X-Git-Url: https://git.sesse.net/?a=blobdiff_plain;f=src%2Faudio_output%2Ffilters.c;h=446f8b65ab17a393d50e6c176478ebe52cc9d63a;hb=7b0761c4359d12de0bad78a1d255a55cb89bc562;hp=d29352f90118a354ccdd443af68b290985908467;hpb=00b6b76217c39d6a7abfa0ada531b743b2aa594a;p=vlc diff --git a/src/audio_output/filters.c b/src/audio_output/filters.c index d29352f901..446f8b65ab 100644 --- a/src/audio_output/filters.c +++ b/src/audio_output/filters.c @@ -1,24 +1,24 @@ /***************************************************************************** * filters.c : audio output filters management ***************************************************************************** - * Copyright (C) 2002-2007 the VideoLAN team + * Copyright (C) 2002-2007 VLC authors and VideoLAN * $Id$ * * Authors: Christophe Massiot * - * This program 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 + * This program is free software; you can redistribute it and/or modify it + * under the terms of the GNU Lesser General Public License as published by + * the Free Software Foundation; either version 2.1 of the License, or * (at your option) any later version. * * This program 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. + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU Lesser General Public License for more details. * - * You should have received a copy of the GNU General Public License - * along with this program; if not, write to the Free Software - * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston MA 02110-1301, USA. + * You should have received a copy of the GNU Lesser General Public License + * along with this program; if not, write to the Free Software Foundation, + * Inc., 51 Franklin Street, Fifth Floor, Boston MA 02110-1301, USA. *****************************************************************************/ /***************************************************************************** @@ -28,362 +28,609 @@ # include "config.h" #endif -#include -#include - -#ifdef HAVE_ALLOCA_H -# include -#endif +#include +#include +#include +#include +#include #include -#include "aout_internal.h" +#include +#include /* for vout_Request */ +#include + #include +#include "aout_internal.h" -/***************************************************************************** - * FindFilter: find an audio filter for a specific transformation - *****************************************************************************/ -static aout_filter_t * FindFilter( aout_instance_t * p_aout, - const audio_sample_format_t * p_input_format, - const audio_sample_format_t * p_output_format ) +static filter_t *FindFilter (vlc_object_t *obj, const char *type, + const char *name, + const audio_sample_format_t *infmt, + const audio_sample_format_t *outfmt) { - static const char typename[] = "audio output"; - aout_filter_t * p_filter; + filter_t *filter = vlc_custom_create (obj, sizeof (*filter), type); + if (unlikely(filter == NULL)) + return NULL; - p_filter = vlc_custom_create( p_aout, sizeof(*p_filter), - VLC_OBJECT_GENERIC, typename ); + filter->fmt_in.audio = *infmt; + filter->fmt_in.i_codec = infmt->i_format; + filter->fmt_out.audio = *outfmt; + filter->fmt_out.i_codec = outfmt->i_format; + filter->p_module = module_need (filter, type, name, false); + if (filter->p_module == NULL) + { + vlc_object_release (filter); + filter = NULL; + } + else + assert (filter->pf_audio_filter != NULL); + return filter; +} + +static filter_t *FindConverter (vlc_object_t *obj, + const audio_sample_format_t *infmt, + const audio_sample_format_t *outfmt) +{ + return FindFilter (obj, "audio converter", NULL, infmt, outfmt); +} - if ( p_filter == NULL ) return NULL; - vlc_object_attach( p_filter, p_aout ); +static filter_t *FindResampler (vlc_object_t *obj, + const audio_sample_format_t *infmt, + const audio_sample_format_t *outfmt) +{ + return FindFilter (obj, "audio resampler", "$audio-resampler", + infmt, outfmt); +} - memcpy( &p_filter->input, p_input_format, sizeof(audio_sample_format_t) ); - memcpy( &p_filter->output, p_output_format, - sizeof(audio_sample_format_t) ); - p_filter->p_module = module_need( p_filter, "audio filter", NULL, false ); - if ( p_filter->p_module == NULL ) +/** + * Destroys a chain of audio filters. + */ +static void aout_FiltersPipelineDestroy(filter_t *const *filters, unsigned n) +{ + for( unsigned i = 0; i < n; i++ ) { - vlc_object_detach( p_filter ); + filter_t *p_filter = filters[i]; + + module_unneed( p_filter, p_filter->p_module ); vlc_object_release( p_filter ); - return NULL; } - - p_filter->b_continuity = false; - - return p_filter; } -/***************************************************************************** - * SplitConversion: split a conversion in two parts - ***************************************************************************** - * Returns the number of conversions required by the first part - 0 if only - * one conversion was asked. - * Beware : p_output_format can be modified during this function if the - * developer passed SplitConversion( toto, titi, titi, ... ). That is legal. - * SplitConversion( toto, titi, toto, ... ) isn't. - *****************************************************************************/ -static int SplitConversion( const audio_sample_format_t * p_input_format, - const audio_sample_format_t * p_output_format, - audio_sample_format_t * p_middle_format ) +static filter_t *TryFormat (vlc_object_t *obj, vlc_fourcc_t codec, + audio_sample_format_t *restrict fmt) { - bool b_format = - (p_input_format->i_format != p_output_format->i_format); - bool b_rate = (p_input_format->i_rate != p_output_format->i_rate); - bool b_channels = - (p_input_format->i_physical_channels - != p_output_format->i_physical_channels) - || (p_input_format->i_original_channels - != p_output_format->i_original_channels); - int i_nb_conversions = b_format + b_rate + b_channels; + audio_sample_format_t output = *fmt; - if ( i_nb_conversions <= 1 ) return 0; + assert (codec != fmt->i_format); + output.i_format = codec; + aout_FormatPrepare (&output); - memcpy( p_middle_format, p_output_format, sizeof(audio_sample_format_t) ); + filter_t *filter = FindConverter (obj, fmt, &output); + if (filter != NULL) + *fmt = output; + return filter; +} - if ( i_nb_conversions == 2 ) +/** + * Allocates audio format conversion filters + * @param obj parent VLC object for new filters + * @param filters table of filters [IN/OUT] + * @param count pointer to the number of filters in the table [IN/OUT] + * @param max size of filters table [IN] + * @param infmt input audio format + * @param outfmt output audio format + * @return 0 on success, -1 on failure + */ +static int aout_FiltersPipelineCreate(vlc_object_t *obj, filter_t **filters, + unsigned *count, unsigned max, + const audio_sample_format_t *restrict infmt, + const audio_sample_format_t *restrict outfmt) +{ + aout_FormatsPrint (obj, "conversion:", infmt, outfmt); + max -= *count; + filters += *count; + + /* There is a lot of second guessing on what the conversion plugins can + * and cannot do. This seems hardly avoidable, the conversion problem need + * to be reduced somehow. */ + audio_sample_format_t input = *infmt; + unsigned n = 0; + + /* Encapsulate or decode non-linear formats */ + if (!AOUT_FMT_LINEAR(infmt) && infmt->i_format != outfmt->i_format) { - if ( !b_format || !b_channels ) + if (n == max) + goto overflow; + + filter_t *f = TryFormat (obj, VLC_CODEC_S32N, &input); + if (f == NULL) + f = TryFormat (obj, VLC_CODEC_FL32, &input); + if (f == NULL) { - p_middle_format->i_rate = p_input_format->i_rate; - aout_FormatPrepare( p_middle_format ); - return 1; + msg_Err (obj, "cannot find %s for conversion pipeline", + "decoder"); + goto error; } - /* !b_rate */ - p_middle_format->i_physical_channels - = p_input_format->i_physical_channels; - p_middle_format->i_original_channels - = p_input_format->i_original_channels; - aout_FormatPrepare( p_middle_format ); - return 1; + filters[n++] = f; } + assert (AOUT_FMT_LINEAR(&input)); - /* i_nb_conversion == 3 */ - p_middle_format->i_rate = p_input_format->i_rate; - aout_FormatPrepare( p_middle_format ); - return 2; -} + /* Remix channels */ + if (infmt->i_physical_channels != outfmt->i_physical_channels + || infmt->i_original_channels != outfmt->i_original_channels) + { /* Remixing currently requires FL32... TODO: S16N */ + if (input.i_format != VLC_CODEC_FL32) + { + if (n == max) + goto overflow; -static void ReleaseFilter( aout_filter_t * p_filter ) -{ - module_unneed( p_filter, p_filter->p_module ); - vlc_object_detach( p_filter ); - vlc_object_release( p_filter ); -} + filter_t *f = TryFormat (obj, VLC_CODEC_FL32, &input); + if (f == NULL) + { + msg_Err (obj, "cannot find %s for conversion pipeline", + "pre-mix converter"); + goto error; + } -/***************************************************************************** - * aout_FiltersCreatePipeline: create a filters pipeline to transform a sample - * format to another - ***************************************************************************** - * pi_nb_filters must be initialized before calling this function - *****************************************************************************/ -int aout_FiltersCreatePipeline( aout_instance_t * p_aout, - aout_filter_t ** pp_filters_start, - int * pi_nb_filters, - const audio_sample_format_t * p_input_format, - const audio_sample_format_t * p_output_format ) -{ - aout_filter_t** pp_filters = pp_filters_start + *pi_nb_filters; - audio_sample_format_t temp_format; - int i_nb_conversions; + filters[n++] = f; + } - if ( AOUT_FMTS_IDENTICAL( p_input_format, p_output_format ) ) - { - msg_Dbg( p_aout, "no need for any filter" ); - return 0; + if (n == max) + goto overflow; + + audio_sample_format_t output; + output.i_format = input.i_format; + output.i_rate = input.i_rate; + output.i_physical_channels = outfmt->i_physical_channels; + output.i_original_channels = outfmt->i_original_channels; + aout_FormatPrepare (&output); + + filter_t *f = FindConverter (obj, &input, &output); + if (f == NULL) + { + msg_Err (obj, "cannot find %s for conversion pipeline", + "remixer"); + goto error; + } + + input = output; + filters[n++] = f; } - aout_FormatsPrint( p_aout, "filter(s)", p_input_format, p_output_format ); + /* Resample */ + if (input.i_rate != outfmt->i_rate) + { /* Resampling works with any linear format, but may be ugly. */ + if (n == max) + goto overflow; - if( *pi_nb_filters + 1 > AOUT_MAX_FILTERS ) - { - msg_Err( p_aout, "max filter reached (%d)", AOUT_MAX_FILTERS ); - intf_UserFatal( p_aout, false, _("Audio filtering failed"), - _("The maximum number of filters (%d) was reached."), - AOUT_MAX_FILTERS ); - return -1; + audio_sample_format_t output = input; + output.i_rate = outfmt->i_rate; + + filter_t *f = FindConverter (obj, &input, &output); + if (f == NULL) + { + msg_Err (obj, "cannot find %s for conversion pipeline", + "resampler"); + goto error; + } + + input = output; + filters[n++] = f; } - /* Try to find a filter to do the whole conversion. */ - pp_filters[0] = FindFilter( p_aout, p_input_format, p_output_format ); - if ( pp_filters[0] != NULL ) + /* Format */ + if (input.i_format != outfmt->i_format) { - msg_Dbg( p_aout, "found a filter for the whole conversion" ); - ++*pi_nb_filters; - return 0; + if (max == 0) + goto overflow; + + filter_t *f = TryFormat (obj, outfmt->i_format, &input); + if (f == NULL) + { + msg_Err (obj, "cannot find %s for conversion pipeline", + "post-mix converter"); + goto error; + } + filters[n++] = f; } - /* We'll have to split the conversion. We always do the downmixing - * before the resampling, because the audio decoder can probably do it - * for us. */ - i_nb_conversions = SplitConversion( p_input_format, - p_output_format, &temp_format ); - if ( !i_nb_conversions ) + msg_Dbg (obj, "conversion pipeline complete"); + *count += n; + return 0; + +overflow: + msg_Err (obj, "maximum of %u conversion filters reached", max); + dialog_Fatal (obj, _("Audio filtering failed"), + _("The maximum number of filters (%u) was reached."), max); +error: + aout_FiltersPipelineDestroy (filters, n); + return -1; +} + +#define aout_FiltersPipelineCreate(obj,f,n,m,i,o) \ + aout_FiltersPipelineCreate(VLC_OBJECT(obj),f,n,m,i,o) + +static inline bool ChangeFiltersString (vlc_object_t *aout, const char *var, + const char *filter, bool add) +{ + return aout_ChangeFilterString (aout, aout, var, filter, add); +} + +/** + * Filters an audio buffer through a chain of filters. + */ +static block_t *aout_FiltersPipelinePlay(filter_t *const *filters, + unsigned count, block_t *block) +{ + /* TODO: use filter chain */ + for (unsigned i = 0; (i < count) && (block != NULL); i++) { - /* There was only one conversion to do, and we already failed. */ - msg_Err( p_aout, "couldn't find a filter for the conversion" ); - return -1; + filter_t *filter = filters[i]; + + /* Please note that p_block->i_nb_samples & i_buffer + * shall be set by the filter plug-in. */ + block = filter->pf_audio_filter (filter, block); } + return block; +} - pp_filters[0] = FindFilter( p_aout, p_input_format, &temp_format ); - if ( pp_filters[0] == NULL && i_nb_conversions == 2 ) +/** Callback for visualization selection */ +static int VisualizationCallback (vlc_object_t *obj, char const *var, + vlc_value_t oldval, vlc_value_t newval, + void *data) +{ + audio_output_t *aout = (audio_output_t *)obj; + const char *mode = newval.psz_string; + + if (!*mode) { - /* Try with only one conversion. */ - SplitConversion( p_input_format, &temp_format, &temp_format ); - pp_filters[0] = FindFilter( p_aout, p_input_format, &temp_format ); + ChangeFiltersString (obj, "audio-visual", "goom", false); + ChangeFiltersString (obj, "audio-visual", "visual", false); + ChangeFiltersString (obj, "audio-visual", "projectm", false); + ChangeFiltersString (obj, "audio-visual", "vsxu", false); } - if ( pp_filters[0] == NULL ) + else if (!strcmp ("goom", mode)) { - msg_Err( p_aout, - "couldn't find a filter for the first part of the conversion" ); - return -1; + ChangeFiltersString (obj, "audio-visual", "visual", false ); + ChangeFiltersString (obj, "audio-visual", "goom", true ); + ChangeFiltersString (obj, "audio-visual", "projectm", false ); + ChangeFiltersString (obj, "audio-visual", "vsxu", false); } - - /* We have the first stage of the conversion. Find a filter for - * the rest. */ - if( *pi_nb_filters + 2 > AOUT_MAX_FILTERS ) + else if (!strcmp ("projectm", mode)) { - ReleaseFilter( pp_filters[0] ); - msg_Err( p_aout, "max filter reached (%d)", AOUT_MAX_FILTERS ); - intf_UserFatal( p_aout, false, _("Audio filtering failed"), - _("The maximum number of filters (%d) was reached."), - AOUT_MAX_FILTERS ); - return -1; + ChangeFiltersString (obj, "audio-visual", "visual", false); + ChangeFiltersString (obj, "audio-visual", "goom", false); + ChangeFiltersString (obj, "audio-visual", "projectm", true); + ChangeFiltersString (obj, "audio-visual", "vsxu", false); } - pp_filters[1] = FindFilter( p_aout, &pp_filters[0]->output, - p_output_format ); - if ( pp_filters[1] == NULL ) + else if (!strcmp ("vsxu", mode)) { - /* Try to split the conversion. */ - i_nb_conversions = SplitConversion( &pp_filters[0]->output, - p_output_format, &temp_format ); - if ( !i_nb_conversions ) - { - ReleaseFilter( pp_filters[0] ); - msg_Err( p_aout, - "couldn't find a filter for the second part of the conversion" ); - return -1; - } - if( *pi_nb_filters + 3 > AOUT_MAX_FILTERS ) - { - ReleaseFilter( pp_filters[0] ); - msg_Err( p_aout, "max filter reached (%d)", AOUT_MAX_FILTERS ); - intf_UserFatal( p_aout, false, _("Audio filtering failed"), - _("The maximum number of filters (%d) was reached."), - AOUT_MAX_FILTERS ); - return -1; - } - pp_filters[1] = FindFilter( p_aout, &pp_filters[0]->output, - &temp_format ); - pp_filters[2] = FindFilter( p_aout, &temp_format, - p_output_format ); - - if ( pp_filters[1] == NULL || pp_filters[2] == NULL ) - { - ReleaseFilter( pp_filters[0] ); - if ( pp_filters[1] != NULL ) - { - ReleaseFilter( pp_filters[1] ); - } - if ( pp_filters[2] != NULL ) - { - ReleaseFilter( pp_filters[2] ); - } - msg_Err( p_aout, - "couldn't find filters for the second part of the conversion" ); - return -1; - } - *pi_nb_filters += 3; - msg_Dbg( p_aout, "found 3 filters for the whole conversion" ); + ChangeFiltersString (obj, "audio-visual", "visual", false); + ChangeFiltersString (obj, "audio-visual", "goom", false); + ChangeFiltersString (obj, "audio-visual", "projectm", false); + ChangeFiltersString (obj, "audio-visual", "vsxu", true); } else { - *pi_nb_filters += 2; - msg_Dbg( p_aout, "found 2 filters for the whole conversion" ); + var_Create (obj, "effect-list", VLC_VAR_STRING); + var_SetString (obj, "effect-list", mode); + + ChangeFiltersString (obj, "audio-visual", "goom", false); + ChangeFiltersString (obj, "audio-visual", "visual", true); + ChangeFiltersString (obj, "audio-visual", "projectm", false); } - return 0; + aout_InputRequestRestart (aout); + (void) var; (void) oldval; (void) data; + return VLC_SUCCESS; } -/***************************************************************************** - * aout_FiltersDestroyPipeline: deallocate a filters pipeline - *****************************************************************************/ -void aout_FiltersDestroyPipeline( aout_instance_t * p_aout, - aout_filter_t ** pp_filters, - int i_nb_filters ) +static int EqualizerCallback (vlc_object_t *obj, char const *var, + vlc_value_t oldval, vlc_value_t newval, + void *data) { - int i; - (void)p_aout; + audio_output_t *aout = (audio_output_t *)obj; + char *mode = newval.psz_string; + bool ret; - for ( i = 0; i < i_nb_filters; i++ ) + if (!*mode) + ret = ChangeFiltersString (obj, "audio-filter", "equalizer", false); + else { - aout_filter_t *p_filter = pp_filters[i]; - - module_unneed( p_filter, p_filter->p_module ); - free( p_filter->p_owner ); - vlc_object_detach( p_filter ); - vlc_object_release( p_filter ); + var_Create (obj, "equalizer-preset", VLC_VAR_STRING); + var_SetString (obj, "equalizer-preset", mode); + ret = ChangeFiltersString (obj, "audio-filter", "equalizer", true); } + + /* That sucks */ + if (ret) + aout_InputRequestRestart (aout); + (void) var; (void) oldval; (void) data; + return VLC_SUCCESS; } -/***************************************************************************** - * aout_FiltersHintBuffers: fill in aout_alloc_t structures to optimize - * buffer allocations - *****************************************************************************/ -void aout_FiltersHintBuffers( aout_instance_t * p_aout, - aout_filter_t ** pp_filters, - int i_nb_filters, aout_alloc_t * p_first_alloc ) +static vout_thread_t *RequestVout (void *data, vout_thread_t *vout, + video_format_t *fmt, bool recycle) +{ + audio_output_t *aout = data; + vout_configuration_t cfg = { + .vout = vout, + .input = NULL, + .change_fmt = true, + .fmt = fmt, + .dpb_size = 1, + }; + + (void) recycle; + return vout_Request (aout, &cfg); +} + +vout_thread_t *aout_filter_RequestVout (filter_t *filter, vout_thread_t *vout, + video_format_t *fmt) +{ + /* NOTE: This only works from audio output. + * If you want to use visualization filters from another place, you will + * need to add a new pf_aout_request_vout callback or store a pointer + * to aout_request_vout_t inside filter_t (i.e. a level of indirection). */ + aout_owner_t *owner = aout_owner ((audio_output_t *)filter->p_parent); + aout_request_vout_t *req = &owner->request_vout; + + return req->pf_request_vout (req->p_private, vout, fmt, + owner->recycle_vout); +} + +static filter_t *CreateFilter (vlc_object_t *parent, const char *name, + const audio_sample_format_t *restrict infmt, + const audio_sample_format_t *restrict outfmt, + bool visu) { - int i; + filter_t *filter = vlc_custom_create (parent, sizeof (*filter), + "audio filter"); + if (unlikely(filter == NULL)) + return NULL; - (void)p_aout; /* unused */ + /*filter->p_owner = NOT NEEDED;*/ + filter->fmt_in.i_codec = infmt->i_format; + filter->fmt_in.audio = *infmt; + filter->fmt_out.i_codec = outfmt->i_format; + filter->fmt_out.audio = *outfmt; - for ( i = i_nb_filters - 1; i >= 0; i-- ) + if (!visu) { - aout_filter_t * p_filter = pp_filters[i]; + filter->p_module = module_need (filter, "audio filter", name, true); + if (filter->p_module != NULL) + return filter; - int i_output_size = p_filter->output.i_bytes_per_frame - * p_filter->output.i_rate * AOUT_MAX_INPUT_RATE - / p_filter->output.i_frame_length; - int i_input_size = p_filter->input.i_bytes_per_frame - * p_filter->input.i_rate * AOUT_MAX_INPUT_RATE - / p_filter->input.i_frame_length; + /* If probing failed, formats shall not have been modified. */ + assert (AOUT_FMTS_IDENTICAL(&filter->fmt_in.audio, infmt)); + assert (AOUT_FMTS_IDENTICAL(&filter->fmt_out.audio, outfmt)); + } - p_first_alloc->i_bytes_per_sec = __MAX( p_first_alloc->i_bytes_per_sec, - i_output_size ); + filter->p_module = module_need (filter, "visualization2", name, true); + if (filter->p_module != NULL) + return filter; - if ( p_filter->b_in_place ) - { - p_first_alloc->i_bytes_per_sec = __MAX( - p_first_alloc->i_bytes_per_sec, - i_input_size ); - p_filter->output_alloc.i_alloc_type = AOUT_ALLOC_NONE; - } - else + vlc_object_release (filter); + return NULL; +} + +/** + * Sets up the audio filters. + */ +int aout_FiltersNew (audio_output_t *aout, + const audio_sample_format_t *restrict infmt, + const audio_sample_format_t *restrict outfmt, + const aout_request_vout_t *request_vout) +{ + aout_owner_t *owner = aout_owner (aout); + + /* Prepare format structure */ + aout_FormatPrint (aout, "input", infmt); + audio_sample_format_t input_format = *infmt; + audio_sample_format_t output_format = *outfmt; + + /* Now add user filters */ + owner->nb_filters = 0; + owner->rate_filter = NULL; + owner->resampler = NULL; + + var_AddCallback (aout, "visual", VisualizationCallback, NULL); + var_AddCallback (aout, "equalizer", EqualizerCallback, NULL); + + if (!AOUT_FMT_LINEAR(outfmt)) + { /* Non-linear output: just convert formats, no filters/visu */ + if (!AOUT_FMTS_IDENTICAL(infmt, outfmt)) { - /* We're gonna need a buffer allocation. */ - memcpy( &p_filter->output_alloc, p_first_alloc, - sizeof(aout_alloc_t) ); - p_first_alloc->i_alloc_type = AOUT_ALLOC_STACK; - p_first_alloc->i_bytes_per_sec = i_input_size; + aout_FormatsPrint (aout, "pass-through:", infmt, outfmt); + owner->filters[0] = FindConverter(VLC_OBJECT(aout), infmt, outfmt); + if (owner->filters[0] == NULL) + { + msg_Err (aout, "cannot setup pass-through"); + goto error; + } + owner->nb_filters++; } + return 0; } -} -/***************************************************************************** - * aout_FiltersPlay: play a buffer - *****************************************************************************/ -void aout_FiltersPlay( aout_instance_t * p_aout, - aout_filter_t ** pp_filters, - int i_nb_filters, aout_buffer_t ** pp_input_buffer ) -{ - int i; + const char *scaletempo = + var_InheritBool (aout, "audio-time-stretch") ? "scaletempo" : NULL; + char *filters = var_InheritString (aout, "audio-filter"); + char *visual = var_InheritString (aout, "audio-visual"); - for( i = 0; i < i_nb_filters; i++ ) + if (request_vout != NULL) + owner->request_vout = *request_vout; + else { - aout_filter_t * p_filter = pp_filters[i]; - aout_buffer_t * p_output_buffer; - - /* Resamplers can produce slightly more samples than (i_in_nb * - * p_filter->output.i_rate / p_filter->input.i_rate) so we need - * slightly bigger buffers. */ - aout_BufferAlloc( &p_filter->output_alloc, - ((mtime_t)(*pp_input_buffer)->i_nb_samples + 2) - * 1000000 / p_filter->input.i_rate, - *pp_input_buffer, p_output_buffer ); - if( p_output_buffer == NULL ) - return; - - /* Please note that p_output_buffer->i_nb_samples & i_nb_bytes - * shall be set by the filter plug-in. */ - if( (*pp_input_buffer)->i_nb_samples > 0 ) + owner->request_vout.pf_request_vout = RequestVout; + owner->request_vout.p_private = aout; + } + owner->recycle_vout = (visual != NULL) && *visual; + + /* parse user filter lists */ + const char *list[AOUT_MAX_FILTERS]; + unsigned n = 0; + + if (scaletempo != NULL) + list[n++] = scaletempo; + if (filters != NULL) + { + char *p = filters, *name; + while ((name = strsep (&p, " :")) != NULL && n < AOUT_MAX_FILTERS) + list[n++] = name; + } + if (visual != NULL && n < AOUT_MAX_FILTERS) + list[n++] = visual; + + for (unsigned i = 0; i < n; i++) + { + const char *name = list[i]; + + if (owner->nb_filters >= AOUT_MAX_FILTERS) { - p_filter->pf_do_work( p_aout, p_filter, *pp_input_buffer, - p_output_buffer ); + msg_Err (aout, "maximum of %u filters reached", AOUT_MAX_FILTERS); + msg_Err (aout, "cannot add user filter %s (skipped)", name); + break; } - else + + filter_t *filter = CreateFilter (VLC_OBJECT(aout), name, + &input_format, &output_format, + name == visual); + if (filter == NULL) { - p_output_buffer->i_nb_bytes = 0; - p_output_buffer->i_nb_samples = 0; + msg_Dbg (aout, "cannot add user filter %s (skipped)", name); + continue; } - if( !p_filter->b_in_place ) + /* convert to the filter input format if necessary */ + if (aout_FiltersPipelineCreate (aout, owner->filters, + &owner->nb_filters, + AOUT_MAX_FILTERS - 1, + &input_format, &filter->fmt_in.audio)) { - aout_BufferFree( *pp_input_buffer ); - *pp_input_buffer = p_output_buffer; + msg_Err (aout, "cannot add user filter %s (skipped)", name); + module_unneed (filter, filter->p_module); + vlc_object_release (filter); + continue; } + + assert (owner->nb_filters < AOUT_MAX_FILTERS); + owner->filters[owner->nb_filters++] = filter; + input_format = filter->fmt_out.audio; + + if (name == scaletempo) + owner->rate_filter = filter; + } + free (visual); + free (filters); + + /* convert to the output format (minus resampling) if necessary */ + output_format.i_rate = input_format.i_rate; + if (aout_FiltersPipelineCreate (aout, owner->filters, &owner->nb_filters, + AOUT_MAX_FILTERS, + &input_format, &output_format)) + { + msg_Err (aout, "cannot setup filtering pipeline"); + goto error; + } + input_format = output_format; + + /* insert the resampler */ + output_format.i_rate = outfmt->i_rate; + assert (AOUT_FMTS_IDENTICAL(&output_format, outfmt)); + owner->resampler = FindResampler (VLC_OBJECT(aout), &input_format, + &output_format); + if (owner->resampler == NULL && input_format.i_rate != outfmt->i_rate) + { + msg_Err (aout, "cannot setup a resampler"); + goto error; } + if (owner->rate_filter == NULL) + owner->rate_filter = owner->resampler; + owner->resampling = 0; + + return 0; - assert( (*pp_input_buffer) == NULL || (*pp_input_buffer)->i_alloc_type != AOUT_ALLOC_STACK ); +error: + aout_FiltersPipelineDestroy (owner->filters, owner->nb_filters); + var_DelCallback (aout, "equalizer", EqualizerCallback, NULL); + var_DelCallback (aout, "visual", VisualizationCallback, NULL); + return -1; } -/***************************************************************************** - * aout_filter_RequestVout - *****************************************************************************/ -vout_thread_t *aout_filter_RequestVout( aout_filter_t *p_filter, - vout_thread_t *p_vout, video_format_t *p_fmt ) +/** + * Destroys the audio filters. + */ +void aout_FiltersDelete (audio_output_t *aout) { - if( !p_filter->request_vout.pf_request_vout ) - return NULL; - return p_filter->request_vout.pf_request_vout( p_filter->request_vout.p_private, - p_vout, p_fmt, true ); + aout_owner_t *owner = aout_owner (aout); + + if (owner->resampler != NULL) + aout_FiltersPipelineDestroy (&owner->resampler, 1); + aout_FiltersPipelineDestroy (owner->filters, owner->nb_filters); + var_DelCallback (aout, "equalizer", EqualizerCallback, NULL); + var_DelCallback (aout, "visual", VisualizationCallback, NULL); + + /* XXX We need to update recycle_vout before calling + * aout_FiltersPipelineDestroy(). + * FIXME There may be a race condition if audio-visual is updated between + * aout_FiltersDestroy() and the next aout_FiltersNew(). + */ + char *visual = var_InheritString (aout, "audio-visual"); + owner->recycle_vout = (visual != NULL) && *visual; + free (visual); +} + +bool aout_FiltersAdjustResampling (audio_output_t *aout, int adjust) +{ + aout_owner_t *owner = aout_owner (aout); + + if (owner->resampler == NULL) + return false; + + if (adjust) + owner->resampling += adjust; + else + owner->resampling = 0; + return owner->resampling != 0; } +block_t *aout_FiltersPlay (audio_output_t *aout, block_t *block, int rate) +{ + aout_owner_t *owner = aout_owner (aout); + int nominal_rate = 0; + + if (rate != INPUT_RATE_DEFAULT) + { + filter_t *rate_filter = owner->rate_filter; + + if (rate_filter == NULL) + goto drop; /* Without linear, non-nominal rate is impossible. */ + + /* Override input rate */ + nominal_rate = rate_filter->fmt_in.audio.i_rate; + rate_filter->fmt_in.audio.i_rate = + (nominal_rate * INPUT_RATE_DEFAULT) / rate; + } + + block = aout_FiltersPipelinePlay (owner->filters, owner->nb_filters, + block); + if (owner->resampler != NULL) + { /* NOTE: the resampler needs to run even if resampling is 0. + * The decoder and output rates can still be different. */ + owner->resampler->fmt_in.audio.i_rate += owner->resampling; + block = aout_FiltersPipelinePlay (&owner->resampler, 1, block); + owner->resampler->fmt_in.audio.i_rate -= owner->resampling; + } + + if (nominal_rate != 0) + { /* Restore input rate */ + assert (owner->rate_filter != NULL); + owner->rate_filter->fmt_in.audio.i_rate = nominal_rate; + } + return block; + +drop: + block_Release (block); + return NULL; +}