From http://codespeak.net/pipermail/icalendar-dev/2010-April/000152.html:

Lines get folded in the middle of multi-octet sequences (checked out
code from svn today). Consider this case:
import icalendar
ical = icalendar.Calendar()
ical.add('summary', u'a' + u'ą'*100)
ical.as_string().decode('utf-8')
...
UnicodeDecodeError: 'utf8' codec can't decode bytes in position 90-91:
invalid data

I have attached a diff of a simple one-line fix.

As I see in the code you actually try not to split a multi-octet
character but you don't recalculate the slice after finding the new
end position. Could you confirm this?

Submitted by Rimvydas Naktinis.
===================================================================
--- src/icalendar/parser.py (revision 73587)
+++ src/icalendar/parser.py (working copy)
@@ -456,6 +456,7 @@
                     else:
                         end -= 1
 
+            slice = self[start:end]
             new_lines.append(slice)
             if end == l_line:
                 # Done