- f = urlopen(url, max_age)
- doc = etree.parse(f)
- f.close()
- return doc
-
-def grab_json(url, max_age):
- f = urlopen(url, max_age)
- doc = json.load(f)
- f.close()
- return doc
-
-def download_rtmp(filename, vbase, vpath):
- if vpath.endswith(".flv"):
- vpath = vpath[:-4]
- cmd = [
- "rtmpdump",
- "-o", filename,
- "-r", vbase,
- "-y", vpath,
- ]
- try:
- p = subprocess.Popen(cmd)
- ret = p.wait()
- if ret != 0:
- print >>sys.stderr, "rtmpdump exited with error code:", ret
- return False
- except OSError, e:
- print >>sys.stderr, "Failed to run rtmpdump!", e
- return False
- except KeyboardInterrupt:
- print "Cancelled", cmd
- try:
- p.terminate()
- p.wait()
- except KeyboardInterrupt:
- p.send_signal(signal.SIGKILL)
- p.wait()
-
-def download_urllib(filename, url):
- print "Downloading: %s -> %s" % (url, filename)
- try:
- src = urllib.urlopen(url)
- dst = open(filename, "w")
- shutil.copyfileobj(src, dst)
- return True
- except KeyboardInterrupt:
- print "\nCancelled", url
- finally:
- src.close()
- dst.close()
- return False
+ 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):
+ 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
+
+def exec_subprocess(cmd):
+ 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
+
+
+def convert_flv_mp4(orig_filename):
+ 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
+
+def convert_filename(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):
+ 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
+
+def download_urllib(filename, url, referrer=None):
+ 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
+
+def natural_sort(l, key=None):
+ 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)
+
+def append_to_qs(url, params):
+ 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