source: phonelux_scrappers/scrappers/ledikom_scrapper.py@ 895cd87

Last change on this file since 895cd87 was 895cd87, checked in by Marko <Marko@…>, 21 months ago

Refactored code

  • Property mode set to 100644
File size: 6.8 KB
Line 
1import json
2import unicodedata
3from datetime import datetime
4import psycopg2
5import config_read
6from bs4 import BeautifulSoup
7from selenium import webdriver
8import requests
9
10import sys
11
12from classes.phoneoffer import PhoneOffer
13
14file_path = 'outputfile.txt'
15sys.stdout = open(file_path, "w")
16
17offer_shop = "Ledikom" # offer shop
18last_updated = datetime.now().date()
19is_validated = False
20
21# Ledikom phone offers that are already in database
22
23offers = json.loads(unicodedata.normalize('NFKD', requests.get('http://localhost:8080/phoneoffer/shop/ledikom').text))
24
25database_offers = []
26
27for 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
38new_offers = []
39
40ledikom_phone_urls = [
41 'https://ledikom.mk/c/416/uredi/apple/iphone?limit=96',
42 'https://ledikom.mk/c/421/uredi/samsung/telefoni?limit=96',
43 'https://ledikom.mk/c/424/mobilni-telefoni/xiaomi/telefoni?limit=96',
44 'https://ledikom.mk/c/430/uredi/huawei/telefoni?limit=96',
45 'https://ledikom.mk/c/441/uredi/oneplus/telefoni?limit=96',
46 'https://ledikom.mk/c/413/uredi/google/telefoni?limit=96',
47 'https://ledikom.mk/c/411/uredi/honor/telefoni?limit=96',
48 'https://ledikom.mk/c/460/uredi/nokia/telefoni?limit=96',
49 'https://ledikom.mk/c/461/uredi/asus/telefoni?limit=96',
50 'https://ledikom.mk/c/488/proizvodi/oppo/telefoni?limit=96'
51]
52
53for ledikom_url in ledikom_phone_urls:
54
55 # selenium is used because of the dynamic content of the page
56 driver1 = webdriver.Safari(executable_path='/usr/bin/safaridriver')
57 driver1.get(ledikom_url)
58 ledikom_html = driver1.page_source
59
60 # closing the driver so the safari instance can pair with another webdriver session
61 driver1.close()
62
63 soup1 = BeautifulSoup(ledikom_html, 'html.parser')
64
65 phones = soup1.find('div', {'id': 'content'}) \
66 .find('div', {'class': 'container'}).find('div', {'class': 'row'}).find('div', {'class': 'item-display'}) \
67 .find_all('div', {'class': 'item-in-grid'})
68
69 if len(phones) == 0:
70 continue
71
72 for phone in phones:
73 offer_url = 'https://ledikom.mk' + phone.find('a').get('href')
74 image_url = phone.find('a').find('img').get('src')
75 temp_offer_name = phone.find('div', {'class': 'item-name'}).find('a').get_text().strip()
76 offer_name = ' '.join(temp_offer_name.split())
77 brand = offer_name.split(' ')[0]
78 price = int(phone.find('span', {'class': 'price'}).get_text().replace('ден.', '')
79 .replace('ден', '')
80 .replace('.', '').strip())
81
82 driver1 = webdriver.Safari(executable_path='/usr/bin/safaridriver')
83 driver1.get(offer_url)
84 # getting offer page html
85 offer_html = driver1.page_source
86 driver1.close()
87
88 soup2 = BeautifulSoup(offer_html, 'html.parser')
89
90 specifications = soup2.find('div', {'id': 'content'}).find('section', {'class': 'padding-section'}) \
91 .find_all('div', {'class': 'container'})[1].find('div', {'class': 'col-md-7'}) \
92 .find_all('div', {'class': 'row'})
93
94 color = None
95 rom_memory = None
96 ram_memory = None
97 back_camera = None
98 operating_system = None
99 chipset = None
100 battery = None
101 cpu = None
102 front_camera = None
103 offer_shop_code = None
104 offer_description = None
105
106 if len(specifications) != 0:
107 colors_tags = specifications[0].find('div', {'class': 'col-md-12 col-xs-12'}).find_all('a')
108 temp_colors = []
109 for color_tag in colors_tags:
110 temp_colors.append(color_tag.get_text().strip())
111 color = ','.join(temp_colors)
112
113 if len(specifications) >= 2:
114 temp_rom = specifications[1].find('div', {'class': 'col-md-12 col-xs-12'}).find_all('a')
115 rom_list = []
116 for rom in temp_rom:
117 rom_list.append(rom.get('title'))
118 rom_memory = ','.join(rom_list)
119
120 if len(specifications) >= 3:
121 temp_ram = specifications[2].find('div', {'class': 'col-md-12 col-xs-12'}).find_all('a')
122 ram_list = []
123 for ram in temp_ram:
124 ram_list.append(ram.get('title'))
125
126 ram_memory = ','.join(ram_list)
127
128 if 'Xiaomi' in brand:
129 temp = color
130 color = rom_memory
131 rom_memory = temp
132
133 temp = ram_memory
134 ram_memory = color
135 color = temp
136
137 new_offers.append(PhoneOffer(offer_shop, offer_name, price, ram_memory, rom_memory,
138 color, front_camera, back_camera, chipset, battery, operating_system, cpu,
139 image_url,
140 offer_url, last_updated, is_validated, offer_description, offer_shop_code))
141
142for new_offer in new_offers:
143 flag = False
144 flag_price = False
145 offer_id = None
146
147 for old_offer in database_offers:
148
149 if new_offer.offer_name == old_offer.offer_name:
150 flag = True
151 if new_offer.price != old_offer.price:
152 flag_price = True
153 offer_id = old_offer.offer_id
154
155 if flag:
156 # print('ALREADY IN DATABASE')
157 # print(new_offer)
158 # if it's already in database, check PRICE and if it's changed, change it !!!!!!
159 if flag_price:
160 print('PRICE CHANGED!') # CHANGE PRICE
161 print('offer id: ' + str(offer_id))
162 headers = {'Content-type': 'application/json'}
163 requests.put('http://localhost:8080/phoneoffer/' + str(offer_id) + '/changeprice/' + str(new_offer.price),
164 headers=headers)
165 else:
166 print('ADDED') # ADD OFFER
167 print(new_offer)
168 headers = {'Content-type': 'application/json'}
169 requests.post('http://localhost:8080/phoneoffer/addoffer',
170 headers=headers, data=json.dumps(new_offer.__dict__, default=str))
171
172print('------------------------------------')
173
174for old_offer in database_offers:
175 flag = False
176 for new_offer in new_offers:
177 if old_offer.offer_name == new_offer.offer_name:
178 flag = True
179
180 if not flag:
181 print('OFFER DELETED')
182 print(old_offer)
183 # DELETE OFFER
184 requests.delete('http://localhost:8080/phoneoffer/deleteoffer/' + str(old_offer.offer_id))
Note: See TracBrowser for help on using the repository browser.