Changeset 895cd87 for phonelux_scrappers/scrappers/tehnomarket_scrapper.py

Timestamp:

10/01/22 22:55:27 (21 months ago)

Author:

Marko <Marko@…>

Branches:

master

Children:

fd5b100

Parents:

48f3030

Message:

Refactored code

File:

• phonelux_scrappers/scrappers/tehnomarket_scrapper.py (modified) (4 diffs)

phonelux_scrappers/scrappers/tehnomarket_scrapper.py

@@ +1 @@
+import json
 import unicodedata
 from datetime import datetime
@@ -6 +7 @@
 from selenium import webdriver
 import requests
+import sys
 
-import sys
+from classes.phoneoffer import PhoneOffer
 
 file_path = 'outputfile.txt'
 sys.stdout = open(file_path, "w")
 
-# Call to read the configuration file and connect to database
-cinfo = config_read.get_databaseconfig("../postgresdb.config")
-db_connection = psycopg2.connect(
-    database=cinfo[0],
-    host=cinfo[1],
-    user=cinfo[2],
-    password=cinfo[3]
-)
-cur = db_connection.cursor()
 
-
-def scrape_function(driver1, i):
+def scrape_function(driver1, i, new_offers):
     offer_shop = "Tehnomarket"  # offer shop
     last_updated = datetime.now().date()
@@ -59 +51 @@
             offer_shop_code = details[4].strip()
 
+            back_camera = None
+            operating_system = None
+            chipset = None
+            battery = None
+            ram_memory = None
+            rom_memory = None
+            cpu = None
+            front_camera = None
+            color = None
+
             specifications = []
             for info in soup2.find_all('span', {'class': 'info'}):
                 specifications.append(info.get_text())
 
-            print(brand)
-            print(offer_name)
-            print()
-            print()
-
             offer_description = '\n'.join(specifications)
 
-            insert_script = 'INSERT INTO phone_offers (offer_shop, brand, offer_name, price, image_url, offer_url,' \
-                            'offer_description, offer_shop_code, last_updated, is_validated)' \
-                            ' VALUES (%s, %s, %s, %s, %s, %s, %s, %s, %s, %s);'
-            insert_value = (offer_shop, brand, offer_name, price, image_url, offer_url, offer_description,
-                            offer_shop_code, last_updated, is_validated)
-            cur.execute(insert_script, insert_value)
-            db_connection.commit()
+            new_offers.append(PhoneOffer(offer_shop, offer_name, price, ram_memory, rom_memory,
+                                         color, front_camera, back_camera, chipset, battery, operating_system, cpu,
+                                         image_url,
+                                         offer_url, last_updated, is_validated, offer_description, offer_shop_code))
     else:
         driver1.implicitly_wait(30)
-        scrape_function(driver1, i)
+        scrape_function(driver1, i, new_offers)
 
+
+# Tehnomarket phone offers that are already in database
+
+offers = json.loads(
+    unicodedata.normalize('NFKD', requests.get('http://localhost:8080/phoneoffer/shop/tehnomarket').text))
+
+database_offers = []
+
+for offer in offers:
+    phoneOffer = PhoneOffer(offer['id'], offer['offer_shop'], offer['offer_name'], offer['price'],
+                            offer['ram_memory'],
+                            offer['rom_memory'], offer['color'], offer['front_camera'], offer['back_camera'],
+                            offer['chipset'], offer['battery'], offer['operating_system'], offer['cpu'],
+                            offer['image_url'],
+                            offer['offer_url'], offer['last_updated'], offer['is_validated'],
+                            offer['offer_description'],
+                            offer['offer_shop_code'])
+    database_offers.append(phoneOffer)
+
+new_offers = []
 
 for i in range(1, 6):
@@ -90 +104 @@
     driver1.get(tehnomarket_url)
 
-    scrape_function(driver1, i)
+    scrape_function(driver1, i, new_offers)
+
     # closing the driver so the safari instance can pair with another webdriver session
     driver1.close()
 
-cur.close()
-db_connection.close()
+for new_offer in new_offers:
+    flag = False
+    flag_price = False
+    offer_id = None
+
+    for old_offer in database_offers:
+
+        if new_offer.offer_shop_code == old_offer.offer_shop_code:
+            flag = True
+            if new_offer.price != old_offer.price:
+                flag_price = True
+                offer_id = old_offer.offer_id
+
+    if flag:
+        # print('ALREADY IN DATABASE')
+        # print(new_offer)
+        # if it's already in database, check PRICE and if it's changed, change it !!!!!!
+        if flag_price:
+            print('PRICE CHANGED!')  # CHANGE PRICE
+            print('offer id: ' + str(offer_id))
+            headers = {'Content-type': 'application/json'}
+            requests.put('http://localhost:8080/phoneoffer/' + str(offer_id) + '/changeprice/' + str(new_offer.price),
+                         headers=headers)
+    else:
+        print('ADDED')  # ADD OFFER
+        print(new_offer)
+        headers = {'Content-type': 'application/json'}
+        requests.post('http://localhost:8080/phoneoffer/addoffer',
+                      headers=headers, data=json.dumps(new_offer.__dict__, default=str))
+
+print('------------------------------------')
+
+for old_offer in database_offers:
+    flag = False
+    for new_offer in new_offers:
+        if old_offer.offer_shop_code == new_offer.offer_shop_code:
+            flag = True
+
+    if not flag:
+        print('OFFER DELETED')
+        print(old_offer)
+        # DELETE OFFER
+        requests.delete('http://localhost:8080/phoneoffer/deleteoffer/' + str(old_offer.offer_id))