source: phonelux_scrappers/scrappers/ledikom_scrapper.py@ dfd5d87

Last change on this file since dfd5d87 was b68ae8d, checked in by Marko <Marko@…>, 23 months ago

Created spring app, edited setec_scrapper

  • Property mode set to 100644
File size: 4.6 KB
Line 
1import unicodedata
2from datetime import datetime
3
4import psycopg2
5import config_read
6from bs4 import BeautifulSoup
7from selenium import webdriver
8import requests
9
10import sys
11
12file_path = 'outputfile.txt'
13sys.stdout = open(file_path, "w")
14
15# Call to read the configuration file and connect to database
16cinfo = config_read.get_databaseconfig("../postgresdb.config")
17db_connection = psycopg2.connect(
18 database=cinfo[0],
19 host=cinfo[1],
20 user=cinfo[2],
21 password=cinfo[3]
22)
23cur = db_connection.cursor()
24
25offer_shop = "Ledikom" # offer shop
26last_updated = datetime.now().date()
27is_validated = False
28
29ledikom_phone_urls = [
30 'https://ledikom.mk/c/416/uredi/apple/iphone?limit=96',
31 'https://ledikom.mk/c/421/uredi/samsung/telefoni?limit=96',
32 'https://ledikom.mk/c/424/mobilni-telefoni/xiaomi/telefoni?limit=96',
33 'https://ledikom.mk/c/430/uredi/huawei/telefoni?limit=96',
34 'https://ledikom.mk/c/441/uredi/oneplus/telefoni?limit=96',
35 'https://ledikom.mk/c/413/uredi/google/telefoni?limit=96',
36 'https://ledikom.mk/c/411/uredi/honor/telefoni?limit=96',
37 'https://ledikom.mk/c/460/uredi/nokia/telefoni?limit=96',
38 'https://ledikom.mk/c/461/uredi/asus/telefoni?limit=96',
39 'https://ledikom.mk/c/488/proizvodi/oppo/telefoni?limit=96'
40]
41
42for ledikom_url in ledikom_phone_urls:
43
44 # selenium is used because of the dynamic content of the page
45 driver1 = webdriver.Safari(executable_path='/usr/bin/safaridriver')
46 driver1.get(ledikom_url)
47 ledikom_html = driver1.page_source
48
49 # closing the driver so the safari instance can pair with another webdriver session
50 driver1.close()
51
52 soup1 = BeautifulSoup(ledikom_html, 'html.parser')
53
54 phones = soup1.find('div', {'id': 'content'}) \
55 .find('div', {'class': 'container'}).find('div', {'class': 'row'}).find('div', {'class': 'item-display'}) \
56 .find_all('div', {'class': 'item-in-grid'})
57
58 if len(phones) == 0:
59 continue
60
61 for phone in phones:
62 offer_url = 'https://ledikom.mk' + phone.find('a').get('href')
63 image_url = phone.find('a').find('img').get('src')
64 temp_offer_name = phone.find('div', {'class': 'item-name'}).find('a').get_text().strip()
65 offer_name = ' '.join(temp_offer_name.split())
66 brand = offer_name.split(' ')[0]
67 price = int(phone.find('span', {'class': 'price'}).get_text().replace('ден.', '').replace('.', '').strip())
68
69 driver1 = webdriver.Safari(executable_path='/usr/bin/safaridriver')
70 driver1.get(offer_url)
71 # getting offer page html
72 offer_html = driver1.page_source
73 driver1.close()
74
75 soup2 = BeautifulSoup(offer_html, 'html.parser')
76
77 specifications = soup2.find('div', {'id': 'content'}).find('section', {'class': 'padding-section'}) \
78 .find_all('div', {'class': 'container'})[1].find('div', {'class': 'col-md-7'}) \
79 .find_all('div', {'class': 'row'})
80
81 color = None
82 rom_memory = None
83 ram_memory = None
84
85 if len(specifications) != 0:
86 colors_tags = specifications[0].find('div', {'class': 'col-md-12 col-xs-12'}).find_all('a')
87 temp_colors = []
88 for color_tag in colors_tags:
89 temp_colors.append(color_tag.get_text().strip())
90 color = ','.join(temp_colors)
91
92 if len(specifications) >= 2:
93 temp_rom = specifications[1].find('div', {'class': 'col-md-12 col-xs-12'}).find_all('a')
94 rom_list = []
95 for rom in temp_rom:
96 rom_list.append(rom.get('title'))
97 rom_memory = ','.join(rom_list)
98
99 if len(specifications) >= 3:
100 temp_ram = specifications[2].find('div', {'class': 'col-md-12 col-xs-12'}).find_all('a')
101 ram_list = []
102 for ram in temp_ram:
103 ram_list.append(ram.get('title'))
104
105 ram_memory = ','.join(ram_list)
106
107 if 'Xiaomi' in brand:
108 temp = color
109 color = rom_memory
110 rom_memory = temp
111
112 temp = ram_memory
113 ram_memory = color
114 color = temp
115
116 insert_script = 'INSERT INTO phone_offers (offer_shop, brand, offer_name, price, image_url, offer_url,' \
117 'ram_memory, rom_memory, color, last_updated, is_validated)' \
118 ' VALUES (%s, %s, %s, %s, %s, %s, %s, %s, %s, %s, %s);'
119 insert_value = (offer_shop, brand, offer_name, price, image_url, offer_url, ram_memory,
120 rom_memory, color, last_updated, is_validated)
121 cur.execute(insert_script, insert_value)
122 db_connection.commit()
123
124cur.close()
125db_connection.close()
Note: See TracBrowser for help on using the repository browser.