From a63dfffd3d125aaa0b9129326d5a6350647b2dd0 Mon Sep 17 00:00:00 2001
From: Zdenek Styblik <stybla@turnovfree.net>
Date: Thu, 6 Jun 2024 16:45:41 +0200
Subject: [PATCH] Fix bug in parse_news() - entries without link not being
 filtered out

I don't know what I was doing, but code in question doesn't make sense.
---
 phpbb2slack.py            |  7 ++++---
 rss2irc.py                |  5 +++--
 tests/files/phpbb-rss.xml |  8 ++++++++
 tests/files/rss.xml       |  4 ++++
 tests/test_phpbb2slack.py | 23 +++++++++++++++++++++++
 tests/test_rss2irc.py     | 21 +++++++++++++++++++++
 6 files changed, 63 insertions(+), 5 deletions(-)

diff --git a/phpbb2slack.py b/phpbb2slack.py
index b879cee..9a0a3a0 100755
--- a/phpbb2slack.py
+++ b/phpbb2slack.py
@@ -247,11 +247,12 @@ def parse_news(data: str, authors: List[str]) -> Dict:
     feed = feedparser.parse(data)
     for entry in feed["entries"]:
         link = entry.pop("link", None)
-        title = entry.pop("title", None)
-        author_detail = entry.pop("author_detail", {"name": None})
-        if not "link" and not "title":
+        title = entry.pop("title", "No title")
+        if not link:
+            # If we don't have a link, there is nothing we can do.
             continue
 
+        author_detail = entry.pop("author_detail", {"name": None})
         if authors and author_detail["name"] not in authors:
             continue
 
diff --git a/rss2irc.py b/rss2irc.py
index 48617f3..8c55af4 100755
--- a/rss2irc.py
+++ b/rss2irc.py
@@ -270,10 +270,11 @@ def parse_news(data: str) -> Dict[str, Tuple[str, str]]:
     feed = feedparser.parse(data)
     for entry in feed["entries"]:
         link = entry.pop("link", "")
-        title = entry.pop("title", "")
-        if not "link" and not "title":
+        if not link:
+            # If we don't have a link, there is nothing we can do.
             continue
 
+        title = entry.pop("title", "No title")
         category = entry.pop("category", "")
         news[link] = (title, category)
 
diff --git a/tests/files/phpbb-rss.xml b/tests/files/phpbb-rss.xml
index 64d1c34..3442b3c 100644
--- a/tests/files/phpbb-rss.xml
+++ b/tests/files/phpbb-rss.xml
@@ -29,5 +29,13 @@ Does anyone else have the same situation?]]></content:encoded>
       <content:encoded><![CDATA[I ordered a new something yesterday and still have not received an email. i have emailed twice somewhere and still nothing....so now trying something.]]></content:encoded>
       <slash:comments>1</slash:comments>
     </item>
+    <item>
+      <title>Broken entry</title>
+      <pubDate>Fri, 14 Feb 2020 20:50:36 +0000</pubDate>
+      <author>invalid@example.com (otherUser)</author>
+      <dc:creator>otherUser</dc:creator>
+      <content:encoded><![CDATA[I ordered a new something yesterday and still have not received an email. i have emailed twice somewhere and still nothing....so now trying something.]]></content:encoded>
+      <slash:comments>1</slash:comments>
+    </item>
   </channel>
 </rss>
diff --git a/tests/files/rss.xml b/tests/files/rss.xml
index 426ce7e..4226dac 100644
--- a/tests/files/rss.xml
+++ b/tests/files/rss.xml
@@ -19,5 +19,9 @@
             <description>Item2 description</description>
             <pubDate>Sun, 02 Feb 2020 11:26:26 -0500</pubDate>
         </item>
+        <item>
+            <description>Item3 description - purposefully broken</description>
+            <pubDate>Sun, 02 Feb 2020 11:25:26 -0500</pubDate>
+        </item>
     </channel>
 </rss>
diff --git a/tests/test_phpbb2slack.py b/tests/test_phpbb2slack.py
index e31883c..b3e2b73 100644
--- a/tests/test_phpbb2slack.py
+++ b/tests/test_phpbb2slack.py
@@ -331,6 +331,29 @@ def test_main_cache_hit(
     assert len(fixture_http_server.requests) == 0
 
 
+def test_parse_news():
+    """Test parse_news()."""
+    expected_news = {
+        "https://phpbb.example.com/threads/something-of-something.424837/": {
+            "title": "Some other problem",
+            "category": None,
+            "comments_cnt": 0,
+        },
+        "https://phpbb.example.com/threads/something-not-received.424836/": {
+            "title": "Something not received",
+            "category": None,
+            "comments_cnt": 1,
+        },
+    }
+
+    rss_fname = os.path.join(SCRIPT_PATH, "files", "phpbb-rss.xml")
+    with open(rss_fname, "rb") as fhandle:
+        rss_data = fhandle.read().decode("utf-8")
+
+    result = phpbb2slack.parse_news(rss_data, [])
+    assert result == expected_news
+
+
 @pytest.mark.parametrize(
     "cache,expected_cache",
     [
diff --git a/tests/test_rss2irc.py b/tests/test_rss2irc.py
index 34c99df..32e9545 100644
--- a/tests/test_rss2irc.py
+++ b/tests/test_rss2irc.py
@@ -337,6 +337,27 @@ def test_main_cache_hit(
     assert sorted(output) == sorted(expected_output)
 
 
+def test_parse_news():
+    """Test parse_news()."""
+    expected_news = {
+        "http://www.example.com/scan.php?page=news_item&px=item1": (
+            "Item1",
+            "",
+        ),
+        "http://www.example.com/scan.php?page=news_item&px=item2": (
+            "Item2",
+            "",
+        ),
+    }
+
+    rss_fname = os.path.join(SCRIPT_PATH, "files", "rss.xml")
+    with open(rss_fname, "rb") as fhandle:
+        rss_data = fhandle.read().decode("utf-8")
+
+    result = rss2irc.parse_news(rss_data)
+    assert result == expected_news
+
+
 def test_scrub_items():
     """Test scrub_items()."""
     logging.basicConfig(level=logging.CRITICAL)