2 from types import StringTypes
4 from StringIO import StringIO
5 from sfa.util.faults import InvalidXML
10 def filter_value(key, value):
12 if isinstance(value, str):
14 value = value.replace('*', '')
15 xpath = 'contains(%s, "%s")' % (key, value)
17 xpath = '%s="%s"' % (key, value)
25 for (key, value) in filter.items():
30 if isinstance(value, str):
31 filter_list.append(XpathFilter.filter_value(key, value))
32 elif isinstance(value, list):
33 stmt = ' or '.join([XpathFilter.filter_value(key, str(val)) for val in value])
34 filter_list.append(stmt)
36 xpath = ' and '.join(filter_list)
37 xpath = '[' + xpath + ']'
41 def __init__(self, node, namespaces):
44 self.namespaces = namespaces
45 self.attrib = node.attrib
48 def xpath(self, xpath, namespaces=None):
50 namespaces = self.namespaces
51 elems = self.node.xpath(xpath, namespaces=namespaces)
52 return [XmlNode(elem, namespaces) for elem in elems]
54 def add_element(self, name, **kwds):
55 element = etree.SubElement(self.node, name, **kwds)
56 return XmlNode(element, self.namespaces)
58 def remove_elements(name):
60 Removes all occurences of an element from the tree. Start at
61 specified root_node if specified, otherwise start at tree's root.
64 if not element_name.startswith('//'):
65 element_name = '//' + element_name
66 elements = self.node.xpath('%s ' % name, namespaces=self.namespaces)
67 for element in elements:
68 parent = element.getparent()
69 parent.remove(element)
72 self.node.remove(element)
74 def set(self, key, value):
75 self.node.set(key, value)
77 def set_text(self, text):
81 del self.node.attrib[key]
83 def iterchildren(self):
84 return self.node.iterchildren()
87 return etree.tostring(self.node, encoding='UTF-8', pretty_print=True)
94 def __init__(self, xml=None, namespaces=None):
96 self.namespaces = namespaces
97 self.default_namespace = None
99 if isinstance(xml, basestring):
101 if isinstance(xml, XmlNode):
103 self.namespaces = xml.namespaces
104 elif isinstance(xml, etree._ElementTree) or isinstance(xml, etree._Element):
105 self.parse_xml(etree.tostring(xml))
107 def parse_xml(self, xml):
109 parse rspec into etree
111 parser = etree.XMLParser(remove_blank_text=True)
113 tree = etree.parse(xml, parser)
115 # 'rspec' file doesnt exist. 'rspec' is proably an xml string
117 tree = etree.parse(StringIO(xml), parser)
119 raise InvalidXML(str(e))
120 root = tree.getroot()
121 self.namespaces = dict(root.nsmap)
123 if 'default' not in self.namespaces and None in self.namespaces:
124 # If the 'None' exist, then it's pointing to the default namespace. This makes
125 # it hard for us to write xpath queries for the default naemspace because lxml
126 # wont understand a None prefix. We will just associate the default namespeace
127 # with a key named 'default'.
128 self.namespaces['default'] = self.namespaces.pop(None)
131 self.namespaces['default'] = 'default'
133 self.root = XmlNode(root, self.namespaces)
135 for key in self.root.attrib.keys():
136 if key.endswith('schemaLocation'):
137 # schema location should be at the end of the list
138 schema_parts = self.root.attrib[key].split(' ')
139 self.schema = schema_parts[1]
140 namespace, schema = schema_parts[0], schema_parts[1]
143 def parse_dict(self, d, root_tag_name='xml', element = None):
145 if self.root is None:
146 self.parse_xml('<%s/>' % root_tag_name)
153 # handle repeating fields
154 for (key, value) in d.items():
155 if isinstance(value, list):
158 if isinstance(val, dict):
159 child_element = etree.SubElement(element, key)
160 self.parse_dict(val, key, child_element)
161 elif isinstance(val, basestring):
162 child_element = etree.SubElement(element, key).text = val
164 elif isinstance(value, int):
165 d[key] = unicode(d[key])
169 # element.attrib.update will explode if DateTimes are in the
172 # looks like iteritems won't stand side-effects
174 if not isinstance(d[k],StringTypes):
177 element.attrib.update(d)
179 def validate(self, schema):
181 Validate against rng schema
183 relaxng_doc = etree.parse(schema)
184 relaxng = etree.RelaxNG(relaxng_doc)
185 if not relaxng(self.root):
186 error = relaxng.error_log.last_error
187 message = "%s (line %s)" % (error.message, error.line)
188 raise InvalidXML(message)
191 def xpath(self, xpath, namespaces=None):
193 namespaces = self.namespaces
194 return self.root.xpath(xpath, namespaces=namespaces)
196 def set(self, key, value, node=None):
199 return node.set(key, value)
201 def remove_attribute(self, name, node=None):
204 node.remove_attribute(name)
207 def add_element(self, name, **kwds):
209 Wrapper around etree.SubElement(). Adds an element to
210 specified parent node. Adds element to root node is parent is
214 xmlnode = parent.add_element(name, *kwds)
217 def remove_elements(self, name, node = None):
219 Removes all occurences of an element from the tree. Start at
220 specified root_node if specified, otherwise start at tree's root.
225 node.remove_elements(name)
227 def attributes_list(self, elem):
228 # convert a list of attribute tags into list of tuples
229 # (tagnme, text_value)
233 opts.append((e.tag, str(e.text).strip()))
236 def get_element_attributes(self, elem=None, depth=0):
238 elem = self.root_node
239 if not hasattr(elem, 'attrib'):
240 # this is probably not an element node with attribute. could be just and an
241 # attribute, return it
243 attrs = dict(elem.attrib)
244 attrs['text'] = str(elem.text).strip()
245 attrs['parent'] = elem.getparent()
246 if isinstance(depth, int) and depth > 0:
247 for child_elem in list(elem):
248 key = str(child_elem.tag)
250 attrs[key] = [self.get_element_attributes(child_elem, depth-1)]
252 attrs[key].append(self.get_element_attributes(child_elem, depth-1))
254 attrs['child_nodes'] = list(elem)
257 def merge(self, in_xml):
264 return etree.tostring(self.root.node, encoding='UTF-8', pretty_print=True)
266 # XXX smbaker, for record.load_from_string
267 def todict(self, elem=None):
271 d.update(elem.attrib)
272 d['text'] = elem.text
273 for child in elem.iterchildren():
274 if child.tag not in d:
276 d[child.tag].append(self.todict(child))
278 if len(d)==1 and ("text" in d):
283 def save(self, filename):
284 f = open(filename, 'w')
285 f.write(self.toxml())
289 #if __name__ == '__main__':
290 # rspec = RSpec('/tmp/resources.rspec')