| 1 |
21
|
aaronmk
|
# XPath-based XML tree manipulation
|
| 2 |
|
|
|
| 3 |
|
|
from xml.dom import Node
|
| 4 |
|
|
|
| 5 |
|
|
from Parser import Parser
|
| 6 |
|
|
import xml_util
|
| 7 |
|
|
|
| 8 |
|
|
class XpathElem:
|
| 9 |
|
|
def __init__(self, name, value=None, attrs=None, is_attr=False,
|
| 10 |
|
|
is_ptr=False):
|
| 11 |
|
|
if attrs == None: attrs = []
|
| 12 |
|
|
self.name = name
|
| 13 |
|
|
self.value = value
|
| 14 |
|
|
self.attrs = attrs
|
| 15 |
|
|
self.is_attr = is_attr
|
| 16 |
|
|
self.is_ptr = is_ptr
|
| 17 |
|
|
|
| 18 |
|
|
def __repr__(self):
|
| 19 |
|
|
str_ = ''
|
| 20 |
|
|
if self.is_attr: str_ += '@'
|
| 21 |
|
|
str_ += self.name+repr(self.attrs)+'='+repr(self.value)
|
| 22 |
|
|
if self.is_ptr: str_ += '->'
|
| 23 |
|
|
return str_
|
| 24 |
|
|
|
| 25 |
|
|
def __eq__(self, other): return self.__dict__ == other.__dict__
|
| 26 |
|
|
|
| 27 |
|
|
class XpathParser(Parser):
|
| 28 |
|
|
def _main(self):
|
| 29 |
|
|
self._match_str('/', required=True)
|
| 30 |
|
|
return self._path()
|
| 31 |
|
|
|
| 32 |
|
|
def _path(self):
|
| 33 |
|
|
tree = []
|
| 34 |
|
|
while True:
|
| 35 |
|
|
elem = XpathElem(is_attr=self._match_str('@'), name=self._fields())
|
| 36 |
|
|
if self._match_str('['):
|
| 37 |
|
|
elem.attrs = self._attrs()
|
| 38 |
|
|
self._match_str(']', required=True)
|
| 39 |
|
|
elem.is_ptr = self._match_str('->')
|
| 40 |
|
|
tree.append(elem)
|
| 41 |
|
|
if not self._match_str('/'): break
|
| 42 |
|
|
return tree
|
| 43 |
|
|
|
| 44 |
|
|
def _fields(self):
|
| 45 |
|
|
if self._match_str('{'):
|
| 46 |
|
|
tree = []
|
| 47 |
|
|
while True:
|
| 48 |
|
|
tree.append(self._field())
|
| 49 |
|
|
if not self._match_str(','): break
|
| 50 |
|
|
self._match_str('}', required=True)
|
| 51 |
|
|
tree = tuple(tree)
|
| 52 |
|
|
tree = tree[0] # just use first field for now
|
| 53 |
|
|
else: tree = self._field()
|
| 54 |
|
|
return tree
|
| 55 |
|
|
|
| 56 |
|
|
def _attrs(self):
|
| 57 |
|
|
tree = []
|
| 58 |
|
|
while True:
|
| 59 |
|
|
path = self._path()
|
| 60 |
|
|
self._match_str('=', required=True)
|
| 61 |
|
|
path[-1].value = self._value()
|
| 62 |
|
|
tree.append(path)
|
| 63 |
|
|
if not self._match_str(','): break
|
| 64 |
|
|
return tree
|
| 65 |
|
|
|
| 66 |
|
|
def _field(self):
|
| 67 |
|
|
return self._name()
|
| 68 |
|
|
|
| 69 |
|
|
def _name(self): return self._match_re(r'[\w.]+', required=True)
|
| 70 |
|
|
|
| 71 |
|
|
def _value(self): return self._match_re(r'[\w.|]+', required=True)
|
| 72 |
|
|
|
| 73 |
|
|
def set_id(path, id_, has_type_containers=True):
|
| 74 |
|
|
if has_type_containers: id_level = 1
|
| 75 |
|
|
else: id_level = 0
|
| 76 |
|
|
path[id_level].attrs.append([XpathElem('id', id_, is_attr=True)])
|
| 77 |
|
|
|
| 78 |
|
|
def get(doc, path, create=False, parent=None):
|
| 79 |
|
|
if parent == None: parent = doc.documentElement
|
| 80 |
|
|
for elem in path:
|
| 81 |
|
|
node = None
|
| 82 |
|
|
if elem.is_attr: node = parent.getAttributeNode(elem.name)
|
| 83 |
|
|
elif elem.name == '.': node = parent
|
| 84 |
|
|
else: node = xml_util.by_tag_name(parent, elem.name)
|
| 85 |
|
|
if node != None and elem.value != None\
|
| 86 |
|
|
and xml_util.value(node) != elem.value: node = None
|
| 87 |
|
|
for attr in elem.attrs:
|
| 88 |
|
|
if get(doc, attr, parent=node) == None: node = None; break
|
| 89 |
|
|
if node == None:
|
| 90 |
|
|
if not create: return None
|
| 91 |
|
|
if elem.is_attr:
|
| 92 |
|
|
parent.setAttribute(elem.name, '')
|
| 93 |
|
|
node = parent.getAttributeNode(elem.name)
|
| 94 |
|
|
else: node = parent.appendChild(doc.createElement(elem.name))
|
| 95 |
|
|
if elem.value != None:
|
| 96 |
|
|
if node.nodeType == Node.ELEMENT_NODE:
|
| 97 |
|
|
node.appendChild(doc.createTextNode(elem.value))
|
| 98 |
|
|
else: node.nodeValue = elem.value
|
| 99 |
|
|
for attr in elem.attrs: get(doc, attr, create, node)
|
| 100 |
|
|
if elem.is_ptr:
|
| 101 |
|
|
pass
|
| 102 |
|
|
parent = node
|
| 103 |
|
|
return parent
|
