[895cd87] | 1 | import json
|
---|
[b68ae8d] | 2 | import unicodedata
|
---|
| 3 | from datetime import datetime
|
---|
| 4 | import psycopg2
|
---|
| 5 | import config_read
|
---|
| 6 | from bs4 import BeautifulSoup
|
---|
| 7 | from selenium import webdriver
|
---|
| 8 | import requests
|
---|
| 9 |
|
---|
| 10 | import sys
|
---|
| 11 |
|
---|
[895cd87] | 12 | from classes.phoneoffer import PhoneOffer
|
---|
| 13 |
|
---|
[b68ae8d] | 14 | file_path = 'outputfile.txt'
|
---|
| 15 | sys.stdout = open(file_path, "w")
|
---|
| 16 |
|
---|
| 17 | offer_shop = "Neptun" # offer shop
|
---|
| 18 | last_updated = datetime.now().date()
|
---|
| 19 | is_validated = False
|
---|
| 20 |
|
---|
[895cd87] | 21 | # Neptun phone offers that are already in database
|
---|
| 22 |
|
---|
| 23 | offers = json.loads(unicodedata.normalize('NFKD', requests.get('http://localhost:8080/phoneoffer/shop/neptun').text))
|
---|
| 24 |
|
---|
| 25 | database_offers = []
|
---|
| 26 |
|
---|
| 27 | for offer in offers:
|
---|
| 28 | phoneOffer = PhoneOffer(offer['id'], offer['offer_shop'], offer['offer_name'], offer['price'],
|
---|
| 29 | offer['ram_memory'],
|
---|
| 30 | offer['rom_memory'], offer['color'], offer['front_camera'], offer['back_camera'],
|
---|
| 31 | offer['chipset'], offer['battery'], offer['operating_system'], offer['cpu'],
|
---|
| 32 | offer['image_url'],
|
---|
| 33 | offer['offer_url'], offer['last_updated'], offer['is_validated'],
|
---|
| 34 | offer['offer_description'],
|
---|
| 35 | offer['offer_shop_code'])
|
---|
| 36 | database_offers.append(phoneOffer)
|
---|
| 37 |
|
---|
| 38 | new_offers = []
|
---|
| 39 |
|
---|
[b68ae8d] | 40 | for i in range(1, 11):
|
---|
[895cd87] | 41 | neptun_url = 'https://www.neptun.mk/mobilni_telefoni.nspx?page=' + str(i)
|
---|
[b68ae8d] | 42 |
|
---|
| 43 | # selenium is used because of the dynamic content of the page
|
---|
| 44 | driver1 = webdriver.Safari(executable_path='/usr/bin/safaridriver')
|
---|
| 45 | driver1.get(neptun_url)
|
---|
| 46 | neptun_html = driver1.page_source
|
---|
| 47 |
|
---|
| 48 | # closing the driver so the safari instance can pair with another webdriver session
|
---|
| 49 | driver1.close()
|
---|
| 50 |
|
---|
| 51 | # response1 = requests.get(neptun_url)
|
---|
| 52 | soup1 = BeautifulSoup(neptun_html, 'html.parser')
|
---|
| 53 |
|
---|
| 54 | phones = soup1.find('div', {'id': 'mainContainer'}).find('div',
|
---|
| 55 | {'class': 'col-lg-9 col-md-9 col-sm-8 col-fix-main'}) \
|
---|
| 56 | .find_all('div', {'class': 'ng-scope product-list-item-grid'})
|
---|
| 57 |
|
---|
| 58 | for phone in phones:
|
---|
| 59 | offer_url = 'https://www.neptun.mk' + phone.find('a').get('href')
|
---|
| 60 | offer_name = phone.find('a').find('h2').get_text().replace('MOB.TEL.', '').strip()
|
---|
| 61 | brand = offer_name.split(' ')[0].strip().capitalize()
|
---|
| 62 | image_url = 'https://www.neptun.mk' + phone.find('a').find('div', {'class': 'row'}).find('img').get('src')
|
---|
| 63 | price = int(
|
---|
| 64 | phone.find('div', {'class': 'col-sm-12 static'}).find('div', {'class': 'product-list-item__prices pt35'})
|
---|
| 65 | .find('div', {'class': 'row'}).find('div', {'class': 'newPriceModel'}) \
|
---|
| 66 | .find('span', {'class': 'product-price__amount--value ng-binding'}).get_text().replace('.', ''))
|
---|
| 67 |
|
---|
| 68 | driver1 = webdriver.Safari(executable_path='/usr/bin/safaridriver')
|
---|
| 69 | driver1.get(offer_url)
|
---|
| 70 | offer_html = driver1.page_source
|
---|
| 71 | # closing the driver so the safari instance can pair with another webdriver session
|
---|
| 72 | driver1.close()
|
---|
| 73 |
|
---|
| 74 | soup2 = BeautifulSoup(offer_html, 'html.parser')
|
---|
| 75 |
|
---|
| 76 | offer_shop_code = soup2.find('div', {'ng-if': 'showProductDetails'}) \
|
---|
| 77 | .find('div', {'class': 'product-details-first-row'}).find('span', {
|
---|
| 78 | 'ng-bind': 'model.CodeNumber'}).get_text().strip()
|
---|
| 79 |
|
---|
| 80 | specifications_table = \
|
---|
| 81 | soup2.find('div', {'id': 'mainContainer'}).find('div', {'ng-if': 'showProductDetails'}).find_all('ul')[-1]
|
---|
| 82 | specifications = specifications_table.get_text(separator='\n').strip().split("\n")
|
---|
| 83 |
|
---|
| 84 | offer_description = specifications_table.get_text(separator='\n').strip()
|
---|
| 85 |
|
---|
[895cd87] | 86 | back_camera = None
|
---|
[b68ae8d] | 87 | operating_system = None
|
---|
| 88 | chipset = None
|
---|
| 89 | battery = None
|
---|
| 90 | ram_memory = None
|
---|
| 91 | rom_memory = None
|
---|
| 92 | cpu = None
|
---|
[895cd87] | 93 | front_camera = None
|
---|
| 94 | color = None
|
---|
| 95 |
|
---|
[b68ae8d] | 96 | for specification in specifications:
|
---|
| 97 | if 'Батерија:' in specification:
|
---|
| 98 | battery = specification.split('Батерија:')[1]
|
---|
| 99 |
|
---|
| 100 | if 'CPU:' in specification:
|
---|
| 101 | cpu = specification.split('CPU:')[1]
|
---|
| 102 |
|
---|
| 103 | if 'Chipset:' in specification:
|
---|
| 104 | chipset = specification.split('Chipset:')[1]
|
---|
| 105 |
|
---|
| 106 | if 'RAM Меморија:' in specification:
|
---|
| 107 | ram_memory = specification.split('RAM Меморија:')[1]
|
---|
| 108 | continue
|
---|
| 109 |
|
---|
| 110 | if 'ROM Меморија:' in specification:
|
---|
| 111 | rom_memory = specification.split('ROM Меморија:')[1]
|
---|
| 112 | continue
|
---|
| 113 |
|
---|
| 114 | if 'ROM:' in specification:
|
---|
| 115 | rom_memory = specification.split('ROM:')[1]
|
---|
| 116 |
|
---|
| 117 | if 'RAM:' in specification:
|
---|
| 118 | ram_memory = specification.split('RAM:')[1]
|
---|
| 119 |
|
---|
| 120 | if 'iOS' in specification or 'Android' in specification:
|
---|
| 121 | operating_system = specification
|
---|
| 122 |
|
---|
[895cd87] | 123 | new_offers.append(PhoneOffer(offer_shop, offer_name, price, ram_memory, rom_memory,
|
---|
| 124 | color, front_camera, back_camera, chipset, battery, operating_system, cpu,
|
---|
| 125 | image_url,
|
---|
| 126 | offer_url, last_updated, is_validated, offer_description, offer_shop_code))
|
---|
| 127 |
|
---|
| 128 | for new_offer in new_offers:
|
---|
| 129 | flag = False
|
---|
| 130 | flag_price = False
|
---|
| 131 | offer_id = None
|
---|
| 132 |
|
---|
| 133 | for old_offer in database_offers:
|
---|
| 134 |
|
---|
| 135 | if new_offer.offer_shop_code == old_offer.offer_shop_code:
|
---|
| 136 | flag = True
|
---|
| 137 | if new_offer.price != old_offer.price:
|
---|
| 138 | flag_price = True
|
---|
| 139 | offer_id = old_offer.offer_id
|
---|
| 140 |
|
---|
| 141 | if flag:
|
---|
| 142 | # print('ALREADY IN DATABASE')
|
---|
| 143 | # print(new_offer)
|
---|
| 144 | # if it's already in database, check PRICE and if it's changed, change it !!!!!!
|
---|
| 145 | if flag_price:
|
---|
| 146 | print('PRICE CHANGED!') # CHANGE PRICE
|
---|
| 147 | print('offer id: ' + str(offer_id))
|
---|
| 148 | headers = {'Content-type': 'application/json'}
|
---|
| 149 | requests.put('http://localhost:8080/phoneoffer/' + str(offer_id) + '/changeprice/' + str(new_offer.price),
|
---|
| 150 | headers=headers)
|
---|
| 151 | else:
|
---|
| 152 | print('ADDED') # ADD OFFER
|
---|
| 153 | print(new_offer)
|
---|
| 154 | headers = {'Content-type': 'application/json'}
|
---|
| 155 | requests.post('http://localhost:8080/phoneoffer/addoffer',
|
---|
| 156 | headers=headers, data=json.dumps(new_offer.__dict__, default=str))
|
---|
| 157 |
|
---|
| 158 | print('------------------------------------')
|
---|
| 159 |
|
---|
| 160 | for old_offer in database_offers:
|
---|
| 161 | flag = False
|
---|
| 162 | for new_offer in new_offers:
|
---|
| 163 | if old_offer.offer_shop_code == new_offer.offer_shop_code:
|
---|
| 164 | flag = True
|
---|
| 165 |
|
---|
| 166 | if not flag:
|
---|
| 167 | print('OFFER DELETED')
|
---|
| 168 | print(old_offer)
|
---|
| 169 | # DELETE OFFER
|
---|
| 170 | requests.delete('http://localhost:8080/phoneoffer/deleteoffer/' + str(old_offer.offer_id))
|
---|