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

import os
import re
import subprocess
import time

from .common import FileDownloader
from ..compat import compat_str
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, compat_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('[rtmpdump] Downloaded %s bytes' % prevsize)
            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('[rtmpdump] Downloaded %s bytes' % fsize)
            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
1cc79574	7	from ..compat import compat_str
3bc2ddcc	8	from ..utils import (
f8271158	9	Popen,
7798fad5	10	check_executable,
9e105a85	11	encodeArgument,
f8271158	12	encodeFilename,
4c83c967	13	get_exe_version,
3bc2ddcc JMF	14	)
	15
	16
4c83c967 PH	17	def rtmpdump_version():
	18	return get_exe_version(
	19	'rtmpdump', ['--help'], r'(?i)RTMPDump\s*v?([0-9a-zA-Z._-]+)')
	20
	21
3bc2ddcc JMF	22	class RtmpFD(FileDownloader):
	23	def real_download(self, filename, info_dict):
	24	def run_rtmpdump(args):
	25	start = time.time()
9b0b6275 S	26	resume_percent = None
9b0b6275 S	27	resume_downloaded_data_len = None
d3c93ec2	28	proc = Popen(args, stderr=subprocess.PIPE)
3bc2ddcc	29	cursor_in_new_line = True
9b0b6275 S	30	proc_stderr_closed = False
9b0b6275 S	31	try:
ddd8486a S	32	while not proc_stderr_closed:
	33	# read line from stderr
	34	line = ''
	35	while True:
	36	char = proc.stderr.read(1)
	37	if not char:
	38	proc_stderr_closed = True
	39	break
	40	if char in [b'\r', b'\n']:
	41	break
	42	line += char.decode('ascii', 'replace')
	43	if not line:
	44	# proc_stderr_closed is True
	45	continue
	46	mobj = re.search(r'([0-9]+\.[0-9]{3}) kB / [0-9]+\.[0-9]{2} sec \(([0-9]{1,2}\.[0-9])%\)', line)
3bc2ddcc	47	if mobj:
2514d263	48	downloaded_data_len = int(float(mobj.group(1)) * 1024)
ddd8486a S	49	percent = float(mobj.group(2))
	50	if not resume_percent:
	51	resume_percent = percent
	52	resume_downloaded_data_len = downloaded_data_len
3bc2ddcc	53	time_now = time.time()
ddd8486a S	54	eta = self.calc_eta(start, time_now, 100 - resume_percent, percent - resume_percent)
	55	speed = self.calc_speed(start, time_now, downloaded_data_len - resume_downloaded_data_len)
	56	data_len = None
	57	if percent > 0:
	58	data_len = int(downloaded_data_len * 100 / percent)
3bc2ddcc	59	self._hook_progress({
ddd8486a	60	'status': 'downloading',
3bc2ddcc	61	'downloaded_bytes': downloaded_data_len,
ddd8486a	62	'total_bytes_estimate': data_len,
3bc2ddcc JMF	63	'tmpfilename': tmpfilename,
3bc2ddcc JMF	64	'filename': filename,
ddd8486a	65	'eta': eta,
5cda4eda	66	'elapsed': time_now - start,
3bc2ddcc	67	'speed': speed,
3ba7740d	68	}, info_dict)
5cda4eda	69	cursor_in_new_line = False
ddd8486a S	70	else:
	71	# no percent for live streams
	72	mobj = re.search(r'([0-9]+\.[0-9]{3}) kB / [0-9]+\.[0-9]{2} sec', line)
	73	if mobj:
	74	downloaded_data_len = int(float(mobj.group(1)) * 1024)
	75	time_now = time.time()
	76	speed = self.calc_speed(start, time_now, downloaded_data_len)
	77	self._hook_progress({
	78	'downloaded_bytes': downloaded_data_len,
	79	'tmpfilename': tmpfilename,
	80	'filename': filename,
	81	'status': 'downloading',
	82	'elapsed': time_now - start,
	83	'speed': speed,
3ba7740d	84	}, info_dict)
ddd8486a S	85	cursor_in_new_line = False
	86	elif self.params.get('verbose', False):
	87	if not cursor_in_new_line:
	88	self.to_screen('')
	89	cursor_in_new_line = True
	90	self.to_screen('[rtmpdump] ' + line)
f5b1bca9	91	if not cursor_in_new_line:
	92	self.to_screen('')
	93	return proc.wait()
	94	except BaseException: # Including KeyboardInterrupt
f0c9fb96	95	proc.kill(timeout=None)
f5b1bca9	96	raise
3bc2ddcc JMF	97
3bc2ddcc JMF	98	url = info_dict['url']
d800609c S	99	player_url = info_dict.get('player_url')
	100	page_url = info_dict.get('page_url')
	101	app = info_dict.get('app')
	102	play_path = info_dict.get('play_path')
	103	tc_url = info_dict.get('tc_url')
	104	flash_version = info_dict.get('flash_version')
3bc2ddcc	105	live = info_dict.get('rtmp_live', False)
d800609c S	106	conn = info_dict.get('rtmp_conn')
d800609c S	107	protocol = info_dict.get('rtmp_protocol')
7bb3ceb4	108	real_time = info_dict.get('rtmp_real_time', False)
7906d199	109	no_resume = info_dict.get('no_resume', False)
f101079a	110	continue_dl = self.params.get('continuedl', True)
3dee7826	111
3bc2ddcc JMF	112	self.report_destination(filename)
	113	tmpfilename = self.temp_name(filename)
	114	test = self.params.get('test', False)
	115
	116	# Check for rtmpdump first
7798fad5	117	if not check_executable('rtmpdump', ['-h']):
beb4b92a	118	self.report_error('RTMP download detected but "rtmpdump" could not be run. Please install')
3bc2ddcc JMF	119	return False
	120
	121	# Download using rtmpdump. rtmpdump returns exit code 2 when
17cc1534	122	# the connection was interrupted and resuming appears to be
3bc2ddcc	123	# possible. This is part of rtmpdump's normal usage, AFAIK.
2a15a98a PH	124	basic_args = [
2a15a98a PH	125	'rtmpdump', '--verbose', '-r', url,
9e105a85	126	'-o', tmpfilename]
3bc2ddcc JMF	127	if player_url is not None:
	128	basic_args += ['--swfVfy', player_url]
	129	if page_url is not None:
	130	basic_args += ['--pageUrl', page_url]
082c6c86 S	131	if app is not None:
082c6c86 S	132	basic_args += ['--app', app]
3bc2ddcc JMF	133	if play_path is not None:
	134	basic_args += ['--playpath', play_path]
	135	if tc_url is not None:
156fc83a	136	basic_args += ['--tcUrl', tc_url]
3bc2ddcc JMF	137	if test:
3bc2ddcc JMF	138	basic_args += ['--stop', '1']
082c6c86 S	139	if flash_version is not None:
082c6c86 S	140	basic_args += ['--flashVer', flash_version]
3bc2ddcc JMF	141	if live:
3bc2ddcc JMF	142	basic_args += ['--live']
eb451334 S	143	if isinstance(conn, list):
	144	for entry in conn:
	145	basic_args += ['--conn', entry]
	146	elif isinstance(conn, compat_str):
3bc2ddcc	147	basic_args += ['--conn', conn]
087ca2cb JMF	148	if protocol is not None:
087ca2cb JMF	149	basic_args += ['--protocol', protocol]
0865f397 PH	150	if real_time:
0865f397 PH	151	basic_args += ['--realtime']
3dee7826 PH	152
	153	args = basic_args
	154	if not no_resume and continue_dl and not live:
	155	args += ['--resume']
	156	if not live and continue_dl:
	157	args += ['--skip', '1']
3bc2ddcc	158
9e105a85	159	args = [encodeArgument(a) for a in args]
3bc2ddcc	160
9e105a85	161	self._debug_cmd(args, exe='rtmpdump')
3bc2ddcc	162
35241756 S	163	RD_SUCCESS = 0
	164	RD_FAILED = 1
	165	RD_INCOMPLETE = 2
52d6a9a6	166	RD_NO_CONNECT = 3
35241756	167
f16f4877 S	168	started = time.time()
f16f4877 S	169
ddd8486a S	170	try:
	171	retval = run_rtmpdump(args)
	172	except KeyboardInterrupt:
	173	if not info_dict.get('is_live'):
	174	raise
	175	retval = RD_SUCCESS
	176	self.to_screen('\n[rtmpdump] Interrupted by user')
3bc2ddcc	177
52d6a9a6	178	if retval == RD_NO_CONNECT:
8dec03ec	179	self.report_error('[rtmpdump] Could not connect to RTMP server.')
52d6a9a6 S	180	return False
52d6a9a6 S	181
40fcba5e	182	while retval in (RD_INCOMPLETE, RD_FAILED) and not test and not live:
3bc2ddcc	183	prevsize = os.path.getsize(encodeFilename(tmpfilename))
f16f4877	184	self.to_screen('[rtmpdump] Downloaded %s bytes' % prevsize)
5f6a1245	185	time.sleep(5.0) # This seems to be needed
9e105a85 S	186	args = basic_args + ['--resume']
	187	if retval == RD_FAILED:
	188	args += ['--skip', '1']
	189	args = [encodeArgument(a) for a in args]
	190	retval = run_rtmpdump(args)
3bc2ddcc	191	cursize = os.path.getsize(encodeFilename(tmpfilename))
35241756	192	if prevsize == cursize and retval == RD_FAILED:
3bc2ddcc	193	break
7af808a5	194	# Some rtmp streams seem abort after ~ 99.8%. Don't complain for those
35241756	195	if prevsize == cursize and retval == RD_INCOMPLETE and cursize > 1024:
8dec03ec	196	self.to_screen('[rtmpdump] Could not download the whole video. This can happen for some advertisements.')
35241756	197	retval = RD_SUCCESS
3bc2ddcc	198	break
35241756	199	if retval == RD_SUCCESS or (test and retval == RD_INCOMPLETE):
3bc2ddcc	200	fsize = os.path.getsize(encodeFilename(tmpfilename))
f16f4877	201	self.to_screen('[rtmpdump] Downloaded %s bytes' % fsize)
3bc2ddcc JMF	202	self.try_rename(tmpfilename, filename)
	203	self._hook_progress({
	204	'downloaded_bytes': fsize,
	205	'total_bytes': fsize,
	206	'filename': filename,
	207	'status': 'finished',
f16f4877	208	'elapsed': time.time() - started,
3ba7740d	209	}, info_dict)
3bc2ddcc JMF	210	return True
3bc2ddcc JMF	211	else:
8dec03ec S	212	self.to_stderr('\n')
8dec03ec S	213	self.report_error('rtmpdump exited with code %d' % retval)
3bc2ddcc	214	return False