]> code.delx.au - webdl/blobdiff - common.py
webdl: refactored HLS download and convert to mp4 logic
[webdl] / common.py
index f51ec5305f60e04eda6686e8cdcf480c188d9c28..accd6dd506eab83f6e6d2acf14ca94241f000731 100644 (file)
--- a/common.py
+++ b/common.py
@@ -166,39 +166,47 @@ def exec_subprocess(cmd):
     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
+def avconv_remux(infile, outfile):
+    print "Converting %s to mp4" % infile
     cmd = [
         "avconv",
-        "-i", flv_filename,
+        "-i", infile,
         "-acodec", "copy",
         "-vcodec", "copy",
-        mp4_filename,
+        outfile,
     ]
     if not exec_subprocess(cmd):
-        return
+        # failed, error has already been logged
+        return False
     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)
+        flv_size = os.stat(infile).st_size
+        mp4_size = os.stat(outfile).st_size
+        if abs(flv_size - mp4_size) < 0.1 * flv_size:
+            os.unlink(infile)
+            return True
         else:
-            print >>sys.stderr, "The size of", mp4_filename, "is suspicious, did avconv fail?"
+            print >>sys.stderr, "The size of", outfile, "is suspicious, did avconv fail?"
+            return False
     except Exception, e:
-        print "Conversion failed", e
+        print >>sys.stderr, "Conversion failed", e
+        return False
 
-def convert_filename(filename):
-    if os.path.splitext(filename.lower())[1] in (".mp4", ".flv"):
-        f = open(filename)
+def convert_to_mp4(filename):
+    with open(filename) as f:
         fourcc = f.read(4)
-        f.close()
-        if fourcc == "FLV\x01":
-            convert_flv_mp4(filename)
+    basename, ext = os.path.splitext(filename)
+
+    if ext == ".mp4" and fourcc == "FLV\x01":
+        os.rename(filename, basename + ".flv")
+        ext = ".flv"
+        filename = basename + ext
+
+    if ext in (".flv", ".ts"):
+        filename_mp4 = basename + ".mp4"
+        return avconv_remux(filename, filename_mp4)
+
+    return ext == ".mp4"
+
 
 def download_rtmp(filename, vbase, vpath, hash_url=None):
     filename = sanify_filename(filename)
@@ -214,8 +222,7 @@ def download_rtmp(filename, vbase, vpath, hash_url=None):
     if hash_url is not None:
         cmd += ["--swfVfy", hash_url]
     if exec_subprocess(cmd):
-        convert_filename(filename)
-        return True
+        return convert_to_mp4(filename)
     else:
         return False
 
@@ -246,8 +253,91 @@ def download_urllib(filename, url, referrer=None):
         except:
             pass
 
-    convert_filename(filename)
-    return True
+    return convert_to_mp4(filename)
+
+def download_hls_get_stream(url):
+    def parse_bandwidth(line):
+        params = line.split(":", 1)[1].split(",")
+        for kv in params:
+            k, v = kv.split("=", 1)
+            if k == "BANDWIDTH":
+                return int(v)
+        return 0
+
+    m3u8 = grab_text(url, 0)
+    best_bandwidth = None
+    best_url = None
+    for line in m3u8.split("\n"):
+        if line.startswith("#EXT-X-STREAM-INF:"):
+            bandwidth = parse_bandwidth(line)
+            if best_bandwidth is None or bandwidth > best_bandwidth:
+                best_bandwidth = bandwidth
+                best_url = None
+        elif not line.startswith("#"):
+            if best_url is None:
+                best_url = line.strip()
+
+    if not best_url:
+        raise Exception("Failed to find best stream for HLS: " + url)
+
+    return best_url
+
+def download_hls_segments(outf, url):
+    m3u8 = grab_text(url, 0)
+
+    fail_if_not_last_segment = None
+    for line in m3u8.split("\n"):
+        if not line.strip() or line.startswith("#"):
+            continue
+
+        if fail_if_not_last_segment:
+            raise e
+
+        try:
+            download_hls_fetch_segment(outf, line)
+        except urllib2.HTTPError, e:
+            fail_if_not_last_segment = e
+            continue
+        sys.stdout.write(".")
+        sys.stdout.flush()
+
+    sys.stdout.write("\n")
+
+def download_hls_fetch_segment(outf, segment_url):
+    try:
+        src = _urlopen(segment_url)
+        shutil.copyfileobj(src, outf)
+    except:
+        raise
+    finally:
+        try:
+            src.close()
+        except:
+            pass
+
+def download_hls(filename, m3u8_master_url, hack_url_func=None):
+    if hack_url_func is None:
+        hack_url_func = lambda url: url
+
+    tmpdir = tempfile.mkdtemp(prefix="webdl-hls")
+
+    print "Downloading: %s" % filename
+
+    try:
+        best_stream_url = download_hls_get_stream(hack_url_func(m3u8_master_url))
+        ts_file = open(filename, "w")
+        download_hls_segments(ts_file, hack_url_func(best_stream_url))
+    except KeyboardInterrupt:
+        print "\nCancelled", m3u8_master_url
+        return False
+    finally:
+        shutil.rmtree(tmpdir)
+        try:
+            ts_file.close()
+        except:
+            pass
+
+    return convert_to_mp4(filename)
 
 def natural_sort(l, key=None):
     ignore_list = ["a", "the"]