| 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 | # file_path = 'outputfile.txt'
|
|---|
| 10 | # sys.stdout = open(file_path, "w")
|
|---|
| 11 |
|
|---|
| 12 | # Call to read the configuration file and connect to database
|
|---|
| 13 | cinfo = config_read.get_databaseconfig("../postgresdb.config")
|
|---|
| 14 | db_connection = psycopg2.connect(
|
|---|
| 15 | database=cinfo[0],
|
|---|
| 16 | host=cinfo[1],
|
|---|
| 17 | user=cinfo[2],
|
|---|
| 18 | password=cinfo[3]
|
|---|
| 19 | )
|
|---|
| 20 | cur = db_connection.cursor()
|
|---|
| 21 |
|
|---|
| 22 | mobitech_url = "https://mobitech.mk/shop/"
|
|---|
| 23 |
|
|---|
| 24 | response1 = requests.get(mobitech_url)
|
|---|
| 25 |
|
|---|
| 26 | soup1 = BeautifulSoup(response1.content, 'html.parser')
|
|---|
| 27 |
|
|---|
| 28 | phones = soup1.find_all('div', {'class': 'jet-woo-products__inner-box'})
|
|---|
| 29 |
|
|---|
| 30 | offer_shop = "Mobitech" # offer shop
|
|---|
| 31 | is_validated = False
|
|---|
| 32 |
|
|---|
| 33 | for phone in phones:
|
|---|
| 34 | offer_url = phone.find('h5', {'class': 'jet-woo-product-title'}).find('a').get('href') # url
|
|---|
| 35 | image_url = phone.find('div', {'class': 'jet-woo-product-thumbnail'}).find('img').get('src') # image
|
|---|
| 36 | brand = phone.find_next('div', {'class': 'jet-woo-product-categories'}).find('a').get_text().strip() # brand
|
|---|
| 37 | offer_name = phone.find('h5', {'class': 'jet-woo-product-title'}).find('a').get_text().strip() # offer_name
|
|---|
| 38 | if brand not in offer_name:
|
|---|
| 39 | offer_name = brand+" "+offer_name
|
|---|
| 40 | temp_prices = phone.find('div', {'class': 'jet-woo-product-price'}).find_all('bdi')
|
|---|
| 41 | price = int(float(temp_prices[len(temp_prices) - 1].get_text().replace("ден", "").replace(",", "").strip())) # price
|
|---|
| 42 | last_updated = datetime.now().date() # offer last_updated date
|
|---|
| 43 |
|
|---|
| 44 | response2 = requests.get(offer_url)
|
|---|
| 45 | soup2 = BeautifulSoup(response2.content, 'html.parser')
|
|---|
| 46 |
|
|---|
| 47 | specifications = soup2.find_all('h2', {'class': 'elementor-heading-title elementor-size-default'})
|
|---|
| 48 |
|
|---|
| 49 | ram_memory = ""
|
|---|
| 50 | rom_memory = ""
|
|---|
| 51 | battery = ""
|
|---|
| 52 | back_camera = ""
|
|---|
| 53 | operating_system = ""
|
|---|
| 54 |
|
|---|
| 55 | for specification in specifications:
|
|---|
| 56 | # rom memory
|
|---|
| 57 | if specification.get_text().startswith("Меморија:"):
|
|---|
| 58 | rom_memory = specification.get_text().split("Меморија:")[1].strip()
|
|---|
| 59 | if rom_memory == "Нема" or rom_memory == "/":
|
|---|
| 60 | rom_memory = None
|
|---|
| 61 |
|
|---|
| 62 | # ram memory
|
|---|
| 63 | if specification.get_text().startswith("РАМ Меморија:"):
|
|---|
| 64 | ram_memory = specification.get_text().split("РАМ Меморија:")[1].strip()
|
|---|
| 65 | if ram_memory == "Нема" or ram_memory == "/":
|
|---|
| 66 | ram_memory = None
|
|---|
| 67 |
|
|---|
| 68 | # camera
|
|---|
| 69 | if specification.get_text().startswith("Камера:"):
|
|---|
| 70 | back_camera = specification.get_text().split("Камера:")[1].strip()
|
|---|
| 71 | if back_camera == "Нема":
|
|---|
| 72 | back_camera = None
|
|---|
| 73 |
|
|---|
| 74 | # operating system
|
|---|
| 75 | if specification.get_text().startswith("Оперативен систем:"):
|
|---|
| 76 | operating_system = specification.get_text().split("Оперативен систем:")[1].split(",")[0].strip()
|
|---|
| 77 | if operating_system == "Нема":
|
|---|
| 78 | operating_system = None
|
|---|
| 79 |
|
|---|
| 80 | # battery
|
|---|
| 81 | if specification.get_text().startswith("Батерија:"):
|
|---|
| 82 | battery = specification.get_text().split("Батерија:")[1].strip()
|
|---|
| 83 | if battery == "Нема":
|
|---|
| 84 | battery = None
|
|---|
| 85 |
|
|---|
| 86 | insert_script = 'INSERT INTO phone_offers (offer_shop, brand, offer_name, price, image_url, offer_url, ram_memory,' \
|
|---|
| 87 | ' rom_memory, battery, back_camera, last_updated, operating_system, is_validated)' \
|
|---|
| 88 | ' VALUES (%s, %s, %s, %s, %s, %s, %s, %s, %s, %s, %s, %s, %s);'
|
|---|
| 89 | insert_value = (offer_shop, brand, offer_name, price, image_url, offer_url, ram_memory,
|
|---|
| 90 | rom_memory, battery, back_camera, last_updated, operating_system, is_validated)
|
|---|
| 91 | cur.execute(insert_script, insert_value)
|
|---|
| 92 | db_connection.commit()
|
|---|
| 93 |
|
|---|
| 94 | cur.close()
|
|---|
| 95 | db_connection.close()
|
|---|
