added Hasaki crawler

hasaki_crawler_engine/conf.json

@@ -21,6 +21,5 @@
   "database": "analytics",
   "db_host": "redshift-cluster-1.cdqj58hfx4p7.ap-southeast-1.redshift.amazonaws.com",
   "db_port": "5439",
-  "crawler_main": "1",
+  "log_loc": "/home/ubuntu/logs/hasaki_crawler.log"
-  "crawler_slave_no": ""
 }

hasaki_crawler_engine/hasaki_categories.py

@@ -10,21 +10,17 @@ from hasaki_db_writer import hasaki_db_writer
 from Util import translate_text_to_english
 
 
-###### Looger ######
-logname = '/home/ubuntu/logs/hasaki_crawler.log'
-#logname = 'hasaki_crawler.log'
-logging.basicConfig(filename=logname,
-                    filemode='a',
-                    format='%(asctime)s,%(msecs)d %(name)s %(levelname)s: %(message)s',
-                    datefmt="%Y-%m-%d %H:%M:%S",
-                    level=logging.INFO)
-
-
 
 
 
 class HasakiCategories:
     def __init__(self, config):
+        ###### Looger ######
+        logging.basicConfig(filename=config.get("log_loc"),
+                            filemode='a',
+                            format='%(asctime)s,%(msecs)d %(name)s %(levelname)s: %(message)s',
+                            datefmt="%Y-%m-%d %H:%M:%S",
+                            level=logging.INFO)
         logging.info("Initializing HasakiSubCategories")
         self.master_category = []
         self.config = config

hasaki_crawler_engine/hasaki_category_products.py

@@ -7,17 +7,15 @@ from playwright.sync_api import sync_playwright
 from hasaki_db_writer import hasaki_db_writer
 from Util import translate_text_to_english
 
-###### Looger ######
+
-logname = '/home/ubuntu/logs/hasaki_crawler.log'
+class HasakiCategoryProducts:
-#logname = 'hasaki_crawler.log'
+    def __init__(self, config):
-logging.basicConfig(filename=logname,
+        ###### Looger ######
+        logging.basicConfig(filename=config.get("log_loc"),
                             filemode='a',
                             format='%(asctime)s,%(msecs)d %(name)s %(levelname)s: %(message)s',
                             datefmt="%Y-%m-%d %H:%M:%S",
                             level=logging.INFO)
-
-class HasakiCategoryProducts:
-    def __init__(self, config):
         logging.info("Initializing HasakiCategoryProducts........")
         self.config = config
         self.crawler_name = self.config.get("crawler_name")

hasaki_crawler_engine/hasaki_crawler.py

@@ -13,16 +13,6 @@ from email.message import EmailMessage
 config = {}
 
 
-###### Looger ######
-logname = '/home/ubuntu/logs/hasaki_crawler.log'
-#logname = 'hasaki_crawler.log'
-logging.basicConfig(filename=logname,
-                    filemode='a',
-                    format='%(asctime)s,%(msecs)d %(name)s %(levelname)s: %(message)s',
-                    datefmt="%Y-%m-%d %H:%M:%S",
-                    level=logging.INFO)
-
-
 
 
 def main(cur):
@@ -181,6 +171,13 @@ if __name__ == "__main__":
         logging.info("Config file loaded.......")
         logging.info(config)
 
+        ###### Looger ######
+        logging.basicConfig(filename=config.get("log_loc"),
+                            filemode='a',
+                            format='%(asctime)s,%(msecs)d %(name)s %(levelname)s: %(message)s',
+                            datefmt="%Y-%m-%d %H:%M:%S",
+                            level=logging.INFO)
+
         conn = sqlite3.connect('process_tracker.db')
         conn.isolation_level = None
 

hasaki_crawler_engine/hasaki_db_writer.py

@@ -1,17 +1,15 @@
 import logging
 import psycopg2
 
-###### Looger ######
+
-logname = '/home/ubuntu/logs/hasaki_crawler.log'
+class hasaki_db_writer:
-#logname = 'hasaki_crawler.log'
+    def __init__(self, config):
-logging.basicConfig(filename=logname,
+        ###### Looger ######
+        logging.basicConfig(filename=config.get("log_loc"),
                             filemode='a',
                             format='%(asctime)s,%(msecs)d %(name)s %(levelname)s: %(message)s',
                             datefmt="%Y-%m-%d %H:%M:%S",
                             level=logging.INFO)
-
-class hasaki_db_writer:
-    def __init__(self, config):
         self.config = config
         self.conn = psycopg2.connect(database=self.config.get('database'), user=self.config.get('db_user'), password=self.config.get('db_pass'), host=self.config.get('db_host'), port=self.config.get('db_port'))
         self.conn.autocommit = True

hasaki_crawler_engine/hasaki_product_info.py

@@ -19,17 +19,15 @@ from webdriver_manager.chrome import ChromeDriverManager
 import brotli
 import json
 
-###### Looger ######
+
-logname = '/home/ubuntu/logs/hasaki_crawler.log'
+class HasakiProductInfo:
-#logname = 'hasaki_crawler.log'
+    def __init__(self, config):
-logging.basicConfig(filename=logname,
+        ###### Looger ######
+        logging.basicConfig(filename=config.get("log_loc"),
                             filemode='a',
                             format='%(asctime)s,%(msecs)d %(name)s %(levelname)s: %(message)s',
                             datefmt="%Y-%m-%d %H:%M:%S",
                             level=logging.INFO)
-
-class HasakiProductInfo:
-    def __init__(self, config):
         logging.info("Initializing HasakiProductInfo")
         self.pattern = r'[' + string.punctuation + ']'
         self.config = config