Update bot detection

leonghui · Nov 3, 2024 · 2666cbf · 2666cbf
1 parent 3dd3717
commit 2666cbf
Show file tree

Hide file tree

Showing 2 changed files with 4 additions and 3 deletions.
diff --git a/amazon_feed.py b/amazon_feed.py
@@ -9,7 +9,7 @@
 from flask import abort
 from requests.exceptions import JSONDecodeError, RequestException
 
-from amazon_feed_data import AmazonListingQuery, AmazonItemQuery
+from amazon_feed_data import AmazonListingQuery, AmazonItemQuery, BOT_PATTERN
 from json_feed_data import JsonFeedTopLevel, JsonFeedItem, JSONFEED_VERSION_URL
 
 
@@ -92,12 +92,12 @@ def get_response_dict(url, query):
 
     # return HTTP error code
     if not response.ok:
-        if response.status_code == 503 or response.text.find("captcha") >= 0:
+        if response.status_code == 503 or re.search(BOT_PATTERN, response.text):
             bot_msg = f'"{query.query_str}" - API paywall triggered, resetting session'
             reset_query_session(query)
 
             logger.warning(bot_msg)
-            abort(503, description=bot_msg)
+            abort(429, description=bot_msg)
         else:
             logger.error(f'"{query.query_str}" - error from source')
             logger.debug(f'"{query.query_str}" - dumping response: {response.text}')

diff --git a/amazon_feed_data.py b/amazon_feed_data.py
@@ -7,6 +7,7 @@
 
 
 ASIN_PATTERN = r"^(B[\dA-Z]{9}|\d{9}(X|\d))$"
+BOT_PATTERN = r"automated access|captcha"
 
 
 class UnavailabilityText(str, Enum):  # allow comparison with strings
Original file line number	Diff line number	Diff line change
Expand Up		@@ -7,6 +7,7 @@


		ASIN_PATTERN = r"^(B[\dA-Z]{9}\|\d{9}(X\|\d))$"
		BOT_PATTERN = r"automated access\|captcha"


		class UnavailabilityText(str, Enum): # allow comparison with strings
Expand Down