Add item authors to RSS feeds (#7012)

* Add item authors to RSS feeds

* Add docstring to warehouse.rss.views._format_author

* Nit: single character variable names

Co-authored-by: Dustin Ingram <[email protected]>

Author: scop

tests/unit/rss/test_views.py

@@ -35,11 +35,14 @@ def test_rss_updates(db_request):
     release1.created = datetime.date(2011, 1, 1)
     release2 = ReleaseFactory.create(project=project2)
     release2.created = datetime.date(2012, 1, 1)
+    release2.author_email = "[email protected]"
     release3 = ReleaseFactory.create(project=project1)
     release3.created = datetime.date(2013, 1, 1)
 
     assert rss.rss_updates(db_request) == {
-        "latest_releases": [release3, release2, release1]
+        "latest_releases": tuple(
+            zip((release3, release2, release1), (None, "[email protected]", None))
+        )
     }
     assert db_request.response.content_type == "text/xml"
 
@@ -64,5 +67,42 @@ def test_rss_packages(db_request):
     project3.created = datetime.date(2013, 1, 1)
     ReleaseFactory.create(project=project3)
 
-    assert rss.rss_packages(db_request) == {"newest_projects": [project3, project1]}
+    assert rss.rss_packages(db_request) == {
+        "newest_projects": tuple(zip((project3, project1), (None, None)))
+    }
     assert db_request.response.content_type == "text/xml"
+
+
+def test_format_author(db_request):
+    db_request.find_service = pretend.call_recorder(
+        lambda *args, **kwargs: pretend.stub(
+            enabled=False, csp_policy=pretend.stub(), merge=lambda _: None
+        )
+    )
+
+    db_request.session = pretend.stub()
+
+    project = ProjectFactory.create()
+    release = ReleaseFactory.create(project=project)
+
+    release.author_email = "[email protected]"
+    assert rss._format_author(release) == release.author_email
+
+    release.author_email = "No Reply <[email protected]>"
+    assert rss._format_author(release) == "[email protected]"
+
+    for invalid in (None, "", "UNKNOWN", "[email protected], UNKNOWN"):
+        release.author_email = invalid
+        assert rss._format_author(release) is None
+
+    release.author_email = (
+        # simple, no spaces
+        "[email protected],"
+        # space after
+        "[email protected] ,"
+        # space before, incl realname
+        " No Reply <[email protected]>,"
+        # two spaces before, angle brackets
+        "  <[email protected]>"
+    )
+    assert rss._format_author(release) == ", ".join(["[email protected]"] * 4)

warehouse/rss/views.py

@@ -10,6 +10,8 @@
 # See the License for the specific language governing permissions and
 # limitations under the License.
 
+from email.utils import getaddresses
+
 from pyramid.view import view_config
 from sqlalchemy.orm import joinedload
 
@@ -18,6 +20,32 @@
 from warehouse.xml import XML_CSP
 
 
+def _format_author(release):
+    """
+    Format release author suitably for inclusion in an RSS feed.
+
+    Release author names and emails are hard to match robustly, mainly
+    because there may be multiple in both, and especially the names may
+    contain pretty much anything. So stick with just the emails with some
+    rudimentary sanity checks.
+
+    Even though the spec says "the email address" and thus assumes a single
+    author, we let multiple pass, comma separated.
+
+    http://www.rssboard.org/rss-specification#ltauthorgtSubelementOfLtitemgt
+    """
+    author_emails = []
+    for _, author_email in getaddresses([release.author_email or ""]):
+        if "@" not in author_email:
+            # Require all valid looking
+            return None
+        author_emails.append(author_email)
+
+    if not author_emails:
+        return None
+    return ", ".join(author_emails)
+
+
 @view_config(
     route_name="rss.updates",
     renderer="rss/updates.xml",
@@ -42,8 +70,9 @@ def rss_updates(request):
         .limit(40)
         .all()
     )
+    release_authors = [_format_author(release) for release in latest_releases]
 
-    return {"latest_releases": latest_releases}
+    return {"latest_releases": tuple(zip(latest_releases, release_authors))}
 
 
 @view_config(
@@ -70,5 +99,8 @@ def rss_packages(request):
         .limit(40)
         .all()
     )
+    project_authors = [
+        _format_author(project.releases[0]) for project in newest_projects
+    ]
 
-    return {"newest_projects": newest_projects}
+    return {"newest_projects": tuple(zip(newest_projects, project_authors))}

warehouse/templates/rss/packages.xml

@@ -2,12 +2,13 @@
 {% block title %}PyPI newest packages{% endblock %}
 {% block description %}Newest packages registered at the Python Package Index{% endblock %}
 {% block items -%}
-  {% for project in newest_projects %}
+  {% for project, author in newest_projects %}
     <item>
       <title>{{ project.name }} added to PyPI</title>
       <link>{{ request.route_url('packaging.project', name=project.normalized_name) }}</link>
       <guid>{{ request.route_url('packaging.project', name=project.normalized_name) }}</guid>
       <description>{{ project.releases[0].summary }}</description>
+      {% if author %}<author>{{ author }}</author>{% endif %}
       <pubDate>{{ project.created|format_rfc822_datetime() }}</pubDate>
     </item>
   {%- endfor %}

warehouse/templates/rss/updates.xml

@@ -2,11 +2,12 @@
 {% block title %}PyPI recent updates{% endblock %}
 {% block description %}Recent updates to the Python Package Index{% endblock %}
 {% block items -%}
-  {% for release in latest_releases %}
+  {% for release, author in latest_releases %}
     <item>
       <title>{{ release.project.name }} {{ release.version }}</title>
       <link>{{ request.route_url('packaging.release', name=release.project.normalized_name, version=release.version) }}</link>
       <description>{{ release.summary }}</description>
+      {% if author %}<author>{{ author }}</author>{% endif %}
       <pubDate>{{ release.created|format_rfc822_datetime() }}</pubDate>
     </item>
   {%- endfor %}