[895cd87] | 1 | import json
|
---|
[47f4eaf] | 2 | import traceback
|
---|
[b68ae8d] | 3 | import unicodedata
|
---|
| 4 | from datetime import datetime
|
---|
| 5 | import psycopg2
|
---|
| 6 | import config_read
|
---|
| 7 | from bs4 import BeautifulSoup
|
---|
| 8 | from selenium import webdriver
|
---|
| 9 | import requests
|
---|
| 10 |
|
---|
| 11 | import sys
|
---|
| 12 |
|
---|
[895cd87] | 13 | from classes.phoneoffer import PhoneOffer
|
---|
| 14 |
|
---|
[b68ae8d] | 15 | file_path = 'outputfile.txt'
|
---|
| 16 | sys.stdout = open(file_path, "w")
|
---|
| 17 |
|
---|
| 18 |
|
---|
| 19 | offer_shop = "Handy" # offer shop
|
---|
| 20 | last_updated = datetime.now().date()
|
---|
| 21 | is_validated = False
|
---|
| 22 |
|
---|
[47f4eaf] | 23 | # Call to read the configuration file and connect to database
|
---|
| 24 | cinfo = config_read.get_databaseconfig("../postgresdb.config")
|
---|
| 25 | db_connection = psycopg2.connect(
|
---|
| 26 | database=cinfo[0],
|
---|
| 27 | host=cinfo[1],
|
---|
| 28 | user=cinfo[2],
|
---|
| 29 | password=cinfo[3]
|
---|
| 30 | )
|
---|
| 31 | cur = db_connection.cursor()
|
---|
| 32 |
|
---|
| 33 | try:
|
---|
| 34 | # Handy phone offers that are already in database
|
---|
| 35 | offers = json.loads(unicodedata.normalize('NFKD', requests.get('http://localhost:8080/phoneoffer/shop/handy').text))
|
---|
| 36 |
|
---|
| 37 | database_offers = []
|
---|
| 38 |
|
---|
| 39 | for offer in offers:
|
---|
| 40 | phoneOffer = PhoneOffer(offer['id'], offer['offer_shop'], offer['offer_name'], offer['price'],
|
---|
| 41 | offer['ram_memory'],
|
---|
| 42 | offer['rom_memory'], offer['color'], offer['front_camera'], offer['back_camera'],
|
---|
| 43 | offer['chipset'], offer['battery'], offer['operating_system'], offer['cpu'],
|
---|
| 44 | offer['image_url'],
|
---|
| 45 | offer['offer_url'], offer['last_updated'], offer['is_validated'],
|
---|
| 46 | offer['offer_description'],
|
---|
| 47 | offer['offer_shop_code'])
|
---|
| 48 | database_offers.append(phoneOffer)
|
---|
| 49 |
|
---|
| 50 | new_offers = []
|
---|
| 51 |
|
---|
| 52 | handy_url = 'https://www.handy.mk/telefoni?page=6'
|
---|
| 53 |
|
---|
| 54 | response1 = requests.get(handy_url)
|
---|
| 55 | soup1 = BeautifulSoup(response1.content, 'html.parser')
|
---|
| 56 |
|
---|
| 57 | phones = soup1.find_all('li', {'data-hook': 'product-list-grid-item'})
|
---|
| 58 |
|
---|
| 59 | for phone in phones:
|
---|
| 60 | offer_url = phone.find('a').get('href')
|
---|
| 61 | offer_name = phone.find('div', {'data-hook': 'not-image-container'})\
|
---|
| 62 | .find('h3', {'data-hook': 'product-item-name'}).get_text().strip()
|
---|
| 63 | brand = offer_name.split(' ')[0].capitalize()
|
---|
| 64 | price = int(float(phone.find('div', {'data-hook': 'not-image-container'}).find('div', {'data-hook': "product-item-product-details"})\
|
---|
| 65 | .find('span', {'data-hook': 'product-item-price-to-pay'}).get_text().strip().replace('ден', '').replace('.', '').replace(',', '.')))
|
---|
| 66 |
|
---|
| 67 | response2 = requests.get(offer_url)
|
---|
| 68 | soup2 = BeautifulSoup(response2.text, 'html.parser')
|
---|
| 69 |
|
---|
| 70 | back_camera = None
|
---|
| 71 | operating_system = None
|
---|
| 72 | chipset = None
|
---|
| 73 | battery = None
|
---|
| 74 | ram_memory = None
|
---|
| 75 | rom_memory = None
|
---|
| 76 | cpu = None
|
---|
| 77 | front_camera = None
|
---|
| 78 | offer_shop_code = None
|
---|
| 79 | color = None
|
---|
| 80 | image_url = None
|
---|
| 81 |
|
---|
| 82 | color_section = soup2.find('section', {'data-hook': 'product-colors-title-section'})
|
---|
| 83 | if color_section is not None:
|
---|
| 84 | temp_colors = color_section.find('fieldset', {'class': 'ColorPickerbase3548966286__container'})\
|
---|
| 85 | .find_all('input', {'type': 'radio'})
|
---|
| 86 | colors_list = []
|
---|
| 87 | for temp_color in temp_colors:
|
---|
| 88 | colors_list.append(temp_color.get('aria-label'))
|
---|
| 89 | color = ','.join(colors_list)
|
---|
| 90 |
|
---|
| 91 | rows = soup2.find('div', {'data-hook': 'info-section-description'}).find_all('li')
|
---|
| 92 |
|
---|
| 93 | if len(rows) == 0:
|
---|
| 94 | rows = soup2.find('div', {'data-hook': 'info-section-description'}).find_all('tr')
|
---|
| 95 |
|
---|
| 96 | specifications = []
|
---|
| 97 |
|
---|
| 98 | for row in rows:
|
---|
| 99 | specifications.append(unicodedata.normalize('NFKD', row.get_text().strip()))
|
---|
| 100 |
|
---|
| 101 | offer_description = '\n'.join(specifications)
|
---|
| 102 |
|
---|
| 103 | new_offers.append(PhoneOffer(offer_shop, offer_name, price, ram_memory, rom_memory,
|
---|
| 104 | color, front_camera, back_camera, chipset, battery, operating_system, cpu,
|
---|
| 105 | image_url,
|
---|
| 106 | offer_url, last_updated, is_validated, offer_description, offer_shop_code))
|
---|
| 107 |
|
---|
| 108 | for new_offer in new_offers:
|
---|
| 109 | flag = False
|
---|
| 110 | flag_price = False
|
---|
| 111 | offer_id = None
|
---|
| 112 |
|
---|
| 113 | for old_offer in database_offers:
|
---|
| 114 |
|
---|
| 115 | if new_offer.offer_name == old_offer.offer_name:
|
---|
| 116 | flag = True
|
---|
| 117 | if new_offer.price != old_offer.price:
|
---|
| 118 | flag_price = True
|
---|
| 119 | offer_id = old_offer.offer_id
|
---|
| 120 |
|
---|
| 121 | if flag:
|
---|
| 122 | # print('ALREADY IN DATABASE')
|
---|
| 123 | # print(new_offer)
|
---|
| 124 | # if it's already in database, check PRICE and if it's changed, change it !!!!!!
|
---|
| 125 | if flag_price:
|
---|
| 126 | print('PRICE CHANGED!') # CHANGE PRICE
|
---|
| 127 | print('offer id: ' + str(offer_id))
|
---|
| 128 | headers = {'Content-type': 'application/json'}
|
---|
| 129 | requests.put('http://localhost:8080/phoneoffer/' + str(offer_id) + '/changeprice/' + str(new_offer.price),
|
---|
| 130 | headers=headers)
|
---|
| 131 | else:
|
---|
| 132 | print('ADDED') # ADD OFFER
|
---|
| 133 | print(new_offer)
|
---|
| 134 | headers = {'Content-type': 'application/json'}
|
---|
| 135 | requests.post('http://localhost:8080/phoneoffer/addoffer',
|
---|
| 136 | headers=headers, data=json.dumps(new_offer.__dict__, default=str))
|
---|
| 137 |
|
---|
| 138 | print('------------------------------------')
|
---|
[895cd87] | 139 |
|
---|
| 140 | for old_offer in database_offers:
|
---|
[47f4eaf] | 141 | flag = False
|
---|
| 142 | for new_offer in new_offers:
|
---|
| 143 | if old_offer.offer_name == new_offer.offer_name:
|
---|
| 144 | flag = True
|
---|
| 145 |
|
---|
| 146 | if not flag:
|
---|
| 147 | print('OFFER DELETED')
|
---|
| 148 | print(old_offer)
|
---|
| 149 | # DELETE OFFER
|
---|
| 150 | requests.delete('http://localhost:8080/phoneoffer/deleteoffer/' + str(old_offer.offer_id))
|
---|
| 151 | except Exception:
|
---|
| 152 | traceback.print_exc()
|
---|
| 153 | insert_script = 'INSERT INTO scrapper_info (store, recieved_at, status)' \
|
---|
| 154 | ' VALUES (%s, %s, %s);'
|
---|
| 155 | insert_value = (offer_shop, last_updated, 'failed')
|
---|
| 156 | cur.execute(insert_script, insert_value)
|
---|
| 157 | db_connection.commit()
|
---|
| 158 | cur.close()
|
---|
| 159 | db_connection.close()
|
---|
| 160 | else:
|
---|
| 161 | insert_script = 'INSERT INTO scrapper_info (store, recieved_at, status)' \
|
---|
| 162 | ' VALUES (%s, %s, %s);'
|
---|
| 163 | insert_value = (offer_shop, last_updated, 'success')
|
---|
| 164 | cur.execute(insert_script, insert_value)
|
---|
| 165 | db_connection.commit()
|
---|
| 166 | cur.close()
|
---|
| 167 | db_connection.close()
|
---|
[895cd87] | 168 |
|
---|
| 169 |
|
---|
[b68ae8d] | 170 |
|
---|