From 499eca68d9fafee7f72294299bcf15ba3654b581 Mon Sep 17 00:00:00 2001
From: Grant Gainey <ggainey@redhat.com>
Date: Fri, 12 Jul 2024 13:25:58 -0400
Subject: [PATCH] Fix urlize_quoted_hrefs() to make links again.

Also now correctly escapes JSON values for the web.

fixes #5563.

(cherry picked from commit 672a6233d210f62fc4c407531461d6e13012ab67)
---
 CHANGES/5563.bugfix                    |  1 +
 pulpcore/app/templatetags/pulp_urls.py | 11 +++--
 pulpcore/tests/unit/test_pulp_urls.py  | 58 ++++++++++++++++++++++++++
 3 files changed, 66 insertions(+), 4 deletions(-)
 create mode 100644 CHANGES/5563.bugfix
 create mode 100644 pulpcore/tests/unit/test_pulp_urls.py

diff --git a/CHANGES/5563.bugfix b/CHANGES/5563.bugfix
new file mode 100644
index 0000000000..c30c686005
--- /dev/null
+++ b/CHANGES/5563.bugfix
@@ -0,0 +1 @@
+Browsable HREFs now have clickable links again.
diff --git a/pulpcore/app/templatetags/pulp_urls.py b/pulpcore/app/templatetags/pulp_urls.py
index 4aa1a8cb17..a94e968bae 100644
--- a/pulpcore/app/templatetags/pulp_urls.py
+++ b/pulpcore/app/templatetags/pulp_urls.py
@@ -3,7 +3,7 @@
 from django import template
 from django.conf import settings
 from django.utils.encoding import force_str
-from django.utils.html import escape
+from django.utils.html import escape, smart_urlquote
 from django.utils.safestring import SafeData, mark_safe
 
 register = template.Library()
@@ -20,6 +20,8 @@
     ("&quot;", "&quot;"),
 ]
 word_split_re = re.compile(r"(\s+)")
+api_root_prefix = r"^" + settings.V3_API_ROOT.replace("/", r"\/")
+href_re = re.compile(api_root_prefix, re.IGNORECASE)
 
 
 @register.filter(needs_autoescape=True)
@@ -57,6 +59,9 @@ def trim_url(x, limit=trim_url_limit):
             url = None
             nofollow_attr = ' rel="nofollow"' if nofollow else ""
 
+            if href_re.match(middle):
+                url = smart_urlquote(middle)
+
             # Check if it's a real URL
             if url and ("{" in url or "%7B" in url):
                 url = None
@@ -74,8 +79,6 @@ def trim_url(x, limit=trim_url_limit):
                     words[i] = mark_safe(word)
                 elif autoescape:
                     words[i] = escape(word)
-        elif safe_input:
-            words[i] = mark_safe(word)
-        elif autoescape:
+        else:
             words[i] = escape(word)
     return "".join(words)
diff --git a/pulpcore/tests/unit/test_pulp_urls.py b/pulpcore/tests/unit/test_pulp_urls.py
new file mode 100644
index 0000000000..cfaa9bd232
--- /dev/null
+++ b/pulpcore/tests/unit/test_pulp_urls.py
@@ -0,0 +1,58 @@
+import pytest
+from django.conf import settings
+
+from pulpcore.app.templatetags import pulp_urls
+
+pytestmark = pytest.mark.usefixtures("fake_domain")
+
+
+def test_urlize_quoted_hrefs_basic_url():
+    """
+    text starts with API_ROOT, defaults. Should be made clickable
+    """
+    txt = settings.V3_API_ROOT + "foo/bar/"
+    ret = pulp_urls.urlize_quoted_hrefs(txt)
+    assert ret == f'<a href="{txt}" rel="nofollow">{txt}</a>'
+
+
+def test_urlize_quoted_hrefs_nofollow():
+    """
+    text starts with API_ROOT, defaults. Should be made clickable
+    """
+    txt = settings.V3_API_ROOT + "foo/bar/"
+    ret = pulp_urls.urlize_quoted_hrefs(txt, nofollow=False)
+    assert ret == f'<a href="{txt}">{txt}</a>'
+
+
+def test_urlize_quoted_hrefs_trim():
+    """
+    text starts with API_ROOT, defaults. Should be made clickable
+    """
+    txt = settings.V3_API_ROOT + "foo/bar/"
+    trim_txt = txt[0] + "..."
+    ret = pulp_urls.urlize_quoted_hrefs(txt, trim_url_limit=4)
+    assert ret == f'<a href="{txt}" rel="nofollow">{trim_txt}</a>'
+
+
+def test_urlize_quoted_hrefs_basic_url_xss():
+    """
+    text starts with API_ROOT, includes XSS, defaults. Should be made clickable, escape XSS
+    """
+    txt = settings.V3_API_ROOT + "foo/bar/<script>alert('ALERT!')</script>blech/"
+    escapified_linked_text = (
+        '<a href="' + settings.V3_API_ROOT + "foo/bar/"
+        '%3Cscript%3Ealert(&#x27;ALERT!&#x27;)%3C/script%3Eblech/" '
+        'rel="nofollow">' + settings.V3_API_ROOT + "foo/bar/&lt;script&gt;"
+        "alert(&#x27;ALERT!&#x27;)&lt;/script&gt;blech/</a>"
+    )
+    ret = pulp_urls.urlize_quoted_hrefs(txt)
+    assert ret == escapified_linked_text
+
+
+def test_urlize_quoted_hrefs_basic_escape():
+    """
+    text contains XSS. Expect escaped
+    """
+    txt = "foo/bar/<script>alert('ALERT!')</script>blech/"
+    ret = pulp_urls.urlize_quoted_hrefs(txt)
+    assert ret == "foo/bar/&lt;script&gt;alert(&#x27;ALERT!&#x27;)&lt;/script&gt;blech/"