simple utility to process niconico comment XML file.

nicoutil.py

# -* coding: utf-8 -*-

import sys
import os
import re
import argparse
import shutil
from xml.etree.ElementTree import *

parser = argparse.ArgumentParser(description='simple utility to process niconico comment XML file.')
parser.add_argument('file', type=file, nargs='+', help='XML file to process')
parser.add_argument('-s', '--shift', type=int, nargs=1, help='time to shift all comments (millisecond)')
parser.add_argument('-r', '--recent', type=int, nargs=1, help='delete recent RECENT comments')
parser.add_argument('-n', '--rename', action='store_true', help='search in directory and rename input file to proper name.')
args = parser.parse_args()

for argfile in args.file:
    try:
        filepath = argfile.name.decode('shift-jis')
        argfile.close()

        filename = os.path.basename(filepath)

        if args.rename:
            if not filename.endswith('.xml'):
                raise IOError('%s: file extension must be .xml' % filepath)
            idmatch = re.match(r'^.*\((.*)?\)\[.*?\]\.xml$', filename)
            if idmatch is None:
                idmatch = re.match(r'^.*-([^-]*)\.xml$', filename)
                if idmatch is None:
                    raise IOError('%s: filename must include video id.' % filepath)
            videoid = idmatch.group(1)
            dirname = os.path.dirname(filepath)
            dirfiles = os.listdir(dirname)
            renamed = False
            for dirfile in dirfiles:
                dirfilename = os.path.basename(dirfile)
                if dirfilename.endswith('.mp4'):
                    diridmatch = re.match(r'^.*-([^-]*)\.mp4', dirfilename)
                    if diridmatch is not None:
                        if diridmatch.group(1) == videoid:
                            dirfilename = re.sub(r'\.mp4$', r'.xml', dirfilename)
                            dstfilepath = filepath.replace(filename, dirfilename)
                            shutil.copy(filepath.encode('shift-jis'), dstfilepath.encode('shift-jis'))
                            filepath = dstfilepath
                            renamed = True
                            break
            if not renamed:
                raise IOError('%s: destination filename was not found' % filepath)

        if filename.endswith('.backup.xml'):
            backupfile = filepath
            xmlfile = re.sub(r'\.backup\.xml$', r'.xml', backupfile)
            tree = parse(backupfile.encode('shift-jis'))
        elif filename.endswith('.xml'):
            xmlfile = filepath
            backupfile = re.sub(r'\.xml$', r'.backup.xml', xmlfile)
            if os.path.isfile(backupfile.encode('shift-jis')):
                tree = parse(backupfile.encode('shift-jis'))
            else:
                tree = parse(xmlfile.encode('shift-jis'))
                os.rename(xmlfile.encode('shift-jis'), backupfile.encode('shift-jis'))
        else:
            raise IOError('%s: file extension must be .xml' % filepath)

        root = tree.getroot()

        if args.shift is not None:
            for element in root.getiterator('chat'):
                vpos = int(element.attrib['vpos'])
                element.set('vpos', str(max(vpos - int(args.shift[0] / 10), 0)))

        if args.recent is not None:
            maxno = 0
            for element in root.getiterator('chat'):
                if maxno < int(element.attrib['no']):
                    maxno = int(element.attrib['no'])
            for element in root.getiterator('chat'):
                no = int(element.attrib['no'])
                if no > maxno - args.recent[0]:
                    root.remove(element)

        tree.write(xmlfile.encode('shift-jis'))
    except:
        print 'error:', sys.exc_info()[1]