From 861e5f958929f24dad0a05f44148528248165295 Mon Sep 17 00:00:00 2001
From: Deimos <deimos@tildes.net>
Date: Mon, 19 Nov 2018 14:12:09 -0700
Subject: [PATCH] Remove utm_ query params from link topics

Adds the start of a "url transformation" system, initially only removing
any utm_ parameters from link topics.
---
 tildes/tests/test_url_transform.py  | 20 +++++++++++++++++
 tildes/tildes/lib/url_transform.py  | 34 +++++++++++++++++++++++++++++
 tildes/tildes/models/topic/topic.py |  3 ++-
 3 files changed, 56 insertions(+), 1 deletion(-)
 create mode 100644 tildes/tests/test_url_transform.py
 create mode 100644 tildes/tildes/lib/url_transform.py

diff --git a/tildes/tests/test_url_transform.py b/tildes/tests/test_url_transform.py
new file mode 100644
index 0000000..833a053
--- /dev/null
+++ b/tildes/tests/test_url_transform.py
@@ -0,0 +1,20 @@
+# Copyright (c) 2018 Tildes contributors <code@tildes.net>
+# SPDX-License-Identifier: AGPL-3.0-or-later
+
+from tildes.lib.url_transform import apply_url_transformations
+
+
+def test_remove_utm_query_params():
+    """Ensure that utm query params are removed but others are left."""
+    url = "http://example.com/path?utm_source=tildes&utm_campaign=test&something=ok"
+    cleaned_url = apply_url_transformations(url)
+
+    assert cleaned_url == "http://example.com/path?something=ok"
+
+
+def test_non_utm_params_unaffected():
+    """Ensure that non-utm_ query params aren't removed."""
+    url = "http://example.com/path?one=x&two=y&three=z"
+    cleaned_url = apply_url_transformations(url)
+
+    assert cleaned_url == url
diff --git a/tildes/tildes/lib/url_transform.py b/tildes/tildes/lib/url_transform.py
new file mode 100644
index 0000000..51f4794
--- /dev/null
+++ b/tildes/tildes/lib/url_transform.py
@@ -0,0 +1,34 @@
+# Copyright (c) 2018 Tildes contributors <code@tildes.net>
+# SPDX-License-Identifier: AGPL-3.0-or-later
+
+"""Functions related to transforming URLs (sanitization, cleanup, etc.)."""
+
+from urllib.parse import parse_qs, urlencode, urlparse, urlunparse
+
+
+def apply_url_transformations(url: str) -> str:
+    """Apply all applicable transformations to a url.
+
+    This method should generally be the only one imported/used from this module, unless
+    there is a specific reason for needing to apply a subset of transformations.
+    """
+    url = remove_utm_query_params(url)
+
+    return url
+
+
+def remove_utm_query_params(url: str) -> str:
+    """Remove any utm_* query parameters from a url."""
+    parsed = urlparse(url)
+
+    query_params = parse_qs(parsed.query)
+
+    cleaned_params = {
+        param: value
+        for param, value in query_params.items()
+        if not param.startswith("utm_")
+    }
+
+    parsed = parsed._replace(query=urlencode(cleaned_params, doseq=True))
+
+    return urlunparse(parsed)
diff --git a/tildes/tildes/models/topic/topic.py b/tildes/tildes/models/topic/topic.py
index 8dc4a73..58b4c8c 100644
--- a/tildes/tildes/models/topic/topic.py
+++ b/tildes/tildes/models/topic/topic.py
@@ -32,6 +32,7 @@ from tildes.lib.id import id_to_id36
 from tildes.lib.markdown import convert_markdown_to_safe_html
 from tildes.lib.string import convert_to_url_slug
 from tildes.lib.url import get_domain_from_url, is_tweet
+from tildes.lib.url_transform import apply_url_transformations
 from tildes.metrics import incr_counter
 from tildes.models import DatabaseModel
 from tildes.models.group import Group
@@ -207,7 +208,7 @@ class Topic(DatabaseModel):
         """Create a new link topic."""
         new_topic = cls._create_base_topic(group, author, title)
         new_topic.topic_type = TopicType.LINK
-        new_topic.link = link
+        new_topic.link = apply_url_transformations(link)
         new_topic.original_url = link
 
         incr_counter("topics", type="link")