[allura] branch dw/8383 created (now 9610f08)

[di...@apache.org]

This is an automated email from the ASF dual-hosted git repository.

dill0wn pushed a change to branch dw/8383
in repository https://gitbox.apache.org/repos/asf/allura.git.


      at 9610f08  8383 strip bad unicode chars from /activity/feed.rss

This branch includes the following new commits:

     new 9610f08  8383 strip bad unicode chars from /activity/feed.rss

The 1 revisions listed above as "new" are entirely new to this
repository and will be described in separate emails.  The revisions
listed as "add" were already present in the repository and have only
been added to this reference.

[allura] 01/01: 8383 strip bad unicode chars from /activity/feed.rss

[di...@apache.org]

This is an automated email from the ASF dual-hosted git repository.

dill0wn pushed a commit to branch dw/8383
in repository https://gitbox.apache.org/repos/asf/allura.git

commit 9610f089871eed357fb3c7b8375b846bc9522cdc
Author: Dillon Walls <di...@slashdotmedia.com>
AuthorDate: Mon Dec 7 23:29:51 2020 +0000

    8383 strip bad unicode chars from /activity/feed.rss
---
 Allura/allura/lib/helpers.py        | 9 +++++++++
 Allura/allura/tests/test_helpers.py | 7 +++++++
 ForgeActivity/forgeactivity/main.py | 2 +-
 3 files changed, 17 insertions(+), 1 deletion(-)

diff --git a/Allura/allura/lib/helpers.py b/Allura/allura/lib/helpers.py
index fd660ad..2f4d2f5 100644
--- a/Allura/allura/lib/helpers.py
+++ b/Allura/allura/lib/helpers.py
@@ -158,6 +158,15 @@ def escape_json(data):
     return json.dumps(data).replace('<', '\\u003C')
 
 
+def strip_bad_unicode(s):
+    """
+    xml doesn't like some control characters: https://www.w3.org/TR/REC-xml/#charsets
+    :param s:
+    :return:
+    """
+    return re.sub('[\x00-\x08\x0B\x0C\x0E-\x1F]', '', s)
+
+
 def monkeypatch(*objs):
     def patchem(func):
         for obj in objs:
diff --git a/Allura/allura/tests/test_helpers.py b/Allura/allura/tests/test_helpers.py
index 0206b02..f6fc852 100644
--- a/Allura/allura/tests/test_helpers.py
+++ b/Allura/allura/tests/test_helpers.py
@@ -89,6 +89,13 @@ def test_escape_json():
     assert_equals(outputdata, outputsample)
 
 
+def test_strip_bad_unicode():
+    inputdata = 'Hello\x08World'
+    outputsample = 'HelloWorld'
+    outputdata = h.strip_bad_unicode(inputdata)
+    assert_equals(outputdata, outputsample)
+
+
 def test_really_unicode():
     here_dir = path.dirname(__file__)
     s = h.really_unicode(b'asdf')
diff --git a/ForgeActivity/forgeactivity/main.py b/ForgeActivity/forgeactivity/main.py
index 0ee91d5..4ca24c4 100644
--- a/ForgeActivity/forgeactivity/main.py
+++ b/ForgeActivity/forgeactivity/main.py
@@ -178,7 +178,7 @@ class ForgeActivityController(BaseController):
             ),
                 link=url,
                 pubdate=t.published,
-                description=t.obj.activity_extras.get('summary'),
+                description=h.strip_bad_unicode(t.obj.activity_extras.get('summary', '')),
                 unique_id=url_id,
                 author_name=t.actor.activity_name,
                 author_link=h.absurl(t.actor.activity_url))