Merge pull request #6392 from dstftw/generalized-fragmented-fd

Generalized fragmented media file downloader
This commit is contained in:
Sergey M. 2015-08-01 20:22:49 +06:00
commit 75e8b2ac87
3 changed files with 165 additions and 124 deletions

View File

@ -7,8 +7,7 @@ import os
import time import time
import xml.etree.ElementTree as etree import xml.etree.ElementTree as etree
from .common import FileDownloader from .fragment import FragmentFD
from .http import HttpFD
from ..compat import ( from ..compat import (
compat_urlparse, compat_urlparse,
compat_urllib_error, compat_urllib_error,
@ -16,8 +15,6 @@ from ..compat import (
from ..utils import ( from ..utils import (
struct_pack, struct_pack,
struct_unpack, struct_unpack,
encodeFilename,
sanitize_open,
xpath_text, xpath_text,
) )
@ -226,16 +223,13 @@ def _add_ns(prop):
return '{http://ns.adobe.com/f4m/1.0}%s' % prop return '{http://ns.adobe.com/f4m/1.0}%s' % prop
class HttpQuietDownloader(HttpFD): class F4mFD(FragmentFD):
def to_screen(self, *args, **kargs):
pass
class F4mFD(FileDownloader):
""" """
A downloader for f4m manifests or AdobeHDS. A downloader for f4m manifests or AdobeHDS.
""" """
FD_NAME = 'f4m'
def _get_unencrypted_media(self, doc): def _get_unencrypted_media(self, doc):
media = doc.findall(_add_ns('media')) media = doc.findall(_add_ns('media'))
if not media: if not media:
@ -288,7 +282,7 @@ class F4mFD(FileDownloader):
def real_download(self, filename, info_dict): def real_download(self, filename, info_dict):
man_url = info_dict['url'] man_url = info_dict['url']
requested_bitrate = info_dict.get('tbr') requested_bitrate = info_dict.get('tbr')
self.to_screen('[download] Downloading f4m manifest') self.to_screen('[%s] Downloading f4m manifest' % self.FD_NAME)
manifest = self.ydl.urlopen(man_url).read() manifest = self.ydl.urlopen(man_url).read()
doc = etree.fromstring(manifest) doc = etree.fromstring(manifest)
@ -320,67 +314,20 @@ class F4mFD(FileDownloader):
# For some akamai manifests we'll need to add a query to the fragment url # For some akamai manifests we'll need to add a query to the fragment url
akamai_pv = xpath_text(doc, _add_ns('pv-2.0')) akamai_pv = xpath_text(doc, _add_ns('pv-2.0'))
self.report_destination(filename) ctx = {
http_dl = HttpQuietDownloader( 'filename': filename,
self.ydl, 'total_frags': total_frags,
{
'continuedl': True,
'quiet': True,
'noprogress': True,
'ratelimit': self.params.get('ratelimit', None),
'test': self.params.get('test', False),
} }
)
tmpfilename = self.temp_name(filename) self._prepare_frag_download(ctx)
(dest_stream, tmpfilename) = sanitize_open(tmpfilename, 'wb')
dest_stream = ctx['dest_stream']
write_flv_header(dest_stream) write_flv_header(dest_stream)
if not live: if not live:
write_metadata_tag(dest_stream, metadata) write_metadata_tag(dest_stream, metadata)
# This dict stores the download progress, it's updated by the progress self._start_frag_download(ctx)
# hook
state = {
'status': 'downloading',
'downloaded_bytes': 0,
'frag_index': 0,
'frag_count': total_frags,
'filename': filename,
'tmpfilename': tmpfilename,
}
start = time.time()
def frag_progress_hook(s):
if s['status'] not in ('downloading', 'finished'):
return
frag_total_bytes = s.get('total_bytes', 0)
if s['status'] == 'finished':
state['downloaded_bytes'] += frag_total_bytes
state['frag_index'] += 1
estimated_size = (
(state['downloaded_bytes'] + frag_total_bytes) /
(state['frag_index'] + 1) * total_frags)
time_now = time.time()
state['total_bytes_estimate'] = estimated_size
state['elapsed'] = time_now - start
if s['status'] == 'finished':
progress = self.calc_percent(state['frag_index'], total_frags)
else:
frag_downloaded_bytes = s['downloaded_bytes']
frag_progress = self.calc_percent(frag_downloaded_bytes,
frag_total_bytes)
progress = self.calc_percent(state['frag_index'], total_frags)
progress += frag_progress / float(total_frags)
state['eta'] = self.calc_eta(
start, time_now, estimated_size, state['downloaded_bytes'] + frag_downloaded_bytes)
state['speed'] = s.get('speed')
self._hook_progress(state)
http_dl.add_progress_hook(frag_progress_hook)
frags_filenames = [] frags_filenames = []
while fragments_list: while fragments_list:
@ -391,9 +338,9 @@ class F4mFD(FileDownloader):
url += '?' + akamai_pv.strip(';') url += '?' + akamai_pv.strip(';')
if info_dict.get('extra_param_to_segment_url'): if info_dict.get('extra_param_to_segment_url'):
url += info_dict.get('extra_param_to_segment_url') url += info_dict.get('extra_param_to_segment_url')
frag_filename = '%s-%s' % (tmpfilename, name) frag_filename = '%s-%s' % (ctx['tmpfilename'], name)
try: try:
success = http_dl.download(frag_filename, {'url': url}) success = ctx['dl'].download(frag_filename, {'url': url})
if not success: if not success:
return False return False
with open(frag_filename, 'rb') as down: with open(frag_filename, 'rb') as down:
@ -425,20 +372,9 @@ class F4mFD(FileDownloader):
msg = 'Missed %d fragments' % (fragments_list[0][1] - (frag_i + 1)) msg = 'Missed %d fragments' % (fragments_list[0][1] - (frag_i + 1))
self.report_warning(msg) self.report_warning(msg)
dest_stream.close() self._finish_frag_download(ctx)
elapsed = time.time() - start
self.try_rename(tmpfilename, filename)
for frag_file in frags_filenames: for frag_file in frags_filenames:
os.remove(frag_file) os.remove(frag_file)
fsize = os.path.getsize(encodeFilename(filename))
self._hook_progress({
'downloaded_bytes': fsize,
'total_bytes': fsize,
'filename': filename,
'status': 'finished',
'elapsed': elapsed,
})
return True return True

View File

@ -0,0 +1,110 @@
from __future__ import division, unicode_literals
import os
import time
from .common import FileDownloader
from .http import HttpFD
from ..utils import (
encodeFilename,
sanitize_open,
)
class HttpQuietDownloader(HttpFD):
def to_screen(self, *args, **kargs):
pass
class FragmentFD(FileDownloader):
"""
A base file downloader class for fragmented media (e.g. f4m/m3u8 manifests).
"""
def _prepare_and_start_frag_download(self, ctx):
self._prepare_frag_download(ctx)
self._start_frag_download(ctx)
def _prepare_frag_download(self, ctx):
self.to_screen('[%s] Total fragments: %d' % (self.FD_NAME, ctx['total_frags']))
self.report_destination(ctx['filename'])
dl = HttpQuietDownloader(
self.ydl,
{
'continuedl': True,
'quiet': True,
'noprogress': True,
'ratelimit': self.params.get('ratelimit', None),
'test': self.params.get('test', False),
}
)
tmpfilename = self.temp_name(ctx['filename'])
dest_stream, tmpfilename = sanitize_open(tmpfilename, 'wb')
ctx.update({
'dl': dl,
'dest_stream': dest_stream,
'tmpfilename': tmpfilename,
})
def _start_frag_download(self, ctx):
total_frags = ctx['total_frags']
# This dict stores the download progress, it's updated by the progress
# hook
state = {
'status': 'downloading',
'downloaded_bytes': 0,
'frag_index': 0,
'frag_count': total_frags,
'filename': ctx['filename'],
'tmpfilename': ctx['tmpfilename'],
}
start = time.time()
ctx['started'] = start
def frag_progress_hook(s):
if s['status'] not in ('downloading', 'finished'):
return
frag_total_bytes = s.get('total_bytes', 0)
if s['status'] == 'finished':
state['downloaded_bytes'] += frag_total_bytes
state['frag_index'] += 1
estimated_size = (
(state['downloaded_bytes'] + frag_total_bytes) /
(state['frag_index'] + 1) * total_frags)
time_now = time.time()
state['total_bytes_estimate'] = estimated_size
state['elapsed'] = time_now - start
if s['status'] == 'finished':
progress = self.calc_percent(state['frag_index'], total_frags)
else:
frag_downloaded_bytes = s['downloaded_bytes']
frag_progress = self.calc_percent(frag_downloaded_bytes,
frag_total_bytes)
progress = self.calc_percent(state['frag_index'], total_frags)
progress += frag_progress / float(total_frags)
state['eta'] = self.calc_eta(
start, time_now, estimated_size, state['downloaded_bytes'] + frag_downloaded_bytes)
state['speed'] = s.get('speed')
self._hook_progress(state)
ctx['dl'].add_progress_hook(frag_progress_hook)
return start
def _finish_frag_download(self, ctx):
ctx['dest_stream'].close()
elapsed = time.time() - ctx['started']
self.try_rename(ctx['tmpfilename'], ctx['filename'])
fsize = os.path.getsize(encodeFilename(ctx['filename']))
self._hook_progress({
'downloaded_bytes': fsize,
'total_bytes': fsize,
'filename': ctx['filename'],
'status': 'finished',
'elapsed': elapsed,
})

View File

@ -4,12 +4,11 @@ import os
import re import re
import subprocess import subprocess
from ..postprocessor.ffmpeg import FFmpegPostProcessor
from .common import FileDownloader from .common import FileDownloader
from ..compat import ( from .fragment import FragmentFD
compat_urlparse,
compat_urllib_request, from ..compat import compat_urlparse
) from ..postprocessor.ffmpeg import FFmpegPostProcessor
from ..utils import ( from ..utils import (
encodeArgument, encodeArgument,
encodeFilename, encodeFilename,
@ -51,54 +50,50 @@ class HlsFD(FileDownloader):
return False return False
class NativeHlsFD(FileDownloader): class NativeHlsFD(FragmentFD):
""" A more limited implementation that does not require ffmpeg """ """ A more limited implementation that does not require ffmpeg """
def real_download(self, filename, info_dict): FD_NAME = 'hlsnative'
url = info_dict['url']
self.report_destination(filename)
tmpfilename = self.temp_name(filename)
self.to_screen( def real_download(self, filename, info_dict):
'[hlsnative] %s: Downloading m3u8 manifest' % info_dict['id']) man_url = info_dict['url']
data = self.ydl.urlopen(url).read() self.to_screen('[%s] Downloading m3u8 manifest' % self.FD_NAME)
s = data.decode('utf-8', 'ignore') manifest = self.ydl.urlopen(man_url).read()
segment_urls = []
s = manifest.decode('utf-8', 'ignore')
fragment_urls = []
for line in s.splitlines(): for line in s.splitlines():
line = line.strip() line = line.strip()
if line and not line.startswith('#'): if line and not line.startswith('#'):
segment_url = ( segment_url = (
line line
if re.match(r'^https?://', line) if re.match(r'^https?://', line)
else compat_urlparse.urljoin(url, line)) else compat_urlparse.urljoin(man_url, line))
segment_urls.append(segment_url) fragment_urls.append(segment_url)
# We only download the first fragment during the test
is_test = self.params.get('test', False) if self.params.get('test', False):
remaining_bytes = self._TEST_FILE_SIZE if is_test else None
byte_counter = 0
with open(tmpfilename, 'wb') as outf:
for i, segurl in enumerate(segment_urls):
self.to_screen(
'[hlsnative] %s: Downloading segment %d / %d' %
(info_dict['id'], i + 1, len(segment_urls)))
seg_req = compat_urllib_request.Request(segurl)
if remaining_bytes is not None:
seg_req.add_header('Range', 'bytes=0-%d' % (remaining_bytes - 1))
segment = self.ydl.urlopen(seg_req).read()
if remaining_bytes is not None:
segment = segment[:remaining_bytes]
remaining_bytes -= len(segment)
outf.write(segment)
byte_counter += len(segment)
if remaining_bytes is not None and remaining_bytes <= 0:
break break
self._hook_progress({ ctx = {
'downloaded_bytes': byte_counter,
'total_bytes': byte_counter,
'filename': filename, 'filename': filename,
'status': 'finished', 'total_frags': len(fragment_urls),
}) }
self.try_rename(tmpfilename, filename)
self._prepare_and_start_frag_download(ctx)
frags_filenames = []
for i, frag_url in enumerate(fragment_urls):
frag_filename = '%s-Frag%d' % (ctx['tmpfilename'], i)
success = ctx['dl'].download(frag_filename, {'url': frag_url})
if not success:
return False
with open(frag_filename, 'rb') as down:
ctx['dest_stream'].write(down.read())
frags_filenames.append(frag_filename)
self._finish_frag_download(ctx)
for frag_file in frags_filenames:
os.remove(frag_file)
return True return True