[895cd87] | 1 | import json
|
---|
[b68ae8d] | 2 | from datetime import datetime
|
---|
| 3 |
|
---|
| 4 | import psycopg2
|
---|
| 5 | import config_read
|
---|
| 6 | from bs4 import BeautifulSoup
|
---|
| 7 | import requests
|
---|
| 8 | import unicodedata
|
---|
| 9 | import sys
|
---|
[895cd87] | 10 | from classes.phoneoffer import PhoneOffer
|
---|
[b68ae8d] | 11 |
|
---|
[895cd87] | 12 | file_path = 'outputfile.txt'
|
---|
| 13 | sys.stdout = open(file_path, "w")
|
---|
[b68ae8d] | 14 |
|
---|
| 15 |
|
---|
| 16 | offer_shop = "Akcija" # offer shop
|
---|
| 17 | last_updated = datetime.now().date()
|
---|
| 18 | is_validated = False
|
---|
| 19 |
|
---|
[895cd87] | 20 | # Akcija phone offers that are already in database
|
---|
| 21 |
|
---|
| 22 | offers = json.loads(unicodedata.normalize('NFKD', requests.get('http://localhost:8080/phoneoffer/shop/akcija').text))
|
---|
| 23 |
|
---|
| 24 | database_offers = []
|
---|
| 25 |
|
---|
| 26 | for offer in offers:
|
---|
| 27 | phoneOffer = PhoneOffer(offer['id'], offer['offer_shop'], offer['offer_name'], offer['price'],
|
---|
| 28 | offer['ram_memory'],
|
---|
| 29 | offer['rom_memory'], offer['color'], offer['front_camera'], offer['back_camera'],
|
---|
| 30 | offer['chipset'], offer['battery'], offer['operating_system'], offer['cpu'],
|
---|
| 31 | offer['image_url'],
|
---|
| 32 | offer['offer_url'], offer['last_updated'], offer['is_validated'],
|
---|
| 33 | offer['offer_description'],
|
---|
| 34 | offer['offer_shop_code'])
|
---|
| 35 | database_offers.append(phoneOffer)
|
---|
| 36 |
|
---|
| 37 | new_offers = []
|
---|
| 38 |
|
---|
[b68ae8d] | 39 | i = 0
|
---|
| 40 | while i <= 20:
|
---|
| 41 | akcija_url = "https://akcija.com.mk/listing/" + str(i) + "?category=mobilnitelefoni"
|
---|
| 42 | response1 = requests.get(akcija_url)
|
---|
| 43 | response1.encoding = 'utf-8'
|
---|
| 44 | soup1 = BeautifulSoup(response1.text, 'html.parser')
|
---|
| 45 |
|
---|
| 46 | phones = soup1.find_all('div', {'class', 'product-item__body pb-xl-2'})
|
---|
| 47 |
|
---|
| 48 | for phone in phones:
|
---|
| 49 | offer_name = phone.find('h5', {'class': 'mb-1 product-item__title'}).find('a') \
|
---|
| 50 | .get_text().replace('Паметен телефон', '').strip()
|
---|
| 51 | brand = offer_name.split(' ')[0]
|
---|
| 52 |
|
---|
| 53 | if brand not in offer_name:
|
---|
| 54 | offer_name = brand + " " + offer_name
|
---|
| 55 |
|
---|
| 56 | offer_url = phone.find('h5', {'class': 'mb-1 product-item__title'}).find('a').get('href')
|
---|
| 57 | image_url = phone.find('div', {'class', 'mb-2'}).find('img').get('src')
|
---|
| 58 | price = int(phone.find('div', {'class', 'flex-center-between mb-1 pt-xl-2'}) \
|
---|
| 59 | .find('ins').get_text().split(' ')[0].strip())
|
---|
| 60 |
|
---|
| 61 | response2 = requests.get(offer_url)
|
---|
| 62 | response2.encoding = 'utf-8'
|
---|
| 63 | soup2 = BeautifulSoup(response2.text, 'html.parser')
|
---|
| 64 |
|
---|
[895cd87] | 65 | back_camera = None
|
---|
| 66 | operating_system = None
|
---|
| 67 | chipset = None
|
---|
| 68 | battery = None
|
---|
| 69 | ram_memory = None
|
---|
| 70 | rom_memory = None
|
---|
| 71 | cpu = None
|
---|
| 72 | front_camera = None
|
---|
| 73 | color = None
|
---|
| 74 | offer_shop_code = None
|
---|
| 75 |
|
---|
[b68ae8d] | 76 | specifications = soup2.find('main', {'id': 'content'}) \
|
---|
| 77 | .find_all('div', {'class', 'container'})[1].find('div', {'class', 'mb-14'}) \
|
---|
| 78 | .find('div', {'class', 'col-md-6 col-lg-4 col-xl-4 mb-md-6 mb-lg-0'}).find_all('p')
|
---|
| 79 |
|
---|
| 80 | offer_description = ''
|
---|
| 81 | for specification in specifications:
|
---|
| 82 | if 'Код за нарачка' in str(specification.get_text(separator='\n').replace('NBSP', '').strip()):
|
---|
| 83 | continue
|
---|
| 84 | offer_description += unicodedata.normalize('NFKD',
|
---|
| 85 | str(specification.get_text(separator='\n').strip())) + "\n"
|
---|
| 86 |
|
---|
[895cd87] | 87 | new_offers.append(PhoneOffer(offer_shop, offer_name, price, ram_memory, rom_memory,
|
---|
| 88 | color, front_camera, back_camera, chipset, battery, operating_system, cpu,
|
---|
| 89 | image_url,
|
---|
| 90 | offer_url, last_updated, is_validated, offer_description, offer_shop_code))
|
---|
[b68ae8d] | 91 | i += 20
|
---|
| 92 |
|
---|
[895cd87] | 93 | for new_offer in new_offers:
|
---|
| 94 | flag = False
|
---|
| 95 | flag_price = False
|
---|
| 96 | offer_id = None
|
---|
| 97 |
|
---|
| 98 | for old_offer in database_offers:
|
---|
| 99 |
|
---|
| 100 | if new_offer.offer_name == old_offer.offer_name:
|
---|
| 101 | flag = True
|
---|
| 102 | if new_offer.price != old_offer.price:
|
---|
| 103 | flag_price = True
|
---|
| 104 | offer_id = old_offer.offer_id
|
---|
| 105 |
|
---|
| 106 | if flag:
|
---|
| 107 | # print('ALREADY IN DATABASE')
|
---|
| 108 | # print(new_offer)
|
---|
| 109 | # if it's already in database, check PRICE and if it's changed, change it !!!!!!
|
---|
| 110 | if flag_price:
|
---|
| 111 | print('PRICE CHANGED!') # CHANGE PRICE
|
---|
| 112 | print('offer id: ' + str(offer_id))
|
---|
| 113 | headers = {'Content-type': 'application/json'}
|
---|
| 114 | requests.put('http://localhost:8080/phoneoffer/' + str(offer_id) + '/changeprice/' + str(new_offer.price),
|
---|
| 115 | headers=headers)
|
---|
| 116 | else:
|
---|
| 117 | print('ADDED') # ADD OFFER
|
---|
| 118 | print(new_offer)
|
---|
| 119 | headers = {'Content-type': 'application/json'}
|
---|
| 120 | requests.post('http://localhost:8080/phoneoffer/addoffer',
|
---|
| 121 | headers=headers, data=json.dumps(new_offer.__dict__, default=str))
|
---|
| 122 |
|
---|
| 123 | print('------------------------------------')
|
---|
| 124 |
|
---|
| 125 | for old_offer in database_offers:
|
---|
| 126 | flag = False
|
---|
| 127 | for new_offer in new_offers:
|
---|
| 128 | if old_offer.offer_name == new_offer.offer_name:
|
---|
| 129 | flag = True
|
---|
| 130 |
|
---|
| 131 | if not flag:
|
---|
| 132 | print('OFFER DELETED')
|
---|
| 133 | print(old_offer)
|
---|
| 134 | # DELETE OFFER
|
---|
| 135 | requests.delete('http://localhost:8080/phoneoffer/deleteoffer/' + str(old_offer.offer_id))
|
---|