def fetch_text(xml_file): assert os.path.exists(xml_file) def listener(path, element, filter): listener.text += element.text listener.text = '' grabber = XmlGrabber() grabber.addXmlElementFilter(XmlPathFilter([ ".*/samenvatting", ".*/beschrijving", ".*/annotatie"])) grabber.grab(xml_file, listener) return listener.text
def fetch_text(xml_file): assert os.path.exists(xml_file) def listener(path, element, filter): listener.text += element.text listener.text = '' grabber = XmlGrabber() grabber.addXmlElementFilter( XmlPathFilter( [".*/samenvatting", ".*/beschrijving", ".*/annotatie"])) grabber.grab(xml_file, listener) return listener.text
def fetch_text(xml_file): assert os.path.exists(xml_file) output_file_name = xml_file + ".txt" with codecs.open(output_file_name, "w", "utf-8") as f: def listener(path, element, filter): f.write(element.text) grabber = XmlGrabber() grabber.addXmlPathFilter(XmlPathFilter([".*/samenvatting"])) grabber.grab(xml_file, listener) return output_file_name
def test_xmlgrabber(self): global counter def listener1(xml_path, element, filter): global counter counter += 1 counter = 0 grabber = XmlGrabber() grabber.addXmlElementFilter(XmlPathFilter([ ".*/selectiepositiedragerannotatie" ])) grabber.grab(xml_file, listener1) self.assertEqual(0, counter) counter = 0 grabber = XmlGrabber(False) grabber.addXmlElementFilter(XmlPathFilter(".*/selectiepositiedragerannotatie" )) grabber.grab(xml_file, listener1) self.assertEqual(1, counter)
def test_xml_elements_collector(self): def f1(): g = XmlGrabber() b = XmlElementsCollector('root') g.addXmlElementFilter(TrueXmlElementFilter()) g.grab(xml_file, b) self.assertRaises(NotImplementedError, f1) class MyXmlElementsCollector(XmlElementsCollector): def grabbed(self, path, element, filter): self.root_element().add_kid(element) g = XmlGrabber() b = MyXmlElementsCollector('root') g.addXmlElementFilter(TrueXmlElementFilter()) g.grab(xml_file, b) self.assertEqual(len(str(b.root_element())), 10026)
def test_xml_elements_collector(self): def f1(): g = XmlGrabber() b = XmlElementsCollector('root') g.addXmlElementFilter(TrueXmlElementFilter()) g.grab(xml_file, b) self.assertRaises( NotImplementedError, f1) class MyXmlElementsCollector(XmlElementsCollector): def grabbed(self, path, element, filter): self.root_element().add_kid(element) g = XmlGrabber() b = MyXmlElementsCollector('root') g.addXmlElementFilter(TrueXmlElementFilter()) g.grab(xml_file, b) self.assertEqual(len(str(b.root_element())), 10026)
def test_xmlgrabber(self): global counter def listener1(xml_path, element, filter): global counter counter += 1 counter = 0 grabber = XmlGrabber() grabber.addXmlElementFilter( XmlPathFilter([".*/selectiepositiedragerannotatie"])) grabber.grab(xml_file, listener1) self.assertEqual(0, counter) counter = 0 grabber = XmlGrabber(False) grabber.addXmlElementFilter( XmlPathFilter(".*/selectiepositiedragerannotatie")) grabber.grab(xml_file, listener1) self.assertEqual(1, counter)
def f1(): g = XmlGrabber() b = XmlElementsCollector('root') g.addXmlElementFilter(TrueXmlElementFilter()) g.grab(xml_file, b)