[b68ae8d] | 1 | from datetime import datetime
|
---|
| 2 |
|
---|
| 3 | import psycopg2
|
---|
| 4 | import config_read
|
---|
| 5 | from bs4 import BeautifulSoup
|
---|
| 6 | import requests
|
---|
| 7 |
|
---|
| 8 | # import sys
|
---|
| 9 | #
|
---|
| 10 | # file_path = 'outputfile.txt'
|
---|
| 11 | # sys.stdout = open(file_path, "w")
|
---|
| 12 |
|
---|
| 13 | # Call to read the configuration file and connect to database
|
---|
| 14 | cinfo = config_read.get_databaseconfig("../postgresdb.config")
|
---|
| 15 | db_connection = psycopg2.connect(
|
---|
| 16 | database=cinfo[0],
|
---|
| 17 | host=cinfo[1],
|
---|
| 18 | user=cinfo[2],
|
---|
| 19 | password=cinfo[3]
|
---|
| 20 | )
|
---|
| 21 | cur = db_connection.cursor()
|
---|
| 22 |
|
---|
| 23 | offer_shop = "Mobi Go" # offer shop
|
---|
| 24 | last_updated = datetime.now().date()
|
---|
| 25 | is_validated = False
|
---|
| 26 | for i in range(1, 6):
|
---|
| 27 | mobigo_url = "https://mobigo.mk/page/" + str(i) + "/"
|
---|
| 28 |
|
---|
| 29 | response1 = requests.get(mobigo_url)
|
---|
| 30 |
|
---|
| 31 | soup1 = BeautifulSoup(response1.content, 'html.parser')
|
---|
| 32 |
|
---|
| 33 | phone_sections = soup1.find_all('ul', {'class': 'recent-posts'})
|
---|
| 34 | phones = phone_sections[len(phone_sections) - 1].find_all('li')
|
---|
| 35 |
|
---|
| 36 | for phone in phones:
|
---|
| 37 | offer_url = phone.find('div', {'class', 'post-thumb'}).find('a').get('href') # offer url
|
---|
| 38 | image_url = phone.find('div', {'class', 'post-thumb'}).find('a').find('img').get('src') # image url
|
---|
| 39 | offer_name = phone.find('div', {'class', 'post-content'}).find_all('h2')[0].get_text().strip() # offer_name
|
---|
| 40 |
|
---|
| 41 | if "Watch" in offer_name or "Tab" in offer_name: # if the product is watch or tablet, continue
|
---|
| 42 | continue
|
---|
| 43 |
|
---|
| 44 | price = int(float(phone.find('div', {'class', 'post-content'}).find_all('h2')[1] \
|
---|
| 45 | .get_text().replace('ден.', '').replace('.', '').strip())) # price
|
---|
| 46 |
|
---|
| 47 | response2 = requests.get(offer_url)
|
---|
| 48 | soup2 = BeautifulSoup(response2.content, 'html.parser')
|
---|
| 49 |
|
---|
| 50 | brand = soup2.find('a', {'rel': 'category tag'}).get_text().strip() # brand
|
---|
| 51 |
|
---|
| 52 | if brand not in offer_name:
|
---|
| 53 | offer_name = brand + " " + offer_name
|
---|
| 54 |
|
---|
| 55 | specifications = soup2.find('table', {'id': 'singlet'}).find_all('tr')
|
---|
| 56 |
|
---|
| 57 | ram_memory = ""
|
---|
| 58 | rom_memory = ""
|
---|
| 59 | battery = ""
|
---|
| 60 | back_camera = ""
|
---|
| 61 | front_camera = ""
|
---|
| 62 | chipset = ""
|
---|
| 63 | operating_system = ""
|
---|
| 64 |
|
---|
| 65 | for specification in specifications:
|
---|
| 66 | if specification.find('td') == None:
|
---|
| 67 | continue
|
---|
| 68 |
|
---|
| 69 | # operating system
|
---|
| 70 | if specification.find('td').get_text() == "Платформа":
|
---|
| 71 | if specification.find('i').get_text() != "/":
|
---|
| 72 | operating_system = specification.find('i').get_text().strip()
|
---|
| 73 | else:
|
---|
| 74 | operating_system = None
|
---|
| 75 |
|
---|
| 76 | # chipset
|
---|
| 77 | if specification.find('td').get_text() == "Chipset":
|
---|
| 78 | if specification.find('i').get_text() != "/":
|
---|
| 79 | chipset = specification.find('i').get_text().strip()
|
---|
| 80 | else:
|
---|
| 81 | chipset = None
|
---|
| 82 |
|
---|
| 83 | # ram and rom memory
|
---|
| 84 | if specification.find('td').get_text() == "Меморија":
|
---|
| 85 | if specification.find('i').get_text() != "/":
|
---|
| 86 | rom_memory = specification.find('i').get_text().replace(',', '').split(' ')[0].strip()
|
---|
| 87 | ram_memory = specification.find('i').get_text().replace(',', '').split(' ')[1].strip()
|
---|
| 88 | else:
|
---|
| 89 | rom_memory = None
|
---|
| 90 | ram_memory = None
|
---|
| 91 |
|
---|
| 92 | # back camera
|
---|
| 93 | if specification.find('td').get_text() == "Главна Камера":
|
---|
| 94 | if specification.find('i').get_text() != "/":
|
---|
| 95 | back_camera = specification.find('i').get_text().strip()
|
---|
| 96 | else:
|
---|
| 97 | back_camera = None
|
---|
| 98 |
|
---|
| 99 | # front camera
|
---|
| 100 | if specification.find('td').get_text() == "Селфи Камера":
|
---|
| 101 | if specification.find('i').get_text() != "/":
|
---|
| 102 | front_camera = specification.find('i').get_text().strip()
|
---|
| 103 | else:
|
---|
| 104 | front_camera = None
|
---|
| 105 |
|
---|
| 106 | # battery
|
---|
| 107 | if specification.find('td').get_text() == "Батерија":
|
---|
| 108 | if specification.find('i').get_text() != "/":
|
---|
| 109 | battery = specification.find('i').get_text().strip()
|
---|
| 110 | else:
|
---|
| 111 | battery = None
|
---|
| 112 |
|
---|
| 113 | insert_script = 'INSERT INTO phone_offers (offer_shop, brand, offer_name, price, image_url, offer_url, ram_memory,' \
|
---|
| 114 | ' rom_memory, battery, back_camera, front_camera, chipset, operating_system, last_updated, is_validated)' \
|
---|
| 115 | ' VALUES (%s, %s, %s, %s, %s, %s, %s, %s, %s, %s, %s, %s, %s, %s, %s);'
|
---|
| 116 | insert_value = (offer_shop, brand, offer_name, price, image_url, offer_url, ram_memory,
|
---|
| 117 | rom_memory, battery, back_camera, front_camera, chipset, operating_system, last_updated, is_validated)
|
---|
| 118 | cur.execute(insert_script, insert_value)
|
---|
| 119 | db_connection.commit()
|
---|
| 120 |
|
---|
| 121 | cur.close()
|
---|
| 122 | db_connection.close()
|
---|