annotate libmpcodecs/vf_ass.c @ 35039:848835e1b053

vf_ass: add support for rendering on YUY2 and UYVY images. Patch by Xidorn Quan [quanxunzhen gmail com].

/*
 * Copyright (C) 2006 Evgeniy Stepanov <[e-mail]>
 *
 * 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 "config.h"

#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#include <inttypes.h>
#include <assert.h>

#include "config.h"
#include "mp_msg.h"
#include "help_mp.h"
#include "mpcommon.h"
#include "img_format.h"
#include "mp_image.h"
#include "vd.h"
#include "vf.h"

#include "libvo/fastmemcpy.h"
#include "libavutil/intreadwrite.h"
#include "sub/sub.h"
#include "m_option.h"
#include "m_struct.h"

#include "sub/ass_mp.h"
#include "sub/eosd.h"

#define _r(c)  ((c)>>24)
#define _g(c)  (((c)>>16)&0xFF)
#define _b(c)  (((c)>>8)&0xFF)
#define _a(c)  ((c)&0xFF)
#define rgba2y(c)  ( (( 263*_r(c) + 516*_g(c) + 100*_b(c)) >> 10) + 16  )
#define rgba2u(c)  ( ((-152*_r(c) - 298*_g(c) + 450*_b(c)) >> 10) + 128 )
#define rgba2v(c)  ( (( 450*_r(c) - 376*_g(c) -  73*_b(c)) >> 10) + 128 )

typedef void (*copy_from_image_func)(struct vf_instance *vf,
                                     int first_row, int last_row);
typedef void (*copy_to_image_func)(struct vf_instance *vf);

static const struct vf_priv_s {
    int outh, outw;

    int is_planar;
    unsigned int outfmt;

    // 1 = auto-added filter: insert only if chain does not support EOSD already
    // 0 = insert always
    int auto_insert;

    unsigned char *planes[3];
    unsigned char *dirty_rows;

    copy_from_image_func copy_from_image;
    copy_to_image_func copy_to_image;
} vf_priv_dflt;

static void copy_from_image_yuv420p(struct vf_instance *, int, int);
static void copy_to_image_yuv420p(struct vf_instance *);
static void copy_from_image_yuv422(struct vf_instance *, int, int);
static void copy_to_image_yuv422(struct vf_instance *);

static int config(struct vf_instance *vf,
                  int width, int height, int d_width, int d_height,
                  unsigned int flags, unsigned int outfmt)
{
    struct mp_eosd_settings res = {0};

    switch (outfmt) {
    case IMGFMT_YV12:
    case IMGFMT_I420:
    case IMGFMT_IYUV:
        vf->priv->is_planar = 1;
        vf->priv->copy_from_image = copy_from_image_yuv420p;
        vf->priv->copy_to_image = copy_to_image_yuv420p;
        break;
    case IMGFMT_UYVY:
    case IMGFMT_YUY2:
        vf->priv->is_planar = 0;
        vf->priv->copy_from_image = copy_from_image_yuv422;
        vf->priv->copy_to_image = copy_to_image_yuv422;
        break;
    default:
        return 0;
    }

    vf->priv->outfmt = outfmt;
    vf->priv->outh = height + ass_top_margin + ass_bottom_margin;
    vf->priv->outw = width;

    if (!opt_screen_size_x && !opt_screen_size_y) {
        d_width  = d_width  * vf->priv->outw / width;
        d_height = d_height * vf->priv->outh / height;
    }

    if (!vf->priv->is_planar)
        vf->priv->planes[0] = malloc(vf->priv->outw * vf->priv->outh);
    vf->priv->planes[1]  = malloc(vf->priv->outw * vf->priv->outh);
    vf->priv->planes[2]  = malloc(vf->priv->outw * vf->priv->outh);
    vf->priv->dirty_rows = malloc(vf->priv->outh);

    res.w    = vf->priv->outw;
    res.h    = vf->priv->outh;
    res.srcw = width;
    res.srch = height;
    res.mt   = ass_top_margin;
    res.mb   = ass_bottom_margin;
    eosd_configure(&res);

    return vf_next_config(vf, vf->priv->outw, vf->priv->outh, d_width,
                          d_height, flags, outfmt);
}

static void get_image(struct vf_instance *vf, mp_image_t *mpi)
{
    if (mpi->type == MP_IMGTYPE_IPB)
        return;
    if (mpi->flags & MP_IMGFLAG_PRESERVE)
        return;
    if (mpi->imgfmt != vf->priv->outfmt)
        return;                 // colorspace differ

    // width never changes, always try full DR
    mpi->priv = vf->dmpi = vf_get_image(vf->next, mpi->imgfmt, mpi->type,
                                        mpi->flags | MP_IMGFLAG_READABLE,
                                        FFMAX(mpi->width,  vf->priv->outw),
                                        FFMAX(mpi->height, vf->priv->outh));

    if ( (vf->dmpi->flags & MP_IMGFLAG_DRAW_CALLBACK) &&
        !(vf->dmpi->flags & MP_IMGFLAG_DIRECT)) {
        mp_msg(MSGT_ASS, MSGL_INFO, MSGTR_MPCODECS_FullDRNotPossible);
        return;
    }
    // set up mpi as a cropped-down image of dmpi:
    if (mpi->flags & MP_IMGFLAG_PLANAR) {
        mpi->planes[0] = vf->dmpi->planes[0] +  ass_top_margin                         * vf->dmpi->stride[0];
        mpi->planes[1] = vf->dmpi->planes[1] + (ass_top_margin >> mpi->chroma_y_shift) * vf->dmpi->stride[1];
        mpi->planes[2] = vf->dmpi->planes[2] + (ass_top_margin >> mpi->chroma_y_shift) * vf->dmpi->stride[2];
        mpi->stride[1] = vf->dmpi->stride[1];
        mpi->stride[2] = vf->dmpi->stride[2];
    } else {
        mpi->planes[0] = vf->dmpi->planes[0] +  ass_top_margin                         * vf->dmpi->stride[0];
    }
    mpi->stride[0] = vf->dmpi->stride[0];
    mpi->width     = vf->dmpi->width;
    mpi->flags    |=  MP_IMGFLAG_DIRECT;
    mpi->flags    &= ~MP_IMGFLAG_DRAW_CALLBACK;
//      vf->dmpi->flags &= ~MP_IMGFLAG_DRAW_CALLBACK;
}

static void blank(mp_image_t *mpi, int y1, int y2)
{
    int color[3] = { 16, 128, 128 };    // black (YUV)
    int y;
    unsigned char *dst;
    int chroma_rows = (y2 - y1) >> mpi->chroma_y_shift;

    if (mpi->flags & MP_IMGFLAG_PLANAR) {
    dst = mpi->planes[0] + y1 * mpi->stride[0];
    for (y = 0; y < y2 - y1; ++y) {
        memset(dst, color[0], mpi->w);
        dst += mpi->stride[0];
    }
    dst = mpi->planes[1] + (y1 >> mpi->chroma_y_shift) * mpi->stride[1];
    for (y = 0; y < chroma_rows; ++y) {
        memset(dst, color[1], mpi->chroma_width);
        dst += mpi->stride[1];
    }
    dst = mpi->planes[2] + (y1 >> mpi->chroma_y_shift) * mpi->stride[2];
    for (y = 0; y < chroma_rows; ++y) {
        memset(dst, color[2], mpi->chroma_width);
        dst += mpi->stride[2];
    }
    } else {
        unsigned char packed_color[4];
        int x;
        if (mpi->imgfmt == IMGFMT_UYVY) {
            packed_color[0] = color[1];
            packed_color[1] = color[0];
            packed_color[2] = color[2];
            packed_color[3] = color[0];
        } else {
            packed_color[0] = color[0];
            packed_color[1] = color[1];
            packed_color[2] = color[0];
            packed_color[3] = color[2];
        }
        dst = mpi->planes[0] + y1 * mpi->stride[0];
        for (y = y1; y < y2; ++y) {
            for (x = 0; x < mpi->w / 2; ++x)
                AV_COPY32(dst + 4 * x, packed_color);
            dst += mpi->stride[0];
        }
    }
}

static int prepare_image(struct vf_instance *vf, mp_image_t *mpi)
{
    if (mpi->flags & MP_IMGFLAG_DIRECT ||
	mpi->flags & MP_IMGFLAG_DRAW_CALLBACK) {
        vf->dmpi = mpi->priv;
        if (!vf->dmpi) {
            mp_msg(MSGT_ASS, MSGL_WARN, MSGTR_MPCODECS_FunWhydowegetNULL);
            return 0;
        }
        mpi->priv = NULL;
        // we've used DR, so we're ready...
        if (ass_top_margin)
            blank(vf->dmpi, 0, ass_top_margin);
        if (ass_bottom_margin)
            blank(vf->dmpi, vf->priv->outh - ass_bottom_margin, vf->priv->outh);
        if (!(mpi->flags & MP_IMGFLAG_PLANAR))
            vf->dmpi->planes[1] = mpi->planes[1];       // passthrough rgb8 palette
        return 0;
    }
    // hope we'll get DR buffer:
    vf->dmpi = vf_get_image(vf->next, vf->priv->outfmt, MP_IMGTYPE_TEMP,
                            MP_IMGFLAG_ACCEPT_STRIDE | MP_IMGFLAG_READABLE,
                            vf->priv->outw, vf->priv->outh);

    // copy mpi->dmpi...
    if (mpi->flags & MP_IMGFLAG_PLANAR) {
        memcpy_pic(vf->dmpi->planes[0] +  ass_top_margin * vf->dmpi->stride[0],
		   mpi->planes[0],
		   mpi->w,
		   mpi->h,
		   vf->dmpi->stride[0],
		   mpi->stride[0]);
        memcpy_pic(vf->dmpi->planes[1] + (ass_top_margin >> mpi->chroma_y_shift) * vf->dmpi->stride[1],
		   mpi->planes[1],
		   mpi->w >> mpi->chroma_x_shift,
                   mpi->h >> mpi->chroma_y_shift,
		   vf->dmpi->stride[1],
                   mpi->stride[1]);
        memcpy_pic(vf->dmpi->planes[2] + (ass_top_margin >> mpi->chroma_y_shift) * vf->dmpi->stride[2],
		   mpi->planes[2],
                   mpi->w >> mpi->chroma_x_shift,
                   mpi->h >> mpi->chroma_y_shift,
		   vf->dmpi->stride[2],
                   mpi->stride[2]);
    } else {
        memcpy_pic(vf->dmpi->planes[0] + ass_top_margin * vf->dmpi->stride[0],
		   mpi->planes[0],
                   mpi->w * (vf->dmpi->bpp / 8),
		   mpi->h,
                   vf->dmpi->stride[0],
		   mpi->stride[0]);
        vf->dmpi->planes[1] = mpi->planes[1];   // passthrough rgb8 palette
    }
    if (ass_top_margin)
        blank(vf->dmpi, 0, ass_top_margin);
    if (ass_bottom_margin)
        blank(vf->dmpi, vf->priv->outh - ass_bottom_margin, vf->priv->outh);
    return 0;
}

/**
 * \brief Copy specified rows from render_context.dmpi to render_context.planes, upsampling to 4:4:4
 */
static void copy_from_image_yuv420p(struct vf_instance *vf, int first_row,
                            int last_row)
{
    int pl;
    int i, j, k;
    unsigned char val;
    int chroma_rows;

    first_row  -= (first_row % 2);
    last_row   += (last_row  % 2);
    chroma_rows = (last_row - first_row) / 2;

    assert(first_row >= 0);
    assert(first_row <= last_row);
    assert(last_row  <= vf->priv->outh);

    for (pl = 1; pl < 3; ++pl) {
        int dst_stride = vf->priv->outw;
        int src_stride = vf->dmpi->stride[pl];

        unsigned char *src      = vf->dmpi->planes[pl] + (first_row / 2) * src_stride;
        unsigned char *dst      = vf->priv->planes[pl] +  first_row      * dst_stride;
        unsigned char *dst_next = dst + dst_stride;
        for (i = 0; i < chroma_rows; ++i) {
            if ((vf->priv->dirty_rows[first_row + i * 2    ] == 0) ||
                (vf->priv->dirty_rows[first_row + i * 2 + 1] == 0)) {
                for (j = 0, k = 0; j < vf->dmpi->chroma_width; ++j, k += 2) {
                    val = *(src + j);
                    *(dst + k    ) = val;
                    *(dst + k + 1) = val;
                    *(dst_next + k    ) = val;
                    *(dst_next + k + 1) = val;
                }
            }
            src += src_stride;
            dst      = dst_next + dst_stride;
            dst_next = dst      + dst_stride;
        }
    }
    for (i = first_row; i < last_row; ++i)
        vf->priv->dirty_rows[i] = 1;
}

/**
 * \brief Copy all previously copied rows back to render_context.dmpi
 */
static void copy_to_image_yuv420p(struct vf_instance *vf)
{
    int pl;
    int i, j, k;
    for (pl = 1; pl < 3; ++pl) {
        int dst_stride = vf->dmpi->stride[pl];
        int src_stride = vf->priv->outw;

        unsigned char *dst      = vf->dmpi->planes[pl];
        unsigned char *src      = vf->priv->planes[pl];
        unsigned char *src_next = vf->priv->planes[pl] + src_stride;
        for (i = 0; i < vf->priv->outh / 2; ++i) {
            if ((vf->priv->dirty_rows[i * 2] == 1)) {
                assert(vf->priv->dirty_rows[i * 2 + 1] == 1);
                for (j = 0, k = 0; j < vf->dmpi->chroma_width; ++j, k += 2) {
                    unsigned val = 0;
                    val += *(src + k);
                    val += *(src + k + 1);
                    val += *(src_next + k);
                    val += *(src_next + k + 1);
                    *(dst + j) = val >> 2;
                }
            }
            dst += dst_stride;
            src      = src_next + src_stride;
            src_next = src      + src_stride;
        }
    }
}

static void copy_from_image_yuv422(struct vf_instance *vf,
                                   int first_row, int last_row)
{
    unsigned char *dirty_rows = vf->priv->dirty_rows;
    int src_stride = vf->dmpi->stride[0];
    int dst_stride = vf->priv->outw;
    unsigned char *src = vf->dmpi->planes[0] + first_row * src_stride;
    unsigned char **dst = vf->priv->planes;
    int dst_off = first_row * dst_stride;
    int is_uyvy = vf->priv->outfmt == IMGFMT_UYVY;
    int i, j, k;

    for (i = first_row; i < last_row; ++i) {
        int next_off = dst_off + dst_stride;
        if (!dirty_rows[i]) {
            if (is_uyvy) {
                for (j = dst_off, k = 0; j < next_off; j += 2, k += 4) {
                    dst[0][j    ] = src[k + 1];
                    dst[0][j + 1] = src[k + 3];
                    dst[1][j    ] = src[k    ];
                    dst[1][j + 1] = src[k    ];
                    dst[2][j    ] = src[k + 2];
                    dst[2][j + 1] = src[k + 2];
                }
            } else {
                for (j = dst_off, k = 0; j < next_off; j += 2, k += 4) {
                    dst[0][j    ] = src[k    ];
                    dst[0][j + 1] = src[k + 2];
                    dst[1][j    ] = src[k + 1];
                    dst[1][j + 1] = src[k + 1];
                    dst[2][j    ] = src[k + 3];
                    dst[2][j + 1] = src[k + 3];
                }
            }
        }
        src += src_stride;
        dst_off = next_off;
    }
    for (i = first_row; i < last_row; ++i)
        dirty_rows[i] = 1;
}

static void copy_to_image_yuv422(struct vf_instance *vf)
{
    unsigned char *dirty_rows = vf->priv->dirty_rows;
    int src_stride = vf->priv->outw;
    int dst_stride = vf->dmpi->stride[0];
    int height = vf->priv->outh;
    unsigned char **src = vf->priv->planes;
    unsigned char *dst = vf->dmpi->planes[0];
    int src_off = 0;
    int is_uyvy = vf->priv->outfmt == IMGFMT_UYVY;
    int i, j, k;

    for (i = 0; i < height; ++i) {
        int next_off = src_off + src_stride;
        if (*dirty_rows++) {
#define AVERAGE(a, b) (((unsigned)(a) + (unsigned)(b)) >> 1)
            if (is_uyvy) {
                for (j = src_off, k = 0; j < next_off; j += 2, k += 4) {
                    dst[k    ] = AVERAGE(src[1][j], src[1][j + 1]);
                    dst[k + 1] = src[0][j];
                    dst[k + 2] = AVERAGE(src[2][j], src[2][j + 1]);
                    dst[k + 3] = src[0][j + 1];
                }
            } else {
                for (j = src_off, k = 0; j < next_off; j += 2, k += 4) {
                    dst[k    ] = src[0][j];
                    dst[k + 1] = AVERAGE(src[1][j], src[1][j + 1]);
                    dst[k + 2] = src[0][j + 1];
                    dst[k + 3] = AVERAGE(src[2][j], src[2][j + 1]);
                }
            }
#undef AVERAGE
        }
        src_off = next_off;
        dst += dst_stride;
    }
}

static void my_draw_bitmap(struct vf_instance *vf, unsigned char *bitmap,
                           int bitmap_w, int bitmap_h, int stride,
                           int dst_x, int dst_y, unsigned color)
{
    unsigned char y = rgba2y(color);
    unsigned char u = rgba2u(color);
    unsigned char v = rgba2v(color);
    unsigned opacity = 255 - _a(color);
    unsigned char *src, *dsty, *dstu, *dstv;
    int i, j;
    mp_image_t *dmpi = vf->dmpi;
    int stride_y = vf->priv->is_planar ? dmpi->stride[0] : vf->priv->outw;

    opacity = (0x10203 * opacity + 0x80) >> 8; /* 0x10203 = (1<<32)/(255*255) */
    /* 0 <= opacity <= 0x10101 */
    src = bitmap;
    dsty = vf->priv->is_planar ? dmpi->planes[0] : vf->priv->planes[0];
    dsty += dst_x + dst_y * stride_y;
    dstu = vf->priv->planes[1] + dst_x + dst_y * vf->priv->outw;
    dstv = vf->priv->planes[2] + dst_x + dst_y * vf->priv->outw;
    for (i = 0; i < bitmap_h; ++i) {
        for (j = 0; j < bitmap_w; ++j) {
            unsigned k = src[j];
            if (!k)
                continue;
            k *= opacity; /* 0 <= k <= 0xFFFFFF */
            dsty[j] = (k * y + (0xFFFFFF - k) * dsty[j] + 0x800000) >> 24;
            dstu[j] = (k * u + (0xFFFFFF - k) * dstu[j] + 0x800000) >> 24;
            dstv[j] = (k * v + (0xFFFFFF - k) * dstv[j] + 0x800000) >> 24;
        }
        src  += stride;
        dsty += stride_y;
        dstu += vf->priv->outw;
        dstv += vf->priv->outw;
    }
}

static void render_frame(struct vf_instance *vf, mp_image_t *mpi,
                         struct mp_eosd_image_list *images)
{
    struct mp_eosd_image *img;
    copy_from_image_func copy_from_image = vf->priv->copy_from_image;
    copy_to_image_func copy_to_image = vf->priv->copy_to_image;

    img = eosd_image_first(images);
    if (!img)
        return;
        memset(vf->priv->dirty_rows, 0, vf->priv->outh);        // reset dirty rows
        while (img) {
            copy_from_image(vf, img->dst_y, img->dst_y + img->h);
            my_draw_bitmap(vf, img->bitmap, img->w, img->h, img->stride,
                           img->dst_x, img->dst_y, img->color);
            img = eosd_image_next(images);
        }
        copy_to_image(vf);
}

static int put_image(struct vf_instance *vf, mp_image_t *mpi, double pts)
{
    struct mp_eosd_image_list images;
    eosd_render_frame(pts, &images);
    prepare_image(vf, mpi);
    render_frame(vf, mpi, &images);
    return vf_next_put_image(vf, vf->dmpi, pts);
}

static int query_format(struct vf_instance *vf, unsigned int fmt)
{
    switch (fmt) {
    case IMGFMT_YV12:
    case IMGFMT_I420:
    case IMGFMT_IYUV:
    case IMGFMT_UYVY:
    case IMGFMT_YUY2:
        return vf_next_query_format(vf, fmt) | VFCAP_EOSD;
    }
    return 0;
}

static int control(vf_instance_t *vf, int request, void *data)
{
    switch (request) {
    case VFCTRL_INIT_EOSD:
        return CONTROL_TRUE;
    case VFCTRL_DRAW_EOSD:
        return CONTROL_TRUE;
    }
    return vf_next_control(vf, request, data);
}

static void uninit(struct vf_instance *vf)
{
    if (!vf->priv->is_planar)
        free(vf->priv->planes[0]);
    free(vf->priv->planes[1]);
    free(vf->priv->planes[2]);
    free(vf->priv->dirty_rows);
}

static const unsigned int fmt_list[] = {
    IMGFMT_YV12,
    IMGFMT_I420,
    IMGFMT_IYUV,
    IMGFMT_UYVY,
    IMGFMT_YUY2,
    0
};

static int vf_open(vf_instance_t *vf, char *args)
{
    int flags;
    unsigned outfmt = vf_match_csp(&vf->next, fmt_list, IMGFMT_YV12);
    if (outfmt)
        flags = vf_next_query_format(vf, outfmt);
    if (!outfmt || (vf->priv->auto_insert && flags & VFCAP_EOSD)) {
        uninit(vf);
        return 0;
    }

    if (vf->priv->auto_insert)
        mp_msg(MSGT_ASS, MSGL_INFO, "[ass] auto-open\n");

    vf->config       = config;
    vf->query_format = query_format;
    vf->uninit       = uninit;
    vf->control      = control;
    vf->get_image    = get_image;
    vf->put_image    = put_image;
    vf->default_caps = VFCAP_EOSD;
    return 1;
}

#define ST_OFF(f) M_ST_OFF(struct vf_priv_s,f)
static const m_option_t vf_opts_fields[] = {
    {"auto", ST_OFF(auto_insert), CONF_TYPE_FLAG, 0, 0, 1, NULL},
    {NULL, NULL, 0, 0, 0, 0, NULL}
};

static const m_struct_t vf_opts = {
    "ass",
    sizeof(struct vf_priv_s),
    &vf_priv_dflt,
    vf_opts_fields
};

const vf_info_t vf_info_ass = {
    "Render ASS/SSA subtitles",
    "ass",
    "Evgeniy Stepanov",
    "",
    vf_open,
    &vf_opts
};