web/zing-offline.py

   1 #!/usr/bin/env python
   2 # Licence: public domain
   3 # Original idea from http://projektantos.wordpress.com/\
   4 #   2009/12/20/tải-nhạc-hot-trong-thang-của-zing-bản-viết-bằng-python/
   5
   6 # destination directory, where folder will be placed
   7 DEST_DIR = "~/Music/Zing"
   8
   9 # download information
  10 HOMEPAGE = "http://mp3.zing.vn/mp3/nghe-album/album-hot/nhac-viet.html"
  11 DOWNLOAD_PATTERN = r'"(http://dl\.mp3.*\?filename=([^/]*\.mp3))"'
  12
  13 import sys
  14 import logging
  15 from datetime import date
  16 from os import mkdir, utime
  17 from os.path import expanduser, join, isdir, isfile
  18 from urllib import urlopen, urlretrieve
  19 from re import findall
  20 from calendar import timegm
  21
  22 logging.basicConfig(level=logging.DEBUG,
  23     format="%(asctime)s %(levelname)s %(message)s")
  24 logging.info('Starting.')
  25
  26 # absolute path to destination directory
  27 dir = join(expanduser(DEST_DIR), "Thang%02d" % date.today().month)
  28 if not isdir(dir):
  29     try:
  30         mkdir(dir)
  31     except:
  32         logging.error("Can't create directory '%s'.", dir)
  33         sys.exit(1)
  34
  35 # get the page content
  36 try:
  37     f = urlopen(HOMEPAGE)
  38     page_content = f.read()
  39     f.close()
  40 except:
  41     logging.error("Can't read page '%s'.", HOMEPAGE)
  42     sys.exit(2)
  43
  44 # find all file URLs and download them
  45 count = 0
  46 for link,filename in findall(DOWNLOAD_PATTERN, page_content):
  47     filepath = join(dir, filename)
  48     if isfile(filepath):
  49         logging.info("'%s' already exists, skipped.", filename)
  50         continue
  51     logging.info("Downloading '%s'..." % filename)
  52     try:
  53         info = urlretrieve(link, filepath)[1]
  54         last_modified = info.getdate('Last-Modified')
  55         if last_modified:
  56             last_modified = timegm(last_modified)
  57             utime(filepath, (last_modified, last_modified))
  58         logging.info("Done.")
  59     except:
  60         logging.info("ERROR!")
  61     count += 1
  62
  63 print "=" * 78
  64 if count <= 1:
  65     logging.info("%d file has been downloaded.", count)
  66 else:
  67     logging.info("%d files have been downloaded to '%s'.", count, dir)
  68 print "=" * 78
  69
  70 sys.exit(0)