60 lines
2.7 KiB
Python
60 lines
2.7 KiB
Python
import requests, re, json, os, yaml
|
|
|
|
def get_conf():
|
|
'''return settings in settings.yaml file'''
|
|
with open(os.path.dirname(os.path.realpath(__file__))+"/settings.yaml", 'r') as conf_file:
|
|
settings = yaml.safe_load(conf_file)
|
|
return settings
|
|
|
|
def load_cookies_from_file(file_path):
|
|
with open(file_path, 'r') as file:
|
|
cookies_data = json.load(file)
|
|
|
|
cookies_dict = {}
|
|
|
|
for cookie_data in cookies_data:
|
|
name_raw = cookie_data.get("Name raw", "")
|
|
content_raw = cookie_data.get("Content raw", "")
|
|
cookie_value = f"{content_raw}"
|
|
|
|
if len(cookie_value) > 0:
|
|
cookies_dict[name_raw] = cookie_value
|
|
|
|
return cookies_dict
|
|
|
|
|
|
if __name__ == '__main__':
|
|
|
|
item_regex = re.compile(r'skuAttr\\\":\\\"(([0-9]*:[0-9]*#[a-zA-Z0-9 ]*;?)*)\\\",\\\"skuId\\\":([0-9]*),\\\"skuIdStr\\\":\\\"[0-9]*\\\",\\\"skuPropIds\\\":\\\"[0-9,]*\\\",\\\"skuVal\\\":{\\\"availQuantity\\\":([0-9]*),(\\\"discount\\\":\\\"([0-9]*)\\\",\\\"discountTips\\\":\\\"-[0-9]*%\\\",)?\\\"hideOriPrice\\\":(false|true),\\\"inventory\\\":([0-9]*),\\\"isActivity\\\":(true|false),\\\"optionalWarrantyPrice\\\":\[\],(\\\"skuActivityAmount\\\":{\\\"currency\\\":\\\"([A-Z]*)\\\",\\\"formatedAmount\\\":\\\"[0-9]*,[0-9]*.\\\",\\\"value\\\":([0-9]*\.[0-9]*)},\\\"skuActivityAmountLocal\\\":\\\"[0-9]*,[0-9]*.\|[0-9]*\|[0-9]*\\\",)?\\\"skuAmount\\\":{\\\"currency\\\":\\\"([A-Z]*)\\\",\\\"formatedAmount\\\":\\\"[0-9]*,[0-9]*.\\\",\\\"value\\\":([0-9]*\.[0-9]*)}')
|
|
choice_regex = re.compile(r'businessModel\\\":\\\"CHOICE\\\"')
|
|
#shipping_cost_regex = re.compile(r'')
|
|
|
|
settings = get_conf()
|
|
|
|
session = requests.Session()
|
|
cookies_file_path = './cookies.json'
|
|
cookies = load_cookies_from_file(cookies_file_path)
|
|
session.cookies.update(cookies)
|
|
|
|
extract = dict()
|
|
|
|
for item in settings['id']:
|
|
url = 'https://aliexpress.com/item/'+item+'.html'
|
|
target_page_response = session.get(url)
|
|
|
|
if target_page_response.status_code == 200:
|
|
content = re.findall(item_regex, target_page_response.text)
|
|
is_choice = bool(re.search(choice_regex, target_page_response.text))
|
|
for elem in content:
|
|
key = item+':'+elem[0]
|
|
discount = 0 if len(elem[5]) == 0 else int(elem[5])
|
|
price = float(elem[13]) if len(elem[11]) == 0 else float(elem[11])
|
|
|
|
# skuId, quantity, discount_percentage, price, currency, choice_delivery
|
|
extract[key] = (elem[2], elem[3], discount, price, elem[12], is_choice)
|
|
#print(elem,"\n\n")
|
|
print(extract)
|
|
else:
|
|
print(f'Failed to fetch target page. Status code: {target_page_response.status_code}')
|
|
|