[yt-dlp.git] / yt_dlp / downloader / rtmp.py

import os
import re
import subprocess
import time

from .common import FileDownloader
from ..utils import (
    Popen,
    check_executable,
    encodeArgument,
    encodeFilename,
    get_exe_version,
)


def rtmpdump_version():
    return get_exe_version(
        'rtmpdump', ['--help'], r'(?i)RTMPDump\s*v?([0-9a-zA-Z._-]+)')


class RtmpFD(FileDownloader):
    def real_download(self, filename, info_dict):
        def run_rtmpdump(args):
            start = time.time()
            resume_percent = None
            resume_downloaded_data_len = None
            proc = Popen(args, stderr=subprocess.PIPE)
            cursor_in_new_line = True
            proc_stderr_closed = False
            try:
                while not proc_stderr_closed:
                    # read line from stderr
                    line = ''
                    while True:
                        char = proc.stderr.read(1)
                        if not char:
                            proc_stderr_closed = True
                            break
                        if char in [b'\r', b'\n']:
                            break
                        line += char.decode('ascii', 'replace')
                    if not line:
                        # proc_stderr_closed is True
                        continue
                    mobj = re.search(r'([0-9]+\.[0-9]{3}) kB / [0-9]+\.[0-9]{2} sec \(([0-9]{1,2}\.[0-9])%\)', line)
                    if mobj:
                        downloaded_data_len = int(float(mobj.group(1)) * 1024)
                        percent = float(mobj.group(2))
                        if not resume_percent:
                            resume_percent = percent
                            resume_downloaded_data_len = downloaded_data_len
                        time_now = time.time()
                        eta = self.calc_eta(start, time_now, 100 - resume_percent, percent - resume_percent)
                        speed = self.calc_speed(start, time_now, downloaded_data_len - resume_downloaded_data_len)
                        data_len = None
                        if percent > 0:
                            data_len = int(downloaded_data_len * 100 / percent)
                        self._hook_progress({
                            'status': 'downloading',
                            'downloaded_bytes': downloaded_data_len,
                            'total_bytes_estimate': data_len,
                            'tmpfilename': tmpfilename,
                            'filename': filename,
                            'eta': eta,
                            'elapsed': time_now - start,
                            'speed': speed,
                        }, info_dict)
                        cursor_in_new_line = False
                    else:
                        # no percent for live streams
                        mobj = re.search(r'([0-9]+\.[0-9]{3}) kB / [0-9]+\.[0-9]{2} sec', line)
                        if mobj:
                            downloaded_data_len = int(float(mobj.group(1)) * 1024)
                            time_now = time.time()
                            speed = self.calc_speed(start, time_now, downloaded_data_len)
                            self._hook_progress({
                                'downloaded_bytes': downloaded_data_len,
                                'tmpfilename': tmpfilename,
                                'filename': filename,
                                'status': 'downloading',
                                'elapsed': time_now - start,
                                'speed': speed,
                            }, info_dict)
                            cursor_in_new_line = False
                        elif self.params.get('verbose', False):
                            if not cursor_in_new_line:
                                self.to_screen('')
                            cursor_in_new_line = True
                            self.to_screen('[rtmpdump] ' + line)
                if not cursor_in_new_line:
                    self.to_screen('')
                return proc.wait()
            except BaseException:  # Including KeyboardInterrupt
                proc.kill(timeout=None)
                raise

        url = info_dict['url']
        player_url = info_dict.get('player_url')
        page_url = info_dict.get('page_url')
        app = info_dict.get('app')
        play_path = info_dict.get('play_path')
        tc_url = info_dict.get('tc_url')
        flash_version = info_dict.get('flash_version')
        live = info_dict.get('rtmp_live', False)
        conn = info_dict.get('rtmp_conn')
        protocol = info_dict.get('rtmp_protocol')
        real_time = info_dict.get('rtmp_real_time', False)
        no_resume = info_dict.get('no_resume', False)
        continue_dl = self.params.get('continuedl', True)

        self.report_destination(filename)
        tmpfilename = self.temp_name(filename)
        test = self.params.get('test', False)

        # Check for rtmpdump first
        if not check_executable('rtmpdump', ['-h']):
            self.report_error('RTMP download detected but "rtmpdump" could not be run. Please install')
            return False

        # Download using rtmpdump. rtmpdump returns exit code 2 when
        # the connection was interrupted and resuming appears to be
        # possible. This is part of rtmpdump's normal usage, AFAIK.
        basic_args = [
            'rtmpdump', '--verbose', '-r', url,
            '-o', tmpfilename]
        if player_url is not None:
            basic_args += ['--swfVfy', player_url]
        if page_url is not None:
            basic_args += ['--pageUrl', page_url]
        if app is not None:
            basic_args += ['--app', app]
        if play_path is not None:
            basic_args += ['--playpath', play_path]
        if tc_url is not None:
            basic_args += ['--tcUrl', tc_url]
        if test:
            basic_args += ['--stop', '1']
        if flash_version is not None:
            basic_args += ['--flashVer', flash_version]
        if live:
            basic_args += ['--live']
        if isinstance(conn, list):
            for entry in conn:
                basic_args += ['--conn', entry]
        elif isinstance(conn, str):
            basic_args += ['--conn', conn]
        if protocol is not None:
            basic_args += ['--protocol', protocol]
        if real_time:
            basic_args += ['--realtime']

        args = basic_args
        if not no_resume and continue_dl and not live:
            args += ['--resume']
        if not live and continue_dl:
            args += ['--skip', '1']

        args = [encodeArgument(a) for a in args]

        self._debug_cmd(args, exe='rtmpdump')

        RD_SUCCESS = 0
        RD_FAILED = 1
        RD_INCOMPLETE = 2
        RD_NO_CONNECT = 3

        started = time.time()

        try:
            retval = run_rtmpdump(args)
        except KeyboardInterrupt:
            if not info_dict.get('is_live'):
                raise
            retval = RD_SUCCESS
            self.to_screen('\n[rtmpdump] Interrupted by user')

        if retval == RD_NO_CONNECT:
            self.report_error('[rtmpdump] Could not connect to RTMP server.')
            return False

        while retval in (RD_INCOMPLETE, RD_FAILED) and not test and not live:
            prevsize = os.path.getsize(encodeFilename(tmpfilename))
            self.to_screen(f'[rtmpdump] Downloaded {prevsize} bytes')
            time.sleep(5.0)  # This seems to be needed
            args = [*basic_args, '--resume']
            if retval == RD_FAILED:
                args += ['--skip', '1']
            args = [encodeArgument(a) for a in args]
            retval = run_rtmpdump(args)
            cursize = os.path.getsize(encodeFilename(tmpfilename))
            if prevsize == cursize and retval == RD_FAILED:
                break
            # Some rtmp streams seem abort after ~ 99.8%. Don't complain for those
            if prevsize == cursize and retval == RD_INCOMPLETE and cursize > 1024:
                self.to_screen('[rtmpdump] Could not download the whole video. This can happen for some advertisements.')
                retval = RD_SUCCESS
                break
        if retval == RD_SUCCESS or (test and retval == RD_INCOMPLETE):
            fsize = os.path.getsize(encodeFilename(tmpfilename))
            self.to_screen(f'[rtmpdump] Downloaded {fsize} bytes')
            self.try_rename(tmpfilename, filename)
            self._hook_progress({
                'downloaded_bytes': fsize,
                'total_bytes': fsize,
                'filename': filename,
                'status': 'finished',
                'elapsed': time.time() - started,
            }, info_dict)
            return True
        else:
            self.to_stderr('\n')
            self.report_error('rtmpdump exited with code %d' % retval)
            return False
Commit	Line	Data
3bc2ddcc JMF	1	import os
	2	import re
	3	import subprocess
3bc2ddcc JMF	4	import time
	5
	6	from .common import FileDownloader
	7	from ..utils import (
f8271158	8	Popen,
7798fad5	9	check_executable,
9e105a85	10	encodeArgument,
f8271158	11	encodeFilename,
4c83c967	12	get_exe_version,
3bc2ddcc JMF	13	)
	14
	15
4c83c967 PH	16	def rtmpdump_version():
	17	return get_exe_version(
	18	'rtmpdump', ['--help'], r'(?i)RTMPDump\s*v?([0-9a-zA-Z._-]+)')
	19
	20
3bc2ddcc JMF	21	class RtmpFD(FileDownloader):
	22	def real_download(self, filename, info_dict):
	23	def run_rtmpdump(args):
	24	start = time.time()
9b0b6275 S	25	resume_percent = None
9b0b6275 S	26	resume_downloaded_data_len = None
d3c93ec2	27	proc = Popen(args, stderr=subprocess.PIPE)
3bc2ddcc	28	cursor_in_new_line = True
9b0b6275 S	29	proc_stderr_closed = False
9b0b6275 S	30	try:
ddd8486a S	31	while not proc_stderr_closed:
	32	# read line from stderr
	33	line = ''
	34	while True:
	35	char = proc.stderr.read(1)
	36	if not char:
	37	proc_stderr_closed = True
	38	break
	39	if char in [b'\r', b'\n']:
	40	break
	41	line += char.decode('ascii', 'replace')
	42	if not line:
	43	# proc_stderr_closed is True
	44	continue
	45	mobj = re.search(r'([0-9]+\.[0-9]{3}) kB / [0-9]+\.[0-9]{2} sec \(([0-9]{1,2}\.[0-9])%\)', line)
3bc2ddcc	46	if mobj:
2514d263	47	downloaded_data_len = int(float(mobj.group(1)) * 1024)
ddd8486a S	48	percent = float(mobj.group(2))
	49	if not resume_percent:
	50	resume_percent = percent
	51	resume_downloaded_data_len = downloaded_data_len
3bc2ddcc	52	time_now = time.time()
ddd8486a S	53	eta = self.calc_eta(start, time_now, 100 - resume_percent, percent - resume_percent)
	54	speed = self.calc_speed(start, time_now, downloaded_data_len - resume_downloaded_data_len)
	55	data_len = None
	56	if percent > 0:
	57	data_len = int(downloaded_data_len * 100 / percent)
3bc2ddcc	58	self._hook_progress({
ddd8486a	59	'status': 'downloading',
3bc2ddcc	60	'downloaded_bytes': downloaded_data_len,
ddd8486a	61	'total_bytes_estimate': data_len,
3bc2ddcc JMF	62	'tmpfilename': tmpfilename,
3bc2ddcc JMF	63	'filename': filename,
ddd8486a	64	'eta': eta,
5cda4eda	65	'elapsed': time_now - start,
3bc2ddcc	66	'speed': speed,
3ba7740d	67	}, info_dict)
5cda4eda	68	cursor_in_new_line = False
ddd8486a S	69	else:
	70	# no percent for live streams
	71	mobj = re.search(r'([0-9]+\.[0-9]{3}) kB / [0-9]+\.[0-9]{2} sec', line)
	72	if mobj:
	73	downloaded_data_len = int(float(mobj.group(1)) * 1024)
	74	time_now = time.time()
	75	speed = self.calc_speed(start, time_now, downloaded_data_len)
	76	self._hook_progress({
	77	'downloaded_bytes': downloaded_data_len,
	78	'tmpfilename': tmpfilename,
	79	'filename': filename,
	80	'status': 'downloading',
	81	'elapsed': time_now - start,
	82	'speed': speed,
3ba7740d	83	}, info_dict)
ddd8486a S	84	cursor_in_new_line = False
	85	elif self.params.get('verbose', False):
	86	if not cursor_in_new_line:
	87	self.to_screen('')
	88	cursor_in_new_line = True
	89	self.to_screen('[rtmpdump] ' + line)
f5b1bca9	90	if not cursor_in_new_line:
	91	self.to_screen('')
	92	return proc.wait()
	93	except BaseException: # Including KeyboardInterrupt
f0c9fb96	94	proc.kill(timeout=None)
f5b1bca9	95	raise
3bc2ddcc JMF	96
3bc2ddcc JMF	97	url = info_dict['url']
d800609c S	98	player_url = info_dict.get('player_url')
	99	page_url = info_dict.get('page_url')
	100	app = info_dict.get('app')
	101	play_path = info_dict.get('play_path')
	102	tc_url = info_dict.get('tc_url')
	103	flash_version = info_dict.get('flash_version')
3bc2ddcc	104	live = info_dict.get('rtmp_live', False)
d800609c S	105	conn = info_dict.get('rtmp_conn')
d800609c S	106	protocol = info_dict.get('rtmp_protocol')
7bb3ceb4	107	real_time = info_dict.get('rtmp_real_time', False)
7906d199	108	no_resume = info_dict.get('no_resume', False)
f101079a	109	continue_dl = self.params.get('continuedl', True)
3dee7826	110
3bc2ddcc JMF	111	self.report_destination(filename)
	112	tmpfilename = self.temp_name(filename)
	113	test = self.params.get('test', False)
	114
	115	# Check for rtmpdump first
7798fad5	116	if not check_executable('rtmpdump', ['-h']):
beb4b92a	117	self.report_error('RTMP download detected but "rtmpdump" could not be run. Please install')
3bc2ddcc JMF	118	return False
	119
	120	# Download using rtmpdump. rtmpdump returns exit code 2 when
17cc1534	121	# the connection was interrupted and resuming appears to be
3bc2ddcc	122	# possible. This is part of rtmpdump's normal usage, AFAIK.
2a15a98a PH	123	basic_args = [
2a15a98a PH	124	'rtmpdump', '--verbose', '-r', url,
9e105a85	125	'-o', tmpfilename]
3bc2ddcc JMF	126	if player_url is not None:
	127	basic_args += ['--swfVfy', player_url]
	128	if page_url is not None:
	129	basic_args += ['--pageUrl', page_url]
082c6c86 S	130	if app is not None:
082c6c86 S	131	basic_args += ['--app', app]
3bc2ddcc JMF	132	if play_path is not None:
	133	basic_args += ['--playpath', play_path]
	134	if tc_url is not None:
156fc83a	135	basic_args += ['--tcUrl', tc_url]
3bc2ddcc JMF	136	if test:
3bc2ddcc JMF	137	basic_args += ['--stop', '1']
082c6c86 S	138	if flash_version is not None:
082c6c86 S	139	basic_args += ['--flashVer', flash_version]
3bc2ddcc JMF	140	if live:
3bc2ddcc JMF	141	basic_args += ['--live']
eb451334 S	142	if isinstance(conn, list):
	143	for entry in conn:
	144	basic_args += ['--conn', entry]
14f25df2	145	elif isinstance(conn, str):
3bc2ddcc	146	basic_args += ['--conn', conn]
087ca2cb JMF	147	if protocol is not None:
087ca2cb JMF	148	basic_args += ['--protocol', protocol]
0865f397 PH	149	if real_time:
0865f397 PH	150	basic_args += ['--realtime']
3dee7826 PH	151
	152	args = basic_args
	153	if not no_resume and continue_dl and not live:
	154	args += ['--resume']
	155	if not live and continue_dl:
	156	args += ['--skip', '1']
3bc2ddcc	157
9e105a85	158	args = [encodeArgument(a) for a in args]
3bc2ddcc	159
9e105a85	160	self._debug_cmd(args, exe='rtmpdump')
3bc2ddcc	161
35241756 S	162	RD_SUCCESS = 0
	163	RD_FAILED = 1
	164	RD_INCOMPLETE = 2
52d6a9a6	165	RD_NO_CONNECT = 3
35241756	166
f16f4877 S	167	started = time.time()
f16f4877 S	168
ddd8486a S	169	try:
	170	retval = run_rtmpdump(args)
	171	except KeyboardInterrupt:
	172	if not info_dict.get('is_live'):
	173	raise
	174	retval = RD_SUCCESS
	175	self.to_screen('\n[rtmpdump] Interrupted by user')
3bc2ddcc	176
52d6a9a6	177	if retval == RD_NO_CONNECT:
8dec03ec	178	self.report_error('[rtmpdump] Could not connect to RTMP server.')
52d6a9a6 S	179	return False
52d6a9a6 S	180
40fcba5e	181	while retval in (RD_INCOMPLETE, RD_FAILED) and not test and not live:
3bc2ddcc	182	prevsize = os.path.getsize(encodeFilename(tmpfilename))
add96eb9	183	self.to_screen(f'[rtmpdump] Downloaded {prevsize} bytes')
5f6a1245	184	time.sleep(5.0) # This seems to be needed
add96eb9	185	args = [*basic_args, '--resume']
9e105a85 S	186	if retval == RD_FAILED:
	187	args += ['--skip', '1']
	188	args = [encodeArgument(a) for a in args]
	189	retval = run_rtmpdump(args)
3bc2ddcc	190	cursize = os.path.getsize(encodeFilename(tmpfilename))
35241756	191	if prevsize == cursize and retval == RD_FAILED:
3bc2ddcc	192	break
7af808a5	193	# Some rtmp streams seem abort after ~ 99.8%. Don't complain for those
35241756	194	if prevsize == cursize and retval == RD_INCOMPLETE and cursize > 1024:
8dec03ec	195	self.to_screen('[rtmpdump] Could not download the whole video. This can happen for some advertisements.')
35241756	196	retval = RD_SUCCESS
3bc2ddcc	197	break
35241756	198	if retval == RD_SUCCESS or (test and retval == RD_INCOMPLETE):
3bc2ddcc	199	fsize = os.path.getsize(encodeFilename(tmpfilename))
add96eb9	200	self.to_screen(f'[rtmpdump] Downloaded {fsize} bytes')
3bc2ddcc JMF	201	self.try_rename(tmpfilename, filename)
	202	self._hook_progress({
	203	'downloaded_bytes': fsize,
	204	'total_bytes': fsize,
	205	'filename': filename,
	206	'status': 'finished',
f16f4877	207	'elapsed': time.time() - started,
3ba7740d	208	}, info_dict)
3bc2ddcc JMF	209	return True
3bc2ddcc JMF	210	else:
8dec03ec S	211	self.to_stderr('\n')
8dec03ec S	212	self.report_error('rtmpdump exited with code %d' % retval)
3bc2ddcc	213	return False