Project

General

Profile

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 36 aaronmk
        self._match_str('/') # optional leading /
42 21 aaronmk
        return self._path()
43
44
    def _path(self):
45
        tree = []
46
        while True:
47
            elem = XpathElem(is_attr=self._match_str('@'), name=self._fields())
48 36 aaronmk
49
            # Attrs
50 21 aaronmk
            if self._match_str('['):
51 36 aaronmk
                elem.attrs = []
52
                while True:
53
                    path = self._path()
54
                    if self._match_str('='): set_value(path, self._value())
55
                    elem.attrs.append(path)
56
                    if not self._match_str(','): break
57 21 aaronmk
                self._match_str(']', required=True)
58 36 aaronmk
59 21 aaronmk
            elem.is_ptr = self._match_str('->')
60
            tree.append(elem)
61 36 aaronmk
62
            # Lookahead assertion
63
            if self._match_str('('):
64
                self._match_str('/', required=True) # next / is inside ()
65
                path = self._path()
66
                self._match_str(')', required=True)
67
                elem.attrs.append(path)
68
                tree += path
69
70 21 aaronmk
            if not self._match_str('/'): break
71 32 aaronmk
72
        # Expand * abbrs
73
        elem_idx = 0
74
        for elem in tree:
75
            id_ = backward_id(elem)
76
            if id_ != None: elem = id_[0]; offset = -2
77
            elif elem.is_ptr: offset = 2
78
            else: offset = 1
79
            before, abbr, after = elem.name.partition('*')
80
            if abbr != '':
81
                try: elem.name = before+tree[elem_idx+offset].name+after
82
                except IndexError: pass # no replacement elem
83
            elem_idx += 1
84
85 21 aaronmk
        return tree
86
87
    def _fields(self):
88
        if self._match_str('{'):
89
            tree = []
90
            while True:
91
                tree.append(self._field())
92
                if not self._match_str(','): break
93
            self._match_str('}', required=True)
94
            tree = tuple(tree)
95
            tree = tree[0] # just use first field for now
96
        else: tree = self._field()
97
        return tree
98
99
    def _field(self):
100
        return self._name()
101
102 32 aaronmk
    def _name(self): return self._match_re(r'[\w.*]+', required=True)
103 21 aaronmk
104
    def _value(self): return self._match_re(r'[\w.|]+', required=True)
105
106 26 aaronmk
instance_level = 1
107 22 aaronmk
108 26 aaronmk
def obj(path):
109
    obj_path = deepcopy(path[:instance_level+1])
110
    obj_path[-1].is_ptr = False # prevent pointer w/o target
111
    return obj_path
112
113 22 aaronmk
def set_id(path, id_, has_types=True):
114 26 aaronmk
    if has_types: id_level = instance_level
115 21 aaronmk
    else: id_level = 0
116
    path[id_level].attrs.append([XpathElem('id', id_, is_attr=True)])
117
118 22 aaronmk
def get(doc, path, create=False, last_only=None, parent=None):
119 32 aaronmk
    # Warning: The last_only optimization may put data that should be together
120
    # into separate nodes
121 21 aaronmk
    if parent == None: parent = doc.documentElement
122 22 aaronmk
    if last_only == None: last_only = create
123
    elem_idx = 0
124 21 aaronmk
    for elem in path:
125 22 aaronmk
        # Find possible matches
126
        children = []
127
        if elem.is_attr:
128
            child = parent.getAttributeNode(elem.name)
129
            if child != None: children = [child]
130
        elif elem.name == '.': children = [parent]
131
        else: children = xml_util.by_tag_name(parent, elem.name, last_only)
132
133
        # Check each match
134 21 aaronmk
        node = None
135 22 aaronmk
        for child in children:
136
            is_match = elem.value == None or xml_util.value(child) == elem.value
137
            for attr in elem.attrs:
138
                if not is_match: break
139
                is_match = get(doc, attr, False, last_only, child) != None
140
            if is_match: node = child; break
141
142
        # Create node
143 21 aaronmk
        if node == None:
144
            if not create: return None
145
            if elem.is_attr:
146
                parent.setAttribute(elem.name, '')
147
                node = parent.getAttributeNode(elem.name)
148
            else: node = parent.appendChild(doc.createElement(elem.name))
149 22 aaronmk
            if elem.value != None: xml_util.set_value(doc, node, elem.value)
150
            for attr in elem.attrs: get(doc, attr, create, last_only, node)
151
152
        # Follow pointer
153 21 aaronmk
        if elem.is_ptr:
154 25 aaronmk
            path = deepcopy(path[elem_idx+1:]) # rest of path
155 32 aaronmk
            id_elem = backward_id(path[instance_level])
156
            if id_elem != None:
157 22 aaronmk
                # backward (child-to-parent) pointer with target ID attr
158 32 aaronmk
                set_value(id_elem, xml_util.get_id(node))
159 22 aaronmk
            else: # forward (parent-to-child) pointer
160
                id_ = xml_util.value(node)
161 26 aaronmk
                obj_path = obj(path) # target object
162 25 aaronmk
                if id_ == None or get(doc, obj_path, False, True) == None:
163
                    # no target or target attrs don't match
164 22 aaronmk
                    if not create: return None
165 25 aaronmk
166 22 aaronmk
                    # Use last target object's ID + 1
167 26 aaronmk
                    obj_path[-1].attrs = [] # just get by tag name
168
                    last = get(doc, obj_path, False, True)
169 25 aaronmk
                    if last != None: id_ = str(int(xml_util.get_id(last)) + 1)
170 22 aaronmk
                    else: id_ = '0'
171 25 aaronmk
172
                    # Will append if target attrs didn't match. Place ! in XPath
173
                    # after element to fork at to avoid this.
174 22 aaronmk
                    xml_util.set_value(doc, node, id_)
175 23 aaronmk
                else: last_only = False
176 25 aaronmk
                set_id(path, id_)
177
            return get(doc, path, create, last_only)
178 22 aaronmk
179 21 aaronmk
        parent = node
180 22 aaronmk
        elem_idx += 1
181 21 aaronmk
    return parent