def get_model():
mark = json_data['make']['slug'].lower(
) if json_data['make']['slug'] else None
model = json_data['model']['slug'].lower(
) if json_data['model']['slug'] else None
if mark is not None and model is not None:
return get_model_id(mark, model)
def parse_model(self):
lines = self.parse_from_base_info('Модель', link=True)
if lines is None:
return None
print('model is find')
lines = lines.get_attribute("href").split('/')
if lines[-3] == 'drugaya':
return None
print('data car models', lines[-4], lines[-3])
return get_model_id(lines[-4], lines[-3])
def find_car_data(self):
self.car_dict = dict(
model_id=get_model_id(*self.get_mark_model_name()),
description=self.parser_description(),
createdAt=self.parse_date_created(),
location=self.parse_location(),
gearbox=self.parse_gearbox(),
cleared=self.parse_cleared(),
mileage=self.parse_mileage(),
engine=self.parse_engine(),
seller=self.parse_seller(),
bp_link=self.current_link,
body=self.parse_body(),
fuel=self.parse_fuel(),
image=self.parse_image(),
updatedAt=timezone.now(),
last_site_updatedAt=None,
year=self.parse_year(),
dtp=self.parse_dtp(),
sold=False)
print(self.car_dict)
def get_ad_data(self, url):
"""Returns ad's data"""
data = dict()
soup = self.get_page(url)
price_soup = soup.find('td', text='Цена') if soup.find(
'td', text='Цена') else soup.find('span', text='Цена')
data['price'] = price_soup.find_next_sibling().find('span').find(
'span').text.translate(OD)
year_mil_soup = soup.find('td', text='Год выпуска') if soup.find(
'td', text='Год выпуска') else soup.find('span',
text='Год выпуска')
data['year'] = year_mil_soup.find_next_sibling().find(
'a').text.translate(OD)
data['mileage'] = int(year_mil_soup.find_next_sibling().find(
'span').text.translate(OD)) // 1000
# engine, fuel
engine_soup = soup.find('td', text='Двигатель') if soup.find(
'td', text='Двигатель') else soup.find('span', text='Двигатель')
data['engine'] = engine_soup.find_next_sibling().find(
'strong').text[:3]
fuel = engine_soup.find_next_sibling().find('span').text.replace(
'(', '').replace(')', '').lower()
data['fuel'] = FUEL.get(fuel)
# gearbox
gearbox_soup = soup.find('td', text='КПП') if soup.find(
'td', text='КПП') else soup.find('span', text='КПП')
gearbox = gearbox_soup.find_next_sibling().find('strong').text.lower()
data['gearbox'] = GEARBOX.get(gearbox)
# body
body_soup = soup.find('td', text='Тип кузова') if soup.find(
'td', text='Тип кузова') else soup.find('span', text='Тип кузова')
body = body_soup.find_next_sibling().find(
'strong').text.split()[0].lower()
data['body'] = BODY.get(body)
# location
location_soup = soup.find('td', text='Область') if soup.find(
'td', text='Область') else soup.find('span', text='Область')
location = location_soup.find_next_sibling().find('a').text.lower()
data['location'] = LOCATION_ALL.get(location)
date_soup = soup.find('td', text='Дата добавления') if soup.find(
'td', text='Дата добавления') else soup.find(
'span', text='Дата добавления')
created = date_soup.find_next_sibling().find('span').text
data['last_site_updatedAt'] = TZ.localize(
datetime.strptime(created, '%d.%m.%Y'))
mark_model = soup.find(
'div', class_='rst-uix-page-tree rst-uix-radius').find(
'a').find_next_sibling('a').find_next_sibling(
'a').find_next_sibling('a').text
mark = mark_model.split()[0].lower()
model = mark_model.split()[1].lower()
data['model_id'] = get_model_id(mark, model)
data['description'] = soup.find('div', class_='rst-page-oldcars-item-option-block-container rst-page-oldcars-item-option-block-container-desc rst-uix-block-more').text.strip() if \
soup.find('div', class_='rst-page-oldcars-item-option-block-container rst-page-oldcars-item-option-block-container-desc rst-uix-block-more') else None
images_soup = soup.find_all('a',
class_='rst-uix-float-left rst-uix-radius')
data['image'] = images_soup[0]['href'] if images_soup else None
data['dtp'] = bool(soup.find('em', text="После ДТП"))
data['rst_link'] = url
return data
def find_model(self, data: dict):
return get_model_id(data['markNameEng'], data['modelNameEng'])