |
|
|
last = get(doc, obj_path, False, True)
|
| 168 |
25
|
aaronmk
|
if last != None: id_ = str(int(xml_util.get_id(last)) + 1)
|
| 169 |
22
|
aaronmk
|
else: id_ = '0'
|
| 170 |
25
|
aaronmk
|
|
| 171 |
|
|
# Will append if target attrs didn't match. Place ! in XPath
|
| 172 |
|
|
# after element to fork at to avoid this.
|
| 173 |
22
|
aaronmk
|
xml_util.set_value(doc, node, id_)
|
| 174 |
23
|
aaronmk
|
else: last_only = False
|
| 175 |
25
|
aaronmk
|
set_id(path, id_)
|
| 176 |
|
|
return get(doc, path, create, last_only)
|
| 177 |
22
|
aaronmk
|
|
| 178 |
21
|
aaronmk
|
parent = node
|
| 179 |
22
|
aaronmk
|
elem_idx += 1
|
| 180 |
21
|
aaronmk
|
return parent
|
