def wgetPage(thread, page, targetdir, platform = 'unix', logfile = ''):
"""Call wget to download page of a thread to targetdir for local archive
"""
if not vbutils.isValidURL(thread.url):
return False
# Count up the subdirs in this URL
# Sometimes vBulletin is installed in root
# Sometimes it is in a subdir
# We need to know this number so we can tell wget to chill later
subdirs = (thread.url.count('/') - 3)
print "Found %s sub-dir in the URL ..." % subdirs
if (page == 1):
wget_cmd = "wget -v -nH --cut-dirs=%s -k -K -E -H -p -P %s -S -w 1 --random-wait --no-cache --no-cookies \"%s\"" % (subdirs, targetdir, thread.url)
else:
wget_cmd = "wget -v -nH --cut-dirs=%s -k -K -E -H -p -P %s -N -w 1 --random-wait --no-cache --no-cookies \"%s&page=%s\"" % (subdirs, targetdir, thread.url, page)
# Split wget invocation into list for subprocess
# TODO Apparently shlex.split can't take unicode arguments?
# Is this going to be a problem in the future?
args = shlex.split(str(wget_cmd))
# Are we specifying a special logfile location?
if (logfile != ''):
args.insert(1, '-a')
args.insert(2, logfile)
# Force special character converstion to win32
# e.g. ? becomes @ (see wget man pages for details)
if (platform == 'windows'):
args.insert(1, '--restrict-file-names=windows')
# subprocess.Popen() will start a subprocess
# TODO This would be better with a single data structure
# of fixed length containing all the processes.
# TODO need to catch errors
proc_firstpage = subprocess.Popen(args)
# .communicate() will grab output and exit info
# will also half execution until complete
# TODO Catch errors from communicate()
proc_firstpage.communicate()
# Assuming success!
return True
def init(argv):
"""Init all globals according to options and params in argv
Returns string with cleaned up URL of a thread to archive
"""
params = {}
params = { "utc" : 0,
"localdir" : '.' }
# wget uses _platform for escaping illegal chrs in filenames
if (sys.platform in "win32"):
params["platform"] = "windows"
else:
params["platform"] = "unix"
# params["useragent"] = "HTTP_USER_AGENT:Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.9.0.17) Gecko/2010010604 Ubuntu/9.04 (jaunty) Firefox/3.0.17"
# Check the commandline for arguments
try:
opts, args = getopt.getopt(argv, "dhl:uvw", ["debug", "help", "localdir=", "utc", "verbose", "windows"])
except getopt.GetoptError:
# Found a flag not in our known list
# Returning a short usage message ...
print "Error: unrecognized flag"
usage()
# ... and bye-bye!
sys.exit(2)
# Evaluate commandline options, arguments
for opt, arg in opts:
if opt in ("-h", "--help"):
usage()
sys.exit()
elif opt in ("-v", "--verbose"):
pass
elif opt in ("-d", "--debug"):
pass
elif opt in ("-u", "--utc"):
params["utc"] = 1
elif opt in ("-l", "--localdir"):
# TODO need to validate the directory
# strip trailing slashes
params["localdir"] = arg.rstrip('/')
elif opt in ("-w", "--windows"):
params["platform"] = "windows"
else:
print "opt: %s, arg: %s" % (opt, arg)
# Should be exactly 1 positional argument
# URL to the thread should be the only argument
try:
raw_URL = args[0]
except IndexError:
print "getthread.py: missing URL"
usage()
sys.exit(2)
# Is raw_URL a valid vB thread URL?
print "Validating URL ..."
if not (vbutils.isValidURL(raw_URL)):
print "Error: %s is not a valid vBulletin thread URL." % raw_URL
print
sys.exit(2)
# Clean up URL from the commandline
# Keep the domain, sub dirs, showthread, and &t=
params["url"] = vbutils.cleanURL(raw_URL)
print "Valid thread URL: %s" % url
return params
