import base64
import io
import itertools
-import os
import time
from .fragment import FragmentFD
compat_struct_unpack,
)
from ..utils import (
- encodeFilename,
fix_xml_ampersands,
- sanitize_open,
xpath_text,
)
first_frag_number = fragment_run_entry_table[0]['first']
fragments_counter = itertools.count(first_frag_number)
for segment, fragments_count in segment_run_table['segment_run']:
+ # In some live HDS streams (for example Rai), `fragments_count` is
+ # abnormal and causing out-of-memory errors. It's OK to change the
+ # number of fragments for live streams as they are updated periodically
+ if fragments_count == 4294967295 and boot_info['live']:
+ fragments_count = 2
for _ in range(fragments_count):
res.append((segment, next(fragments_counter)))
media))
-def _add_ns(prop):
- return '{http://ns.adobe.com/f4m/1.0}%s' % prop
+def _add_ns(prop, ver=1):
+ return '{http://ns.adobe.com/f4m/%d.0}%s' % (ver, prop)
+
+
+def get_base_url(manifest):
+ base_url = xpath_text(
+ manifest, [_add_ns('baseURL'), _add_ns('baseURL', 2)],
+ 'base URL', default=None)
+ if base_url:
+ base_url = base_url.strip()
+ return base_url
class F4mFD(FragmentFD):
man_url = info_dict['url']
requested_bitrate = info_dict.get('tbr')
self.to_screen('[%s] Downloading f4m manifest' % self.FD_NAME)
- urlh = self.ydl.urlopen(man_url)
+
+ urlh = self.ydl.urlopen(self._prepare_url(info_dict, man_url))
man_url = urlh.geturl()
# Some manifests may be malformed, e.g. prosiebensat1 generated manifests
# (see https://github.com/rg3/youtube-dl/issues/6215#issuecomment-121704244
rate, media = list(filter(
lambda f: int(f[0]) == requested_bitrate, formats))[0]
- base_url = compat_urlparse.urljoin(man_url, media.attrib['url'])
+ # Prefer baseURL for relative URLs as per 11.2 of F4M 3.0 spec.
+ man_base_url = get_base_url(doc) or man_url
+
+ base_url = compat_urlparse.urljoin(man_base_url, media.attrib['url'])
bootstrap_node = doc.find(_add_ns('bootstrapInfo'))
- boot_info, bootstrap_url = self._parse_bootstrap_node(bootstrap_node, base_url)
+ boot_info, bootstrap_url = self._parse_bootstrap_node(
+ bootstrap_node, man_base_url)
live = boot_info['live']
metadata_node = media.find(_add_ns('metadata'))
if metadata_node is not None:
dest_stream = ctx['dest_stream']
- write_flv_header(dest_stream)
- if not live:
- write_metadata_tag(dest_stream, metadata)
+ if ctx['complete_frags_downloaded_bytes'] == 0:
+ write_flv_header(dest_stream)
+ if not live:
+ write_metadata_tag(dest_stream, metadata)
base_url_parsed = compat_urllib_parse_urlparse(base_url)
self._start_frag_download(ctx)
- frags_filenames = []
+ frag_index = 0
while fragments_list:
seg_i, frag_i = fragments_list.pop(0)
+ frag_index += 1
+ if frag_index <= ctx['fragment_index']:
+ continue
name = 'Seg%d-Frag%d' % (seg_i, frag_i)
query = []
if base_url_parsed.query:
if info_dict.get('extra_param_to_segment_url'):
query.append(info_dict['extra_param_to_segment_url'])
url_parsed = base_url_parsed._replace(path=base_url_parsed.path + name, query='&'.join(query))
- frag_filename = '%s-%s' % (ctx['tmpfilename'], name)
try:
- success = ctx['dl'].download(frag_filename, {'url': url_parsed.geturl()})
+ success, down_data = self._download_fragment(ctx, url_parsed.geturl(), info_dict)
if not success:
return False
- (down, frag_sanitized) = sanitize_open(frag_filename, 'rb')
- down_data = down.read()
- down.close()
reader = FlvReader(down_data)
while True:
try:
break
raise
if box_type == b'mdat':
- dest_stream.write(box_data)
+ self._append_fragment(ctx, box_data)
break
- if live:
- os.remove(encodeFilename(frag_sanitized))
- else:
- frags_filenames.append(frag_sanitized)
except (compat_urllib_error.HTTPError, ) as err:
if live and (err.code == 404 or err.code == 410):
# We didn't keep up with the live window. Continue
self._finish_frag_download(ctx)
- for frag_file in frags_filenames:
- os.remove(encodeFilename(frag_file))
-
return True