[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 | import sys
|
---|
| 10 |
|
---|
[895cd87] | 11 | from classes.phoneoffer import PhoneOffer
|
---|
| 12 |
|
---|
[b68ae8d] | 13 | file_path = 'outputfile.txt'
|
---|
| 14 | sys.stdout = open(file_path, "w")
|
---|
| 15 |
|
---|
| 16 |
|
---|
[895cd87] | 17 | def scrape_function(driver1, i, new_offers):
|
---|
[b68ae8d] | 18 | offer_shop = "Tehnomarket" # offer shop
|
---|
| 19 | last_updated = datetime.now().date()
|
---|
| 20 | is_validated = False
|
---|
| 21 |
|
---|
| 22 | tehnomarket_html = driver1.page_source
|
---|
| 23 | soup1 = BeautifulSoup(tehnomarket_html, 'html.parser')
|
---|
| 24 | active_li = soup1.find('div', {'class': 'adjust-elems pagination pagination-centered'}).find('li',
|
---|
| 25 | {'class': 'active'})
|
---|
| 26 |
|
---|
| 27 | print('page: ' + active_li.get_text())
|
---|
| 28 |
|
---|
| 29 | if int(active_li.get_text().strip()) == i:
|
---|
| 30 | phones = soup1.find('ul', {'class': 'products products-display-grid thumbnails'}).find_all('li', {
|
---|
| 31 | 'class': 'span4 product-fix'})
|
---|
| 32 |
|
---|
| 33 | for phone in phones:
|
---|
| 34 | offer_url = phone.find('a').get('href')
|
---|
| 35 | offer_name = phone.find('div', {'class': 'product-name'}).get_text().strip()
|
---|
| 36 | price = int(phone.find('div', {'class': 'product-price clearfix'}).find('strong') \
|
---|
| 37 | .get_text().replace('ден.', '').replace(',', '').strip())
|
---|
| 38 |
|
---|
| 39 | response2 = requests.get(offer_url)
|
---|
| 40 | soup2 = BeautifulSoup(response2.content, 'html.parser')
|
---|
| 41 |
|
---|
| 42 | image = soup2.find('div', {'id': 'product_gallery'}).find('img')
|
---|
| 43 |
|
---|
| 44 | image_url = None
|
---|
| 45 | if image is not None:
|
---|
| 46 | image_url = image.get('src')
|
---|
| 47 |
|
---|
| 48 | details = soup2.find('div', {'class': 'product-desc'}).get_text().split('\n')
|
---|
| 49 |
|
---|
| 50 | brand = details[2].strip().capitalize()
|
---|
| 51 | offer_shop_code = details[4].strip()
|
---|
| 52 |
|
---|
[895cd87] | 53 | back_camera = None
|
---|
| 54 | operating_system = None
|
---|
| 55 | chipset = None
|
---|
| 56 | battery = None
|
---|
| 57 | ram_memory = None
|
---|
| 58 | rom_memory = None
|
---|
| 59 | cpu = None
|
---|
| 60 | front_camera = None
|
---|
| 61 | color = None
|
---|
| 62 |
|
---|
[b68ae8d] | 63 | specifications = []
|
---|
| 64 | for info in soup2.find_all('span', {'class': 'info'}):
|
---|
| 65 | specifications.append(info.get_text())
|
---|
| 66 |
|
---|
| 67 | offer_description = '\n'.join(specifications)
|
---|
| 68 |
|
---|
[895cd87] | 69 | new_offers.append(PhoneOffer(offer_shop, offer_name, price, ram_memory, rom_memory,
|
---|
| 70 | color, front_camera, back_camera, chipset, battery, operating_system, cpu,
|
---|
| 71 | image_url,
|
---|
| 72 | offer_url, last_updated, is_validated, offer_description, offer_shop_code))
|
---|
[b68ae8d] | 73 | else:
|
---|
| 74 | driver1.implicitly_wait(30)
|
---|
[895cd87] | 75 | scrape_function(driver1, i, new_offers)
|
---|
| 76 |
|
---|
[b68ae8d] | 77 |
|
---|
[895cd87] | 78 | # Tehnomarket phone offers that are already in database
|
---|
| 79 |
|
---|
| 80 | offers = json.loads(
|
---|
| 81 | unicodedata.normalize('NFKD', requests.get('http://localhost:8080/phoneoffer/shop/tehnomarket').text))
|
---|
| 82 |
|
---|
| 83 | database_offers = []
|
---|
| 84 |
|
---|
| 85 | for offer in offers:
|
---|
| 86 | phoneOffer = PhoneOffer(offer['id'], offer['offer_shop'], offer['offer_name'], offer['price'],
|
---|
| 87 | offer['ram_memory'],
|
---|
| 88 | offer['rom_memory'], offer['color'], offer['front_camera'], offer['back_camera'],
|
---|
| 89 | offer['chipset'], offer['battery'], offer['operating_system'], offer['cpu'],
|
---|
| 90 | offer['image_url'],
|
---|
| 91 | offer['offer_url'], offer['last_updated'], offer['is_validated'],
|
---|
| 92 | offer['offer_description'],
|
---|
| 93 | offer['offer_shop_code'])
|
---|
| 94 | database_offers.append(phoneOffer)
|
---|
| 95 |
|
---|
| 96 | new_offers = []
|
---|
[b68ae8d] | 97 |
|
---|
| 98 | for i in range(1, 6):
|
---|
| 99 | tehnomarket_url = 'https://tehnomarket.com.mk/category/4109/mobilni-telefoni#page/' + str(i)
|
---|
| 100 | # print(anhoch_url)
|
---|
| 101 |
|
---|
| 102 | # selenium is used because of the dynamic content of the page
|
---|
| 103 | driver1 = webdriver.Safari(executable_path='/usr/bin/safaridriver')
|
---|
| 104 | driver1.get(tehnomarket_url)
|
---|
| 105 |
|
---|
[895cd87] | 106 | scrape_function(driver1, i, new_offers)
|
---|
| 107 |
|
---|
[b68ae8d] | 108 | # closing the driver so the safari instance can pair with another webdriver session
|
---|
| 109 | driver1.close()
|
---|
| 110 |
|
---|
[895cd87] | 111 | for new_offer in new_offers:
|
---|
| 112 | flag = False
|
---|
| 113 | flag_price = False
|
---|
| 114 | offer_id = None
|
---|
| 115 |
|
---|
| 116 | for old_offer in database_offers:
|
---|
| 117 |
|
---|
| 118 | if new_offer.offer_shop_code == old_offer.offer_shop_code:
|
---|
| 119 | flag = True
|
---|
| 120 | if new_offer.price != old_offer.price:
|
---|
| 121 | flag_price = True
|
---|
| 122 | offer_id = old_offer.offer_id
|
---|
| 123 |
|
---|
| 124 | if flag:
|
---|
| 125 | # print('ALREADY IN DATABASE')
|
---|
| 126 | # print(new_offer)
|
---|
| 127 | # if it's already in database, check PRICE and if it's changed, change it !!!!!!
|
---|
| 128 | if flag_price:
|
---|
| 129 | print('PRICE CHANGED!') # CHANGE PRICE
|
---|
| 130 | print('offer id: ' + str(offer_id))
|
---|
| 131 | headers = {'Content-type': 'application/json'}
|
---|
| 132 | requests.put('http://localhost:8080/phoneoffer/' + str(offer_id) + '/changeprice/' + str(new_offer.price),
|
---|
| 133 | headers=headers)
|
---|
| 134 | else:
|
---|
| 135 | print('ADDED') # ADD OFFER
|
---|
| 136 | print(new_offer)
|
---|
| 137 | headers = {'Content-type': 'application/json'}
|
---|
| 138 | requests.post('http://localhost:8080/phoneoffer/addoffer',
|
---|
| 139 | headers=headers, data=json.dumps(new_offer.__dict__, default=str))
|
---|
| 140 |
|
---|
| 141 | print('------------------------------------')
|
---|
| 142 |
|
---|
| 143 | for old_offer in database_offers:
|
---|
| 144 | flag = False
|
---|
| 145 | for new_offer in new_offers:
|
---|
| 146 | if old_offer.offer_shop_code == new_offer.offer_shop_code:
|
---|
| 147 | flag = True
|
---|
| 148 |
|
---|
| 149 | if not flag:
|
---|
| 150 | print('OFFER DELETED')
|
---|
| 151 | print(old_offer)
|
---|
| 152 | # DELETE OFFER
|
---|
| 153 | requests.delete('http://localhost:8080/phoneoffer/deleteoffer/' + str(old_offer.offer_id))
|
---|