asusidx2xml.py

# -*- coding: utf-8 -*-
import os
import re
import datetime

from ftplib import FTP
from BeautifulSoup import BeautifulSoup

_multilines_re = re.compile(r'<([a-z0-9\-]+)\s+(.*?)\s+</\1>', re.S | re.M)
_singleline_re = re.compile(r'<(product|version)>\s+(.+)$', re.M)

class AsusFTP(object):
    host = 'ftp.asus.com.tw'
    
    def __init__(self):
        BeautifulSoup.NESTABLE_TAGS['path'] = ['version']
        BeautifulSoup.NESTABLE_TAGS['release-date'] = ['version']
        BeautifulSoup.NESTABLE_TAGS['description'] = ['version']
        
        self.ftp = FTP(self.host, timeout=45)
        self.ftp.login()
    
    def __del__(self):
        self.ftp.quit()
    
    def _parse(self, idx):
        idx = idx.replace('<~', '</')
        idx = _multilines_re.sub(r'<\1>\2</\1>', idx)
        idx = _singleline_re.sub(r'<\1 value="\2">', idx)
        return BeautifulSoup(idx)
    
    def get_idx_as_soup(self, path):
        self.ftp.cwd(os.path.dirname(path))
        idx = []
        self.ftp.retrlines('RETR %s' % os.path.basename(path), idx.append)
        return self._parse('\n'.join(idx))

if __name__ == '__main__':
    aftp = AsusFTP()
    
    soup = aftp.get_idx_as_soup('/pub/ASUS/Path.idx')
    platform = soup.findAll('product', value='EeePC')[0]
    platform_path = platform.path.contents[0].strip()
    
    soup = aftp.get_idx_as_soup(platform_path)
    product = soup.findAll('product', value='S101H')[0]
    product_path = product.path.contents[0].strip()
    
    soup = aftp.get_idx_as_soup(product_path)
    versions = soup.findAll('version')
    
    for version in versions:
        print '\n'
        print ' - VERSION: ', version.get('value')
        print '\tDATE: ', datetime.datetime.strptime(version.find('release-date').contents[0].strip(), '%m/%d/%Y')
        print '\tPATH: ', version.find('path').contents[0].strip()
        print '\tDESCIPTION: ', version.find('description').contents[0].strip()