#!/usr/bin/python
"""
podget: download podcast attachments from feeds.
Adam Sampson <ats@offog.org>
"""

import sys, os, re, optparse, feedparser, subprocess, time

USER_AGENT = "podget/1.0"

def get_filename_prefix(feed, entry, raw_filename, options):
	# Media Network often has proper dates in the entry titles or filenames.
	if feed.get("title") == "Media Network Vintage Vault":
		day, month, year = None, None, None

		m = re.search(r'(\d\d)\.(\d\d)\.(\d\d\d\d)', entry.get("title"))
		if m is not None:
			day, month, year = m.group(1, 2, 3)

		m = re.search(r'(\d\d)\.(\d\d)\.(\d\d\d\d)', raw_filename)
		if m is not None:
			day, month, year = m.group(1, 2, 3)

		if day is not None:
			return ("Media Network - %s-%s-%s - "
			        % (year, month, day))

	return ""

def try_unlink(fn):
	try:
		os.unlink(fn)
	except OSError:
		pass

def download_file(url, raw_filename, filename, length, referer, options):
	def is_downloaded(fn):
		try:
			st = os.stat(fn)
		except OSError:
			return False
		if options.nolength or length is None or st.st_size == length:
			return True
		return False

	for dir in [options.outputdir] + options.checkdirs:
		for fn in os.listdir(dir):
			fn = os.path.join(dir, fn)
			if fn.endswith(raw_filename):
				if is_downloaded(fn):
					if options.dryrun:
						print >>sys.stderr, "Already got: " + fn
					return

	output_fn = os.path.join(options.outputdir, filename)

	cmd = [
		"wget",
		"-U", USER_AGENT,
		"-O", output_fn,
		"--referer", referer,
		url,
		]

	if options.dryrun:
		print >>sys.stderr, "Would run: " + " ".join(cmd)
		return

	delay = 1
	while True:
		print >>sys.stderr, "Downloading: " + url
		try_unlink(output_fn)
		rc = subprocess.call(cmd)
		if rc == 0 and is_downloaded(output_fn):
			# Success!
			return

		if delay > 40:
			print >>sys.stderr, "Too many retries; giving up"
			try_unlink(output_fn)
			break

		print >>sys.stderr, "Failed, trying again in %d seconds" % delay
		delay *= 2
		time.sleep(delay)

def handle_enclosure(feed, entry, link, referer, options):
	url = link.get("href")
	if url is None:
		return

	length = link.get("length")
	if length is not None:
		length = int(length)

	filename = os.path.basename(url)

	# Filenames starting with dots are annoying.
	filename = re.sub(r'^\.+', '', filename)

	fn_prefix = get_filename_prefix(feed, entry, filename, options)
	download_file(url, filename, fn_prefix + filename, length, referer, options)

def scan_feed(feed_url, options):
	data = feedparser.parse(feed_url, agent=USER_AGENT)
	entries = data.entries
	if options.reverse:
		entries = reversed(entries)
	for entry in entries:
		links = entry.get("links")
		if links is None:
			continue
		for link in links:
			if link.get("rel") == "enclosure":
				handle_enclosure(data.feed, entry, link, feed_url, options)

if __name__ == "__main__":
	parser = optparse.OptionParser()
	parser.add_option("-n", dest="dryrun",
	                  action="store_true", default=False,
	                  help="print what would be done and exit")
	parser.add_option("-d", dest="outputdir", metavar="DIR",
	                  default=".",
	                  help="directory to store downloaded files in")
	parser.add_option("-D", dest="checkdirs", metavar="DIR",
	                  action="append", default=[],
	                  help="extra directory to check for downloaded files")
	parser.add_option("-L", dest="nolength",
	                  action="store_true", default=False,
	                  help="don't check file lengths against podcast")
	parser.add_option("-r", dest="reverse",
	                  action="store_true", default=False,
	                  help="download oldest items first")
	(options, args) = parser.parse_args()
	for url in args:
		scan_feed(url, options)