added Hasaki crawler

hasaki_crawler_engine/hasaki_product_info.py

@@ -2,6 +2,8 @@ import hashlib
 import logging
 import string
 import re
+
+import playwright
 import psycopg2
 from playwright.sync_api import sync_playwright
 from hasaki_db_writer import hasaki_db_writer
@@ -97,13 +99,19 @@ class HasakiProductInfo:
                     context = browser.new_context(user_agent=random_mobile_ua)
                     page = context.new_page()
 
-                    page.goto(url)
-                    page.reload()
+                    api_requests = {}
 
-                    with page.expect_response("**/wap/v2/product/detail**") as response:
-                        api_requests = response.value.json()
-
-                    browser.close()
+                    try:
+                        page.goto(url, timeout=5000)
+                        with page.expect_response("**/wap/v2/product/detail**") as response:
+                            api_requests = response.value.json()
+                    except playwright._impl._errors.TimeoutError:
+                        logging.info("Timeout occurred. Retrying.....")
+                        page.reload()
+                        with page.expect_response("**/wap/v2/product/detail**") as response:
+                            api_requests = response.value.json()
+                    finally:
+                        browser.close()
 
                     return api_requests
             except Exception as e:

hasaki_crawler_engine/test.py

@@ -1,4 +1,4 @@
-
+import playwright
 from playwright.sync_api import sync_playwright
 from fake_useragent import UserAgent
 import logging
@@ -14,12 +14,17 @@ with sync_playwright() as p:
     context = browser.new_context(user_agent=random_mobile_ua)
     page = context.new_page()
 
-    page.goto("https://hasaki.vn/san-pham/kem-duong-skin1004-lam-diu-da-chiet-xuat-rau-ma-75ml-89637.html")
+    try:
 
-    page.reload()
-
-    with page.expect_response("**/wap/v2/product/detail**") as response:
-        api_requests = response.value.json()
+        page.goto("https://hasaki.vn/san-pham/kem-duong-skin1004-lam-diu-da-chiet-xuat-rau-ma-75ml-89637.html",
+                            timeout=5000)
+        with page.expect_response("**/wap/v2/product/detail**") as response:
+            api_requests = response.value.json()
+    except playwright._impl._errors.TimeoutError:
+        logging.info("Timeout occurred. Retrying.....")
+        page.reload()
+        with page.expect_response("**/wap/v2/product/detail**") as response:
+            api_requests = response.value.json()
 
 
     browser.close()