| 1 | import json
|
|---|
| 2 | import traceback
|
|---|
| 3 | import unicodedata
|
|---|
| 4 | from datetime import datetime
|
|---|
| 5 | import psycopg2
|
|---|
| 6 | import config_read
|
|---|
| 7 | from bs4 import BeautifulSoup
|
|---|
| 8 | from selenium import webdriver
|
|---|
| 9 | import requests
|
|---|
| 10 |
|
|---|
| 11 | import sys
|
|---|
| 12 |
|
|---|
| 13 | from classes.phoneoffer import PhoneOffer
|
|---|
| 14 |
|
|---|
| 15 | file_path = 'outputfile.txt'
|
|---|
| 16 | sys.stdout = open(file_path, "w")
|
|---|
| 17 |
|
|---|
| 18 | offer_shop = "Neptun" # offer shop
|
|---|
| 19 | last_updated = datetime.now().date()
|
|---|
| 20 | is_validated = False
|
|---|
| 21 |
|
|---|
| 22 | # Call to read the configuration file and connect to database
|
|---|
| 23 | cinfo = config_read.get_databaseconfig("../postgresdb.config")
|
|---|
| 24 | db_connection = psycopg2.connect(
|
|---|
| 25 | database=cinfo[0],
|
|---|
| 26 | host=cinfo[1],
|
|---|
| 27 | user=cinfo[2],
|
|---|
| 28 | password=cinfo[3]
|
|---|
| 29 | )
|
|---|
| 30 | cur = db_connection.cursor()
|
|---|
| 31 |
|
|---|
| 32 | try:
|
|---|
| 33 | # Neptun phone offers that are already in database
|
|---|
| 34 | offers = json.loads(unicodedata.normalize('NFKD', requests.get('http://localhost:8080/phoneoffer/shop/neptun').text))
|
|---|
| 35 |
|
|---|
| 36 | database_offers = []
|
|---|
| 37 |
|
|---|
| 38 | for offer in offers:
|
|---|
| 39 | phoneOffer = PhoneOffer(offer['id'], offer['offer_shop'], offer['offer_name'], offer['price'],
|
|---|
| 40 | offer['ram_memory'],
|
|---|
| 41 | offer['rom_memory'], offer['color'], offer['front_camera'], offer['back_camera'],
|
|---|
| 42 | offer['chipset'], offer['battery'], offer['operating_system'], offer['cpu'],
|
|---|
| 43 | offer['image_url'],
|
|---|
| 44 | offer['offer_url'], offer['last_updated'], offer['is_validated'],
|
|---|
| 45 | offer['offer_description'],
|
|---|
| 46 | offer['offer_shop_code'])
|
|---|
| 47 | database_offers.append(phoneOffer)
|
|---|
| 48 |
|
|---|
| 49 | new_offers = []
|
|---|
| 50 |
|
|---|
| 51 | for i in range(1, 11):
|
|---|
| 52 | neptun_url = 'https://www.neptun.mk/mobilni_telefoni.nspx?page=' + str(i)
|
|---|
| 53 |
|
|---|
| 54 | # selenium is used because of the dynamic content of the page
|
|---|
| 55 | driver1 = webdriver.Safari(executable_path='/usr/bin/safaridriver')
|
|---|
| 56 | driver1.get(neptun_url)
|
|---|
| 57 | neptun_html = driver1.page_source
|
|---|
| 58 |
|
|---|
| 59 | # closing the driver so the safari instance can pair with another webdriver session
|
|---|
| 60 | driver1.close()
|
|---|
| 61 |
|
|---|
| 62 | # response1 = requests.get(neptun_url)
|
|---|
| 63 | soup1 = BeautifulSoup(neptun_html, 'html.parser')
|
|---|
| 64 |
|
|---|
| 65 | phones = soup1.find('div', {'id': 'mainContainer'}).find('div',
|
|---|
| 66 | {'class': 'col-lg-9 col-md-9 col-sm-8 col-fix-main'}) \
|
|---|
| 67 | .find_all('div', {'class': 'ng-scope product-list-item-grid'})
|
|---|
| 68 |
|
|---|
| 69 | for phone in phones:
|
|---|
| 70 | offer_url = 'https://www.neptun.mk' + phone.find('a').get('href')
|
|---|
| 71 | offer_name = phone.find('a').find('h2').get_text().replace('MOB.TEL.', '').strip()
|
|---|
| 72 | brand = offer_name.split(' ')[0].strip().capitalize()
|
|---|
| 73 | image_url = 'https://www.neptun.mk' + phone.find('a').find('div', {'class': 'row'}).find('img').get('src')
|
|---|
| 74 | price = int(
|
|---|
| 75 | phone.find('div', {'class': 'col-sm-12 static'}).find('div', {'class': 'product-list-item__prices pt35'})
|
|---|
| 76 | .find('div', {'class': 'row'}).find('div', {'class': 'newPriceModel'}) \
|
|---|
| 77 | .find('span', {'class': 'product-price__amount--value ng-binding'}).get_text().replace('.', ''))
|
|---|
| 78 |
|
|---|
| 79 | driver1 = webdriver.Safari(executable_path='/usr/bin/safaridriver')
|
|---|
| 80 | driver1.get(offer_url)
|
|---|
| 81 | offer_html = driver1.page_source
|
|---|
| 82 | # closing the driver so the safari instance can pair with another webdriver session
|
|---|
| 83 | driver1.close()
|
|---|
| 84 |
|
|---|
| 85 | soup2 = BeautifulSoup(offer_html, 'html.parser')
|
|---|
| 86 |
|
|---|
| 87 | offer_shop_code = soup2.find('div', {'ng-if': 'showProductDetails'}) \
|
|---|
| 88 | .find('div', {'class': 'product-details-first-row'}).find('span', {
|
|---|
| 89 | 'ng-bind': 'model.CodeNumber'}).get_text().strip()
|
|---|
| 90 |
|
|---|
| 91 | specifications_table = \
|
|---|
| 92 | soup2.find('div', {'id': 'mainContainer'}).find('div', {'ng-if': 'showProductDetails'}).find_all('ul')[-1]
|
|---|
| 93 | specifications = specifications_table.get_text(separator='\n').strip().split("\n")
|
|---|
| 94 |
|
|---|
| 95 | offer_description = specifications_table.get_text(separator='\n').strip()
|
|---|
| 96 |
|
|---|
| 97 | back_camera = None
|
|---|
| 98 | operating_system = None
|
|---|
| 99 | chipset = None
|
|---|
| 100 | battery = None
|
|---|
| 101 | ram_memory = None
|
|---|
| 102 | rom_memory = None
|
|---|
| 103 | cpu = None
|
|---|
| 104 | front_camera = None
|
|---|
| 105 | color = None
|
|---|
| 106 |
|
|---|
| 107 | for specification in specifications:
|
|---|
| 108 | if 'Батерија:' in specification:
|
|---|
| 109 | battery = specification.split('Батерија:')[1]
|
|---|
| 110 |
|
|---|
| 111 | if 'CPU:' in specification:
|
|---|
| 112 | cpu = specification.split('CPU:')[1]
|
|---|
| 113 |
|
|---|
| 114 | if 'Chipset:' in specification:
|
|---|
| 115 | chipset = specification.split('Chipset:')[1]
|
|---|
| 116 |
|
|---|
| 117 | if 'RAM Меморија:' in specification:
|
|---|
| 118 | ram_memory = specification.split('RAM Меморија:')[1]
|
|---|
| 119 | continue
|
|---|
| 120 |
|
|---|
| 121 | if 'ROM Меморија:' in specification:
|
|---|
| 122 | rom_memory = specification.split('ROM Меморија:')[1]
|
|---|
| 123 | continue
|
|---|
| 124 |
|
|---|
| 125 | if 'ROM:' in specification:
|
|---|
| 126 | rom_memory = specification.split('ROM:')[1]
|
|---|
| 127 |
|
|---|
| 128 | if 'RAM:' in specification:
|
|---|
| 129 | ram_memory = specification.split('RAM:')[1]
|
|---|
| 130 |
|
|---|
| 131 | if 'iOS' in specification or 'Android' in specification:
|
|---|
| 132 | operating_system = specification
|
|---|
| 133 |
|
|---|
| 134 | new_offers.append(PhoneOffer(offer_shop, offer_name, price, ram_memory, rom_memory,
|
|---|
| 135 | color, front_camera, back_camera, chipset, battery, operating_system, cpu,
|
|---|
| 136 | image_url,
|
|---|
| 137 | offer_url, last_updated, is_validated, offer_description, offer_shop_code))
|
|---|
| 138 |
|
|---|
| 139 | for new_offer in new_offers:
|
|---|
| 140 | flag = False
|
|---|
| 141 | flag_price = False
|
|---|
| 142 | offer_id = None
|
|---|
| 143 |
|
|---|
| 144 | for old_offer in database_offers:
|
|---|
| 145 |
|
|---|
| 146 | if new_offer.offer_shop_code == old_offer.offer_shop_code:
|
|---|
| 147 | flag = True
|
|---|
| 148 | if new_offer.price != old_offer.price:
|
|---|
| 149 | flag_price = True
|
|---|
| 150 | offer_id = old_offer.offer_id
|
|---|
| 151 |
|
|---|
| 152 | if flag:
|
|---|
| 153 | # print('ALREADY IN DATABASE')
|
|---|
| 154 | # print(new_offer)
|
|---|
| 155 | # if it's already in database, check PRICE and if it's changed, change it !!!!!!
|
|---|
| 156 | if flag_price:
|
|---|
| 157 | print('PRICE CHANGED!') # CHANGE PRICE
|
|---|
| 158 | print('offer id: ' + str(offer_id))
|
|---|
| 159 | headers = {'Content-type': 'application/json'}
|
|---|
| 160 | requests.put('http://localhost:8080/phoneoffer/' + str(offer_id) + '/changeprice/' + str(new_offer.price),
|
|---|
| 161 | headers=headers)
|
|---|
| 162 | else:
|
|---|
| 163 | print('ADDED') # ADD OFFER
|
|---|
| 164 | print(new_offer)
|
|---|
| 165 | headers = {'Content-type': 'application/json'}
|
|---|
| 166 | requests.post('http://localhost:8080/phoneoffer/addoffer',
|
|---|
| 167 | headers=headers, data=json.dumps(new_offer.__dict__, default=str))
|
|---|
| 168 |
|
|---|
| 169 | print('------------------------------------')
|
|---|
| 170 |
|
|---|
| 171 | for old_offer in database_offers:
|
|---|
| 172 | flag = False
|
|---|
| 173 | for new_offer in new_offers:
|
|---|
| 174 | if old_offer.offer_shop_code == new_offer.offer_shop_code:
|
|---|
| 175 | flag = True
|
|---|
| 176 |
|
|---|
| 177 | if not flag:
|
|---|
| 178 | print('OFFER DELETED')
|
|---|
| 179 | print(old_offer)
|
|---|
| 180 | # DELETE OFFER
|
|---|
| 181 | requests.delete('http://localhost:8080/phoneoffer/deleteoffer/' + str(old_offer.offer_id))
|
|---|
| 182 | except Exception:
|
|---|
| 183 | traceback.print_exc()
|
|---|
| 184 | insert_script = 'INSERT INTO scrapper_info (store, recieved_at, status)' \
|
|---|
| 185 | ' VALUES (%s, %s, %s);'
|
|---|
| 186 | insert_value = (offer_shop, last_updated, 'failed')
|
|---|
| 187 | cur.execute(insert_script, insert_value)
|
|---|
| 188 | db_connection.commit()
|
|---|
| 189 | cur.close()
|
|---|
| 190 | db_connection.close()
|
|---|
| 191 | else:
|
|---|
| 192 | insert_script = 'INSERT INTO scrapper_info (store, recieved_at, status)' \
|
|---|
| 193 | ' VALUES (%s, %s, %s);'
|
|---|
| 194 | insert_value = (offer_shop, last_updated, 'success')
|
|---|
| 195 | cur.execute(insert_script, insert_value)
|
|---|
| 196 | db_connection.commit()
|
|---|
| 197 | cur.close()
|
|---|
| 198 | db_connection.close()
|
|---|
