Project

General

Profile

1 55 aaronmk
# XPath parsing
2 21 aaronmk
3 55 aaronmk
import copy
4 1009 aaronmk
import warnings
5 21 aaronmk
6
from Parser import Parser
7 892 aaronmk
import util
8 77 aaronmk
import xml_dom
9 21 aaronmk
10 1001 aaronmk
##### Path elements
11
12 21 aaronmk
class XpathElem:
13 1002 aaronmk
    def __init__(self, name='', value=None, is_attr=False):
14 21 aaronmk
        self.name = name
15
        self.value = value
16
        self.is_attr = is_attr
17 97 aaronmk
        self.is_positive = True
18 1238 aaronmk
        self.is_lookup_only = False
19 94 aaronmk
        self.is_ptr = False
20
        self.keys = []
21
        self.attrs = []
22 78 aaronmk
        self.other_branches = [] # temp implementation for split paths
23 21 aaronmk
24
    def __repr__(self):
25
        str_ = ''
26 141 aaronmk
        if not self.is_positive: str_ += '!'
27 21 aaronmk
        if self.is_attr: str_ += '@'
28 1004 aaronmk
        if self.name == '': str_ += '""'
29
        else: str_ += self.name
30 1238 aaronmk
        if self.is_lookup_only: str_ += '?'
31 94 aaronmk
        if self.keys != []: str_ += repr(self.keys)
32
        if self.attrs != []: str_ += ':'+repr(self.attrs)
33
        if self.is_ptr: str_ += '->'
34
        if self.other_branches != []: str_ += '{'+repr(self.other_branches)+'}'
35 25 aaronmk
        if self.value != None: str_ += '='+repr(self.value)
36 21 aaronmk
        return str_
37
38
    def __eq__(self, other): return self.__dict__ == other.__dict__
39
40 1002 aaronmk
empty_elem = XpathElem()
41
42
def elem_is_empty(elem): return elem == empty_elem
43
44 1004 aaronmk
def is_self(elem): return elem.name == '' or elem.name == '.'
45
46
def is_parent(elem): return elem.name == '..'
47
48 1001 aaronmk
##### Paths
49
50 99 aaronmk
def is_positive(path): return path[0].is_positive
51
52 1003 aaronmk
def is_rooted(path): return elem_is_empty(path[0])
53
54 24 aaronmk
def value(path): return path[-1].value
55
56 1294 aaronmk
def set_value(path, value):
57
    '''Caller must make a shallow copy of the path to prevent modifications from
58
    propagating to other copies of the path (a deep copy is not needed)'''
59
    path[-1] = copy.copy(path[-1]) # don't modify other copies of the path
60
    path[-1].value = value
61 22 aaronmk
62 32 aaronmk
def backward_id(elem):
63 171 aaronmk
    if len(elem.keys) >= 1 and value(elem.keys[0]) == None: return elem.keys[0]
64 32 aaronmk
    else: return None
65
66 1001 aaronmk
instance_level = 1 # root's grandchildren
67
68
def obj(path):
69
    obj_path = copy.deepcopy(path[:instance_level+1])
70
    obj_path[-1].is_ptr = False # prevent pointer w/o target
71
    return obj_path
72
73
def set_id(path, id_, has_types=True):
74 1294 aaronmk
    '''Caller must make a shallow copy of the path to prevent modifications from
75
    propagating to other copies of the path (a deep copy is not needed)'''
76 1001 aaronmk
    if has_types: id_level = instance_level
77
    else: id_level = 0 # root's children
78 1004 aaronmk
    if is_self(path[0]): id_level += 1 # explicit root element
79 1294 aaronmk
80
    id_elem = path[id_level] = copy.copy(path[id_level])
81
        # don't modify other copies of the path
82
    id_elem.keys = id_elem.keys[:] # don't modify other copies of the elem
83
    id_elem.keys.append([XpathElem('id', id_, True)])
84 1001 aaronmk
85
def is_id(path): return path[0].is_attr and path[0].name == 'id'
86
87 1070 aaronmk
def is_instance(elem): return elem.keys != [] and is_id(elem.keys[0])
88 1001 aaronmk
89
##### Parsing
90
91 170 aaronmk
def expand_abbr(name, repl):
92
    before, abbr, after = name.partition('*')
93
    if abbr != '': name = before+repl+after
94
    return name
95
96 154 aaronmk
_cache = {}
97
98 76 aaronmk
def parse(str_):
99 154 aaronmk
    try: return _cache[str_]
100
    except KeyError: pass
101
102 76 aaronmk
    parser = Parser(str_)
103 21 aaronmk
104 76 aaronmk
    def _path():
105 97 aaronmk
        is_positive = not parser.str_('!')
106
107 21 aaronmk
        tree = []
108
        while True:
109 38 aaronmk
            # Split path
110 76 aaronmk
            if parser.str_('{'):
111 1070 aaronmk
                last = tree[-1]
112
                last.other_branches = _paths()
113 76 aaronmk
                parser.str_('}', required=True)
114 1070 aaronmk
                if parser.str_('/'): # common subpath after {}
115
                    subpath = _path()
116
                    for branch in last.other_branches: branch += subpath
117
                tree += last.other_branches.pop(0) # use first path for now
118
                break # anything after split path has already been parsed
119 36 aaronmk
120 754 aaronmk
            elem = XpathElem(is_attr=parser.str_('@'), name=_value())
121 38 aaronmk
122 1238 aaronmk
            elem.is_lookup_only = parser.str_('?')
123
124 94 aaronmk
            # Keys used to match nodes
125 76 aaronmk
            if parser.str_('['):
126 94 aaronmk
                elem.keys = _paths()
127 76 aaronmk
                parser.str_(']', required=True)
128 36 aaronmk
129 94 aaronmk
            # Attrs created when no matching node exists
130
            if parser.str_(':'):
131
                parser.str_('[', required=True)
132
                elem.attrs = _paths()
133
                parser.str_(']', required=True)
134
135 76 aaronmk
            elem.is_ptr = parser.str_('->')
136 21 aaronmk
            tree.append(elem)
137 36 aaronmk
138
            # Lookahead assertion
139 76 aaronmk
            if parser.str_('('):
140
                parser.str_('/', required=True) # next / is inside ()
141
                path = _path()
142
                parser.str_(')', required=True)
143 94 aaronmk
                elem.keys.append(path)
144 36 aaronmk
                tree += path
145
146 76 aaronmk
            if not parser.str_('/'): break
147 32 aaronmk
148 97 aaronmk
        tree[0].is_positive = is_positive
149
150
        # Value
151 997 aaronmk
        if parser.str_('='):
152 1009 aaronmk
            if parser.str_('$'): # reference (different from a pointer)
153
                value = _path()
154
            else: value = _value() # literal value
155 997 aaronmk
            set_value(tree, value)
156 85 aaronmk
157 32 aaronmk
        # Expand * abbrs
158 70 aaronmk
        for i in reversed(xrange(len(tree))):
159
            elem = tree[i]
160 171 aaronmk
            if elem.is_ptr: offset = 2
161 32 aaronmk
            else: offset = 1
162 170 aaronmk
            try: repl = tree[i+offset].name
163 171 aaronmk
            except IndexError: pass # no replacement elem
164
            else: elem.name = expand_abbr(elem.name, repl)
165 32 aaronmk
166 21 aaronmk
        return tree
167 76 aaronmk
168 754 aaronmk
    def _value():
169
        if parser.str_('"'):
170
            value = parser.re(r'[^"]*')
171
            parser.str_('"', required=True)
172
        else: value = parser.re(r'(?:(?:\w+:)*[\w.*]+)?')
173
        return value
174
175 94 aaronmk
    def _paths():
176
        paths = []
177
        while True:
178
            paths.append(_path())
179
            if not parser.str_(','): break
180
        return paths
181
182 76 aaronmk
    path = _path()
183
    parser.end()
184 154 aaronmk
    _cache[str_] = path
185 76 aaronmk
    return path
186 21 aaronmk
187 1001 aaronmk
##### Querying/creating XML trees
188 22 aaronmk
189 1005 aaronmk
def get(root, xpath, create=False, last_only=None, limit=1, allow_rooted=True):
190 141 aaronmk
    '''Warning: The last_only optimization may put data that should be together
191
    into separate nodes'''
192 77 aaronmk
    if last_only == None: last_only = create
193 887 aaronmk
    if limit == None or limit > 1: last_only = False
194 892 aaronmk
    if util.is_str(xpath): xpath = parse(xpath)
195 99 aaronmk
196 1005 aaronmk
    # Handle edge cases
197 998 aaronmk
    if xpath == []: return [root]
198 886 aaronmk
    if create and not is_positive(xpath): return []
199 1005 aaronmk
200
    # Define vars
201 998 aaronmk
    doc = root.ownerDocument
202 1005 aaronmk
    if allow_rooted and is_rooted(xpath): root = doc.documentElement
203 165 aaronmk
    elem = xpath[0]
204
205
    # Find possible matches
206
    children = []
207
    if elem.is_attr:
208 998 aaronmk
        child = root.getAttributeNode(elem.name)
209 165 aaronmk
        if child != None: children = [child]
210 1004 aaronmk
    elif is_self(elem): children = [root]
211
    elif is_parent(elem):
212 1000 aaronmk
        parent = xml_dom.parent(root)
213 1013 aaronmk
        if parent == None: return [] # don't try to create doc root's parent
214
        root = parent
215
        children = [root]
216 165 aaronmk
    else:
217 998 aaronmk
        children = xml_dom.by_tag_name(root, elem.name,
218 1013 aaronmk
            last_only and (elem.keys == [] or is_instance(elem)))
219 165 aaronmk
220 1009 aaronmk
    # Retrieve elem value
221
    value_ = elem.value
222
    if util.is_list(value_): # reference (different from a pointer)
223
        targets = get(root, value_)
224
        try: target = targets[0]
225
        except IndexError:
226
            warnings.warn(UserWarning('XPath reference target missing: '
227
                +str(value_)+'\nXPath: '+str(xpath)))
228
            value_ = None
229
        else: value_ = xml_dom.value(target)
230
231 165 aaronmk
    # Check each match
232 885 aaronmk
    nodes = []
233 165 aaronmk
    for child in children:
234 1009 aaronmk
        is_match = value_ == None or xml_dom.value(child) == value_
235 165 aaronmk
        for attr in elem.keys:
236
            if not is_match: break
237 1011 aaronmk
            is_match = ((get(child, attr, False, last_only,
238
                allow_rooted=False) != []) == is_positive(attr))
239 885 aaronmk
        if is_match:
240
            nodes.append(child)
241
            if limit != None and len(nodes) >= limit: break
242 165 aaronmk
243
    # Create node
244 886 aaronmk
    if nodes == []:
245 1239 aaronmk
        if not create or elem.is_lookup_only: return []
246 77 aaronmk
        if elem.is_attr:
247 998 aaronmk
            root.setAttribute(elem.name, '')
248
            node = root.getAttributeNode(elem.name)
249 1013 aaronmk
        elif util.list_eq_is(children, [root]): node = root
250 998 aaronmk
        else: node = root.appendChild(doc.createElement(elem.name))
251 1009 aaronmk
        if value_ != None: xml_dom.set_value(node, value_)
252 886 aaronmk
        nodes.append(node)
253 165 aaronmk
254 1009 aaronmk
    path_value = value(xpath)
255 165 aaronmk
    xpath = xpath[1:] # rest of XPath
256
257 886 aaronmk
    next = []
258
    for node in nodes:
259
        # Create attrs
260
        if create:
261
            for attr in elem.keys + elem.attrs:
262 1011 aaronmk
                get(node, attr, create, last_only, allow_rooted=False)
263 886 aaronmk
264
        # Follow pointer
265
        if elem.is_ptr:
266 998 aaronmk
            root = doc.documentElement
267 886 aaronmk
            xpath = copy.deepcopy(xpath)
268
            id_path = backward_id(xpath[instance_level])
269
            if id_path != None: # backward (child-to-parent) pointer with ID key
270
                id_path[0].name = expand_abbr(id_path[0].name, node.tagName)
271
                set_value(id_path, xml_dom.get_id(node)) # modify xpath's id key
272
            else: # forward (parent-to-child) pointer
273
                id_ = xml_dom.value(node)
274
                obj_xpath = obj(xpath) # target object
275
                if id_ == None or get(root, obj_xpath, False, True) == []:
276
                    # no target or target keys don't match
277
                    if not create: continue
278
279
                    # Use last target object's ID + 1
280
                    obj_xpath[-1].keys = [] # just get by tag name
281
                    last = get(root, obj_xpath, False, True)
282
                    if last != []: id_ = str(int(xml_dom.get_id(last[0])) + 1)
283
                    else: id_ = '0'
284
285
                    # Will append if target keys didn't match.
286
                    # Use lookahead assertion to avoid this.
287
                    xml_dom.set_value(node, id_)
288
                else: last_only = False
289
                set_id(xpath, id_)
290
        else: root = node
291 1005 aaronmk
        next += get(root, xpath, create, last_only, limit, allow_rooted=False)
292 886 aaronmk
293
        for branch in elem.other_branches:
294 1294 aaronmk
            branch = branch[:] # don't modify input!
295 1009 aaronmk
            set_value(branch, path_value)
296 1005 aaronmk
            next += get(node, branch, create, last_only, limit,
297
                allow_rooted=False)
298 165 aaronmk
299
    return next
300 82 aaronmk
301 141 aaronmk
def put_obj(root, xpath, id_, has_types, value=None):
302 892 aaronmk
    if util.is_str(xpath): xpath = parse(xpath)
303 155 aaronmk
304 1294 aaronmk
    xpath = xpath[:] # don't modify input!
305 82 aaronmk
    set_id(xpath, id_, has_types)
306 85 aaronmk
    if value != None: set_value(xpath, value)
307 141 aaronmk
    get(root, xpath, True)
308 133 aaronmk
309 135 aaronmk
def path2xml(xpath, first_branch=True):
310
    root = xml_dom.create_doc().documentElement
311 141 aaronmk
    get(root, xpath, True)
312 135 aaronmk
    return root
313 133 aaronmk
314 141 aaronmk
def xml_set_id(root, id_): xml_dom.set_id(root.firstChild, id_)