def get_text_line_from_xml_element(xml_element):
t = TextLine(bbox_from_string(xml_element.attrib['bbox']))
for text_xml_element in xml_element.findall('./text'):
if len(text_xml_element.attrib) == 0:
continue
t.texts.append(get_text_from_xml_element(text_xml_element))
return t
def get_text_from_xml_element(xml_element):
text = xml_element.text
if isinstance(text, str):
text = text[0][0]
else:
text = ' '
bbox = None
if 'bbox' in xml_element.attrib:
bbox = bbox_from_string(xml_element.attrib['bbox'])
return Text(
bbox, text,
TextStyle(xml_element.attrib['font'],
float(xml_element.attrib['size']),
xml_element.attrib['colourspace'],
xml_element.attrib['ncolour']))
def get_page_from_xml_element(xml_element):
page = Page(bbox_from_string(xml_element.attrib['bbox']))
for text_box_node in xml_element.findall('./textbox'):
page.text_boxes.append(get_text_box_from_xml_element(text_box_node))
return page
def setUp(self):
self.bbox = bbox_from_string('459.840,753.697,462.075,765.210')
def test_five_points(self):
with self.assertRaises(InvalidBboxString):
bbox_from_string('1,2,3,4,5')
def test_two_points(self):
with self.assertRaises(InvalidBboxString):
bbox_from_string('1,2')
def test_one_point(self):
with self.assertRaises(InvalidBboxString):
bbox_from_string('1')