Project

General

Profile

1
# XPath parsing
2

    
3
import copy
4
import warnings
5

    
6
from Parser import Parser
7
import util
8
import xml_dom
9

    
10
##### Path elements
11

    
12
class XpathElem:
13
    def __init__(self, name='', value=None, is_attr=False):
14
        self.name = name
15
        self.value = value
16
        self.is_attr = is_attr
17
        self.is_positive = True
18
        self.is_lookup_only = False
19
        self.is_ptr = False
20
        self.keys = []
21
        self.attrs = []
22
        self.other_branches = [] # temp implementation for split paths
23
    
24
    def __repr__(self):
25
        str_ = ''
26
        if not self.is_positive: str_ += '!'
27
        if self.is_attr: str_ += '@'
28
        if self.name == '': str_ += '""'
29
        else: str_ += self.name
30
        if self.is_lookup_only: str_ += '?'
31
        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
        if self.value != None: str_ += '='+repr(self.value)
36
        return str_
37
    
38
    def __eq__(self, other): return self.__dict__ == other.__dict__
39

    
40
empty_elem = XpathElem()
41

    
42
def elem_is_empty(elem): return elem == empty_elem
43

    
44
def is_self(elem): return elem.name == '' or elem.name == '.'
45

    
46
def is_parent(elem): return elem.name == '..'
47

    
48
##### Paths
49

    
50
def is_positive(path): return path[0].is_positive
51

    
52
def is_rooted(path): return elem_is_empty(path[0])
53

    
54
def value(path): return path[-1].value
55

    
56
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

    
62
def backward_id(elem):
63
    if len(elem.keys) >= 1 and value(elem.keys[0]) == None: return elem.keys[0]
64
    else: return None
65

    
66
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
    '''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
    if has_types: id_level = instance_level
77
    else: id_level = 0 # root's children
78
    if is_self(path[0]): id_level += 1 # explicit root element
79
    
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

    
85
def is_id(path): return path[0].is_attr and path[0].name == 'id'
86

    
87
def is_instance(elem): return elem.keys != [] and is_id(elem.keys[0])
88

    
89
##### Parsing
90

    
91
def expand_abbr(name, repl):
92
    before, abbr, after = name.partition('*')
93
    if abbr != '': name = before+repl+after
94
    return name
95

    
96
_cache = {}
97

    
98
def parse(str_):
99
    try: return _cache[str_]
100
    except KeyError: pass
101
    
102
    parser = Parser(str_)
103
    
104
    def _path():
105
        is_positive = not parser.str_('!')
106
        
107
        tree = []
108
        while True:
109
            # Split path
110
            if parser.str_('{'):
111
                last = tree[-1]
112
                last.other_branches = _paths()
113
                parser.str_('}', required=True)
114
                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
            
120
            elem = XpathElem(is_attr=parser.str_('@'), name=_value())
121
            
122
            elem.is_lookup_only = parser.str_('?')
123
            
124
            # Keys used to match nodes
125
            if parser.str_('['):
126
                elem.keys = _paths()
127
                parser.str_(']', required=True)
128
            
129
            # 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
            elem.is_ptr = parser.str_('->')
136
            tree.append(elem)
137
            
138
            # Lookahead assertion
139
            if parser.str_('('):
140
                parser.str_('/', required=True) # next / is inside ()
141
                path = _path()
142
                parser.str_(')', required=True)
143
                elem.keys.append(path)
144
                tree += path
145
            
146
            if not parser.str_('/'): break
147
        
148
        tree[0].is_positive = is_positive
149
        
150
        # Value
151
        if parser.str_('='):
152
            if parser.str_('$'): # reference (different from a pointer)
153
                value = _path()
154
            else: value = _value() # literal value
155
            set_value(tree, value)
156
        
157
        # Expand * abbrs
158
        for i in reversed(xrange(len(tree))):
159
            elem = tree[i]
160
            if elem.is_ptr: offset = 2
161
            else: offset = 1
162
            try: repl = tree[i+offset].name
163
            except IndexError: pass # no replacement elem
164
            else: elem.name = expand_abbr(elem.name, repl)
165
        
166
        return tree
167
    
168
    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
    def _paths():
176
        paths = []
177
        while True:
178
            paths.append(_path())
179
            if not parser.str_(','): break
180
        return paths
181
    
182
    path = _path()
183
    parser.end()
184
    _cache[str_] = path
185
    return path
186

    
187
##### Querying/creating XML trees
188

    
189
def get(root, xpath, create=False, last_only=None, limit=1, allow_rooted=True):
190
    '''Warning: The last_only optimization may put data that should be together
191
    into separate nodes'''
192
    if last_only == None: last_only = create
193
    if limit == None or limit > 1: last_only = False
194
    if util.is_str(xpath): xpath = parse(xpath)
195
    
196
    # Handle edge cases
197
    if xpath == []: return [root]
198
    if create and not is_positive(xpath): return []
199
    
200
    # Define vars
201
    doc = root.ownerDocument
202
    if allow_rooted and is_rooted(xpath): root = doc.documentElement
203
    elem = xpath[0]
204
    
205
    # Find possible matches
206
    children = []
207
    if elem.is_attr:
208
        child = root.getAttributeNode(elem.name)
209
        if child != None: children = [child]
210
    elif is_self(elem): children = [root]
211
    elif is_parent(elem):
212
        parent = xml_dom.parent(root)
213
        if parent == None: return [] # don't try to create doc root's parent
214
        root = parent
215
        children = [root]
216
    else:
217
        children = xml_dom.by_tag_name(root, elem.name,
218
            last_only and (elem.keys == [] or is_instance(elem)))
219
    
220
    # 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
    # Check each match
232
    nodes = []
233
    for child in children:
234
        is_match = value_ == None or xml_dom.value(child) == value_
235
        for attr in elem.keys:
236
            if not is_match: break
237
            is_match = ((get(child, attr, False, last_only,
238
                allow_rooted=False) != []) == is_positive(attr))
239
        if is_match:
240
            nodes.append(child)
241
            if limit != None and len(nodes) >= limit: break
242
    
243
    # Create node
244
    if nodes == []:
245
        if not create or elem.is_lookup_only: return []
246
        if elem.is_attr:
247
            root.setAttribute(elem.name, '')
248
            node = root.getAttributeNode(elem.name)
249
        elif util.list_eq_is(children, [root]): node = root
250
        else: node = root.appendChild(doc.createElement(elem.name))
251
        if value_ != None: xml_dom.set_value(node, value_)
252
        nodes.append(node)
253
    
254
    path_value = value(xpath)
255
    xpath = xpath[1:] # rest of XPath
256
    
257
    next = []
258
    for node in nodes:
259
        # Create attrs
260
        if create:
261
            for attr in elem.keys + elem.attrs:
262
                get(node, attr, create, last_only, allow_rooted=False)
263
        
264
        # Follow pointer
265
        if elem.is_ptr:
266
            root = doc.documentElement
267
            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
        next += get(root, xpath, create, last_only, limit, allow_rooted=False)
292
        
293
        for branch in elem.other_branches:
294
            branch = branch[:] # don't modify input!
295
            set_value(branch, path_value)
296
            next += get(node, branch, create, last_only, limit,
297
                allow_rooted=False)
298
    
299
    return next
300

    
301
def get_1(*args, **kw_args):
302
    '''Runs get() and returns the first result'''
303
    return util.list_get(get(*args, **kw_args), 0)
304

    
305
def put_obj(root, xpath, id_, has_types, value=None):
306
    if util.is_str(xpath): xpath = parse(xpath)
307
    
308
    xpath = xpath[:] # don't modify input!
309
    set_id(xpath, id_, has_types)
310
    if value != None: set_value(xpath, value)
311
    get(root, xpath, True)
312

    
313
def path2xml(xpath, first_branch=True):
314
    root = xml_dom.create_doc().documentElement
315
    get(root, xpath, True)
316
    return root
317

    
318
def xml_set_id(root, id_): xml_dom.set_id(root.firstChild, id_)
(16-16/17)