Better unicode support as per the RFC(s)

[frenchie/icalparse.git] / icalparse.py

diff --git a/icalparse.py b/icalparse.py
index 3057560..edfca4c 100755 (executable)
--- a/icalparse.py
+++ b/icalparse.py
@@ -26,10 +26,8 @@ import os
 
 
 class InvalidICS(Exception): pass
-class notJoined(Exception): pass
 class IncompleteICS(InvalidICS): pass
 
-
 def lineJoiner(oldcal):
        '''Takes a string containing a calendar and returns an array of its lines'''
 
@@ -43,7 +41,7 @@ def lineJoiner(oldcal):
                oldcal = '\r\n'.join(oldcal)
 
        oldcal = oldcal.replace('\r\n ', '').replace('\r\n\t','')
-       return oldcal.strip().split('\r\n')
+       return [unicode(x, 'utf-8') for x in oldcal.strip().split('\r\n')]
 
 
 def lineFolder(oldcal, length=75):
@@ -56,6 +54,7 @@ def lineFolder(oldcal, length=75):
        sl = length - 1
 
        for line in oldcal:
+               line = line.encode('utf-8')
                # Line fits inside length, do nothing
                if len(line.rstrip()) <= length:
                        cal.append(line)
@@ -72,7 +71,8 @@ def lineFolder(oldcal, length=75):
 
 
 def splitFields(cal):
-       '''Takes a list of lines in a calendar file and returns a list of key, value pairs'''
+       '''Takes a list of lines in a calendar file and returns a list of tuples        
+       as (key, value) pairs'''
 
        ical = [tuple(x.split(':',1)) for x in cal]
 
@@ -85,7 +85,8 @@ def splitFields(cal):
 
 
 def joinFields(ical):
-       '''Takes a list of tuples that make up a calendar file and returns a list of lines'''
+       '''Takes a list of tuples that make up a calendar file and returns it to a
+       list of lines'''
 
        return [':'.join(x) for x in ical]
 
@@ -125,9 +126,7 @@ def getContent(url='',stdin=False):
 
 def getHTTPContent(url='',cache='.httplib2-cache'):
        '''This function attempts to play nice when retrieving content from HTTP
-       services. It's what you should use in a CGI script. It will (by default)
-       slurp the first 20 bytes of the file and check that we are indeed looking
-       at an ICS file before going for broke.'''
+       services. It's what you should use in a CGI script.'''
 
        try:
                import httplib2
@@ -234,6 +233,8 @@ if __name__ == '__main__':
        parser = OptionParser('usage: %prog [options] url')
        parser.add_option('-s', '--stdin', action='store_true', dest='stdin',
                default=False, help='Take a calendar from standard input')
+       parser.add_option('-v', '--verbose', action='store_true', dest='verbose',
+               default=False, help='Be verbose when rules are being applied')
        parser.add_option('-o', '--output', dest='outfile', default='',
                help='Specify output file (defaults to standard output)')
 
@@ -249,6 +250,6 @@ if __name__ == '__main__':
 
        content = getContent(url, options.stdin)
        cal = lineJoiner(content)
-       ical = applyRules(splitFields(cal), generateRules())
+       ical = applyRules(splitFields(cal), generateRules(), options.verbose)
        output = lineFolder(joinFields(ical))
        writeOutput(output, options.outfile)