usr bin python coding utf-8 format pyfeedfilter --title жопа --author

#!/usr/bin/python
#coding=utf-8

# format
# pyfeedfilter --title=жопа --author=Василий

# target: skipped

# на входе stdin, ессестно ф формате фида, rss/atom
# на stdout тот-же фид, но только с выжившими элементами

import feedparser
import sys
import getopt
import feed_marshal
from xml.dom import minidom


#########################################################
def filter_title(entry, title, invert):
    title = title.decode('utf-8')
    return (entry.title.count(title) != 0) ^ invert

#########################################################
def filter_text(entry, text, invert):
    text = text.decode('utf-8')
    return (entry.description.count(text) != 0) ^ invert

#########################################################
def filter_author(entry, author, invert):
    author = author.decode('utf-8')
    return (entry.author.count(author) != 0) ^ invert

#########################################################
# bug: need more enclosures
def filter_mime_type(entry, mime_type, invert):
    mime_type = mime_type.decode('utf-8')
    return (entry.enclosures[0].type.count(mime_type) != 0) ^ invert

#########################################################
# bug: need more enclosures
def filter_mime_size_min(entry, mime_size_min, invert):
    return (int(entry.enclosures[0].length) >= int(mime_size_min)) ^ invert

#########################################################
# bug: need more enclosures
def filter_mime_size_max(entry, mime_size_max, invert):
    return (int(entry.enclosures[0].length) <= int(mime_size_max)) ^ invert

#########################################################
def process(feed_file, opts, filter_func, invertConditions):

    string = feed_file.read()
    xml = minidom.parseString(string)

    parser = feedparser.parse(string)
    entries = parser.entries

    items = xml.getElementsByTagName('item')
    channel = xml.getElementsByTagName('channel')[0]

    m = entries[:]

    # filter
    for opt, arg in opts:
        if filter_func.has_key(opt):
            entries = filter(lambda e: filter_func[opt](e, arg, invertConditions[opt] ), entries)

    # removes filtered nodes
    i = 0
    while i < len(m):
        if entries.count(m[i]) == 0:
            del m[i]
            item = xml.getElementsByTagName('item')[i]
            channel.removeChild(item)
        else:
            i += 1

    return xml


#########################################################
def main():

    usage = '''
Usage: pyfeedfilter [OPTIONS]
    -h, --help          this help

    --title             filter feed title
    --text              filter feed description text
    --author            filter feed author
    --mime-type         filter feed enclosure mime-type
    --mime-size-min     filter feed enclosure mime-type min mime size
    --mime-size-max     filter feed enclosure mime-type max mime size

pyfeedfilter take feed xml from stdin
for inverse match use ^ after =
'''


    try:
        options, args = getopt.getopt(sys.argv[1:],
                                    'h',
                                    ['help', 'title=', 'text=', 'author=',
                                     'mime-type=', 'mime-size-min=', 'mime-size-max='])
    except getopt.GetoptError:
        print 'getopt.GetoptError: ', getopt.GetoptError.msg
    except:
        print "Unknown Error"

    for opt, arg in options:
        if opt in ("-h", "--help"):
            print usage
            exit()

    filter_func = {}
    filter_func['--title'] = filter_title
    filter_func['--text'] = filter_text
    filter_func['--author'] = filter_author
    filter_func['--mime-type'] = filter_mime_type
    filter_func['--mime-size-min'] = filter_mime_size_min
    filter_func['--mime-size-max'] = filter_mime_size_max
    #filter_func[] =

    # invert matches
    invertCond = {}
    for opt, arg in options:
        invertCond[opt] = arg[0] == "^"
        if arg[0] == "^":
            index = options.index( (opt, arg) )
            arg = arg[1:]
            options[ index ] = (opt, arg)

    # processig
    xmlDoc = process(sys.stdin, options, filter_func, invertCond)
    sys.stdout.write( xmlDoc.toxml("UTF-8") )


#########################################################
#########################################################
if __name__ == '__main__':
    main()