def format_authors(authors: bs4.element.Tag) -> List[str]:
""" Retrieves list of authors from the tag. """
authors = str(authors).replace('<dd>',
'').replace('</dd>',
'').replace(', ', ',').strip()
authors = authors.split(',')
return authors
def format_authors(authors: bs4.element.Tag) -> List[str]:
""" Tranforms the raw authors string into a list of authors. """
authors = str(authors)
authors = authors.replace('\n', '').replace(' ', ' ').replace(', ', ',')
authors_out = authors.split(',')
authors_out = [a.strip() for a in authors_out]
return authors_out
def parse_manually(self, parse_object: bs4.element.Tag) -> dict:
"""
Method which is dedicated to manuall parse broken html
Input: parse_object = object which we would parse
Output: dict
"""
list_column_names = [str(v) for v in parse_object.find_all('b')]
parse_object = str(parse_object)
list_column_names.insert(0, '</a>')
list_split = []
for types in list_column_names:
if types in list_column_names:
list_split.append(types)
parse_object = parse_object.replace(types, self.rand)
parse_split = parse_object.split(self.rand)
if '</a>' in list_split:
list_split[0] = sp.status_iasa
list_split = [self.remove_tags(x) for x in list_split]
list_split = [self.remove_special(x) for x in list_split]
list_split = [self.remove_spaces(x) for x in list_split]
list_split = [v for v in list_split if v]
value_dict = {}
if len(parse_split) > 1:
for column_value, value in zip(list_split, parse_split[1:]):
value_dict.update(
self.make_further_check(sp.rechange_iasa[column_value],
value, sp.rechange_phrase))
return value_dict
def format_datetime(date_str: bs4.element.Tag, time: str) -> datetime:
"""
Formatting date & time correctly
:param date_str: string of html, contains date
:param time: contains time (format: XX:XX)
:return: string with the result if there is data, else returns empty string
"""
arr = date_str.split(' ')
year = int(arr[2])
month = get_month(arr[1])
day = int(arr[0])
if time == '':
hour = 20
minutes = 0
else:
time = time.split(':')
hour = int(time[0])
minutes = int(time[1])
return datetime(year, month, day, hour, minutes)
