1
mirror of https://github.com/mpv-player/mpv synced 2024-10-30 04:46:41 +01:00
mpv/stream/stream.c

1093 lines
32 KiB
C
Raw Normal View History

/*
* This file is part of MPlayer.
*
* MPlayer is free software; you can redistribute it and/or modify
* it under the terms of the GNU General Public License as published by
* the Free Software Foundation; either version 2 of the License, or
* (at your option) any later version.
*
* MPlayer is distributed in the hope that it will be useful,
* but WITHOUT ANY WARRANTY; without even the implied warranty of
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
* GNU General Public License for more details.
*
* You should have received a copy of the GNU General Public License along
* with MPlayer; if not, write to the Free Software Foundation, Inc.,
* 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA.
*/
#include <stdio.h>
#include <stdlib.h>
lua: add an utility function for starting processes Because 1) Lua is terrible, and 2) popen() is terrible. Unfortunately, since Unix is also terrible, this turned out more complicated than I hoped. As a consequence and to avoid that this code has to be maintained forever, add a disclaimer that any function in Lua's utils module can disappear any time. The complexity seems a bit ridiculous, especially for a feature so far removed from actual video playback, so if it turns out that we don't really need this function, it will be dropped again. The motivation for this commit is the same as with 8e4fa5fc. Note that there is an "#ifndef __GLIBC__". The GNU people are very special people and thought it'd be convenient to actually declare "environ", even though the POSIX people, which are also very special people, state that no header declares this and that the user has to declare this manually. Since the GNU people overtook the Unix world with their very clever "embrace, extend, extinguish" strategy, but not 100%, and trying to build without _GNU_SOURCE is hopeless; but since there might be Unix environments which support _GNU_SOURCE features partially, this means that in practice "environ" will be randomly declared or not declared by system headers. Also, gcc was written by very clever people too, and prints a warning if an external variable is declared twice (I didn't check, but I suppose redeclaring is legal C, and not even the gcc people are clever enough to only warn against a definitely not legal C construct, although sometimes they do this), ...and since we at mpv hate compiler warnings, we seek to silence them all. Adding a configure test just for a warning seems too radical, so we special-case this against __GLIBC__, which is hopefully not defined on other libcs, especially not libcs which don't implement all aspects of _GNU_SOURCE, and redefine "environ" on systems even if the headers define it already (because they support _GNU_SOURCE - as I mentioned before, the clever GNU people wrote software THAT portable that other libcs just gave up and implemented parts of _GNU_SOURCE, although probably not all), which means that compiling mpv will print a warning about "environ" being redefined, but at least this won't happen on my system, so all is fine. However, should someone complain about this warning, I will force whoever complained about this warning to read this ENTIRE commit message, and if possible, will also force them to eat a printed-out copy of the GNU Manifesto, and if that is not enough, maybe this person could even be forced to convince the very clever POSIX people of not doing crap like this: having the user to manually declare somewhat central symbols - but I doubt it's possible, because the POSIX people are too far gone and only care about maintaining compatibility with old versions of AIX and HP-UX. Oh, also, this code contains some subtle and obvious issues, but writing about this is not fun.
2014-10-19 01:42:28 +02:00
#include <sys/types.h>
#include <unistd.h>
lua: add an utility function for starting processes Because 1) Lua is terrible, and 2) popen() is terrible. Unfortunately, since Unix is also terrible, this turned out more complicated than I hoped. As a consequence and to avoid that this code has to be maintained forever, add a disclaimer that any function in Lua's utils module can disappear any time. The complexity seems a bit ridiculous, especially for a feature so far removed from actual video playback, so if it turns out that we don't really need this function, it will be dropped again. The motivation for this commit is the same as with 8e4fa5fc. Note that there is an "#ifndef __GLIBC__". The GNU people are very special people and thought it'd be convenient to actually declare "environ", even though the POSIX people, which are also very special people, state that no header declares this and that the user has to declare this manually. Since the GNU people overtook the Unix world with their very clever "embrace, extend, extinguish" strategy, but not 100%, and trying to build without _GNU_SOURCE is hopeless; but since there might be Unix environments which support _GNU_SOURCE features partially, this means that in practice "environ" will be randomly declared or not declared by system headers. Also, gcc was written by very clever people too, and prints a warning if an external variable is declared twice (I didn't check, but I suppose redeclaring is legal C, and not even the gcc people are clever enough to only warn against a definitely not legal C construct, although sometimes they do this), ...and since we at mpv hate compiler warnings, we seek to silence them all. Adding a configure test just for a warning seems too radical, so we special-case this against __GLIBC__, which is hopefully not defined on other libcs, especially not libcs which don't implement all aspects of _GNU_SOURCE, and redefine "environ" on systems even if the headers define it already (because they support _GNU_SOURCE - as I mentioned before, the clever GNU people wrote software THAT portable that other libcs just gave up and implemented parts of _GNU_SOURCE, although probably not all), which means that compiling mpv will print a warning about "environ" being redefined, but at least this won't happen on my system, so all is fine. However, should someone complain about this warning, I will force whoever complained about this warning to read this ENTIRE commit message, and if possible, will also force them to eat a printed-out copy of the GNU Manifesto, and if that is not enough, maybe this person could even be forced to convince the very clever POSIX people of not doing crap like this: having the user to manually declare somewhat central symbols - but I doubt it's possible, because the POSIX people are too far gone and only care about maintaining compatibility with old versions of AIX and HP-UX. Oh, also, this code contains some subtle and obvious issues, but writing about this is not fun.
2014-10-19 01:42:28 +02:00
#include <errno.h>
#include <strings.h>
#include <assert.h>
#include <libavutil/common.h>
stream: redo playback abort handling This mechanism originates from MPlayer's way of dealing with blocking network, but it's still useful. On opening and closing, mpv waits for network synchronously, and also some obscure commands and use-cases can lead to such blocking. In these situations, the stream is asynchronously forced to stop by "interrupting" it. The old design interrupting I/O was a bit broken: polling with a callback, instead of actively interrupting it. Change the direction of this. There is no callback anymore, and the player calls mp_cancel_trigger() to force the stream to return. libavformat (via stream_lavf.c) has the old broken design, and fixing it would require fixing libavformat, which won't happen so quickly. So we have to keep that part. But everything above the stream layer is prepared for a better design, and more sophisticated methods than mp_cancel_test() could be easily introduced. There's still one problem: commands are still run in the central playback loop, which we assume can block on I/O in the worst case. That's not a problem yet, because we simply mark some commands as being able to stop playback of the current file ("quit" etc.), so input.c could abort playback as soon as such a command is queued. But there are also commands abort playback only conditionally, and the logic for that is in the playback core and thus "unreachable". For example, "playlist_next" aborts playback only if there's a next file. We don't want it to always abort playback. As a quite ugly hack, abort playback only if at least 2 abort commands are queued - this pretty much happens only if the core is frozen and doesn't react to input.
2014-09-13 14:23:08 +02:00
#include "osdep/atomics.h"
lua: add an utility function for starting processes Because 1) Lua is terrible, and 2) popen() is terrible. Unfortunately, since Unix is also terrible, this turned out more complicated than I hoped. As a consequence and to avoid that this code has to be maintained forever, add a disclaimer that any function in Lua's utils module can disappear any time. The complexity seems a bit ridiculous, especially for a feature so far removed from actual video playback, so if it turns out that we don't really need this function, it will be dropped again. The motivation for this commit is the same as with 8e4fa5fc. Note that there is an "#ifndef __GLIBC__". The GNU people are very special people and thought it'd be convenient to actually declare "environ", even though the POSIX people, which are also very special people, state that no header declares this and that the user has to declare this manually. Since the GNU people overtook the Unix world with their very clever "embrace, extend, extinguish" strategy, but not 100%, and trying to build without _GNU_SOURCE is hopeless; but since there might be Unix environments which support _GNU_SOURCE features partially, this means that in practice "environ" will be randomly declared or not declared by system headers. Also, gcc was written by very clever people too, and prints a warning if an external variable is declared twice (I didn't check, but I suppose redeclaring is legal C, and not even the gcc people are clever enough to only warn against a definitely not legal C construct, although sometimes they do this), ...and since we at mpv hate compiler warnings, we seek to silence them all. Adding a configure test just for a warning seems too radical, so we special-case this against __GLIBC__, which is hopefully not defined on other libcs, especially not libcs which don't implement all aspects of _GNU_SOURCE, and redefine "environ" on systems even if the headers define it already (because they support _GNU_SOURCE - as I mentioned before, the clever GNU people wrote software THAT portable that other libcs just gave up and implemented parts of _GNU_SOURCE, although probably not all), which means that compiling mpv will print a warning about "environ" being redefined, but at least this won't happen on my system, so all is fine. However, should someone complain about this warning, I will force whoever complained about this warning to read this ENTIRE commit message, and if possible, will also force them to eat a printed-out copy of the GNU Manifesto, and if that is not enough, maybe this person could even be forced to convince the very clever POSIX people of not doing crap like this: having the user to manually declare somewhat central symbols - but I doubt it's possible, because the POSIX people are too far gone and only care about maintaining compatibility with old versions of AIX and HP-UX. Oh, also, this code contains some subtle and obvious issues, but writing about this is not fun.
2014-10-19 01:42:28 +02:00
#include "osdep/io.h"
#include "talloc.h"
#include "config.h"
#include "common/common.h"
#include "common/global.h"
#include "misc/bstr.h"
#include "common/msg.h"
#include "options/options.h"
#include "options/path.h"
2010-05-30 15:39:41 +02:00
#include "osdep/timer.h"
#include "stream.h"
#include "options/m_option.h"
#include "options/m_config.h"
#ifdef __MINGW32__
#include <windows.h>
#endif
// Includes additional padding in case sizes get rounded up by sector size.
#define TOTAL_BUFFER_SIZE (STREAM_MAX_BUFFER_SIZE + STREAM_MAX_SECTOR_SIZE)
extern const stream_info_t stream_info_cdda;
extern const stream_info_t stream_info_dvb;
extern const stream_info_t stream_info_tv;
extern const stream_info_t stream_info_pvr;
extern const stream_info_t stream_info_smb;
extern const stream_info_t stream_info_null;
extern const stream_info_t stream_info_memory;
extern const stream_info_t stream_info_mf;
extern const stream_info_t stream_info_ffmpeg;
extern const stream_info_t stream_info_ffmpeg_unsafe;
demux_lavf: add support for libavdevice libavdevice supports various "special" video and audio inputs, such as screen-capture or libavfilter filter graphs. libavdevice inputs are implemented as demuxers. They don't use the custom stream callbacks (in AVFormatContext.pb). Instead, input parameters are passed as filename. This means the mpv stream layer has to be disabled. Do this by adding the pseudo stream handler avdevice://, whose only purpose is passing the filename to demux_lavf, without actually doing anything. Change the logic how the filename is passed to libavformat. Remove handling of the filename from demux_open_lavf() and move it to lavf_check_file(). (This also fixes a possible bug when skipping the "lavf://" prefix.) libavdevice now can be invoked by specifying demuxer and args as in: mpv avdevice://demuxer:args The args are passed as filename to libavformat. When using libavdevice demuxers, their actual meaning is highly implementation specific. They don't refer to actual filenames. Note: libavdevice is disabled by default. There is one problem: libavdevice pulls in libavfilter, which in turn causes symbol clashes with mpv internals. The problem is that libavfilter includes a mplayer filter bridge, which is used to interface with a set of nearly unmodified mplayer filters copied into libavfilter. This filter bridge uses the same symbol names as mplayer/mpv's filter chain, which results in symbol clashes at link-time. This can be prevented by building ffmpeg with --disable-filter=mp, but unfortunately this is not the default. This means linking to libavdevice (which in turn forces linking with libavfilter by default) must be disabled. We try doing this by compiling a test file that defines one of the clashing symbols (vf_mpi_clear). To enable libavdevice input, ffmpeg should be built with the options: --disable-filter=mp and mpv with: --enable-libavdevice Originally, I tried to auto-detect it. But the resulting complications in configure did't seem worth the trouble.
2012-11-30 18:41:04 +01:00
extern const stream_info_t stream_info_avdevice;
extern const stream_info_t stream_info_file;
extern const stream_info_t stream_info_ifo;
extern const stream_info_t stream_info_ifo_dvdnav;
extern const stream_info_t stream_info_dvd;
extern const stream_info_t stream_info_dvdnav;
extern const stream_info_t stream_info_bdmv_dir;
extern const stream_info_t stream_info_bluray;
extern const stream_info_t stream_info_bdnav;
extern const stream_info_t stream_info_rar_filter;
extern const stream_info_t stream_info_rar_entry;
extern const stream_info_t stream_info_edl;
static const stream_info_t *const stream_list[] = {
#if HAVE_CDDA
&stream_info_cdda,
#endif
&stream_info_ffmpeg,
&stream_info_ffmpeg_unsafe,
&stream_info_avdevice,
#if HAVE_DVBIN
&stream_info_dvb,
#endif
#if HAVE_TV
&stream_info_tv,
#endif
#if HAVE_PVR
&stream_info_pvr,
#endif
#if HAVE_LIBSMBCLIENT
&stream_info_smb,
#endif
#if HAVE_DVDREAD
&stream_info_ifo,
&stream_info_dvd,
#endif
#if HAVE_DVDNAV
&stream_info_ifo_dvdnav,
&stream_info_dvdnav,
#endif
#if HAVE_LIBBLURAY
&stream_info_bdmv_dir,
&stream_info_bluray,
&stream_info_bdnav,
#endif
&stream_info_memory,
&stream_info_null,
&stream_info_mf,
&stream_info_edl,
&stream_info_rar_filter,
&stream_info_rar_entry,
&stream_info_file,
NULL
};
static int stream_seek_unbuffered(stream_t *s, int64_t newpos);
static int from_hex(unsigned char c)
{
if (c >= 'a' && c <= 'f')
return c - 'a' + 10;
if (c >= 'A' && c <= 'F')
return c - 'A' + 10;
if (c >= '0' && c <= '9')
return c - '0';
return -1;
}
// Replace escape sequences in an URL (or a part of an URL)
void mp_url_unescape_inplace(char *buf)
{
int len = strlen(buf);
int o = 0;
for (int i = 0; i < len; i++) {
unsigned char c = buf[i];
if (c == '%' && i + 2 < len) { //must have 2 more chars
int c1 = from_hex(buf[i + 1]);
int c2 = from_hex(buf[i + 2]);
if (c1 >= 0 && c2 >= 0) {
c = c1 * 16 + c2;
i = i + 2; //only skip next 2 chars if valid esc
}
}
buf[o++] = c;
}
buf[o++] = '\0';
}
// Escape according to http://tools.ietf.org/html/rfc3986#section-2.1
// Only unreserved characters are not escaped.
// The argument ok (if not NULL) is as follows:
// ok[0] != '~': additional characters that are not escaped
// ok[0] == '~': do not escape anything but these characters
// (can't override the unreserved characters, which are
// never escaped, and '%', which is always escaped)
char *mp_url_escape(void *talloc_ctx, const char *s, const char *ok)
{
int len = strlen(s);
char *buf = talloc_array(talloc_ctx, char, len * 3 + 1);
int o = 0;
for (int i = 0; i < len; i++) {
unsigned char c = s[i];
if ((c >= 'a' && c <= 'z') || (c >= 'A' && c <= 'Z') ||
(c >= '0' && c <= '9') || strchr("-._~", c) ||
(ok && ((ok[0] != '~') == !!strchr(ok, c)) && c != '%'))
{
buf[o++] = c;
} else {
const char hex[] = "0123456789ABCDEF";
buf[o++] = '%';
buf[o++] = hex[c / 16];
buf[o++] = hex[c % 16];
}
}
buf[o++] = '\0';
return buf;
}
static const char *find_url_opt(struct stream *s, const char *opt)
{
for (int n = 0; s->info->url_options && s->info->url_options[n]; n++) {
const char *entry = s->info->url_options[n];
const char *t = strchr(entry, '=');
assert(t);
if (strncmp(opt, entry, t - entry) == 0)
return t + 1;
}
return NULL;
}
static bstr split_next(bstr *s, char end, const char *delim)
{
int idx = bstrcspn(*s, delim);
if (end && (idx >= s->len || s->start[idx] != end))
return (bstr){0};
bstr r = bstr_splice(*s, 0, idx);
*s = bstr_cut(*s, idx + (end ? 1 : 0));
return r;
}
// Parse the stream URL, syntax:
// proto:// [<username>@]<hostname>[:<port>][/<filename>]
// (the proto:// part is already removed from s->path)
// This code originates from times when http code used this, but now it's
// just relict from other stream implementations reusing this code.
static bool parse_url(struct stream *st, struct m_config *config)
{
bstr s = bstr0(st->path);
const char *f_names[4] = {"username", "hostname", "port", "filename"};
bstr f[4];
f[0] = split_next(&s, '@', "@:/");
f[1] = split_next(&s, 0, ":/");
f[2] = bstr_eatstart0(&s, ":") ? split_next(&s, 0, "/") : (bstr){0};
f[3] = bstr_eatstart0(&s, "/") ? s : (bstr){0};
for (int n = 0; n < 4; n++) {
if (f[n].len) {
const char *opt = find_url_opt(st, f_names[n]);
if (!opt) {
MP_ERR(st, "Stream type '%s' accepts no '%s' field in URLs.\n",
st->info->name, f_names[n]);
return false;
}
int r = m_config_set_option(config, bstr0(opt), f[n]);
if (r < 0) {
MP_ERR(st, "Error setting stream option: %s\n",
m_option_strerror(r));
return false;
}
}
}
return true;
}
static stream_t *new_stream(void)
{
return talloc_zero_size(NULL, sizeof(stream_t) + TOTAL_BUFFER_SIZE);
}
static const char *match_proto(const char *url, const char *proto)
{
int l = strlen(proto);
if (l > 0) {
if (strncasecmp(url, proto, l) == 0 && strncmp("://", url + l, 3) == 0)
return url + l + 3;
} else if (!mp_is_url(bstr0(url))) {
return url; // pure filenames
}
return NULL;
}
static int open_internal(const stream_info_t *sinfo, struct stream *underlying,
stream: redo playback abort handling This mechanism originates from MPlayer's way of dealing with blocking network, but it's still useful. On opening and closing, mpv waits for network synchronously, and also some obscure commands and use-cases can lead to such blocking. In these situations, the stream is asynchronously forced to stop by "interrupting" it. The old design interrupting I/O was a bit broken: polling with a callback, instead of actively interrupting it. Change the direction of this. There is no callback anymore, and the player calls mp_cancel_trigger() to force the stream to return. libavformat (via stream_lavf.c) has the old broken design, and fixing it would require fixing libavformat, which won't happen so quickly. So we have to keep that part. But everything above the stream layer is prepared for a better design, and more sophisticated methods than mp_cancel_test() could be easily introduced. There's still one problem: commands are still run in the central playback loop, which we assume can block on I/O in the worst case. That's not a problem yet, because we simply mark some commands as being able to stop playback of the current file ("quit" etc.), so input.c could abort playback as soon as such a command is queued. But there are also commands abort playback only conditionally, and the logic for that is in the playback core and thus "unreachable". For example, "playlist_next" aborts playback only if there's a next file. We don't want it to always abort playback. As a quite ugly hack, abort playback only if at least 2 abort commands are queued - this pretty much happens only if the core is frozen and doesn't react to input.
2014-09-13 14:23:08 +02:00
const char *url, int flags, struct mp_cancel *c,
struct mpv_global *global, struct stream **ret)
{
if (sinfo->stream_filter != !!underlying)
return STREAM_NO_MATCH;
if (sinfo->stream_filter && (flags & STREAM_NO_FILTERS))
return STREAM_NO_MATCH;
if (!sinfo->is_safe && (flags & STREAM_SAFE_ONLY))
return STREAM_UNSAFE;
if (!sinfo->is_network && (flags & STREAM_NETWORK_ONLY))
return STREAM_UNSAFE;
const char *path = NULL;
// Stream filters use the original URL, with no protocol matching at all.
if (!sinfo->stream_filter) {
for (int n = 0; sinfo->protocols && sinfo->protocols[n]; n++) {
path = match_proto(url, sinfo->protocols[n]);
if (path)
break;
}
if (!path)
return STREAM_NO_MATCH;
}
stream_t *s = new_stream();
s->log = mp_log_new(s, global->log, sinfo->name);
s->info = sinfo;
s->opts = global->opts;
stream: redo playback abort handling This mechanism originates from MPlayer's way of dealing with blocking network, but it's still useful. On opening and closing, mpv waits for network synchronously, and also some obscure commands and use-cases can lead to such blocking. In these situations, the stream is asynchronously forced to stop by "interrupting" it. The old design interrupting I/O was a bit broken: polling with a callback, instead of actively interrupting it. Change the direction of this. There is no callback anymore, and the player calls mp_cancel_trigger() to force the stream to return. libavformat (via stream_lavf.c) has the old broken design, and fixing it would require fixing libavformat, which won't happen so quickly. So we have to keep that part. But everything above the stream layer is prepared for a better design, and more sophisticated methods than mp_cancel_test() could be easily introduced. There's still one problem: commands are still run in the central playback loop, which we assume can block on I/O in the worst case. That's not a problem yet, because we simply mark some commands as being able to stop playback of the current file ("quit" etc.), so input.c could abort playback as soon as such a command is queued. But there are also commands abort playback only conditionally, and the logic for that is in the playback core and thus "unreachable". For example, "playlist_next" aborts playback only if there's a next file. We don't want it to always abort playback. As a quite ugly hack, abort playback only if at least 2 abort commands are queued - this pretty much happens only if the core is frozen and doesn't react to input.
2014-09-13 14:23:08 +02:00
s->cancel = c;
s->global = global;
s->url = talloc_strdup(s, url);
s->path = talloc_strdup(s, path);
s->source = underlying;
s->allow_caching = true;
s->is_network = sinfo->is_network;
s->mode = flags & (STREAM_READ | STREAM_WRITE);
if ((s->mode & STREAM_WRITE) && !sinfo->can_write) {
MP_VERBOSE(s, "No write access implemented.\n");
talloc_free(s);
return STREAM_NO_MATCH;
}
// Parse options
if (sinfo->priv_size) {
struct m_obj_desc desc = {
.priv_size = sinfo->priv_size,
.priv_defaults = sinfo->priv_defaults,
.options = sinfo->options,
};
if (sinfo->get_defaults)
desc.priv_defaults = sinfo->get_defaults(s);
struct m_config *config = m_config_from_obj_desc(s, s->log, &desc);
s->priv = config->optstruct;
if (s->info->url_options && !parse_url(s, config)) {
MP_ERR(s, "URL parsing failed on url %s\n", url);
talloc_free(s);
return STREAM_ERROR;
}
}
int r = (sinfo->open)(s);
if (r != STREAM_OK) {
talloc_free(s);
return r;
}
if (!s->read_chunk)
s->read_chunk = 4 * (s->sector_size ? s->sector_size : STREAM_BUFFER_SIZE);
if (!s->fill_buffer)
s->allow_caching = false;
assert(s->seekable == !!s->seek);
cache: make the stream cache a proper stream that wraps other streams Before this commit, the cache was franken-hacked on top of the stream API. You had to use special functions (like cache_stream_fill_buffer() instead of stream_fill_buffer()), which would access the stream in a cached manner. The whole idea about the previous design was that the cache runs in a thread or in a forked process, while the cache awa functions made sure the stream instance looked consistent to the user. If you used the normal functions instead of the special ones while the cache was running, you were out of luck. Make it a bit more reasonable by turning the cache into a stream on its own. This makes it behave exactly like a normal stream. The stream callbacks call into the original (uncached) stream to do work. No special cache functions or redirections are needed. The only different thing about cache streams is that they are created by special functions, instead of being part of the auto_open_streams[] array. To make things simpler, remove the threading implementation, which was messed into the code. The threading code could perhaps be kept, but I don't really want to have to worry about this special case. A proper threaded implementation will be added later. Remove the cache enabling code from stream_radio.c. Since enabling the cache involves replacing the old stream with a new one, the code as-is can't be kept. It would be easily possible to enable the cache by requesting a cache size (which is also much simpler). But nobody uses stream_radio.c and I can't even test this thing, and the cache is probably not really important for it either.
2013-05-24 18:49:09 +02:00
s->uncached_type = s->type;
2014-10-25 16:42:24 +02:00
MP_VERBOSE(s, "Opened: %s\n", url);
if (s->mime_type)
MP_VERBOSE(s, "Mime-type: '%s'\n", s->mime_type);
*ret = s;
return STREAM_OK;
}
stream: redo playback abort handling This mechanism originates from MPlayer's way of dealing with blocking network, but it's still useful. On opening and closing, mpv waits for network synchronously, and also some obscure commands and use-cases can lead to such blocking. In these situations, the stream is asynchronously forced to stop by "interrupting" it. The old design interrupting I/O was a bit broken: polling with a callback, instead of actively interrupting it. Change the direction of this. There is no callback anymore, and the player calls mp_cancel_trigger() to force the stream to return. libavformat (via stream_lavf.c) has the old broken design, and fixing it would require fixing libavformat, which won't happen so quickly. So we have to keep that part. But everything above the stream layer is prepared for a better design, and more sophisticated methods than mp_cancel_test() could be easily introduced. There's still one problem: commands are still run in the central playback loop, which we assume can block on I/O in the worst case. That's not a problem yet, because we simply mark some commands as being able to stop playback of the current file ("quit" etc.), so input.c could abort playback as soon as such a command is queued. But there are also commands abort playback only conditionally, and the logic for that is in the playback core and thus "unreachable". For example, "playlist_next" aborts playback only if there's a next file. We don't want it to always abort playback. As a quite ugly hack, abort playback only if at least 2 abort commands are queued - this pretty much happens only if the core is frozen and doesn't react to input.
2014-09-13 14:23:08 +02:00
struct stream *stream_create(const char *url, int flags,
struct mp_cancel *c, struct mpv_global *global)
{
struct mp_log *log = mp_log_new(NULL, global->log, "!stream");
struct stream *s = NULL;
assert(url);
// Open stream proper
bool unsafe = false;
for (int i = 0; stream_list[i]; i++) {
stream: redo playback abort handling This mechanism originates from MPlayer's way of dealing with blocking network, but it's still useful. On opening and closing, mpv waits for network synchronously, and also some obscure commands and use-cases can lead to such blocking. In these situations, the stream is asynchronously forced to stop by "interrupting" it. The old design interrupting I/O was a bit broken: polling with a callback, instead of actively interrupting it. Change the direction of this. There is no callback anymore, and the player calls mp_cancel_trigger() to force the stream to return. libavformat (via stream_lavf.c) has the old broken design, and fixing it would require fixing libavformat, which won't happen so quickly. So we have to keep that part. But everything above the stream layer is prepared for a better design, and more sophisticated methods than mp_cancel_test() could be easily introduced. There's still one problem: commands are still run in the central playback loop, which we assume can block on I/O in the worst case. That's not a problem yet, because we simply mark some commands as being able to stop playback of the current file ("quit" etc.), so input.c could abort playback as soon as such a command is queued. But there are also commands abort playback only conditionally, and the logic for that is in the playback core and thus "unreachable". For example, "playlist_next" aborts playback only if there's a next file. We don't want it to always abort playback. As a quite ugly hack, abort playback only if at least 2 abort commands are queued - this pretty much happens only if the core is frozen and doesn't react to input.
2014-09-13 14:23:08 +02:00
int r = open_internal(stream_list[i], NULL, url, flags, c, global, &s);
if (r == STREAM_OK)
break;
if (r == STREAM_NO_MATCH || r == STREAM_UNSUPPORTED)
continue;
if (r == STREAM_UNSAFE) {
unsafe = true;
continue;
}
if (r != STREAM_OK) {
mp_err(log, "Failed to open %s.\n", url);
goto done;
}
}
if (!s && unsafe) {
mp_err(log, "\nRefusing to load potentially unsafe URL from a playlist.\n"
"Use --playlist=file or the --load-unsafe-playlists option to "
"load it anyway.\n\n");
goto done;
}
if (!s) {
mp_err(log, "No protocol handler found to open URL %s\n", url);
mp_err(log, "The protocol is either unsupported, or was disabled "
"at compile-time.\n");
goto done;
}
// Open stream filters
for (;;) {
struct stream *new = NULL;
for (int i = 0; stream_list[i]; i++) {
stream: redo playback abort handling This mechanism originates from MPlayer's way of dealing with blocking network, but it's still useful. On opening and closing, mpv waits for network synchronously, and also some obscure commands and use-cases can lead to such blocking. In these situations, the stream is asynchronously forced to stop by "interrupting" it. The old design interrupting I/O was a bit broken: polling with a callback, instead of actively interrupting it. Change the direction of this. There is no callback anymore, and the player calls mp_cancel_trigger() to force the stream to return. libavformat (via stream_lavf.c) has the old broken design, and fixing it would require fixing libavformat, which won't happen so quickly. So we have to keep that part. But everything above the stream layer is prepared for a better design, and more sophisticated methods than mp_cancel_test() could be easily introduced. There's still one problem: commands are still run in the central playback loop, which we assume can block on I/O in the worst case. That's not a problem yet, because we simply mark some commands as being able to stop playback of the current file ("quit" etc.), so input.c could abort playback as soon as such a command is queued. But there are also commands abort playback only conditionally, and the logic for that is in the playback core and thus "unreachable". For example, "playlist_next" aborts playback only if there's a next file. We don't want it to always abort playback. As a quite ugly hack, abort playback only if at least 2 abort commands are queued - this pretty much happens only if the core is frozen and doesn't react to input.
2014-09-13 14:23:08 +02:00
int r = open_internal(stream_list[i], s, s->url, flags, c, global, &new);
if (r == STREAM_OK)
break;
}
if (!new)
break;
s = new;
}
done:
talloc_free(log);
return s;
}
struct stream *stream_open(const char *filename, struct mpv_global *global)
{
stream: redo playback abort handling This mechanism originates from MPlayer's way of dealing with blocking network, but it's still useful. On opening and closing, mpv waits for network synchronously, and also some obscure commands and use-cases can lead to such blocking. In these situations, the stream is asynchronously forced to stop by "interrupting" it. The old design interrupting I/O was a bit broken: polling with a callback, instead of actively interrupting it. Change the direction of this. There is no callback anymore, and the player calls mp_cancel_trigger() to force the stream to return. libavformat (via stream_lavf.c) has the old broken design, and fixing it would require fixing libavformat, which won't happen so quickly. So we have to keep that part. But everything above the stream layer is prepared for a better design, and more sophisticated methods than mp_cancel_test() could be easily introduced. There's still one problem: commands are still run in the central playback loop, which we assume can block on I/O in the worst case. That's not a problem yet, because we simply mark some commands as being able to stop playback of the current file ("quit" etc.), so input.c could abort playback as soon as such a command is queued. But there are also commands abort playback only conditionally, and the logic for that is in the playback core and thus "unreachable". For example, "playlist_next" aborts playback only if there's a next file. We don't want it to always abort playback. As a quite ugly hack, abort playback only if at least 2 abort commands are queued - this pretty much happens only if the core is frozen and doesn't react to input.
2014-09-13 14:23:08 +02:00
return stream_create(filename, STREAM_READ, NULL, global);
}
stream_t *open_output_stream(const char *filename, struct mpv_global *global)
{
stream: redo playback abort handling This mechanism originates from MPlayer's way of dealing with blocking network, but it's still useful. On opening and closing, mpv waits for network synchronously, and also some obscure commands and use-cases can lead to such blocking. In these situations, the stream is asynchronously forced to stop by "interrupting" it. The old design interrupting I/O was a bit broken: polling with a callback, instead of actively interrupting it. Change the direction of this. There is no callback anymore, and the player calls mp_cancel_trigger() to force the stream to return. libavformat (via stream_lavf.c) has the old broken design, and fixing it would require fixing libavformat, which won't happen so quickly. So we have to keep that part. But everything above the stream layer is prepared for a better design, and more sophisticated methods than mp_cancel_test() could be easily introduced. There's still one problem: commands are still run in the central playback loop, which we assume can block on I/O in the worst case. That's not a problem yet, because we simply mark some commands as being able to stop playback of the current file ("quit" etc.), so input.c could abort playback as soon as such a command is queued. But there are also commands abort playback only conditionally, and the logic for that is in the playback core and thus "unreachable". For example, "playlist_next" aborts playback only if there's a next file. We don't want it to always abort playback. As a quite ugly hack, abort playback only if at least 2 abort commands are queued - this pretty much happens only if the core is frozen and doesn't react to input.
2014-09-13 14:23:08 +02:00
return stream_create(filename, STREAM_WRITE, NULL, global);
}
static int stream_reconnect(stream_t *s)
{
#define MAX_RECONNECT_RETRIES 5
#define RECONNECT_SLEEP_MAX_MS 500
if (!s->streaming || s->uncached_stream)
return 0;
if (!s->seekable)
return 0;
stream: redo playback abort handling This mechanism originates from MPlayer's way of dealing with blocking network, but it's still useful. On opening and closing, mpv waits for network synchronously, and also some obscure commands and use-cases can lead to such blocking. In these situations, the stream is asynchronously forced to stop by "interrupting" it. The old design interrupting I/O was a bit broken: polling with a callback, instead of actively interrupting it. Change the direction of this. There is no callback anymore, and the player calls mp_cancel_trigger() to force the stream to return. libavformat (via stream_lavf.c) has the old broken design, and fixing it would require fixing libavformat, which won't happen so quickly. So we have to keep that part. But everything above the stream layer is prepared for a better design, and more sophisticated methods than mp_cancel_test() could be easily introduced. There's still one problem: commands are still run in the central playback loop, which we assume can block on I/O in the worst case. That's not a problem yet, because we simply mark some commands as being able to stop playback of the current file ("quit" etc.), so input.c could abort playback as soon as such a command is queued. But there are also commands abort playback only conditionally, and the logic for that is in the playback core and thus "unreachable". For example, "playlist_next" aborts playback only if there's a next file. We don't want it to always abort playback. As a quite ugly hack, abort playback only if at least 2 abort commands are queued - this pretty much happens only if the core is frozen and doesn't react to input.
2014-09-13 14:23:08 +02:00
if (mp_cancel_test(s->cancel))
return 0;
int64_t pos = s->pos;
int sleep_ms = 5;
for (int retry = 0; retry < MAX_RECONNECT_RETRIES; retry++) {
MP_WARN(s, "Connection lost! Attempting to reconnect (%d)...\n", retry + 1);
if (retry) {
mp_sleep_us(sleep_ms * 1000);
sleep_ms = MPMIN(sleep_ms * 2, RECONNECT_SLEEP_MAX_MS);
}
stream: redo playback abort handling This mechanism originates from MPlayer's way of dealing with blocking network, but it's still useful. On opening and closing, mpv waits for network synchronously, and also some obscure commands and use-cases can lead to such blocking. In these situations, the stream is asynchronously forced to stop by "interrupting" it. The old design interrupting I/O was a bit broken: polling with a callback, instead of actively interrupting it. Change the direction of this. There is no callback anymore, and the player calls mp_cancel_trigger() to force the stream to return. libavformat (via stream_lavf.c) has the old broken design, and fixing it would require fixing libavformat, which won't happen so quickly. So we have to keep that part. But everything above the stream layer is prepared for a better design, and more sophisticated methods than mp_cancel_test() could be easily introduced. There's still one problem: commands are still run in the central playback loop, which we assume can block on I/O in the worst case. That's not a problem yet, because we simply mark some commands as being able to stop playback of the current file ("quit" etc.), so input.c could abort playback as soon as such a command is queued. But there are also commands abort playback only conditionally, and the logic for that is in the playback core and thus "unreachable". For example, "playlist_next" aborts playback only if there's a next file. We don't want it to always abort playback. As a quite ugly hack, abort playback only if at least 2 abort commands are queued - this pretty much happens only if the core is frozen and doesn't react to input.
2014-09-13 14:23:08 +02:00
if (mp_cancel_test(s->cancel))
return 0;
s->eof = 1;
int r = stream_control(s, STREAM_CTRL_RECONNECT, NULL);
if (r == STREAM_UNSUPPORTED)
return 0;
if (r != STREAM_OK)
continue;
cache: make the stream cache a proper stream that wraps other streams Before this commit, the cache was franken-hacked on top of the stream API. You had to use special functions (like cache_stream_fill_buffer() instead of stream_fill_buffer()), which would access the stream in a cached manner. The whole idea about the previous design was that the cache runs in a thread or in a forked process, while the cache awa functions made sure the stream instance looked consistent to the user. If you used the normal functions instead of the special ones while the cache was running, you were out of luck. Make it a bit more reasonable by turning the cache into a stream on its own. This makes it behave exactly like a normal stream. The stream callbacks call into the original (uncached) stream to do work. No special cache functions or redirections are needed. The only different thing about cache streams is that they are created by special functions, instead of being part of the auto_open_streams[] array. To make things simpler, remove the threading implementation, which was messed into the code. The threading code could perhaps be kept, but I don't really want to have to worry about this special case. A proper threaded implementation will be added later. Remove the cache enabling code from stream_radio.c. Since enabling the cache involves replacing the old stream with a new one, the code as-is can't be kept. It would be easily possible to enable the cache by requesting a cache size (which is also much simpler). But nobody uses stream_radio.c and I can't even test this thing, and the cache is probably not really important for it either.
2013-05-24 18:49:09 +02:00
if (stream_seek_unbuffered(s, pos) < 0 && s->pos == pos)
return 1;
}
return 0;
}
static void stream_capture_write(stream_t *s, void *buf, size_t len)
{
if (s->capture_file && len > 0) {
if (fwrite(buf, len, 1, s->capture_file) < 1) {
MP_ERR(s, "Error writing capture file: %s\n", mp_strerror(errno));
stream_set_capture_file(s, NULL);
}
}
}
void stream_set_capture_file(stream_t *s, const char *filename)
{
if (!bstr_equals(bstr0(s->capture_filename), bstr0(filename))) {
if (s->capture_file)
fclose(s->capture_file);
talloc_free(s->capture_filename);
s->capture_file = NULL;
s->capture_filename = NULL;
if (filename) {
s->capture_file = fopen(filename, "wb");
if (s->capture_file) {
s->capture_filename = talloc_strdup(NULL, filename);
if (s->buf_pos < s->buf_len)
stream_capture_write(s, s->buffer, s->buf_len);
} else {
MP_ERR(s, "Error opening capture file: %s\n", mp_strerror(errno));
}
}
}
}
cache: make the stream cache a proper stream that wraps other streams Before this commit, the cache was franken-hacked on top of the stream API. You had to use special functions (like cache_stream_fill_buffer() instead of stream_fill_buffer()), which would access the stream in a cached manner. The whole idea about the previous design was that the cache runs in a thread or in a forked process, while the cache awa functions made sure the stream instance looked consistent to the user. If you used the normal functions instead of the special ones while the cache was running, you were out of luck. Make it a bit more reasonable by turning the cache into a stream on its own. This makes it behave exactly like a normal stream. The stream callbacks call into the original (uncached) stream to do work. No special cache functions or redirections are needed. The only different thing about cache streams is that they are created by special functions, instead of being part of the auto_open_streams[] array. To make things simpler, remove the threading implementation, which was messed into the code. The threading code could perhaps be kept, but I don't really want to have to worry about this special case. A proper threaded implementation will be added later. Remove the cache enabling code from stream_radio.c. Since enabling the cache involves replacing the old stream with a new one, the code as-is can't be kept. It would be easily possible to enable the cache by requesting a cache size (which is also much simpler). But nobody uses stream_radio.c and I can't even test this thing, and the cache is probably not really important for it either.
2013-05-24 18:49:09 +02:00
// Read function bypassing the local stream buffer. This will not write into
// s->buffer, but into buf[0..len] instead.
// Returns < 0 on error, 0 on EOF, and length of bytes read on success.
// Partial reads are possible, even if EOF is not reached.
static int stream_read_unbuffered(stream_t *s, void *buf, int len)
{
int orig_len = len;
s->buf_pos = s->buf_len = 0;
// we will retry even if we already reached EOF previously.
len = s->fill_buffer ? s->fill_buffer(s, buf, len) : -1;
if (len < 0)
len = 0;
if (len == 0) {
// do not retry if this looks like proper eof
int64_t size = -1;
stream_control(s, STREAM_CTRL_GET_SIZE, &size);
if (s->eof || s->pos == size)
goto eof_out;
// just in case this is an error e.g. due to network
// timeout reset and retry
if (!stream_reconnect(s))
goto eof_out;
// make sure EOF is set to ensure no endless loops
s->eof = 1;
cache: make the stream cache a proper stream that wraps other streams Before this commit, the cache was franken-hacked on top of the stream API. You had to use special functions (like cache_stream_fill_buffer() instead of stream_fill_buffer()), which would access the stream in a cached manner. The whole idea about the previous design was that the cache runs in a thread or in a forked process, while the cache awa functions made sure the stream instance looked consistent to the user. If you used the normal functions instead of the special ones while the cache was running, you were out of luck. Make it a bit more reasonable by turning the cache into a stream on its own. This makes it behave exactly like a normal stream. The stream callbacks call into the original (uncached) stream to do work. No special cache functions or redirections are needed. The only different thing about cache streams is that they are created by special functions, instead of being part of the auto_open_streams[] array. To make things simpler, remove the threading implementation, which was messed into the code. The threading code could perhaps be kept, but I don't really want to have to worry about this special case. A proper threaded implementation will be added later. Remove the cache enabling code from stream_radio.c. Since enabling the cache involves replacing the old stream with a new one, the code as-is can't be kept. It would be easily possible to enable the cache by requesting a cache size (which is also much simpler). But nobody uses stream_radio.c and I can't even test this thing, and the cache is probably not really important for it either.
2013-05-24 18:49:09 +02:00
return stream_read_unbuffered(s, buf, orig_len);
eof_out:
s->eof = 1;
return 0;
}
// When reading succeeded we are obviously not at eof.
s->eof = 0;
s->pos += len;
stream_capture_write(s, buf, len);
return len;
}
static int stream_fill_buffer_by(stream_t *s, int64_t len)
{
len = MPMIN(len, s->read_chunk);
len = MPMAX(len, STREAM_BUFFER_SIZE);
if (s->sector_size)
len = s->sector_size;
len = stream_read_unbuffered(s, s->buffer, len);
s->buf_pos = 0;
s->buf_len = len;
return s->buf_len;
}
int stream_fill_buffer(stream_t *s)
{
return stream_fill_buffer_by(s, STREAM_BUFFER_SIZE);
}
// Read between 1..buf_size bytes of data, return how much data has been read.
// Return 0 on EOF, error, of if buf_size was 0.
int stream_read_partial(stream_t *s, char *buf, int buf_size)
{
assert(s->buf_pos <= s->buf_len);
assert(buf_size >= 0);
if (s->buf_pos == s->buf_len && buf_size > 0) {
s->buf_pos = s->buf_len = 0;
// Do a direct read, but only if there's no sector alignment requirement
// Also, small reads will be more efficient with buffering & copying
if (!s->sector_size && buf_size >= STREAM_BUFFER_SIZE)
return stream_read_unbuffered(s, buf, buf_size);
if (!stream_fill_buffer(s))
return 0;
}
int len = FFMIN(buf_size, s->buf_len - s->buf_pos);
memcpy(buf, &s->buffer[s->buf_pos], len);
s->buf_pos += len;
if (len > 0)
s->eof = 0;
return len;
}
int stream_read(stream_t *s, char *mem, int total)
{
int len = total;
while (len > 0) {
int read = stream_read_partial(s, mem, len);
if (read <= 0)
break; // EOF
mem += read;
len -= read;
}
total -= len;
if (total > 0)
s->eof = 0;
return total;
}
// Read ahead at most len bytes without changing the read position. Return a
// pointer to the internal buffer, starting from the current read position.
// Can read ahead at most STREAM_MAX_BUFFER_SIZE bytes.
// The returned buffer becomes invalid on the next stream call, and you must
// not write to it.
struct bstr stream_peek(stream_t *s, int len)
{
assert(len >= 0);
assert(len <= STREAM_MAX_BUFFER_SIZE);
if (s->buf_len - s->buf_pos < len) {
// Move to front to guarantee we really can read up to max size.
int buf_valid = s->buf_len - s->buf_pos;
memmove(s->buffer, &s->buffer[s->buf_pos], buf_valid);
// Fill rest of the buffer.
while (buf_valid < len) {
int chunk = MPMAX(len - buf_valid, STREAM_BUFFER_SIZE);
if (s->sector_size)
chunk = s->sector_size;
assert(buf_valid + chunk <= TOTAL_BUFFER_SIZE);
int read = stream_read_unbuffered(s, &s->buffer[buf_valid], chunk);
if (read == 0)
break; // EOF
buf_valid += read;
}
s->buf_pos = 0;
s->buf_len = buf_valid;
if (s->buf_len)
s->eof = 0;
}
return (bstr){.start = &s->buffer[s->buf_pos],
.len = FFMIN(len, s->buf_len - s->buf_pos)};
}
int stream_write_buffer(stream_t *s, unsigned char *buf, int len)
{
int rd;
if (!s->write_buffer)
return -1;
rd = s->write_buffer(s, buf, len);
if (rd < 0)
return -1;
s->pos += rd;
assert(rd == len && "stream_write_buffer(): unexpected short write");
return rd;
}
static int stream_skip_read(struct stream *s, int64_t len)
{
while (len > 0) {
int x = s->buf_len - s->buf_pos;
if (x == 0) {
if (!stream_fill_buffer_by(s, len))
return 0; // EOF
x = s->buf_len - s->buf_pos;
}
if (x > len)
x = len;
s->buf_pos += x;
len -= x;
}
return 1;
}
// Drop the internal buffer. Note that this will advance the stream position
// (as seen by stream_tell()), because the real stream position is ahead of the
// logical stream position by the amount of buffered but not yet read data.
void stream_drop_buffers(stream_t *s)
{
s->pos = stream_tell(s);
s->buf_pos = s->buf_len = 0;
s->eof = 0;
}
cache: make the stream cache a proper stream that wraps other streams Before this commit, the cache was franken-hacked on top of the stream API. You had to use special functions (like cache_stream_fill_buffer() instead of stream_fill_buffer()), which would access the stream in a cached manner. The whole idea about the previous design was that the cache runs in a thread or in a forked process, while the cache awa functions made sure the stream instance looked consistent to the user. If you used the normal functions instead of the special ones while the cache was running, you were out of luck. Make it a bit more reasonable by turning the cache into a stream on its own. This makes it behave exactly like a normal stream. The stream callbacks call into the original (uncached) stream to do work. No special cache functions or redirections are needed. The only different thing about cache streams is that they are created by special functions, instead of being part of the auto_open_streams[] array. To make things simpler, remove the threading implementation, which was messed into the code. The threading code could perhaps be kept, but I don't really want to have to worry about this special case. A proper threaded implementation will be added later. Remove the cache enabling code from stream_radio.c. Since enabling the cache involves replacing the old stream with a new one, the code as-is can't be kept. It would be easily possible to enable the cache by requesting a cache size (which is also much simpler). But nobody uses stream_radio.c and I can't even test this thing, and the cache is probably not really important for it either.
2013-05-24 18:49:09 +02:00
// Seek function bypassing the local stream buffer.
static int stream_seek_unbuffered(stream_t *s, int64_t newpos)
{
if (newpos != s->pos) {
if (newpos > s->pos && !s->seekable) {
MP_ERR(s, "Cannot seek forward in this stream\n");
return 0;
}
if (newpos < s->pos && !s->seekable) {
MP_ERR(s, "Cannot seek backward in linear streams!\n");
return 1;
}
if (s->seek(s, newpos) <= 0) {
MP_ERR(s, "Seek failed\n");
return 0;
}
stream_drop_buffers(s);
s->pos = newpos;
}
s->eof = 0; // EOF reset when seek succeeds.
return -1;
}
cache: make the stream cache a proper stream that wraps other streams Before this commit, the cache was franken-hacked on top of the stream API. You had to use special functions (like cache_stream_fill_buffer() instead of stream_fill_buffer()), which would access the stream in a cached manner. The whole idea about the previous design was that the cache runs in a thread or in a forked process, while the cache awa functions made sure the stream instance looked consistent to the user. If you used the normal functions instead of the special ones while the cache was running, you were out of luck. Make it a bit more reasonable by turning the cache into a stream on its own. This makes it behave exactly like a normal stream. The stream callbacks call into the original (uncached) stream to do work. No special cache functions or redirections are needed. The only different thing about cache streams is that they are created by special functions, instead of being part of the auto_open_streams[] array. To make things simpler, remove the threading implementation, which was messed into the code. The threading code could perhaps be kept, but I don't really want to have to worry about this special case. A proper threaded implementation will be added later. Remove the cache enabling code from stream_radio.c. Since enabling the cache involves replacing the old stream with a new one, the code as-is can't be kept. It would be easily possible to enable the cache by requesting a cache size (which is also much simpler). But nobody uses stream_radio.c and I can't even test this thing, and the cache is probably not really important for it either.
2013-05-24 18:49:09 +02:00
// Unlike stream_seek, does not try to seek within local buffer.
// Unlike stream_seek_unbuffered(), it still fills the local buffer.
static int stream_seek_long(stream_t *s, int64_t pos)
{
if (s->mode == STREAM_WRITE) {
if (!s->seekable || !s->seek(s, pos))
return 0;
return 1;
}
int64_t newpos = pos;
if (s->sector_size)
newpos = (pos / s->sector_size) * s->sector_size;
MP_TRACE(s, "Seek from %" PRId64 " to %" PRId64
" (with offset %d)\n", s->pos, pos, (int)(pos - newpos));
2013-06-06 20:39:53 +02:00
if (pos >= s->pos && !s->seekable && s->fast_skip) {
// skipping is handled by generic code below
} else if (stream_seek_unbuffered(s, newpos) >= 0) {
return 0;
}
if (pos >= s->pos && stream_skip_read(s, pos - s->pos) > 0)
return 1; // success
// Fill failed, but seek still is a success (partially).
s->eof = 0; // eof should be set only on read
MP_VERBOSE(s, "Seek to/past EOF: no buffer preloaded.\n");
return 1;
}
int stream_seek(stream_t *s, int64_t pos)
{
MP_TRACE(s, "seek to 0x%llX\n", (long long)pos);
if (pos == stream_tell(s))
return 1;
if (pos < 0) {
MP_ERR(s, "Invalid seek to negative position %llx!\n", (long long)pos);
pos = 0;
}
if (pos < s->pos) {
int64_t x = pos - (s->pos - (int)s->buf_len);
if (x >= 0) {
s->buf_pos = x;
s->eof = 0;
return 1;
}
}
cache: make the stream cache a proper stream that wraps other streams Before this commit, the cache was franken-hacked on top of the stream API. You had to use special functions (like cache_stream_fill_buffer() instead of stream_fill_buffer()), which would access the stream in a cached manner. The whole idea about the previous design was that the cache runs in a thread or in a forked process, while the cache awa functions made sure the stream instance looked consistent to the user. If you used the normal functions instead of the special ones while the cache was running, you were out of luck. Make it a bit more reasonable by turning the cache into a stream on its own. This makes it behave exactly like a normal stream. The stream callbacks call into the original (uncached) stream to do work. No special cache functions or redirections are needed. The only different thing about cache streams is that they are created by special functions, instead of being part of the auto_open_streams[] array. To make things simpler, remove the threading implementation, which was messed into the code. The threading code could perhaps be kept, but I don't really want to have to worry about this special case. A proper threaded implementation will be added later. Remove the cache enabling code from stream_radio.c. Since enabling the cache involves replacing the old stream with a new one, the code as-is can't be kept. It would be easily possible to enable the cache by requesting a cache size (which is also much simpler). But nobody uses stream_radio.c and I can't even test this thing, and the cache is probably not really important for it either.
2013-05-24 18:49:09 +02:00
return stream_seek_long(s, pos);
}
int stream_skip(stream_t *s, int64_t len)
{
int64_t target = stream_tell(s) + len;
if (len < 0)
return stream_seek(s, target);
if (len > 2 * STREAM_BUFFER_SIZE && s->seekable) {
// Seek to 1 byte before target - this is the only way to distinguish
// skip-to-EOF and skip-past-EOF in general. Successful seeking means
// absolutely nothing, so test by doing a real read of the last byte.
int r = stream_seek(s, target - 1);
if (r) {
stream_read_char(s);
return !stream_eof(s) && stream_tell(s) == target;
}
return r;
}
return stream_skip_read(s, len);
}
int stream_control(stream_t *s, int cmd, void *arg)
{
if (!s->control)
return STREAM_UNSUPPORTED;
int r = s->control(s, cmd, arg);
if (r == STREAM_UNSUPPORTED) {
// Fallbacks
switch (cmd) {
case STREAM_CTRL_GET_SIZE:
if (s->end_pos > 0) {
*(int64_t *)arg = s->end_pos;
return STREAM_OK;
}
break;
}
}
return r;
}
void free_stream(stream_t *s)
{
if (!s)
return;
stream_set_capture_file(s, NULL);
if (s->close)
s->close(s);
free_stream(s->uncached_stream);
free_stream(s->source);
talloc_free(s);
}
2013-06-21 00:47:58 +02:00
stream_t *open_memory_stream(void *data, int len)
{
assert(len >= 0);
struct mpv_global *dummy = talloc_zero(NULL, struct mpv_global);
dummy->log = mp_null_log;
stream_t *s = stream_open("memory://", dummy);
assert(s);
talloc_steal(s, dummy);
stream_control(s, STREAM_CTRL_SET_CONTENTS, &(bstr){data, len});
2013-06-21 00:47:58 +02:00
return s;
}
static stream_t *open_cache(stream_t *orig, const char *name)
{
stream_t *cache = new_stream();
cache->uncached_type = orig->uncached_type;
cache->uncached_stream = orig;
cache->seekable = true;
cache->mode = STREAM_READ;
cache->read_chunk = 4 * STREAM_BUFFER_SIZE;
cache->url = talloc_strdup(cache, orig->url);
cache->mime_type = talloc_strdup(cache, orig->mime_type);
cache->demuxer = talloc_strdup(cache, orig->demuxer);
cache->lavf_type = talloc_strdup(cache, orig->lavf_type);
cache->safe_origin = orig->safe_origin;
cache->streaming = orig->streaming,
cache->is_network = orig->is_network;
cache->opts = orig->opts;
stream: redo playback abort handling This mechanism originates from MPlayer's way of dealing with blocking network, but it's still useful. On opening and closing, mpv waits for network synchronously, and also some obscure commands and use-cases can lead to such blocking. In these situations, the stream is asynchronously forced to stop by "interrupting" it. The old design interrupting I/O was a bit broken: polling with a callback, instead of actively interrupting it. Change the direction of this. There is no callback anymore, and the player calls mp_cancel_trigger() to force the stream to return. libavformat (via stream_lavf.c) has the old broken design, and fixing it would require fixing libavformat, which won't happen so quickly. So we have to keep that part. But everything above the stream layer is prepared for a better design, and more sophisticated methods than mp_cancel_test() could be easily introduced. There's still one problem: commands are still run in the central playback loop, which we assume can block on I/O in the worst case. That's not a problem yet, because we simply mark some commands as being able to stop playback of the current file ("quit" etc.), so input.c could abort playback as soon as such a command is queued. But there are also commands abort playback only conditionally, and the logic for that is in the playback core and thus "unreachable". For example, "playlist_next" aborts playback only if there's a next file. We don't want it to always abort playback. As a quite ugly hack, abort playback only if at least 2 abort commands are queued - this pretty much happens only if the core is frozen and doesn't react to input.
2014-09-13 14:23:08 +02:00
cache->cancel = orig->cancel;
cache->global = orig->global;
cache->log = mp_log_new(cache, cache->global->log, name);
return cache;
}
static struct mp_cache_opts check_cache_opts(stream_t *stream,
struct mp_cache_opts *opts)
cache: make the stream cache a proper stream that wraps other streams Before this commit, the cache was franken-hacked on top of the stream API. You had to use special functions (like cache_stream_fill_buffer() instead of stream_fill_buffer()), which would access the stream in a cached manner. The whole idea about the previous design was that the cache runs in a thread or in a forked process, while the cache awa functions made sure the stream instance looked consistent to the user. If you used the normal functions instead of the special ones while the cache was running, you were out of luck. Make it a bit more reasonable by turning the cache into a stream on its own. This makes it behave exactly like a normal stream. The stream callbacks call into the original (uncached) stream to do work. No special cache functions or redirections are needed. The only different thing about cache streams is that they are created by special functions, instead of being part of the auto_open_streams[] array. To make things simpler, remove the threading implementation, which was messed into the code. The threading code could perhaps be kept, but I don't really want to have to worry about this special case. A proper threaded implementation will be added later. Remove the cache enabling code from stream_radio.c. Since enabling the cache involves replacing the old stream with a new one, the code as-is can't be kept. It would be easily possible to enable the cache by requesting a cache size (which is also much simpler). But nobody uses stream_radio.c and I can't even test this thing, and the cache is probably not really important for it either.
2013-05-24 18:49:09 +02:00
{
struct mp_cache_opts use_opts = *opts;
if (use_opts.size == -1)
use_opts.size = stream->streaming ? use_opts.def_size : 0;
if (stream->mode != STREAM_READ || !stream->allow_caching || use_opts.size < 1)
use_opts.size = 0;
return use_opts;
cache: make the stream cache a proper stream that wraps other streams Before this commit, the cache was franken-hacked on top of the stream API. You had to use special functions (like cache_stream_fill_buffer() instead of stream_fill_buffer()), which would access the stream in a cached manner. The whole idea about the previous design was that the cache runs in a thread or in a forked process, while the cache awa functions made sure the stream instance looked consistent to the user. If you used the normal functions instead of the special ones while the cache was running, you were out of luck. Make it a bit more reasonable by turning the cache into a stream on its own. This makes it behave exactly like a normal stream. The stream callbacks call into the original (uncached) stream to do work. No special cache functions or redirections are needed. The only different thing about cache streams is that they are created by special functions, instead of being part of the auto_open_streams[] array. To make things simpler, remove the threading implementation, which was messed into the code. The threading code could perhaps be kept, but I don't really want to have to worry about this special case. A proper threaded implementation will be added later. Remove the cache enabling code from stream_radio.c. Since enabling the cache involves replacing the old stream with a new one, the code as-is can't be kept. It would be easily possible to enable the cache by requesting a cache size (which is also much simpler). But nobody uses stream_radio.c and I can't even test this thing, and the cache is probably not really important for it either.
2013-05-24 18:49:09 +02:00
}
bool stream_wants_cache(stream_t *stream, struct mp_cache_opts *opts)
{
struct mp_cache_opts use_opts = check_cache_opts(stream, opts);
return use_opts.size > 0;
}
// return 1 on success, 0 if the cache is disabled/not needed, and -1 on error
// or if the cache is disabled
int stream_enable_cache(stream_t **stream, struct mp_cache_opts *opts)
cache: make the stream cache a proper stream that wraps other streams Before this commit, the cache was franken-hacked on top of the stream API. You had to use special functions (like cache_stream_fill_buffer() instead of stream_fill_buffer()), which would access the stream in a cached manner. The whole idea about the previous design was that the cache runs in a thread or in a forked process, while the cache awa functions made sure the stream instance looked consistent to the user. If you used the normal functions instead of the special ones while the cache was running, you were out of luck. Make it a bit more reasonable by turning the cache into a stream on its own. This makes it behave exactly like a normal stream. The stream callbacks call into the original (uncached) stream to do work. No special cache functions or redirections are needed. The only different thing about cache streams is that they are created by special functions, instead of being part of the auto_open_streams[] array. To make things simpler, remove the threading implementation, which was messed into the code. The threading code could perhaps be kept, but I don't really want to have to worry about this special case. A proper threaded implementation will be added later. Remove the cache enabling code from stream_radio.c. Since enabling the cache involves replacing the old stream with a new one, the code as-is can't be kept. It would be easily possible to enable the cache by requesting a cache size (which is also much simpler). But nobody uses stream_radio.c and I can't even test this thing, and the cache is probably not really important for it either.
2013-05-24 18:49:09 +02:00
{
stream_t *orig = *stream;
struct mp_cache_opts use_opts = check_cache_opts(*stream, opts);
cache: make the stream cache a proper stream that wraps other streams Before this commit, the cache was franken-hacked on top of the stream API. You had to use special functions (like cache_stream_fill_buffer() instead of stream_fill_buffer()), which would access the stream in a cached manner. The whole idea about the previous design was that the cache runs in a thread or in a forked process, while the cache awa functions made sure the stream instance looked consistent to the user. If you used the normal functions instead of the special ones while the cache was running, you were out of luck. Make it a bit more reasonable by turning the cache into a stream on its own. This makes it behave exactly like a normal stream. The stream callbacks call into the original (uncached) stream to do work. No special cache functions or redirections are needed. The only different thing about cache streams is that they are created by special functions, instead of being part of the auto_open_streams[] array. To make things simpler, remove the threading implementation, which was messed into the code. The threading code could perhaps be kept, but I don't really want to have to worry about this special case. A proper threaded implementation will be added later. Remove the cache enabling code from stream_radio.c. Since enabling the cache involves replacing the old stream with a new one, the code as-is can't be kept. It would be easily possible to enable the cache by requesting a cache size (which is also much simpler). But nobody uses stream_radio.c and I can't even test this thing, and the cache is probably not really important for it either.
2013-05-24 18:49:09 +02:00
if (use_opts.size < 1)
return 0;
cache: make the stream cache a proper stream that wraps other streams Before this commit, the cache was franken-hacked on top of the stream API. You had to use special functions (like cache_stream_fill_buffer() instead of stream_fill_buffer()), which would access the stream in a cached manner. The whole idea about the previous design was that the cache runs in a thread or in a forked process, while the cache awa functions made sure the stream instance looked consistent to the user. If you used the normal functions instead of the special ones while the cache was running, you were out of luck. Make it a bit more reasonable by turning the cache into a stream on its own. This makes it behave exactly like a normal stream. The stream callbacks call into the original (uncached) stream to do work. No special cache functions or redirections are needed. The only different thing about cache streams is that they are created by special functions, instead of being part of the auto_open_streams[] array. To make things simpler, remove the threading implementation, which was messed into the code. The threading code could perhaps be kept, but I don't really want to have to worry about this special case. A proper threaded implementation will be added later. Remove the cache enabling code from stream_radio.c. Since enabling the cache involves replacing the old stream with a new one, the code as-is can't be kept. It would be easily possible to enable the cache by requesting a cache size (which is also much simpler). But nobody uses stream_radio.c and I can't even test this thing, and the cache is probably not really important for it either.
2013-05-24 18:49:09 +02:00
stream_t *fcache = open_cache(orig, "file-cache");
if (stream_file_cache_init(fcache, orig, &use_opts) <= 0) {
fcache->uncached_stream = NULL; // don't free original stream
free_stream(fcache);
fcache = orig;
}
cache: make the stream cache a proper stream that wraps other streams Before this commit, the cache was franken-hacked on top of the stream API. You had to use special functions (like cache_stream_fill_buffer() instead of stream_fill_buffer()), which would access the stream in a cached manner. The whole idea about the previous design was that the cache runs in a thread or in a forked process, while the cache awa functions made sure the stream instance looked consistent to the user. If you used the normal functions instead of the special ones while the cache was running, you were out of luck. Make it a bit more reasonable by turning the cache into a stream on its own. This makes it behave exactly like a normal stream. The stream callbacks call into the original (uncached) stream to do work. No special cache functions or redirections are needed. The only different thing about cache streams is that they are created by special functions, instead of being part of the auto_open_streams[] array. To make things simpler, remove the threading implementation, which was messed into the code. The threading code could perhaps be kept, but I don't really want to have to worry about this special case. A proper threaded implementation will be added later. Remove the cache enabling code from stream_radio.c. Since enabling the cache involves replacing the old stream with a new one, the code as-is can't be kept. It would be easily possible to enable the cache by requesting a cache size (which is also much simpler). But nobody uses stream_radio.c and I can't even test this thing, and the cache is probably not really important for it either.
2013-05-24 18:49:09 +02:00
stream_t *cache = open_cache(fcache, "cache");
int res = stream_cache_init(cache, fcache, &use_opts);
cache: make the stream cache a proper stream that wraps other streams Before this commit, the cache was franken-hacked on top of the stream API. You had to use special functions (like cache_stream_fill_buffer() instead of stream_fill_buffer()), which would access the stream in a cached manner. The whole idea about the previous design was that the cache runs in a thread or in a forked process, while the cache awa functions made sure the stream instance looked consistent to the user. If you used the normal functions instead of the special ones while the cache was running, you were out of luck. Make it a bit more reasonable by turning the cache into a stream on its own. This makes it behave exactly like a normal stream. The stream callbacks call into the original (uncached) stream to do work. No special cache functions or redirections are needed. The only different thing about cache streams is that they are created by special functions, instead of being part of the auto_open_streams[] array. To make things simpler, remove the threading implementation, which was messed into the code. The threading code could perhaps be kept, but I don't really want to have to worry about this special case. A proper threaded implementation will be added later. Remove the cache enabling code from stream_radio.c. Since enabling the cache involves replacing the old stream with a new one, the code as-is can't be kept. It would be easily possible to enable the cache by requesting a cache size (which is also much simpler). But nobody uses stream_radio.c and I can't even test this thing, and the cache is probably not really important for it either.
2013-05-24 18:49:09 +02:00
if (res <= 0) {
cache->uncached_stream = NULL; // don't free original stream
free_stream(cache);
if (fcache != orig)
free_stream(fcache);
cache: make the stream cache a proper stream that wraps other streams Before this commit, the cache was franken-hacked on top of the stream API. You had to use special functions (like cache_stream_fill_buffer() instead of stream_fill_buffer()), which would access the stream in a cached manner. The whole idea about the previous design was that the cache runs in a thread or in a forked process, while the cache awa functions made sure the stream instance looked consistent to the user. If you used the normal functions instead of the special ones while the cache was running, you were out of luck. Make it a bit more reasonable by turning the cache into a stream on its own. This makes it behave exactly like a normal stream. The stream callbacks call into the original (uncached) stream to do work. No special cache functions or redirections are needed. The only different thing about cache streams is that they are created by special functions, instead of being part of the auto_open_streams[] array. To make things simpler, remove the threading implementation, which was messed into the code. The threading code could perhaps be kept, but I don't really want to have to worry about this special case. A proper threaded implementation will be added later. Remove the cache enabling code from stream_radio.c. Since enabling the cache involves replacing the old stream with a new one, the code as-is can't be kept. It would be easily possible to enable the cache by requesting a cache size (which is also much simpler). But nobody uses stream_radio.c and I can't even test this thing, and the cache is probably not really important for it either.
2013-05-24 18:49:09 +02:00
} else {
*stream = cache;
}
return res;
}
static uint16_t stream_read_word_endian(stream_t *s, bool big_endian)
{
unsigned int y = stream_read_char(s);
y = (y << 8) | stream_read_char(s);
if (!big_endian)
y = ((y >> 8) & 0xFF) | (y << 8);
return y;
}
// Read characters until the next '\n' (including), or until the buffer in s is
// exhausted.
static int read_characters(stream_t *s, uint8_t *dst, int dstsize, int utf16)
{
if (utf16 == 1 || utf16 == 2) {
uint8_t *cur = dst;
while (1) {
if ((cur - dst) + 8 >= dstsize) // PUT_UTF8 writes max. 8 bytes
return -1; // line too long
uint32_t c;
uint8_t tmp;
GET_UTF16(c, stream_read_word_endian(s, utf16 == 2), return -1;)
if (s->eof)
break; // legitimate EOF; ignore the case of partial reads
PUT_UTF8(c, tmp, *cur++ = tmp;)
if (c == '\n')
break;
}
return cur - dst;
} else {
if (s->buf_pos >= s->buf_len)
stream_fill_buffer(s);
uint8_t *src = s->buffer + s->buf_pos;
int src_len = s->buf_len - s->buf_pos;
uint8_t *end = memchr(src, '\n', src_len);
int len = end ? end - src + 1 : src_len;
if (len > dstsize)
return -1; // line too long
memcpy(dst, src, len);
s->buf_pos += len;
return len;
}
}
// On error, or if the line is larger than max-1, return NULL and unset s->eof.
// On EOF, return NULL, and s->eof will be set.
// Otherwise, return the line (including \n or \r\n at the end of the line).
// If the return value is non-NULL, it's always the same as mem.
// utf16: 0: UTF8 or 8 bit legacy, 1: UTF16-LE, 2: UTF16-BE
unsigned char *stream_read_line(stream_t *s, unsigned char *mem, int max,
int utf16)
{
if (max < 1)
return NULL;
int read = 0;
while (1) {
// Reserve 1 byte of ptr for terminating \0.
int l = read_characters(s, &mem[read], max - read - 1, utf16);
if (l < 0 || memchr(&mem[read], '\0', l)) {
MP_WARN(s, "error reading line\n");
s->eof = false;
return NULL;
}
read += l;
if (l == 0 || (read > 0 && mem[read - 1] == '\n'))
break;
}
mem[read] = '\0';
if (s->eof && read == 0) // legitimate EOF
return NULL;
return mem;
}
static const char *const bom[3] = {"\xEF\xBB\xBF", "\xFF\xFE", "\xFE\xFF"};
// Return utf16 argument for stream_read_line
int stream_skip_bom(struct stream *s)
{
bstr data = stream_peek(s, 4);
for (int n = 0; n < 3; n++) {
if (bstr_startswith0(data, bom[n])) {
stream_skip(s, strlen(bom[n]));
return n;
}
}
return -1; // default to 8 bit codepages
}
// Read the rest of the stream into memory (current pos to EOF), and return it.
// talloc_ctx: used as talloc parent for the returned allocation
// max_size: must be set to >0. If the file is larger than that, it is treated
// as error. This is a minor robustness measure.
// returns: stream contents, or .start/.len set to NULL on error
// If the file was empty, but no error happened, .start will be non-NULL and
// .len will be 0.
// For convenience, the returned buffer is padded with a 0 byte. The padding
// is not included in the returned length.
struct bstr stream_read_complete(struct stream *s, void *talloc_ctx,
int max_size)
{
if (max_size > 1000000000)
abort();
int bufsize;
int total_read = 0;
int padding = 1;
char *buf = NULL;
int64_t size = 0;
stream_control(s, STREAM_CTRL_GET_SIZE, &size);
if (size > max_size)
return (struct bstr){NULL, 0};
if (size > 0)
bufsize = size + padding;
else
bufsize = 1000;
while (1) {
buf = talloc_realloc_size(talloc_ctx, buf, bufsize);
int readsize = stream_read(s, buf + total_read, bufsize - total_read);
total_read += readsize;
if (total_read < bufsize)
break;
if (bufsize > max_size) {
talloc_free(buf);
return (struct bstr){NULL, 0};
}
bufsize = FFMIN(bufsize + (bufsize >> 1), max_size + padding);
}
buf = talloc_realloc_size(talloc_ctx, buf, total_read + padding);
memset(&buf[total_read], 0, padding);
return (struct bstr){buf, total_read};
}
stream: redo playback abort handling This mechanism originates from MPlayer's way of dealing with blocking network, but it's still useful. On opening and closing, mpv waits for network synchronously, and also some obscure commands and use-cases can lead to such blocking. In these situations, the stream is asynchronously forced to stop by "interrupting" it. The old design interrupting I/O was a bit broken: polling with a callback, instead of actively interrupting it. Change the direction of this. There is no callback anymore, and the player calls mp_cancel_trigger() to force the stream to return. libavformat (via stream_lavf.c) has the old broken design, and fixing it would require fixing libavformat, which won't happen so quickly. So we have to keep that part. But everything above the stream layer is prepared for a better design, and more sophisticated methods than mp_cancel_test() could be easily introduced. There's still one problem: commands are still run in the central playback loop, which we assume can block on I/O in the worst case. That's not a problem yet, because we simply mark some commands as being able to stop playback of the current file ("quit" etc.), so input.c could abort playback as soon as such a command is queued. But there are also commands abort playback only conditionally, and the logic for that is in the playback core and thus "unreachable". For example, "playlist_next" aborts playback only if there's a next file. We don't want it to always abort playback. As a quite ugly hack, abort playback only if at least 2 abort commands are queued - this pretty much happens only if the core is frozen and doesn't react to input.
2014-09-13 14:23:08 +02:00
struct mp_cancel {
atomic_bool triggered;
#ifdef __MINGW32__
HANDLE event;
#endif
int wakeup_pipe[2];
stream: redo playback abort handling This mechanism originates from MPlayer's way of dealing with blocking network, but it's still useful. On opening and closing, mpv waits for network synchronously, and also some obscure commands and use-cases can lead to such blocking. In these situations, the stream is asynchronously forced to stop by "interrupting" it. The old design interrupting I/O was a bit broken: polling with a callback, instead of actively interrupting it. Change the direction of this. There is no callback anymore, and the player calls mp_cancel_trigger() to force the stream to return. libavformat (via stream_lavf.c) has the old broken design, and fixing it would require fixing libavformat, which won't happen so quickly. So we have to keep that part. But everything above the stream layer is prepared for a better design, and more sophisticated methods than mp_cancel_test() could be easily introduced. There's still one problem: commands are still run in the central playback loop, which we assume can block on I/O in the worst case. That's not a problem yet, because we simply mark some commands as being able to stop playback of the current file ("quit" etc.), so input.c could abort playback as soon as such a command is queued. But there are also commands abort playback only conditionally, and the logic for that is in the playback core and thus "unreachable". For example, "playlist_next" aborts playback only if there's a next file. We don't want it to always abort playback. As a quite ugly hack, abort playback only if at least 2 abort commands are queued - this pretty much happens only if the core is frozen and doesn't react to input.
2014-09-13 14:23:08 +02:00
};
lua: add an utility function for starting processes Because 1) Lua is terrible, and 2) popen() is terrible. Unfortunately, since Unix is also terrible, this turned out more complicated than I hoped. As a consequence and to avoid that this code has to be maintained forever, add a disclaimer that any function in Lua's utils module can disappear any time. The complexity seems a bit ridiculous, especially for a feature so far removed from actual video playback, so if it turns out that we don't really need this function, it will be dropped again. The motivation for this commit is the same as with 8e4fa5fc. Note that there is an "#ifndef __GLIBC__". The GNU people are very special people and thought it'd be convenient to actually declare "environ", even though the POSIX people, which are also very special people, state that no header declares this and that the user has to declare this manually. Since the GNU people overtook the Unix world with their very clever "embrace, extend, extinguish" strategy, but not 100%, and trying to build without _GNU_SOURCE is hopeless; but since there might be Unix environments which support _GNU_SOURCE features partially, this means that in practice "environ" will be randomly declared or not declared by system headers. Also, gcc was written by very clever people too, and prints a warning if an external variable is declared twice (I didn't check, but I suppose redeclaring is legal C, and not even the gcc people are clever enough to only warn against a definitely not legal C construct, although sometimes they do this), ...and since we at mpv hate compiler warnings, we seek to silence them all. Adding a configure test just for a warning seems too radical, so we special-case this against __GLIBC__, which is hopefully not defined on other libcs, especially not libcs which don't implement all aspects of _GNU_SOURCE, and redefine "environ" on systems even if the headers define it already (because they support _GNU_SOURCE - as I mentioned before, the clever GNU people wrote software THAT portable that other libcs just gave up and implemented parts of _GNU_SOURCE, although probably not all), which means that compiling mpv will print a warning about "environ" being redefined, but at least this won't happen on my system, so all is fine. However, should someone complain about this warning, I will force whoever complained about this warning to read this ENTIRE commit message, and if possible, will also force them to eat a printed-out copy of the GNU Manifesto, and if that is not enough, maybe this person could even be forced to convince the very clever POSIX people of not doing crap like this: having the user to manually declare somewhat central symbols - but I doubt it's possible, because the POSIX people are too far gone and only care about maintaining compatibility with old versions of AIX and HP-UX. Oh, also, this code contains some subtle and obvious issues, but writing about this is not fun.
2014-10-19 01:42:28 +02:00
static void cancel_destroy(void *p)
{
struct mp_cancel *c = p;
#ifdef __MINGW32__
CloseHandle(c->event);
#endif
lua: add an utility function for starting processes Because 1) Lua is terrible, and 2) popen() is terrible. Unfortunately, since Unix is also terrible, this turned out more complicated than I hoped. As a consequence and to avoid that this code has to be maintained forever, add a disclaimer that any function in Lua's utils module can disappear any time. The complexity seems a bit ridiculous, especially for a feature so far removed from actual video playback, so if it turns out that we don't really need this function, it will be dropped again. The motivation for this commit is the same as with 8e4fa5fc. Note that there is an "#ifndef __GLIBC__". The GNU people are very special people and thought it'd be convenient to actually declare "environ", even though the POSIX people, which are also very special people, state that no header declares this and that the user has to declare this manually. Since the GNU people overtook the Unix world with their very clever "embrace, extend, extinguish" strategy, but not 100%, and trying to build without _GNU_SOURCE is hopeless; but since there might be Unix environments which support _GNU_SOURCE features partially, this means that in practice "environ" will be randomly declared or not declared by system headers. Also, gcc was written by very clever people too, and prints a warning if an external variable is declared twice (I didn't check, but I suppose redeclaring is legal C, and not even the gcc people are clever enough to only warn against a definitely not legal C construct, although sometimes they do this), ...and since we at mpv hate compiler warnings, we seek to silence them all. Adding a configure test just for a warning seems too radical, so we special-case this against __GLIBC__, which is hopefully not defined on other libcs, especially not libcs which don't implement all aspects of _GNU_SOURCE, and redefine "environ" on systems even if the headers define it already (because they support _GNU_SOURCE - as I mentioned before, the clever GNU people wrote software THAT portable that other libcs just gave up and implemented parts of _GNU_SOURCE, although probably not all), which means that compiling mpv will print a warning about "environ" being redefined, but at least this won't happen on my system, so all is fine. However, should someone complain about this warning, I will force whoever complained about this warning to read this ENTIRE commit message, and if possible, will also force them to eat a printed-out copy of the GNU Manifesto, and if that is not enough, maybe this person could even be forced to convince the very clever POSIX people of not doing crap like this: having the user to manually declare somewhat central symbols - but I doubt it's possible, because the POSIX people are too far gone and only care about maintaining compatibility with old versions of AIX and HP-UX. Oh, also, this code contains some subtle and obvious issues, but writing about this is not fun.
2014-10-19 01:42:28 +02:00
close(c->wakeup_pipe[0]);
close(c->wakeup_pipe[1]);
}
stream: redo playback abort handling This mechanism originates from MPlayer's way of dealing with blocking network, but it's still useful. On opening and closing, mpv waits for network synchronously, and also some obscure commands and use-cases can lead to such blocking. In these situations, the stream is asynchronously forced to stop by "interrupting" it. The old design interrupting I/O was a bit broken: polling with a callback, instead of actively interrupting it. Change the direction of this. There is no callback anymore, and the player calls mp_cancel_trigger() to force the stream to return. libavformat (via stream_lavf.c) has the old broken design, and fixing it would require fixing libavformat, which won't happen so quickly. So we have to keep that part. But everything above the stream layer is prepared for a better design, and more sophisticated methods than mp_cancel_test() could be easily introduced. There's still one problem: commands are still run in the central playback loop, which we assume can block on I/O in the worst case. That's not a problem yet, because we simply mark some commands as being able to stop playback of the current file ("quit" etc.), so input.c could abort playback as soon as such a command is queued. But there are also commands abort playback only conditionally, and the logic for that is in the playback core and thus "unreachable". For example, "playlist_next" aborts playback only if there's a next file. We don't want it to always abort playback. As a quite ugly hack, abort playback only if at least 2 abort commands are queued - this pretty much happens only if the core is frozen and doesn't react to input.
2014-09-13 14:23:08 +02:00
struct mp_cancel *mp_cancel_new(void *talloc_ctx)
{
struct mp_cancel *c = talloc_ptrtype(talloc_ctx, c);
lua: add an utility function for starting processes Because 1) Lua is terrible, and 2) popen() is terrible. Unfortunately, since Unix is also terrible, this turned out more complicated than I hoped. As a consequence and to avoid that this code has to be maintained forever, add a disclaimer that any function in Lua's utils module can disappear any time. The complexity seems a bit ridiculous, especially for a feature so far removed from actual video playback, so if it turns out that we don't really need this function, it will be dropped again. The motivation for this commit is the same as with 8e4fa5fc. Note that there is an "#ifndef __GLIBC__". The GNU people are very special people and thought it'd be convenient to actually declare "environ", even though the POSIX people, which are also very special people, state that no header declares this and that the user has to declare this manually. Since the GNU people overtook the Unix world with their very clever "embrace, extend, extinguish" strategy, but not 100%, and trying to build without _GNU_SOURCE is hopeless; but since there might be Unix environments which support _GNU_SOURCE features partially, this means that in practice "environ" will be randomly declared or not declared by system headers. Also, gcc was written by very clever people too, and prints a warning if an external variable is declared twice (I didn't check, but I suppose redeclaring is legal C, and not even the gcc people are clever enough to only warn against a definitely not legal C construct, although sometimes they do this), ...and since we at mpv hate compiler warnings, we seek to silence them all. Adding a configure test just for a warning seems too radical, so we special-case this against __GLIBC__, which is hopefully not defined on other libcs, especially not libcs which don't implement all aspects of _GNU_SOURCE, and redefine "environ" on systems even if the headers define it already (because they support _GNU_SOURCE - as I mentioned before, the clever GNU people wrote software THAT portable that other libcs just gave up and implemented parts of _GNU_SOURCE, although probably not all), which means that compiling mpv will print a warning about "environ" being redefined, but at least this won't happen on my system, so all is fine. However, should someone complain about this warning, I will force whoever complained about this warning to read this ENTIRE commit message, and if possible, will also force them to eat a printed-out copy of the GNU Manifesto, and if that is not enough, maybe this person could even be forced to convince the very clever POSIX people of not doing crap like this: having the user to manually declare somewhat central symbols - but I doubt it's possible, because the POSIX people are too far gone and only care about maintaining compatibility with old versions of AIX and HP-UX. Oh, also, this code contains some subtle and obvious issues, but writing about this is not fun.
2014-10-19 01:42:28 +02:00
talloc_set_destructor(c, cancel_destroy);
stream: redo playback abort handling This mechanism originates from MPlayer's way of dealing with blocking network, but it's still useful. On opening and closing, mpv waits for network synchronously, and also some obscure commands and use-cases can lead to such blocking. In these situations, the stream is asynchronously forced to stop by "interrupting" it. The old design interrupting I/O was a bit broken: polling with a callback, instead of actively interrupting it. Change the direction of this. There is no callback anymore, and the player calls mp_cancel_trigger() to force the stream to return. libavformat (via stream_lavf.c) has the old broken design, and fixing it would require fixing libavformat, which won't happen so quickly. So we have to keep that part. But everything above the stream layer is prepared for a better design, and more sophisticated methods than mp_cancel_test() could be easily introduced. There's still one problem: commands are still run in the central playback loop, which we assume can block on I/O in the worst case. That's not a problem yet, because we simply mark some commands as being able to stop playback of the current file ("quit" etc.), so input.c could abort playback as soon as such a command is queued. But there are also commands abort playback only conditionally, and the logic for that is in the playback core and thus "unreachable". For example, "playlist_next" aborts playback only if there's a next file. We don't want it to always abort playback. As a quite ugly hack, abort playback only if at least 2 abort commands are queued - this pretty much happens only if the core is frozen and doesn't react to input.
2014-09-13 14:23:08 +02:00
*c = (struct mp_cancel){.triggered = ATOMIC_VAR_INIT(false)};
#ifdef __MINGW32__
c->event = CreateEventW(NULL, TRUE, FALSE, NULL);
#endif
mp_make_wakeup_pipe(c->wakeup_pipe);
stream: redo playback abort handling This mechanism originates from MPlayer's way of dealing with blocking network, but it's still useful. On opening and closing, mpv waits for network synchronously, and also some obscure commands and use-cases can lead to such blocking. In these situations, the stream is asynchronously forced to stop by "interrupting" it. The old design interrupting I/O was a bit broken: polling with a callback, instead of actively interrupting it. Change the direction of this. There is no callback anymore, and the player calls mp_cancel_trigger() to force the stream to return. libavformat (via stream_lavf.c) has the old broken design, and fixing it would require fixing libavformat, which won't happen so quickly. So we have to keep that part. But everything above the stream layer is prepared for a better design, and more sophisticated methods than mp_cancel_test() could be easily introduced. There's still one problem: commands are still run in the central playback loop, which we assume can block on I/O in the worst case. That's not a problem yet, because we simply mark some commands as being able to stop playback of the current file ("quit" etc.), so input.c could abort playback as soon as such a command is queued. But there are also commands abort playback only conditionally, and the logic for that is in the playback core and thus "unreachable". For example, "playlist_next" aborts playback only if there's a next file. We don't want it to always abort playback. As a quite ugly hack, abort playback only if at least 2 abort commands are queued - this pretty much happens only if the core is frozen and doesn't react to input.
2014-09-13 14:23:08 +02:00
return c;
}
// Request abort.
void mp_cancel_trigger(struct mp_cancel *c)
{
atomic_store(&c->triggered, true);
#ifdef __MINGW32__
SetEvent(c->event);
#endif
write(c->wakeup_pipe[1], &(char){0}, 1);
stream: redo playback abort handling This mechanism originates from MPlayer's way of dealing with blocking network, but it's still useful. On opening and closing, mpv waits for network synchronously, and also some obscure commands and use-cases can lead to such blocking. In these situations, the stream is asynchronously forced to stop by "interrupting" it. The old design interrupting I/O was a bit broken: polling with a callback, instead of actively interrupting it. Change the direction of this. There is no callback anymore, and the player calls mp_cancel_trigger() to force the stream to return. libavformat (via stream_lavf.c) has the old broken design, and fixing it would require fixing libavformat, which won't happen so quickly. So we have to keep that part. But everything above the stream layer is prepared for a better design, and more sophisticated methods than mp_cancel_test() could be easily introduced. There's still one problem: commands are still run in the central playback loop, which we assume can block on I/O in the worst case. That's not a problem yet, because we simply mark some commands as being able to stop playback of the current file ("quit" etc.), so input.c could abort playback as soon as such a command is queued. But there are also commands abort playback only conditionally, and the logic for that is in the playback core and thus "unreachable". For example, "playlist_next" aborts playback only if there's a next file. We don't want it to always abort playback. As a quite ugly hack, abort playback only if at least 2 abort commands are queued - this pretty much happens only if the core is frozen and doesn't react to input.
2014-09-13 14:23:08 +02:00
}
// Restore original state. (Allows reusing a mp_cancel.)
void mp_cancel_reset(struct mp_cancel *c)
{
atomic_store(&c->triggered, false);
#ifdef __MINGW32__
ResetEvent(c->event);
#endif
lua: add an utility function for starting processes Because 1) Lua is terrible, and 2) popen() is terrible. Unfortunately, since Unix is also terrible, this turned out more complicated than I hoped. As a consequence and to avoid that this code has to be maintained forever, add a disclaimer that any function in Lua's utils module can disappear any time. The complexity seems a bit ridiculous, especially for a feature so far removed from actual video playback, so if it turns out that we don't really need this function, it will be dropped again. The motivation for this commit is the same as with 8e4fa5fc. Note that there is an "#ifndef __GLIBC__". The GNU people are very special people and thought it'd be convenient to actually declare "environ", even though the POSIX people, which are also very special people, state that no header declares this and that the user has to declare this manually. Since the GNU people overtook the Unix world with their very clever "embrace, extend, extinguish" strategy, but not 100%, and trying to build without _GNU_SOURCE is hopeless; but since there might be Unix environments which support _GNU_SOURCE features partially, this means that in practice "environ" will be randomly declared or not declared by system headers. Also, gcc was written by very clever people too, and prints a warning if an external variable is declared twice (I didn't check, but I suppose redeclaring is legal C, and not even the gcc people are clever enough to only warn against a definitely not legal C construct, although sometimes they do this), ...and since we at mpv hate compiler warnings, we seek to silence them all. Adding a configure test just for a warning seems too radical, so we special-case this against __GLIBC__, which is hopefully not defined on other libcs, especially not libcs which don't implement all aspects of _GNU_SOURCE, and redefine "environ" on systems even if the headers define it already (because they support _GNU_SOURCE - as I mentioned before, the clever GNU people wrote software THAT portable that other libcs just gave up and implemented parts of _GNU_SOURCE, although probably not all), which means that compiling mpv will print a warning about "environ" being redefined, but at least this won't happen on my system, so all is fine. However, should someone complain about this warning, I will force whoever complained about this warning to read this ENTIRE commit message, and if possible, will also force them to eat a printed-out copy of the GNU Manifesto, and if that is not enough, maybe this person could even be forced to convince the very clever POSIX people of not doing crap like this: having the user to manually declare somewhat central symbols - but I doubt it's possible, because the POSIX people are too far gone and only care about maintaining compatibility with old versions of AIX and HP-UX. Oh, also, this code contains some subtle and obvious issues, but writing about this is not fun.
2014-10-19 01:42:28 +02:00
// Flush it fully.
while (1) {
int r = read(c->wakeup_pipe[0], &(char[256]){0}, 256);
if (r < 0 && errno == EINTR)
continue;
if (r <= 0)
break;
}
stream: redo playback abort handling This mechanism originates from MPlayer's way of dealing with blocking network, but it's still useful. On opening and closing, mpv waits for network synchronously, and also some obscure commands and use-cases can lead to such blocking. In these situations, the stream is asynchronously forced to stop by "interrupting" it. The old design interrupting I/O was a bit broken: polling with a callback, instead of actively interrupting it. Change the direction of this. There is no callback anymore, and the player calls mp_cancel_trigger() to force the stream to return. libavformat (via stream_lavf.c) has the old broken design, and fixing it would require fixing libavformat, which won't happen so quickly. So we have to keep that part. But everything above the stream layer is prepared for a better design, and more sophisticated methods than mp_cancel_test() could be easily introduced. There's still one problem: commands are still run in the central playback loop, which we assume can block on I/O in the worst case. That's not a problem yet, because we simply mark some commands as being able to stop playback of the current file ("quit" etc.), so input.c could abort playback as soon as such a command is queued. But there are also commands abort playback only conditionally, and the logic for that is in the playback core and thus "unreachable". For example, "playlist_next" aborts playback only if there's a next file. We don't want it to always abort playback. As a quite ugly hack, abort playback only if at least 2 abort commands are queued - this pretty much happens only if the core is frozen and doesn't react to input.
2014-09-13 14:23:08 +02:00
}
// Return whether the caller should abort.
// For convenience, c==NULL is allowed.
bool mp_cancel_test(struct mp_cancel *c)
{
return c ? atomic_load(&c->triggered) : false;
stream: redo playback abort handling This mechanism originates from MPlayer's way of dealing with blocking network, but it's still useful. On opening and closing, mpv waits for network synchronously, and also some obscure commands and use-cases can lead to such blocking. In these situations, the stream is asynchronously forced to stop by "interrupting" it. The old design interrupting I/O was a bit broken: polling with a callback, instead of actively interrupting it. Change the direction of this. There is no callback anymore, and the player calls mp_cancel_trigger() to force the stream to return. libavformat (via stream_lavf.c) has the old broken design, and fixing it would require fixing libavformat, which won't happen so quickly. So we have to keep that part. But everything above the stream layer is prepared for a better design, and more sophisticated methods than mp_cancel_test() could be easily introduced. There's still one problem: commands are still run in the central playback loop, which we assume can block on I/O in the worst case. That's not a problem yet, because we simply mark some commands as being able to stop playback of the current file ("quit" etc.), so input.c could abort playback as soon as such a command is queued. But there are also commands abort playback only conditionally, and the logic for that is in the playback core and thus "unreachable". For example, "playlist_next" aborts playback only if there's a next file. We don't want it to always abort playback. As a quite ugly hack, abort playback only if at least 2 abort commands are queued - this pretty much happens only if the core is frozen and doesn't react to input.
2014-09-13 14:23:08 +02:00
}
#ifdef __MINGW32__
void *mp_cancel_get_event(struct mp_cancel *c)
{
return c->event;
}
#endif
lua: add an utility function for starting processes Because 1) Lua is terrible, and 2) popen() is terrible. Unfortunately, since Unix is also terrible, this turned out more complicated than I hoped. As a consequence and to avoid that this code has to be maintained forever, add a disclaimer that any function in Lua's utils module can disappear any time. The complexity seems a bit ridiculous, especially for a feature so far removed from actual video playback, so if it turns out that we don't really need this function, it will be dropped again. The motivation for this commit is the same as with 8e4fa5fc. Note that there is an "#ifndef __GLIBC__". The GNU people are very special people and thought it'd be convenient to actually declare "environ", even though the POSIX people, which are also very special people, state that no header declares this and that the user has to declare this manually. Since the GNU people overtook the Unix world with their very clever "embrace, extend, extinguish" strategy, but not 100%, and trying to build without _GNU_SOURCE is hopeless; but since there might be Unix environments which support _GNU_SOURCE features partially, this means that in practice "environ" will be randomly declared or not declared by system headers. Also, gcc was written by very clever people too, and prints a warning if an external variable is declared twice (I didn't check, but I suppose redeclaring is legal C, and not even the gcc people are clever enough to only warn against a definitely not legal C construct, although sometimes they do this), ...and since we at mpv hate compiler warnings, we seek to silence them all. Adding a configure test just for a warning seems too radical, so we special-case this against __GLIBC__, which is hopefully not defined on other libcs, especially not libcs which don't implement all aspects of _GNU_SOURCE, and redefine "environ" on systems even if the headers define it already (because they support _GNU_SOURCE - as I mentioned before, the clever GNU people wrote software THAT portable that other libcs just gave up and implemented parts of _GNU_SOURCE, although probably not all), which means that compiling mpv will print a warning about "environ" being redefined, but at least this won't happen on my system, so all is fine. However, should someone complain about this warning, I will force whoever complained about this warning to read this ENTIRE commit message, and if possible, will also force them to eat a printed-out copy of the GNU Manifesto, and if that is not enough, maybe this person could even be forced to convince the very clever POSIX people of not doing crap like this: having the user to manually declare somewhat central symbols - but I doubt it's possible, because the POSIX people are too far gone and only care about maintaining compatibility with old versions of AIX and HP-UX. Oh, also, this code contains some subtle and obvious issues, but writing about this is not fun.
2014-10-19 01:42:28 +02:00
// The FD becomes readable if mp_cancel_test() would return true.
// Don't actually read from it, just use it for poll().
int mp_cancel_get_fd(struct mp_cancel *c)
{
return c->wakeup_pipe[0];
}
2014-06-30 12:49:01 +02:00
void stream_print_proto_list(struct mp_log *log)
{
int count = 0;
mp_info(log, "Protocols:\n\n");
for (int i = 0; stream_list[i]; i++) {
const stream_info_t *stream_info = stream_list[i];
if (!stream_info->protocols)
continue;
for (int j = 0; stream_info->protocols[j]; j++) {
if (*stream_info->protocols[j] == '\0')
continue;
mp_info(log, " %s://\n", stream_info->protocols[j]);
count++;
}
}
mp_info(log, "\nTotal: %d protocols\n", count);
}