Merge and deduplicate .ics/.ical files

merge_ics.py

#!/usr/bin/env python3
# -*- coding: utf-8 -*-

"""
Merges and deduplicates one or more .ics / ical files.

For usage information, call with --help.

Author: Jan Schlüter
"""

import sys
import io
from argparse import ArgumentParser


def opts_parser():
    usage =\
"""Merges and deduplicates one or more .ics / ical files.
In case of duplicated entries (by UID or DTSTART+SUMMARY),
files listed earlier take precedence.
"""
    parser = ArgumentParser(description=usage)
    parser.add_argument('infile', nargs='+',
            type=str,
            help='The .ics or ical files to read.')
    parser.add_argument('outfile',
            type=str,
            help='The .ics or ical file to write.')
    parser.add_argument('--verbose', '-v',
            action='store_true',
            help='If given, report all duplicated entries.')
    return parser


def read_objects(fn):
    """
    Reads calendar objects from the given .ics / ical file and yields
    them. Global calendar properties (e.g., PRODID) are also treated as
    objects.
    """
    vobject = []
    vobject_type = None
    with io.open(fn, newline='') as f:
        line = f.readline()
        if line != "BEGIN:VCALENDAR\r\n":
            raise ValueError("%s does not start with BEGIN:VCALENDAR" % fn)
        for line in f:
            if line == "END:VCALENDAR\r\n":
                return
            if not vobject_type:
                # we're on global scope, check if we have a new object
                if line.startswith("BEGIN:"):
                    vobject_type = line[6:].rstrip("\r\n")
                    vobject = [line]
                else:
                    yield line
            else:
                # we're in an object, check if it ended
                vobject.append(line)
                if line == "END:%s\r\n" % vobject_type:
                    yield "".join(vobject)
                    vobject = []
                    vobject_type = None


def get_identifiers(vobject):
    """
    Return a set of identifiers for a given calendar object string or
    global calendar property. For the latter, uses the property name.
    For objects, uses the object type and UID/TZID if present, and a
    combination of object type, DTSTART and SUMMARY if present.
    """
    def read_property(name):
        """Returns a property value of our vobject, or "" if not present."""
        start_pos = vobject.find(name)
        if start_pos < 0:
            return ""
        end_pos = vobject.find("\r\n", start_pos)
        value = vobject[start_pos:end_pos]
        # unfold folded lines (linebreak + single whitespace)
        while vobject[end_pos + 2] in " \t":
            start_pos = end_pos + 3
            end_pos = vobject.find("\r\n", start_pos)
            value += vobject[start_pos:end_pos]
        return value

    if not vobject.startswith("BEGIN:"):
        # it's a global property, use its name
        property_name = vobject.split(":", 1)[0]
        return {property_name}
    else:
        vobject_type = vobject[6:].split("\r\n", 1)[0]
        if vobject_type == "VTIMEZONE":
            # for timezones, use their identifier
            return {(vobject_type, read_property("TZID"))}
        else:
            # for events, todos, ... use their uid or dtstart+summary
            identifiers = set()
            uid = read_property("UID")
            if uid:
                identifiers.add((vobject_type, uid))
            dtstart = read_property("DTSTART")
            summary = read_property("SUMMARY")
            if dtstart or summary:
                identifiers.add((vobject_type, dtstart, summary))
            return identifiers

    
def main():
    # parse command line
    parser = opts_parser()
    options = parser.parse_args()

    # remember identifiers of calendar objects already written
    ids_written = set()

    # open output file
    with io.open(options.outfile, 'w', newline='') as f:
        f.write("BEGIN:VCALENDAR\r\n")
        # iterate over calendar objects in input files
        for fn in options.infile:
            for vobject in read_objects(fn):
                # get the set of identifiers for the current object
                identifiers = get_identifiers(vobject)
                # only write it if none of the identifiers occurred before
                if not (identifiers & ids_written):
                    f.write(vobject)
                elif options.verbose:
                    print("already written:", identifiers & ids_written)
                # remember the identifiers
                ids_written.update(identifiers)
        f.write("END:VCALENDAR\r\n")

if __name__ == "__main__":
    main()