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 1345 aaronmk
import xpath_func
92
93 170 aaronmk
def expand_abbr(name, repl):
94
    before, abbr, after = name.partition('*')
95
    if abbr != '': name = before+repl+after
96
    return name
97
98 154 aaronmk
_cache = {}
99
100 76 aaronmk
def parse(str_):
101 154 aaronmk
    try: return _cache[str_]
102
    except KeyError: pass
103
104 76 aaronmk
    parser = Parser(str_)
105 21 aaronmk
106 76 aaronmk
    def _path():
107 97 aaronmk
        is_positive = not parser.str_('!')
108
109 21 aaronmk
        tree = []
110
        while True:
111 38 aaronmk
            # Split path
112 76 aaronmk
            if parser.str_('{'):
113 1070 aaronmk
                last = tree[-1]
114
                last.other_branches = _paths()
115 76 aaronmk
                parser.str_('}', required=True)
116 1070 aaronmk
                if parser.str_('/'): # common subpath after {}
117
                    subpath = _path()
118
                    for branch in last.other_branches: branch += subpath
119
                tree += last.other_branches.pop(0) # use first path for now
120
                break # anything after split path has already been parsed
121 36 aaronmk
122 754 aaronmk
            elem = XpathElem(is_attr=parser.str_('@'), name=_value())
123 38 aaronmk
124 1238 aaronmk
            elem.is_lookup_only = parser.str_('?')
125
126 94 aaronmk
            # Keys used to match nodes
127 76 aaronmk
            if parser.str_('['):
128 94 aaronmk
                elem.keys = _paths()
129 76 aaronmk
                parser.str_(']', required=True)
130 36 aaronmk
131 94 aaronmk
            # Attrs created when no matching node exists
132
            if parser.str_(':'):
133
                parser.str_('[', required=True)
134
                elem.attrs = _paths()
135
                parser.str_(']', required=True)
136
137 76 aaronmk
            elem.is_ptr = parser.str_('->')
138 21 aaronmk
            tree.append(elem)
139 36 aaronmk
140
            # Lookahead assertion
141 76 aaronmk
            if parser.str_('('):
142
                parser.str_('/', required=True) # next / is inside ()
143
                path = _path()
144
                parser.str_(')', required=True)
145 94 aaronmk
                elem.keys.append(path)
146 36 aaronmk
                tree += path
147
148 76 aaronmk
            if not parser.str_('/'): break
149 32 aaronmk
150 97 aaronmk
        tree[0].is_positive = is_positive
151
152
        # Value
153 997 aaronmk
        if parser.str_('='):
154 1009 aaronmk
            if parser.str_('$'): # reference (different from a pointer)
155
                value = _path()
156
            else: value = _value() # literal value
157 997 aaronmk
            set_value(tree, value)
158 85 aaronmk
159 32 aaronmk
        # Expand * abbrs
160 70 aaronmk
        for i in reversed(xrange(len(tree))):
161
            elem = tree[i]
162 171 aaronmk
            if elem.is_ptr: offset = 2
163 32 aaronmk
            else: offset = 1
164 170 aaronmk
            try: repl = tree[i+offset].name
165 171 aaronmk
            except IndexError: pass # no replacement elem
166
            else: elem.name = expand_abbr(elem.name, repl)
167 32 aaronmk
168 21 aaronmk
        return tree
169 76 aaronmk
170 754 aaronmk
    def _value():
171
        if parser.str_('"'):
172
            value = parser.re(r'[^"]*')
173
            parser.str_('"', required=True)
174
        else: value = parser.re(r'(?:(?:\w+:)*[\w.*]+)?')
175
        return value
176
177 94 aaronmk
    def _paths():
178
        paths = []
179
        while True:
180
            paths.append(_path())
181
            if not parser.str_(','): break
182
        return paths
183
184 76 aaronmk
    path = _path()
185
    parser.end()
186 1342 aaronmk
    path = xpath_func.process(path)
187 154 aaronmk
    _cache[str_] = path
188 76 aaronmk
    return path
189 21 aaronmk
190 1001 aaronmk
##### Querying/creating XML trees
191 22 aaronmk
192 1005 aaronmk
def get(root, xpath, create=False, last_only=None, limit=1, allow_rooted=True):
193 141 aaronmk
    '''Warning: The last_only optimization may put data that should be together
194
    into separate nodes'''
195 77 aaronmk
    if last_only == None: last_only = create
196 887 aaronmk
    if limit == None or limit > 1: last_only = False
197 892 aaronmk
    if util.is_str(xpath): xpath = parse(xpath)
198 99 aaronmk
199 1005 aaronmk
    # Handle edge cases
200 998 aaronmk
    if xpath == []: return [root]
201 886 aaronmk
    if create and not is_positive(xpath): return []
202 1005 aaronmk
203
    # Define vars
204 998 aaronmk
    doc = root.ownerDocument
205 1005 aaronmk
    if allow_rooted and is_rooted(xpath): root = doc.documentElement
206 165 aaronmk
    elem = xpath[0]
207
208
    # Find possible matches
209
    children = []
210
    if elem.is_attr:
211 998 aaronmk
        child = root.getAttributeNode(elem.name)
212 165 aaronmk
        if child != None: children = [child]
213 1004 aaronmk
    elif is_self(elem): children = [root]
214
    elif is_parent(elem):
215 1000 aaronmk
        parent = xml_dom.parent(root)
216 1013 aaronmk
        if parent == None: return [] # don't try to create doc root's parent
217
        root = parent
218
        children = [root]
219 165 aaronmk
    else:
220 998 aaronmk
        children = xml_dom.by_tag_name(root, elem.name,
221 1013 aaronmk
            last_only and (elem.keys == [] or is_instance(elem)))
222 165 aaronmk
223 1009 aaronmk
    # Retrieve elem value
224
    value_ = elem.value
225
    if util.is_list(value_): # reference (different from a pointer)
226
        targets = get(root, value_)
227
        try: target = targets[0]
228
        except IndexError:
229
            warnings.warn(UserWarning('XPath reference target missing: '
230
                +str(value_)+'\nXPath: '+str(xpath)))
231
            value_ = None
232
        else: value_ = xml_dom.value(target)
233
234 165 aaronmk
    # Check each match
235 885 aaronmk
    nodes = []
236 165 aaronmk
    for child in children:
237 1009 aaronmk
        is_match = value_ == None or xml_dom.value(child) == value_
238 165 aaronmk
        for attr in elem.keys:
239
            if not is_match: break
240 1011 aaronmk
            is_match = ((get(child, attr, False, last_only,
241
                allow_rooted=False) != []) == is_positive(attr))
242 885 aaronmk
        if is_match:
243
            nodes.append(child)
244
            if limit != None and len(nodes) >= limit: break
245 165 aaronmk
246
    # Create node
247 886 aaronmk
    if nodes == []:
248 1239 aaronmk
        if not create or elem.is_lookup_only: return []
249 77 aaronmk
        if elem.is_attr:
250 998 aaronmk
            root.setAttribute(elem.name, '')
251
            node = root.getAttributeNode(elem.name)
252 1013 aaronmk
        elif util.list_eq_is(children, [root]): node = root
253 998 aaronmk
        else: node = root.appendChild(doc.createElement(elem.name))
254 1009 aaronmk
        if value_ != None: xml_dom.set_value(node, value_)
255 886 aaronmk
        nodes.append(node)
256 165 aaronmk
257 1009 aaronmk
    path_value = value(xpath)
258 165 aaronmk
    xpath = xpath[1:] # rest of XPath
259
260 886 aaronmk
    next = []
261
    for node in nodes:
262
        # Create attrs
263
        if create:
264
            for attr in elem.keys + elem.attrs:
265 1011 aaronmk
                get(node, attr, create, last_only, allow_rooted=False)
266 886 aaronmk
267
        # Follow pointer
268
        if elem.is_ptr:
269 998 aaronmk
            root = doc.documentElement
270 886 aaronmk
            xpath = copy.deepcopy(xpath)
271
            id_path = backward_id(xpath[instance_level])
272
            if id_path != None: # backward (child-to-parent) pointer with ID key
273
                id_path[0].name = expand_abbr(id_path[0].name, node.tagName)
274
                set_value(id_path, xml_dom.get_id(node)) # modify xpath's id key
275
            else: # forward (parent-to-child) pointer
276
                id_ = xml_dom.value(node)
277
                obj_xpath = obj(xpath) # target object
278
                if id_ == None or get(root, obj_xpath, False, True) == []:
279
                    # no target or target keys don't match
280
                    if not create: continue
281
282
                    # Use last target object's ID + 1
283
                    obj_xpath[-1].keys = [] # just get by tag name
284
                    last = get(root, obj_xpath, False, True)
285
                    if last != []: id_ = str(int(xml_dom.get_id(last[0])) + 1)
286
                    else: id_ = '0'
287
288
                    # Will append if target keys didn't match.
289
                    # Use lookahead assertion to avoid this.
290
                    xml_dom.set_value(node, id_)
291
                else: last_only = False
292
                set_id(xpath, id_)
293
        else: root = node
294 1005 aaronmk
        next += get(root, xpath, create, last_only, limit, allow_rooted=False)
295 886 aaronmk
296
        for branch in elem.other_branches:
297 1294 aaronmk
            branch = branch[:] # don't modify input!
298 1009 aaronmk
            set_value(branch, path_value)
299 1005 aaronmk
            next += get(node, branch, create, last_only, limit,
300
                allow_rooted=False)
301 165 aaronmk
302
    return next
303 82 aaronmk
304 1320 aaronmk
def get_1(*args, **kw_args):
305
    '''Runs get() and returns the first result'''
306
    return util.list_get(get(*args, **kw_args), 0)
307
308 141 aaronmk
def put_obj(root, xpath, id_, has_types, value=None):
309 892 aaronmk
    if util.is_str(xpath): xpath = parse(xpath)
310 155 aaronmk
311 1294 aaronmk
    xpath = xpath[:] # don't modify input!
312 82 aaronmk
    set_id(xpath, id_, has_types)
313 85 aaronmk
    if value != None: set_value(xpath, value)
314 141 aaronmk
    get(root, xpath, True)
315 133 aaronmk
316 135 aaronmk
def path2xml(xpath, first_branch=True):
317
    root = xml_dom.create_doc().documentElement
318 141 aaronmk
    get(root, xpath, True)
319 135 aaronmk
    return root
320 133 aaronmk
321 141 aaronmk
def xml_set_id(root, id_): xml_dom.set_id(root.firstChild, id_)