Globus

import re
import os
import requests as req

from functions.supportfunctions import dump
from functions.SQLHelper import SQLHelper


bank_name = 'globus_bank'

name_db = 'coins.db'
cur_dir = '/Users/sergeyilyin/Google Drive/cbr/coins'
path_db = os.path.join(cur_dir, name_db)

db = SQLHelper(path_db)

bank = db["banks_list"].find({"eng_name": bank_name})
urls = db["bank_info"].find({"bank_id":bank[0]["id"]})
htmls = [url["href_page"] for url in urls]


# print(htmls)

def parser(html_list):
    # cont=[]
    cat_list = []
    name_list = []
    sell_price_list = []
    buy_price_list = []
    for i in html_list:
        # print(i)
        resp = req.get(i)
        html = resp.text
        html = re.sub(r'<!--.*?-->', '', html, flags=re.DOTALL)
        # print('Обращаемся к серверу банка')
        # print(html)
        re_cat = re.compile(r"product-artikul.*?(\d{4}-\d{4})", re.DOTALL)
        re_name = re.compile(r"product-name.*?(\d*?[-А-Яа-я()\s]*)</a>", re.DOTALL)
        re_sell_price = re.compile(r"class=\"price giftd-price\">([а-яА-Я\s]+|\d+?\s+?\d+)", re.DOTALL)
        re_buy_price = re.compile(r"span class=\"price\">([а-яА-Я\s]+|\d+?\s+?\d+)", re.DOTALL)

        cat_list.extend(re_cat.findall(html))
        name_list.extend(re_name.findall(html))
        sell_price_list.extend(re_sell_price.findall(html))
        for i in sell_price_list:
            if float(i.strip().replace(' ','')) == True:
                float(i.strip().replace(' ',''))
            else: None
            print(i)
        buy_price_list.extend(re_buy_price.findall(html))

        # print(sell_price_list)

    cont = {'sell': [{"cat_number": c, "coin_name": cn, "price": p}
                     for c, cn, p in zip(cat_list, name_list, sell_price_list)],
            'buy': [{"cat_number": c, "coin_name": cn, "price": p}
                     for c, cn, p in zip(cat_list, name_list, buy_price_list)]}
            # print(cat_list)

    # print(cont)

    return cont


        # cont = {'sell': [{"cat_number": i[0], "price": float(i[1])} for i in re_cat.findall(html)]}

        # cont = {"sell": [{"cat_number": c, "price": p}
        #                  for c, p in zip(cat_number, price)]}

        # cont.extend(list(re_cat.findall(html)))

    # print('Формируем пары значений: каталожный номер - цена')

    # cont=set(cont)

    # cont = {"sell":[{"cat_number":(i[0]),"price": float(i[1].replace(' ',''))} for i in cont]}
    # return cont

# print(parser(htmls))

cont = parser(htmls)

# print(cont)

# print('Сохраняем дамп')

# dump(cont, bank_name)

print('Парсер закончил работу')