[b68ae8d] | 1 | import time
|
---|
| 2 | from datetime import datetime
|
---|
| 3 | import psycopg2
|
---|
| 4 | from selenium import webdriver
|
---|
| 5 | from selenium.webdriver.common.by import By
|
---|
| 6 | from selenium.webdriver.support.ui import WebDriverWait
|
---|
| 7 | from selenium.webdriver.support import expected_conditions as EC
|
---|
| 8 |
|
---|
| 9 | import config_read
|
---|
| 10 | from bs4 import BeautifulSoup
|
---|
| 11 | import requests
|
---|
| 12 | import unicodedata
|
---|
| 13 | import sys
|
---|
| 14 |
|
---|
| 15 | file_path = '../outputfile.txt'
|
---|
| 16 | sys.stdout = open(file_path, "w")
|
---|
| 17 |
|
---|
| 18 | # Call to read the configuration file and connect to database
|
---|
| 19 | cinfo = config_read.get_databaseconfig("../postgresdb.config")
|
---|
| 20 | db_connection = psycopg2.connect(
|
---|
| 21 | database=cinfo[0],
|
---|
| 22 | host=cinfo[1],
|
---|
| 23 | user=cinfo[2],
|
---|
| 24 | password=cinfo[3]
|
---|
| 25 | )
|
---|
| 26 | cur = db_connection.cursor()
|
---|
| 27 |
|
---|
| 28 |
|
---|
| 29 | def scrape_function(driver1, i):
|
---|
| 30 | offer_shop = "Anhoch" # offer shop
|
---|
| 31 | last_updated = datetime.now().date()
|
---|
| 32 | is_validated = False
|
---|
| 33 |
|
---|
| 34 | anhoch_html = driver1.page_source
|
---|
| 35 | soup1 = BeautifulSoup(anhoch_html, 'html.parser')
|
---|
| 36 | active_li = soup1.find('div', {'class': 'adjust-elems pagination pagination-centered'}).find('li', {'class': 'active'})
|
---|
| 37 |
|
---|
| 38 | li_element = int(active_li.get_text().strip())
|
---|
| 39 |
|
---|
| 40 | print('page: '+str(li_element))
|
---|
| 41 |
|
---|
| 42 | if int(active_li.get_text().strip()) == i:
|
---|
| 43 | phones = soup1.find('section', {'id': 'main'}).find('div', {'class': 'span8'}) \
|
---|
| 44 | .find('div', {'class': 'products'}).find_all('li')
|
---|
| 45 | for phone in phones:
|
---|
| 46 | offer_url = phone.find('a').get('href')
|
---|
| 47 | image_url = phone.find('a').find('img').get('src')
|
---|
| 48 | offer_name = phone.find('div', {'class': 'product-name'}).find('a').get_text().strip()
|
---|
| 49 | price = int(phone.get('data-price'))
|
---|
| 50 | brand = phone.find('div', {'class': 'product-price'}).find_all('div')[2].find('strong').get_text().strip()
|
---|
| 51 |
|
---|
| 52 | response2 = requests.get(offer_url)
|
---|
| 53 | soup2 = BeautifulSoup(response2.content, 'html.parser')
|
---|
| 54 | offer_shop_code = soup2.find('div', {'class': 'product-desc'}).get_text().strip().split('\n')[3]
|
---|
| 55 |
|
---|
| 56 | offer_description = soup2.find('div', {'class': 'description'}) \
|
---|
| 57 | .find('div', {'class': 'tab-content'}).find('pre').get_text().strip()
|
---|
| 58 |
|
---|
| 59 | print(offer_name)
|
---|
| 60 | print(brand)
|
---|
| 61 | print()
|
---|
| 62 | print()
|
---|
| 63 |
|
---|
| 64 | # insert_script = 'INSERT INTO phone_offers (offer_shop, brand, offer_name , price, image_url, offer_url,' \
|
---|
| 65 | # 'offer_shop_code, offer_description, last_updated, is_validated)' \
|
---|
| 66 | # ' VALUES (%s, %s, %s, %s, %s, %s, %s, %s, %s, %s);'
|
---|
| 67 | # insert_value = (offer_shop, brand, offer_name, price, image_url, offer_url,
|
---|
| 68 | # offer_shop_code, offer_description, last_updated, is_validated)
|
---|
| 69 | # cur.execute(insert_script, insert_value)
|
---|
| 70 | # db_connection.commit()
|
---|
| 71 | else:
|
---|
| 72 | driver1.implicitly_wait(5)
|
---|
| 73 | scrape_function(driver1, i)
|
---|
| 74 |
|
---|
| 75 |
|
---|
| 76 | for i in range(1, 19):
|
---|
| 77 | anhoch_url = "https://www.anhoch.com/category/3017/smartfoni-i-mobilni-tel#page/"+str(i)
|
---|
| 78 | # print(anhoch_url)
|
---|
| 79 |
|
---|
| 80 | # selenium is used because of the dynamic content of the page
|
---|
| 81 | driver1 = webdriver.Safari(executable_path='/usr/bin/safaridriver')
|
---|
| 82 | driver1.get(anhoch_url)
|
---|
| 83 |
|
---|
| 84 | scrape_function(driver1, i)
|
---|
| 85 | # closing the driver so the safari instance can pair with another webdriver session
|
---|
| 86 | driver1.close()
|
---|
| 87 |
|
---|
| 88 |
|
---|
| 89 | cur.close()
|
---|
| 90 | db_connection.close()
|
---|