| 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 | import sys
|
|---|
| 11 |
|
|---|
| 12 | from classes.phoneoffer import PhoneOffer
|
|---|
| 13 |
|
|---|
| 14 | file_path = 'outputfile.txt'
|
|---|
| 15 | sys.stdout = open(file_path, "w")
|
|---|
| 16 |
|
|---|
| 17 |
|
|---|
| 18 | def scrape_function(driver1, i, new_offers):
|
|---|
| 19 | offer_shop = "Tehnomarket" # offer shop
|
|---|
| 20 | last_updated = datetime.now().date()
|
|---|
| 21 | is_validated = False
|
|---|
| 22 |
|
|---|
| 23 | tehnomarket_html = driver1.page_source
|
|---|
| 24 | soup1 = BeautifulSoup(tehnomarket_html, 'html.parser')
|
|---|
| 25 | active_li = soup1.find('div', {'class': 'adjust-elems pagination pagination-centered'}).find('li',
|
|---|
| 26 | {'class': 'active'})
|
|---|
| 27 |
|
|---|
| 28 | print('page: ' + active_li.get_text())
|
|---|
| 29 |
|
|---|
| 30 | if int(active_li.get_text().strip()) == i:
|
|---|
| 31 | phones = soup1.find('ul', {'class': 'products products-display-grid thumbnails'}).find_all('li', {
|
|---|
| 32 | 'class': 'span4 product-fix'})
|
|---|
| 33 |
|
|---|
| 34 | for phone in phones:
|
|---|
| 35 | offer_url = phone.find('a').get('href')
|
|---|
| 36 | offer_name = phone.find('div', {'class': 'product-name'}).get_text().strip()
|
|---|
| 37 | price = int(phone.find('div', {'class': 'product-price clearfix'}).find('strong') \
|
|---|
| 38 | .get_text().replace('ден.', '').replace(',', '').strip())
|
|---|
| 39 |
|
|---|
| 40 | response2 = requests.get(offer_url)
|
|---|
| 41 | soup2 = BeautifulSoup(response2.content, 'html.parser')
|
|---|
| 42 |
|
|---|
| 43 | image = soup2.find('div', {'id': 'product_gallery'}).find('img')
|
|---|
| 44 |
|
|---|
| 45 | image_url = None
|
|---|
| 46 | if image is not None:
|
|---|
| 47 | image_url = image.get('src')
|
|---|
| 48 |
|
|---|
| 49 | details = soup2.find('div', {'class': 'product-desc'}).get_text().split('\n')
|
|---|
| 50 |
|
|---|
| 51 | brand = details[2].strip().capitalize()
|
|---|
| 52 | offer_shop_code = details[4].strip()
|
|---|
| 53 |
|
|---|
| 54 | back_camera = None
|
|---|
| 55 | operating_system = None
|
|---|
| 56 | chipset = None
|
|---|
| 57 | battery = None
|
|---|
| 58 | ram_memory = None
|
|---|
| 59 | rom_memory = None
|
|---|
| 60 | cpu = None
|
|---|
| 61 | front_camera = None
|
|---|
| 62 | color = None
|
|---|
| 63 |
|
|---|
| 64 | specifications = []
|
|---|
| 65 | for info in soup2.find_all('span', {'class': 'info'}):
|
|---|
| 66 | specifications.append(info.get_text())
|
|---|
| 67 |
|
|---|
| 68 | offer_description = '\n'.join(specifications)
|
|---|
| 69 |
|
|---|
| 70 | new_offers.append(PhoneOffer(offer_shop, offer_name, price, ram_memory, rom_memory,
|
|---|
| 71 | color, front_camera, back_camera, chipset, battery, operating_system, cpu,
|
|---|
| 72 | image_url,
|
|---|
| 73 | offer_url, last_updated, is_validated, offer_description, offer_shop_code))
|
|---|
| 74 | else:
|
|---|
| 75 | driver1.implicitly_wait(30)
|
|---|
| 76 | scrape_function(driver1, i, new_offers)
|
|---|
| 77 |
|
|---|
| 78 |
|
|---|
| 79 | # Call to read the configuration file and connect to database
|
|---|
| 80 | cinfo = config_read.get_databaseconfig("../postgresdb.config")
|
|---|
| 81 | db_connection = psycopg2.connect(
|
|---|
| 82 | database=cinfo[0],
|
|---|
| 83 | host=cinfo[1],
|
|---|
| 84 | user=cinfo[2],
|
|---|
| 85 | password=cinfo[3]
|
|---|
| 86 | )
|
|---|
| 87 | cur = db_connection.cursor()
|
|---|
| 88 |
|
|---|
| 89 | try:
|
|---|
| 90 | # Tehnomarket phone offers that are already in database
|
|---|
| 91 | offers = json.loads(
|
|---|
| 92 | unicodedata.normalize('NFKD', requests.get('http://localhost:8080/phoneoffer/shop/tehnomarket').text))
|
|---|
| 93 |
|
|---|
| 94 | database_offers = []
|
|---|
| 95 |
|
|---|
| 96 | for offer in offers:
|
|---|
| 97 | phoneOffer = PhoneOffer(offer['id'], offer['offer_shop'], offer['offer_name'], offer['price'],
|
|---|
| 98 | offer['ram_memory'],
|
|---|
| 99 | offer['rom_memory'], offer['color'], offer['front_camera'], offer['back_camera'],
|
|---|
| 100 | offer['chipset'], offer['battery'], offer['operating_system'], offer['cpu'],
|
|---|
| 101 | offer['image_url'],
|
|---|
| 102 | offer['offer_url'], offer['last_updated'], offer['is_validated'],
|
|---|
| 103 | offer['offer_description'],
|
|---|
| 104 | offer['offer_shop_code'])
|
|---|
| 105 | database_offers.append(phoneOffer)
|
|---|
| 106 |
|
|---|
| 107 | new_offers = []
|
|---|
| 108 |
|
|---|
| 109 | for i in range(1, 6):
|
|---|
| 110 | tehnomarket_url = 'https://tehnomarket.com.mk/category/4109/mobilni-telefoni#page/' + str(i)
|
|---|
| 111 | # print(anhoch_url)
|
|---|
| 112 |
|
|---|
| 113 | # selenium is used because of the dynamic content of the page
|
|---|
| 114 | driver1 = webdriver.Safari(executable_path='/usr/bin/safaridriver')
|
|---|
| 115 | driver1.get(tehnomarket_url)
|
|---|
| 116 |
|
|---|
| 117 | scrape_function(driver1, i, new_offers)
|
|---|
| 118 |
|
|---|
| 119 | # closing the driver so the safari instance can pair with another webdriver session
|
|---|
| 120 | driver1.close()
|
|---|
| 121 |
|
|---|
| 122 | for new_offer in new_offers:
|
|---|
| 123 | flag = False
|
|---|
| 124 | flag_price = False
|
|---|
| 125 | offer_id = None
|
|---|
| 126 |
|
|---|
| 127 | for old_offer in database_offers:
|
|---|
| 128 |
|
|---|
| 129 | if new_offer.offer_shop_code == old_offer.offer_shop_code:
|
|---|
| 130 | flag = True
|
|---|
| 131 | if new_offer.price != old_offer.price:
|
|---|
| 132 | flag_price = True
|
|---|
| 133 | offer_id = old_offer.offer_id
|
|---|
| 134 |
|
|---|
| 135 | if flag:
|
|---|
| 136 | # print('ALREADY IN DATABASE')
|
|---|
| 137 | # print(new_offer)
|
|---|
| 138 | # if it's already in database, check PRICE and if it's changed, change it !!!!!!
|
|---|
| 139 | if flag_price:
|
|---|
| 140 | print('PRICE CHANGED!') # CHANGE PRICE
|
|---|
| 141 | print('offer id: ' + str(offer_id))
|
|---|
| 142 | headers = {'Content-type': 'application/json'}
|
|---|
| 143 | requests.put('http://localhost:8080/phoneoffer/' + str(offer_id) + '/changeprice/' + str(new_offer.price),
|
|---|
| 144 | headers=headers)
|
|---|
| 145 | else:
|
|---|
| 146 | print('ADDED') # ADD OFFER
|
|---|
| 147 | print(new_offer)
|
|---|
| 148 | headers = {'Content-type': 'application/json'}
|
|---|
| 149 | requests.post('http://localhost:8080/phoneoffer/addoffer',
|
|---|
| 150 | headers=headers, data=json.dumps(new_offer.__dict__, default=str))
|
|---|
| 151 |
|
|---|
| 152 | print('------------------------------------')
|
|---|
| 153 |
|
|---|
| 154 | for old_offer in database_offers:
|
|---|
| 155 | flag = False
|
|---|
| 156 | for new_offer in new_offers:
|
|---|
| 157 | if old_offer.offer_shop_code == new_offer.offer_shop_code:
|
|---|
| 158 | flag = True
|
|---|
| 159 |
|
|---|
| 160 | if not flag:
|
|---|
| 161 | print('OFFER DELETED')
|
|---|
| 162 | print(old_offer)
|
|---|
| 163 | # DELETE OFFER
|
|---|
| 164 | requests.delete('http://localhost:8080/phoneoffer/deleteoffer/' + str(old_offer.offer_id))
|
|---|
| 165 | except Exception:
|
|---|
| 166 | traceback.print_exc()
|
|---|
| 167 | insert_script = 'INSERT INTO scrapper_info (store, recieved_at, status)' \
|
|---|
| 168 | ' VALUES (%s, %s, %s);'
|
|---|
| 169 | insert_value = ('Tehnomarket', datetime.now().date(), 'failed')
|
|---|
| 170 | cur.execute(insert_script, insert_value)
|
|---|
| 171 | db_connection.commit()
|
|---|
| 172 | cur.close()
|
|---|
| 173 | db_connection.close()
|
|---|
| 174 | else:
|
|---|
| 175 | insert_script = 'INSERT INTO scrapper_info (store, recieved_at, status)' \
|
|---|
| 176 | ' VALUES (%s, %s, %s);'
|
|---|
| 177 | insert_value = ('Tehnomarket', datetime.now().date(), 'success')
|
|---|
| 178 | cur.execute(insert_script, insert_value)
|
|---|
| 179 | db_connection.commit()
|
|---|
| 180 | cur.close()
|
|---|
| 181 | db_connection.close()
|
|---|
