Changeset 47f4eaf for phonelux_scrappers/scrappers/tehnomarket_scrapper.py

Timestamp:

11/20/22 16:34:52 (20 months ago)

Author:

Marko <Marko@…>

Branches:

master

Parents:

ffd50db

Message:

Final features implemented

File:

• phonelux_scrappers/scrappers/tehnomarket_scrapper.py (modified) (2 diffs)

phonelux_scrappers/scrappers/tehnomarket_scrapper.py

@@ -1 +1 @@
 import json
+import traceback
 import unicodedata
 from datetime import datetime
@@ -76 +77 @@
 
 
-# Tehnomarket phone offers that are already in database
+# 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()
 
-offers = json.loads(
-    unicodedata.normalize('NFKD', requests.get('http://localhost:8080/phoneoffer/shop/tehnomarket').text))
+try:
+    # 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 = []
+    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)
+    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 = []
+    new_offers = []
 
-for i in range(1, 6):
-    tehnomarket_url = 'https://tehnomarket.com.mk/category/4109/mobilni-telefoni#page/' + str(i)
-    # print(anhoch_url)
+    for i in range(1, 6):
+        tehnomarket_url = 'https://tehnomarket.com.mk/category/4109/mobilni-telefoni#page/' + str(i)
+        # print(anhoch_url)
 
-    # selenium is used because of the dynamic content of the page
-    driver1 = webdriver.Safari(executable_path='/usr/bin/safaridriver')
-    driver1.get(tehnomarket_url)
+        # selenium is used because of the dynamic content of the page
+        driver1 = webdriver.Safari(executable_path='/usr/bin/safaridriver')
+        driver1.get(tehnomarket_url)
 
-    scrape_function(driver1, i, new_offers)
+        scrape_function(driver1, i, new_offers)
 
-    # closing the driver so the safari instance can pair with another webdriver session
-    driver1.close()
+        # closing the driver so the safari instance can pair with another webdriver session
+        driver1.close()
 
-for new_offer in new_offers:
-    flag = False
-    flag_price = False
-    offer_id = None
+    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 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))
+        if not flag:
+            print('OFFER DELETED')
+            print(old_offer)
+            # DELETE OFFER
+            requests.delete('http://localhost:8080/phoneoffer/deleteoffer/' + str(old_offer.offer_id))
+except Exception:
+    traceback.print_exc()
+    insert_script = 'INSERT INTO scrapper_info (store, recieved_at, status)' \
+                    ' VALUES (%s, %s, %s);'
+    insert_value = ('Tehnomarket', datetime.now().date(), 'failed')
+    cur.execute(insert_script, insert_value)
+    db_connection.commit()
+    cur.close()
+    db_connection.close()
+else:
+    insert_script = 'INSERT INTO scrapper_info (store, recieved_at, status)' \
+                    ' VALUES (%s, %s, %s);'
+    insert_value = ('Tehnomarket', datetime.now().date(), 'success')
+    cur.execute(insert_script, insert_value)
+    db_connection.commit()
+    cur.close()
+    db_connection.close()