X-Git-Url: https://code.delx.au/webdl/blobdiff_plain/8d6b793245e362dce6ff957cdd3e245c4292d268..b6ad3911d2593698d5ed405520aa2cc0e6ce2ea0:/sbs.py diff --git a/sbs.py b/sbs.py index d622e40..4af3632 100644 --- a/sbs.py +++ b/sbs.py @@ -1,27 +1,26 @@ #!/usr/bin/env python # vim:ts=4:sts=4:sw=4:noet -from common import grab_json, grab_xml, download_rtmp, download_urllib, Node +from common import grab_text, grab_html, grab_json, grab_xml, download_rtmp, download_urllib, Node, append_to_qs import collections +import urlparse BASE = "http://www.sbs.com.au" -MENU_URL = "/api/video_feed/f/dYtmxB/%s?startIndex=%d" -VIDEO_URL = BASE + "/api/video_feed/f/dYtmxB/CxeOeDELXKEv/%s?form=json" +VIDEO_MENU = BASE + "/ondemand/js/video-menu" +VIDEO_URL = BASE + "/ondemand/video/single/%s" +VIDEO_MAGIC = { + "v": "2.5.14", + "fp": "MAC 11,1,102,55", + "r": "FLQDD", + "g": "YNANAXRIYFYO", +} +SWF_URL = "http://resources.sbs.com.au/vod/theplatform/core/current/swf/flvPlayer.swf" NS = { "smil": "http://www.w3.org/2005/SMIL21/Language", } -SECTIONS = [ - "section-sbstv", - "section-programs", -] - -CATEGORY_MAP = { - "Factual": "Documentary", -} - class SbsNode(Node): def __init__(self, title, parent, video_id): @@ -31,76 +30,85 @@ class SbsNode(Node): self.can_download = True def download(self): - doc = grab_json(VIDEO_URL % self.video_id, 0) - best_url = None - best_bitrate = 0 - for d in doc["media$content"]: - bitrate = d["plfile$bitrate"] - if bitrate > best_bitrate or best_url is None: - best_bitrate = bitrate - best_url = d["plfile$url"] - - doc = grab_xml(best_url, 3600) - if doc.xpath("//smil:meta/@base", namespaces=NS): - vbase = doc.xpath("//smil:meta/@base", namespaces=NS)[0] - vpath = doc.xpath("//smil:video/@src", namespaces=NS)[0] - ext = vpath.rsplit(".", 1)[1] - filename = self.title + "." + ext - return download_rtmp(filename, vbase, vpath) - else: - from lxml import etree - url = doc.xpath("//smil:video/@src", namespaces=NS)[0] - ext = url.rsplit(".", 1)[1] - filename = self.title + "." + ext - url += "?v=2.5.14&fp=MAC%2011,1,102,55&r=FLQDD&g=YNANAXRIYFYO" - return download_urllib(filename, url) - -def fill_entry(get_catnode, entry): - title = entry["title"] - video_id = entry["id"] - info = collections.defaultdict(list) - for d in entry["media$categories"]: - if not d.has_key("media$scheme"): - continue - info[d["media$scheme"]].append(d["media$name"]) - - if "Section/Promos" in info.get("Section", []): - # ignore promos - return - - for category in info.get("Genre", ["$UnknownCategory$"]): - category = CATEGORY_MAP.get(category, category) - parent_node = get_catnode(category) - SbsNode(title, parent_node, video_id) - - -def fill_section(get_catnode, section): - index = 1 - while True: - doc = grab_json(BASE + MENU_URL % (section, index), 3600) + doc = grab_html(VIDEO_URL % self.video_id, 0) + meta_video = doc.xpath("//meta[@property='og:video']")[0] + swf_url = meta_video.attrib["content"] + swf_url_qs = urlparse.parse_qs(urlparse.urlparse(swf_url).query) + desc_url = swf_url_qs["v"][0] + + doc = grab_text(desc_url, 0) + doc_qs = urlparse.parse_qs(doc) + desc_url = doc_qs["releaseUrl"][0] + + doc = grab_xml(desc_url, 0) + error = doc.xpath("//smil:param[@name='exception']/@value", namespaces=NS) + if error: + raise Exception("Error downloading, SBS said: " + error[0]) + + video = doc.xpath("//smil:video", namespaces=NS)[0] + video_url = video.attrib["src"] + if not video_url: + raise Exception("Unsupported video '%s': %s" % (self.title, desc_url)) + ext = urlparse.urlsplit(video_url).path.rsplit(".", 1)[1] + filename = self.title + "." + ext + video_url = append_to_qs(video_url, VIDEO_MAGIC) + return download_urllib(filename, video_url, referrer=SWF_URL) + +class SbsNavNode(Node): + def __init__(self, title, parent, url): + Node.__init__(self, title, parent) + self.url = url + + def fill_children(self): + try: + doc = grab_json(BASE + self.url, 3600) + except ValueError: + # SBS sends XML as an error message :\ + return if len(doc.get("entries", [])) == 0: - break + return for entry in doc["entries"]: - fill_entry(get_catnode, entry) - index += doc["itemsPerPage"] + self.fill_entry(entry) + + def fill_entry(self, entry): + title = entry["title"] + video_id = entry["id"] + SbsNode(title, self, video_id) -class SbsRoot(Node): - def __init__(self, parent=None): +class SbsRootNode(Node): + def __init__(self, parent): Node.__init__(self, "SBS", parent) - self.catnodes = {} - def get_catnode(self, name): + def fill_children(self): + menu = grab_json(VIDEO_MENU, 3600, skip_assignment=True) + for name in menu.keys(): + self.fill_category(self, menu[name]) + + def create_nav_node(self, name, parent, cat_data, url_key): try: - return self.catnodes[name] + url = cat_data[url_key] except KeyError: - n = Node(name, self) - self.catnodes[name] = n - return n - - def fill_children(self): - for section in SECTIONS: - fill_section(self.get_catnode, section) + return + if url.strip(): + SbsNavNode(name, parent, url) + + def fill_category(self, parent, cat_data): + if not cat_data.has_key("children"): + name = cat_data["name"] + self.create_nav_node(name, parent, cat_data, "url") + return + + node = Node(cat_data["name"], parent) + self.create_nav_node("-Featured", node, cat_data, "furl") + self.create_nav_node("-Latest", node, cat_data, "url") + self.create_nav_node("-Most Popular", node, cat_data, "purl") + + children = cat_data.get("children", []) + if isinstance(children, dict): + children = [children[k] for k in sorted(children.keys())] + for child_cat in children: + self.fill_category(node, child_cat) def fill_nodes(root_node): - SbsRoot(root_node) + SbsRootNode(root_node)