Read Open Document Table and put in a list of named tuples

odt_read.py

import xml.etree.ElementTree, re, zipfile, collections

def get_ns(tag):
	ns = re.findall(r'\{(.*)\}(.*)', tag)[0]	#Assume only one
	return ns


def one_or_default(it, default=None):
	try:
		result = it.__next__()
		try:
			extra = it.__next__()
			raise Exception(f'Too much data: {extra}')
		except StopIteration:
			return result

	except StopIteration:
		return default

def any_differs(it, value):
	for i in it:
		if i != value:
			return True

	return False

def ods_to_named_tuples(filename):
	#Fuggly solution, probably gives you the first table if you have multiple sheets, didn't test
	with zipfile.ZipFile(filename) as z:
		with z.open('content.xml') as infile:
			ns_map = dict()
			table = None
			for ev, v in xml.etree.ElementTree.iterparse(infile, events='start-ns end'.split()):
				if ev == 'start-ns':
					key, value = v
					ns_map[key] = value
				elif ev == 'end':
					ns, tag = get_ns(v.tag)
					if tag == 'table' and ns == ns_map['table']:
						table = v
						break

			assert table, f'Could not find any table in {infile}'

			result = list()
			cell_type = None
			for r in table.findall('table:table-row', ns_map):
				row = [one_or_default((t.text.strip() for t in c.findall('text:p', ns_map)), '') for c in r.findall('table:table-cell', ns_map)]
				while row and not row[-1]:
					del row[-1]

				if any_differs(row, ''):
					if cell_type:
						missing_cells = len(cell_type._fields) - len(row)
						result.append(cell_type(*(row + [''] * missing_cells)))
					else:
						cell_type = collections.namedtuple('cell', row)

			return result

for row in ods_to_named_tuples('/home/devilholk/operators.ods'):
	print(row)

#Example Output
#  cell(op='floor_divide_modulus', rank='5', python_name='', base_type='binary', category='arithmetic.composit', python_symbol='', python_function='divmod', python_library='builtins')
#  cell(op='exponentiation', rank='7', python_name='pow', base_type='binary', category='arithmetic.exponent', python_symbol='**', python_function='', python_library='')
#  cell(op='matrix.multiply', rank='5', python_name='matmul', base_type='binary', category='arithmetic.matrix', python_symbol='@', python_function='', python_library='')
#  cell(op='modulus', rank='5', python_name='mod', base_type='binary', category='arithmetic.modulus', python_symbol='%', python_function='', python_library='')
#  cell(op='divide', rank='5', python_name='truediv', base_type='binary', category='arithmetic.scaling', python_symbol='/', python_function='', python_library='')
#  cell(op='floor_divide', rank='5', python_name='floordiv', base_type='binary', category='arithmetic.scaling', python_symbol='//', python_function='', python_library='')