From bab98d75d76a17961c20c9abda5dc430d5cbb0c4 Mon Sep 17 00:00:00 2001
From: Ori <ori.gold@perimeterx.com>
Date: Thu, 16 Nov 2023 13:11:15 -0800
Subject: [PATCH] adding raw_url and updating px_metadata

---
 perimeterx/px_activities_client.py |  2 ++
 perimeterx/px_api.py               |  3 +++
 perimeterx/px_context.py           | 11 +++++++++++
 px_metadata.json                   |  3 ++-
 4 files changed, 18 insertions(+), 1 deletion(-)

diff --git a/perimeterx/px_activities_client.py b/perimeterx/px_activities_client.py
index 7a44d75..8bb383b 100644
--- a/perimeterx/px_activities_client.py
+++ b/perimeterx/px_activities_client.py
@@ -53,6 +53,8 @@ def send_to_perimeterx(activity_type, ctx, config, detail):
             'client_uuid': ctx.uuid,
             'request_id': ctx.request_id
         }
+        if ctx.raw_url:
+            _details['raw_url'] = ctx.raw_url
 
         if len(detail.keys()) > 0:
             _details = dict(list(_details.items()) + list(detail.items()))
diff --git a/perimeterx/px_api.py b/perimeterx/px_api.py
index 0886299..772135a 100644
--- a/perimeterx/px_api.py
+++ b/perimeterx/px_api.py
@@ -229,6 +229,9 @@ def prepare_risk_body(ctx, config):
         logger.debug('attaching px_cookie to request')
         body['additional']['px_cookie'] = ctx.decoded_cookie
 
+    if ctx.raw_url:
+        body['additional']['raw_url'] = ctx.raw_url
+
     return body
 
 
diff --git a/perimeterx/px_context.py b/perimeterx/px_context.py
index d17bc1e..f8399d2 100644
--- a/perimeterx/px_context.py
+++ b/perimeterx/px_context.py
@@ -66,6 +66,8 @@ def __init__(self, request, config):
         else:
             uri = request.path
             full_url = request.url
+
+        raw_url = request.url if request.url != full_url else None
         hostname = request.host
         is_filtered = is_filtered_request(config, request)
         sensitive_route = sum(1 for _ in filter(lambda sensitive_route_item: re.match(sensitive_route_item, uri), config.sensitive_routes_regex)) > 0 or sum(1 for _ in filter(lambda sensitive_route_item: uri == sensitive_route_item, config.sensitive_routes)) > 0
@@ -79,6 +81,7 @@ def __init__(self, request, config):
         self._user_agent = user_agent
         self._full_url = full_url
         self._uri = uri
+        self._raw_url = raw_url
         self._hostname = hostname
         self._px_cookies = px_cookies
         self._cookie_names = request_cookie_names
@@ -215,6 +218,14 @@ def uri(self):
     def uri(self, uri):
         self._uri = uri
 
+    @property
+    def raw_url(self):
+        return self._raw_url
+
+    @raw_url.setter
+    def raw_url(self, raw_url):
+        self._raw_url = raw_url
+
     @property
     def hostname(self):
         return self._hostname
diff --git a/px_metadata.json b/px_metadata.json
index 7823f83..1c5183b 100644
--- a/px_metadata.json
+++ b/px_metadata.json
@@ -35,6 +35,7 @@
         "vid_extraction"
     ],
     "excluded_tests": [
-        "test_url_decode_reserved_characters_with_character\\[;-"
+        "test_url_decode_reserved_characters_send_decoded_url\\[;-",
+        "test_url_decode_reserved_characters_send_additional_raw_url\\[(/|:)-"
     ]
 }
\ No newline at end of file