[utils] Skip invalid/non HTML entities (Closes #7518)

[youtube-dl] / youtube_dl / utils.py

diff --git a/youtube_dl/utils.py b/youtube_dl/utils.py
index efd5f4ae111448ca9933136b151ca1bf6a0505a8..b7013a6aaef400633d45ac33c4a7fcb949802be7 100644 (file)
--- a/youtube_dl/utils.py
+++ b/youtube_dl/utils.py
@@ -366,7 +366,7 @@ def sanitize_path(s):
     if drive_or_unc:
         norm_path.pop(0)
     sanitized_path = [
-        path_part if path_part in ['.', '..'] else re.sub('(?:[/<>:"\\|\\\\?\\*]|\.$)', '#', path_part)
+        path_part if path_part in ['.', '..'] else re.sub('(?:[/<>:"\\|\\\\?\\*]|[\s.]$)', '#', path_part)
         for path_part in norm_path]
     if drive_or_unc:
         sanitized_path.insert(0, drive_or_unc + os.path.sep)
@@ -396,7 +396,11 @@ def _htmlentity_transform(entity):
             numstr = '0%s' % numstr
         else:
             base = 10
-        return compat_chr(int(numstr, base))
+        # See https://github.com/rg3/youtube-dl/issues/7518
+        try:
+            return compat_chr(int(numstr, base))
+        except ValueError:
+            pass
 
     # Unknown entity in name, return its literal representation
     return ('&%s;' % entity)
@@ -911,7 +915,8 @@ def unified_strdate(date_str, day_first=True):
         timetuple = email.utils.parsedate_tz(date_str)
         if timetuple:
             upload_date = datetime.datetime(*timetuple[:6]).strftime('%Y%m%d')
-    return compat_str(upload_date)
+    if upload_date is not None:
+        return compat_str(upload_date)
 
 
 def determine_ext(url, default_ext='unknown_video'):