From 42a0d85e7f46c8ae39480ce5b1a045ba2b7395a5 Mon Sep 17 00:00:00 2001
From: naumansharifwork <157045300+naumansharifwork@users.noreply.github.com>
Date: Mon, 16 Sep 2024 00:03:09 +0500
Subject: [PATCH] updated utils.py to include post url and get_cookies
 functions (#113)

---
 clean/utils.py | 58 ++++++++++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 58 insertions(+)

diff --git a/clean/utils.py b/clean/utils.py
index dda7195..b1b8689 100644
--- a/clean/utils.py
+++ b/clean/utils.py
@@ -287,3 +287,61 @@ def get_repeated_asset_url(self, objects: List[MetadataDict]):
         else:
             seen_urls.add(asset_url)
     return repeated_urls
+
+
+@retry(tries=3, delay=15, backoff=2)
+def post_url(
+    url, user_agent="Big Local News (biglocalnews.org)", session=None, **kwargs
+):
+    """Request the provided URL and return a response object.
+
+    Args:
+        url (str): the url to be requested
+        user_agent (str): the user-agent header passed with the request (default: biglocalnews.org)
+        session: a session object to use when making the request. optional
+    """
+    logger.debug(f"Requesting {url}")
+
+    # Set the headers
+    if "headers" not in kwargs:
+        kwargs["headers"] = {}
+    kwargs["headers"]["User-Agent"] = user_agent
+
+    # Go get it
+    if session is not None:
+        logger.debug(f"Requesting with session {session}")
+        response = session.post(url, **kwargs)
+    else:
+        response = requests.post(url, **kwargs)
+    logger.debug(f"Response code: {response.status_code}")
+
+    # Verify that the response is 200
+    assert response.ok
+
+    # Return the response
+    return response
+
+
+@retry(tries=3, delay=15, backoff=2)
+def get_cookies(url, user_agent="Big Local News (biglocalnews.org)", **kwargs):
+    """Request the provided URL and return cookie object.
+
+    Args:
+        url (str): the url to be requested
+        user_agent (str): the user-agent header passed with the request (default: biglocalnews.org)
+    """
+    logger.debug(f"Requesting {url}")
+
+    # Set the headers
+    if "headers" not in kwargs:
+        kwargs["headers"] = {}
+    kwargs["headers"]["User-Agent"] = user_agent
+    response = requests.get(url, **kwargs)
+
+    # Verify that the response is 200
+    assert response.ok
+
+    cookies = response.cookies.get_dict()
+
+    # Return the response
+    return cookies