From b6c2067c714de5765fb790fded960ecac37b3fb6 Mon Sep 17 00:00:00 2001
From: Bob Mottram <bob@libreserver.org>
Date: Wed, 30 Mar 2022 19:13:40 +0100
Subject: [PATCH] Handle fractional seconds within post published date

---
 inbox.py    | 11 ++++++++---
 newswire.py | 17 +++++++++++++++++
 posts.py    |  4 ++++
 3 files changed, 29 insertions(+), 3 deletions(-)

diff --git a/inbox.py b/inbox.py
index fe8196713..ff00fecd9 100644
--- a/inbox.py
+++ b/inbox.py
@@ -2306,11 +2306,16 @@ def _valid_post_content(base_dir: str, nickname: str, domain: str,
 
     if not message_json['object'].get('published'):
         return False
-    if 'T' not in message_json['object']['published']:
+    published = message_json['object']['published']
+    if 'T' not in published:
         return False
-    if 'Z' not in message_json['object']['published']:
+    if 'Z' not in published:
         return False
-    if not valid_post_date(message_json['object']['published'], 90, debug):
+    if '.' in published:
+        # converts 2022-03-30T17:37:58.734Z into 2022-03-30T17:37:58Z
+        published = published.split('.')[0] + 'Z'
+        message_json['object']['published'] = published
+    if not valid_post_date(published, 90, debug):
         return False
 
     summary = None
diff --git a/newswire.py b/newswire.py
index 36097404d..e5f4a1740 100644
--- a/newswire.py
+++ b/newswire.py
@@ -268,6 +268,14 @@ def _valid_feed_date(pub_date: str, debug: bool = False) -> bool:
     # convert from YY-MM-DD HH:MM:SS+00:00 to
     # YY-MM-DDTHH:MM:SSZ
     post_date = pub_date.replace(' ', 'T').replace('+00:00', 'Z')
+    if '.' in post_date:
+        ending = post_date.split('.')[1]
+        timezone_str = ''
+        for ending_char in ending:
+            if not ending_char.isdigit():
+                timezone_str += ending_char
+        if timezone_str:
+            post_date = post_date.split('.')[0] + timezone_str
     return valid_post_date(post_date, 90, debug)
 
 
@@ -320,6 +328,15 @@ def parse_feed_date(pub_date: str, unique_string_identifier: str) -> str:
         if 'UT' in pub_date and 'UT' not in date_format:
             continue
 
+        # remove any fraction of a second
+        if '.' in pub_date:
+            ending = pub_date.split('.')[1]
+            timezone_str = ''
+            for ending_char in ending:
+                if not ending_char.isdigit():
+                    timezone_str += ending_char
+            if timezone_str:
+                pub_date = pub_date.split('.')[0] + timezone_str
         try:
             published_date = datetime.strptime(pub_date, date_format)
         except BaseException:
diff --git a/posts.py b/posts.py
index f587f5d3d..d53778500 100644
--- a/posts.py
+++ b/posts.py
@@ -4926,6 +4926,10 @@ def download_announce(session, base_dir: str, http_prefix: str,
                              base_dir, nickname, domain, post_id,
                              recent_posts_cache)
             return None
+        if '.' in announced_json['published'] and \
+           'Z' in announced_json['published']:
+            announced_json['published'] = \
+                announced_json['published'].split('.')[0] + 'Z'
         if not valid_post_date(announced_json['published'], 90, debug):
             print('WARN: announced post is not recently published ' +
                   str(announced_json))