CACHE_DIR = os.path.expanduser("~/.cache/webdl")
USER_AGENT = "Mozilla/5.0 (Macintosh; Intel Mac OS X 10.6; rv:21.0) Gecko/20100101 Firefox/21.0"
class Node(object):
CACHE_DIR = os.path.expanduser("~/.cache/webdl")
USER_AGENT = "Mozilla/5.0 (Macintosh; Intel Mac OS X 10.6; rv:21.0) Gecko/20100101 Firefox/21.0"
class Node(object):
- def __init__(self, title, parent=None):
- self.title = title
- if parent:
- parent.children.append(self)
- self.parent = parent
- self.children = []
- self.can_download = False
+ def __init__(self, title, parent=None):
+ self.title = title
+ if parent:
+ parent.children.append(self)
+ self.parent = parent
+ self.children = []
+ self.can_download = False
valid_chars = frozenset("-_.()!@#%^ abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ0123456789")
def sanify_filename(filename):
valid_chars = frozenset("-_.()!@#%^ abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ0123456789")
def sanify_filename(filename):
- filename = filename.encode("ascii", "ignore")
- filename = "".join(c for c in filename if c in valid_chars)
- return filename
+ filename = filename.encode("ascii", "ignore")
+ filename = "".join(c for c in filename if c in valid_chars)
+ return filename
cookiejar = cookielib.CookieJar()
urlopener = urllib2.build_opener(urllib2.HTTPCookieProcessor(cookiejar))
def _urlopen(url, referrer=None):
cookiejar = cookielib.CookieJar()
urlopener = urllib2.build_opener(urllib2.HTTPCookieProcessor(cookiejar))
def _urlopen(url, referrer=None):
- req = urllib2.Request(url)
- req.add_header("User-Agent", USER_AGENT)
- if referrer:
- req.add_header("Referer", referrer)
- return urlopener.open(req)
+ req = urllib2.Request(url)
+ req.add_header("User-Agent", USER_AGENT)
+ if referrer:
+ req.add_header("Referer", referrer)
+ return urlopener.open(req)
-### print url
- if not os.path.isdir(CACHE_DIR):
- os.makedirs(CACHE_DIR)
-
- if max_age <= 0:
- return _urlopen(url)
-
- filename = hashlib.md5(url).hexdigest()
- filename = os.path.join(CACHE_DIR, filename)
- if os.path.exists(filename):
- file_age = int(time.time()) - os.path.getmtime(filename)
- if file_age < max_age:
- return open(filename)
-
- src = _urlopen(url)
- dst = open(filename, "w")
- try:
- shutil.copyfileobj(src, dst)
- except Exception, e:
- try:
- os.unlink(filename)
- except OSError:
- pass
- raise e
- src.close()
- dst.close()
-
- return open(filename)
+### print url
+ if not os.path.isdir(CACHE_DIR):
+ os.makedirs(CACHE_DIR)
+
+ if max_age <= 0:
+ return _urlopen(url)
+
+ filename = hashlib.md5(url).hexdigest()
+ filename = os.path.join(CACHE_DIR, filename)
+ if os.path.exists(filename):
+ file_age = int(time.time()) - os.path.getmtime(filename)
+ if file_age < max_age:
+ return open(filename)
+
+ src = _urlopen(url)
+ dst = open(filename, "w")
+ try:
+ shutil.copyfileobj(src, dst)
+ except Exception, e:
+ try:
+ os.unlink(filename)
+ except OSError:
+ pass
+ raise e
+ src.close()
+ dst.close()
+
+ return open(filename)
- f = urlopen(url, max_age)
- text = f.read().decode("utf-8")
- f.close()
- return text
+ f = urlopen(url, max_age)
+ text = f.read().decode("utf-8")
+ f.close()
+ return text
- f = urlopen(url, max_age)
- doc = html.parse(f, html.HTMLParser(encoding="utf-8", recover=True))
- f.close()
- return doc
+ f = urlopen(url, max_age)
+ doc = html.parse(f, html.HTMLParser(encoding="utf-8", recover=True))
+ f.close()
+ return doc
- f = urlopen(url, max_age)
- doc = etree.parse(f, etree.XMLParser(encoding="utf-8", recover=True))
- f.close()
- return doc
+ f = urlopen(url, max_age)
+ doc = etree.parse(f, etree.XMLParser(encoding="utf-8", recover=True))
+ f.close()
+ return doc
def grab_json(url, max_age, skip_assignment=False, skip_function=False):
def grab_json(url, max_age, skip_assignment=False, skip_function=False):
- f = urlopen(url, max_age)
- if skip_assignment:
- text = f.read()
- pos = text.find("=")
- doc = json.loads(text[pos+1:])
- elif skip_function:
- text = f.read()
- pos = text.find("(")
- rpos = text.rfind(")")
- doc = json.loads(text[pos+1:rpos])
- else:
- doc = json.load(f)
- f.close()
- return doc
+ f = urlopen(url, max_age)
+ if skip_assignment:
+ text = f.read()
+ pos = text.find("=")
+ doc = json.loads(text[pos+1:])
+ elif skip_function:
+ text = f.read()
+ pos = text.find("(")
+ rpos = text.rfind(")")
+ doc = json.loads(text[pos+1:rpos])
+ else:
+ doc = json.load(f)
+ f.close()
+ return doc
- try:
- p = subprocess.Popen(cmd)
- ret = p.wait()
- if ret != 0:
- print >>sys.stderr, cmd[0], "exited with error code:", ret
- return False
- else:
- return True
- except OSError, e:
- print >>sys.stderr, "Failed to run", cmd[0], e
- except KeyboardInterrupt:
- print "Cancelled", cmd
- try:
- p.terminate()
- p.wait()
- except KeyboardInterrupt:
- p.send_signal(signal.SIGKILL)
- p.wait()
- return False
+ try:
+ p = subprocess.Popen(cmd)
+ ret = p.wait()
+ if ret != 0:
+ print >>sys.stderr, cmd[0], "exited with error code:", ret
+ return False
+ else:
+ return True
+ except OSError, e:
+ print >>sys.stderr, "Failed to run", cmd[0], e
+ except KeyboardInterrupt:
+ print "Cancelled", cmd
+ try:
+ p.terminate()
+ p.wait()
+ except KeyboardInterrupt:
+ p.send_signal(signal.SIGKILL)
+ p.wait()
+ return False
- basename = os.path.splitext(orig_filename)[0]
- flv_filename = basename + ".flv"
- mp4_filename = basename + ".mp4"
- if orig_filename != flv_filename:
- os.rename(orig_filename, flv_filename)
- print "Converting %s to mp4" % flv_filename
- cmd = [
- "ffmpeg",
- "-i", flv_filename,
- "-acodec", "copy",
- "-vcodec", "copy",
- mp4_filename,
- ]
- if not exec_subprocess(cmd):
- return
- try:
- flv_size = os.stat(flv_filename).st_size
- mp4_size = os.stat(mp4_filename).st_size
- if abs(flv_size - mp4_size) < 0.05 * flv_size:
- os.unlink(flv_filename)
- else:
- print >>sys.stderr, "The size of", mp4_filename, "is suspicious, did ffmpeg fail?"
- except Exception, e:
- print "Conversion failed", e
+ basename = os.path.splitext(orig_filename)[0]
+ flv_filename = basename + ".flv"
+ mp4_filename = basename + ".mp4"
+ if orig_filename != flv_filename:
+ os.rename(orig_filename, flv_filename)
+ print "Converting %s to mp4" % flv_filename
+ cmd = [
+ "avconv",
+ "-i", flv_filename,
+ "-acodec", "copy",
+ "-vcodec", "copy",
+ mp4_filename,
+ ]
+ if not exec_subprocess(cmd):
+ return
+ try:
+ flv_size = os.stat(flv_filename).st_size
+ mp4_size = os.stat(mp4_filename).st_size
+ if abs(flv_size - mp4_size) < 0.05 * flv_size:
+ os.unlink(flv_filename)
+ else:
+ print >>sys.stderr, "The size of", mp4_filename, "is suspicious, did avconv fail?"
+ except Exception, e:
+ print "Conversion failed", e
- if os.path.splitext(filename.lower())[1] in (".mp4", ".flv"):
- f = open(filename)
- fourcc = f.read(4)
- f.close()
- if fourcc == "FLV\x01":
- convert_flv_mp4(filename)
+ if os.path.splitext(filename.lower())[1] in (".mp4", ".flv"):
+ f = open(filename)
+ fourcc = f.read(4)
+ f.close()
+ if fourcc == "FLV\x01":
+ convert_flv_mp4(filename)
def download_rtmp(filename, vbase, vpath, hash_url=None):
def download_rtmp(filename, vbase, vpath, hash_url=None):
- filename = sanify_filename(filename)
- print "Downloading: %s" % filename
- if vpath.endswith(".flv"):
- vpath = vpath[:-4]
- cmd = [
- "rtmpdump",
- "-o", filename,
- "-r", vbase,
- "-y", vpath,
- ]
- if hash_url is not None:
- cmd += ["--swfVfy", hash_url]
- if exec_subprocess(cmd):
- convert_filename(filename)
- return True
- else:
- return False
+ filename = sanify_filename(filename)
+ print "Downloading: %s" % filename
+ if vpath.endswith(".flv"):
+ vpath = vpath[:-4]
+ cmd = [
+ "rtmpdump",
+ "-o", filename,
+ "-r", vbase,
+ "-y", vpath,
+ ]
+ if hash_url is not None:
+ cmd += ["--swfVfy", hash_url]
+ if exec_subprocess(cmd):
+ convert_filename(filename)
+ return True
+ else:
+ return False
- filename = sanify_filename(filename)
- print "Downloading: %s" % filename
- try:
- src = _urlopen(url, referrer)
- dst = open(filename, "w")
- while True:
- buf = src.read(1024*1024)
- if not buf:
- break
- dst.write(buf)
- sys.stdout.write(".")
- sys.stdout.flush()
- print
- except KeyboardInterrupt:
- print "\nCancelled", url
- return False
- finally:
- try:
- src.close()
- except:
- pass
- try:
- dst.close()
- except:
- pass
-
- convert_filename(filename)
- return True
+ filename = sanify_filename(filename)
+ print "Downloading: %s" % filename
+ try:
+ src = _urlopen(url, referrer)
+ dst = open(filename, "w")
+ while True:
+ buf = src.read(1024*1024)
+ if not buf:
+ break
+ dst.write(buf)
+ sys.stdout.write(".")
+ sys.stdout.flush()
+ print
+ except KeyboardInterrupt:
+ print "\nCancelled", url
+ return False
+ finally:
+ try:
+ src.close()
+ except:
+ pass
+ try:
+ dst.close()
+ except:
+ pass
+
+ convert_filename(filename)
+ return True
- ignore_list = ["a", "the"]
- def key_func(k):
- if key is not None:
- k = key(k)
- k = k.lower()
- newk = []
- for c in re.split("([0-9]+)", k):
- c = c.strip()
- if c.isdigit():
- newk.append(int(c))
- else:
- for subc in c.split():
- if subc not in ignore_list:
- newk.append(subc)
- return newk
-
- return sorted(l, key=key_func)
+ ignore_list = ["a", "the"]
+ def key_func(k):
+ if key is not None:
+ k = key(k)
+ k = k.lower()
+ newk = []
+ for c in re.split("([0-9]+)", k):
+ c = c.strip()
+ if c.isdigit():
+ newk.append(int(c))
+ else:
+ for subc in c.split():
+ if subc not in ignore_list:
+ newk.append(subc)
+ return newk
+
+ return sorted(l, key=key_func)
- r = list(urlparse.urlsplit(url))
- qs = urlparse.parse_qs(r[3])
- for k, v in params.iteritems():
- if v is not None:
- qs[k] = v
- elif qs.has_key(k):
- del qs[k]
- r[3] = urllib.urlencode(qs, True)
- url = urlparse.urlunsplit(r)
- return url
+ r = list(urlparse.urlsplit(url))
+ qs = urlparse.parse_qs(r[3])
+ for k, v in params.iteritems():
+ if v is not None:
+ qs[k] = v
+ elif qs.has_key(k):
+ del qs[k]
+ r[3] = urllib.urlencode(qs, True)
+ url = urlparse.urlunsplit(r)
+ return url