/*
 * Copyright (c) 2016, Alliance for Open Media. All rights reserved
 *
 * This source code is subject to the terms of the BSD 2 Clause License and
 * the Alliance for Open Media Patent License 1.0. If the BSD 2 Clause License
 * was not distributed with this source code in the LICENSE file, you can
 * obtain it at www.aomedia.org/license/software. If the Alliance for Open
 * Media Patent License 1.0 was not distributed with this source code in the
 * PATENTS file, you can obtain it at www.aomedia.org/license/patent.
 */

#include <assert.h>

#include "./aom_scale_rtcd.h"
#include "./aom_config.h"

#include "aom/aom_integer.h"

#include "av1/common/blockd.h"
#include "av1/common/reconinter.h"
#include "av1/common/reconintra.h"

#if CONFIG_AOM_HIGHBITDEPTH
void av1_highbd_build_inter_predictor(
    const uint8_t *src, int src_stride, uint8_t *dst, int dst_stride,
    const MV *src_mv, const struct scale_factors *sf, int w, int h, int ref,
    const InterpKernel *kernel, enum mv_precision precision, int x, int y,
    int bd) {
  const int is_q4 = precision == MV_PRECISION_Q4;
  const MV mv_q4 = { is_q4 ? src_mv->row : src_mv->row * 2,
                     is_q4 ? src_mv->col : src_mv->col * 2 };
  MV32 mv = av1_scale_mv(&mv_q4, x, y, sf);
  const int subpel_x = mv.col & SUBPEL_MASK;
  const int subpel_y = mv.row & SUBPEL_MASK;

  src += (mv.row >> SUBPEL_BITS) * src_stride + (mv.col >> SUBPEL_BITS);

  high_inter_predictor(src, src_stride, dst, dst_stride, subpel_x, subpel_y, sf,
                       w, h, ref, kernel, sf->x_step_q4, sf->y_step_q4, bd);
}
#endif  // CONFIG_AOM_HIGHBITDEPTH

void av1_build_inter_predictor(const uint8_t *src, int src_stride,
                                uint8_t *dst, int dst_stride, const MV *src_mv,
                                const struct scale_factors *sf, int w, int h,
                                int ref, const InterpKernel *kernel,
                                enum mv_precision precision, int x, int y) {
  const int is_q4 = precision == MV_PRECISION_Q4;
  const MV mv_q4 = { is_q4 ? src_mv->row : src_mv->row * 2,
                     is_q4 ? src_mv->col : src_mv->col * 2 };
  MV32 mv = av1_scale_mv(&mv_q4, x, y, sf);
  const int subpel_x = mv.col & SUBPEL_MASK;
  const int subpel_y = mv.row & SUBPEL_MASK;

  src += (mv.row >> SUBPEL_BITS) * src_stride + (mv.col >> SUBPEL_BITS);

  inter_predictor(src, src_stride, dst, dst_stride, subpel_x, subpel_y, sf, w,
                  h, ref, kernel, sf->x_step_q4, sf->y_step_q4);
}

void build_inter_predictors(MACROBLOCKD *xd, int plane, int block, int bw,
                            int bh, int x, int y, int w, int h, int mi_x,
                            int mi_y) {
  struct macroblockd_plane *const pd = &xd->plane[plane];
  const MODE_INFO *mi = xd->mi[0];
  const int is_compound = has_second_ref(&mi->mbmi);
  const InterpKernel *kernel = av1_filter_kernels[mi->mbmi.interp_filter];
  int ref;

  for (ref = 0; ref < 1 + is_compound; ++ref) {
    const struct scale_factors *const sf = &xd->block_refs[ref]->sf;
    struct buf_2d *const pre_buf = &pd->pre[ref];
    struct buf_2d *const dst_buf = &pd->dst;
    uint8_t *const dst = dst_buf->buf + dst_buf->stride * y + x;
    const MV mv = mi->mbmi.sb_type < BLOCK_8X8
                      ? average_split_mvs(pd, mi, ref, block)
                      : mi->mbmi.mv[ref].as_mv;

    // TODO(jkoleszar): This clamping is done in the incorrect place for the
    // scaling case. It needs to be done on the scaled MV, not the pre-scaling
    // MV. Note however that it performs the subsampling aware scaling so
    // that the result is always q4.
    // mv_precision precision is MV_PRECISION_Q4.
    const MV mv_q4 = clamp_mv_to_umv_border_sb(
        xd, &mv, bw, bh, pd->subsampling_x, pd->subsampling_y);

    uint8_t *pre;
    MV32 scaled_mv;
    int xs, ys, subpel_x, subpel_y;
    const int is_scaled = av1_is_scaled(sf);

    if (is_scaled) {
      pre = pre_buf->buf + scaled_buffer_offset(x, y, pre_buf->stride, sf);
      scaled_mv = av1_scale_mv(&mv_q4, mi_x + x, mi_y + y, sf);
      xs = sf->x_step_q4;
      ys = sf->y_step_q4;
    } else {
      pre = pre_buf->buf + (y * pre_buf->stride + x);
      scaled_mv.row = mv_q4.row;
      scaled_mv.col = mv_q4.col;
      xs = ys = 16;
    }
    subpel_x = scaled_mv.col & SUBPEL_MASK;
    subpel_y = scaled_mv.row & SUBPEL_MASK;
    pre += (scaled_mv.row >> SUBPEL_BITS) * pre_buf->stride +
           (scaled_mv.col >> SUBPEL_BITS);

#if CONFIG_AOM_HIGHBITDEPTH
    if (xd->cur_buf->flags & YV12_FLAG_HIGHBITDEPTH) {
      high_inter_predictor(pre, pre_buf->stride, dst, dst_buf->stride, subpel_x,
                           subpel_y, sf, w, h, ref, kernel, xs, ys, xd->bd);
    } else {
      inter_predictor(pre, pre_buf->stride, dst, dst_buf->stride, subpel_x,
                      subpel_y, sf, w, h, ref, kernel, xs, ys);
    }
#else
    inter_predictor(pre, pre_buf->stride, dst, dst_buf->stride, subpel_x,
                    subpel_y, sf, w, h, ref, kernel, xs, ys);
#endif  // CONFIG_AOM_HIGHBITDEPTH
  }
}

void av1_build_inter_predictor_sub8x8(MACROBLOCKD *xd, int plane, int i,
                                       int ir, int ic, int mi_row, int mi_col) {
  struct macroblockd_plane *const pd = &xd->plane[plane];
  MODE_INFO *const mi = xd->mi[0];
  const BLOCK_SIZE plane_bsize = get_plane_block_size(mi->mbmi.sb_type, pd);
  const int width = 4 * num_4x4_blocks_wide_lookup[plane_bsize];
  const int height = 4 * num_4x4_blocks_high_lookup[plane_bsize];

  uint8_t *const dst = &pd->dst.buf[(ir * pd->dst.stride + ic) << 2];
  int ref;
  const int is_compound = has_second_ref(&mi->mbmi);
  const InterpKernel *kernel = av1_filter_kernels[mi->mbmi.interp_filter];

  for (ref = 0; ref < 1 + is_compound; ++ref) {
    const uint8_t *pre =
        &pd->pre[ref].buf[(ir * pd->pre[ref].stride + ic) << 2];
#if CONFIG_AOM_HIGHBITDEPTH
    if (xd->cur_buf->flags & YV12_FLAG_HIGHBITDEPTH) {
      av1_highbd_build_inter_predictor(
          pre, pd->pre[ref].stride, dst, pd->dst.stride,
          &mi->bmi[i].as_mv[ref].as_mv, &xd->block_refs[ref]->sf, width, height,
          ref, kernel, MV_PRECISION_Q3, mi_col * MI_SIZE + 4 * ic,
          mi_row * MI_SIZE + 4 * ir, xd->bd);
    } else {
      av1_build_inter_predictor(
          pre, pd->pre[ref].stride, dst, pd->dst.stride,
          &mi->bmi[i].as_mv[ref].as_mv, &xd->block_refs[ref]->sf, width, height,
          ref, kernel, MV_PRECISION_Q3, mi_col * MI_SIZE + 4 * ic,
          mi_row * MI_SIZE + 4 * ir);
    }
#else
    av1_build_inter_predictor(
        pre, pd->pre[ref].stride, dst, pd->dst.stride,
        &mi->bmi[i].as_mv[ref].as_mv, &xd->block_refs[ref]->sf, width, height,
        ref, kernel, MV_PRECISION_Q3, mi_col * MI_SIZE + 4 * ic,
        mi_row * MI_SIZE + 4 * ir);
#endif  // CONFIG_AOM_HIGHBITDEPTH
  }
}

static void build_inter_predictors_for_planes(MACROBLOCKD *xd, BLOCK_SIZE bsize,
                                              int mi_row, int mi_col,
                                              int plane_from, int plane_to) {
  int plane;
  const int mi_x = mi_col * MI_SIZE;
  const int mi_y = mi_row * MI_SIZE;
  for (plane = plane_from; plane <= plane_to; ++plane) {
    const struct macroblockd_plane *pd = &xd->plane[plane];
    const int bw = 4 * num_4x4_blocks_wide_lookup[bsize] >> pd->subsampling_x;
    const int bh = 4 * num_4x4_blocks_high_lookup[bsize] >> pd->subsampling_y;

    if (xd->mi[0]->mbmi.sb_type < BLOCK_8X8) {
      const PARTITION_TYPE bp = bsize - xd->mi[0]->mbmi.sb_type;
      const int have_vsplit = bp != PARTITION_HORZ;
      const int have_hsplit = bp != PARTITION_VERT;
      const int num_4x4_w = 2 >> ((!have_vsplit) | pd->subsampling_x);
      const int num_4x4_h = 2 >> ((!have_hsplit) | pd->subsampling_y);
      const int pw = 8 >> (have_vsplit | pd->subsampling_x);
      const int ph = 8 >> (have_hsplit | pd->subsampling_y);
      int x, y;
      assert(bp != PARTITION_NONE && bp < PARTITION_TYPES);
      assert(bsize == BLOCK_8X8);
      assert(pw * num_4x4_w == bw && ph * num_4x4_h == bh);
      for (y = 0; y < num_4x4_h; ++y)
        for (x = 0; x < num_4x4_w; ++x)
          build_inter_predictors(xd, plane, y * 2 + x, bw, bh, 4 * x, 4 * y, pw,
                                 ph, mi_x, mi_y);
    } else {
      build_inter_predictors(xd, plane, 0, bw, bh, 0, 0, bw, bh, mi_x, mi_y);
    }
  }
}

void av1_build_inter_predictors_sby(MACROBLOCKD *xd, int mi_row, int mi_col,
                                     BLOCK_SIZE bsize) {
  build_inter_predictors_for_planes(xd, bsize, mi_row, mi_col, 0, 0);
}

void av1_build_inter_predictors_sbp(MACROBLOCKD *xd, int mi_row, int mi_col,
                                     BLOCK_SIZE bsize, int plane) {
  build_inter_predictors_for_planes(xd, bsize, mi_row, mi_col, plane, plane);
}

void av1_build_inter_predictors_sbuv(MACROBLOCKD *xd, int mi_row, int mi_col,
                                      BLOCK_SIZE bsize) {
  build_inter_predictors_for_planes(xd, bsize, mi_row, mi_col, 1,
                                    MAX_MB_PLANE - 1);
}

void av1_build_inter_predictors_sb(MACROBLOCKD *xd, int mi_row, int mi_col,
                                    BLOCK_SIZE bsize) {
  build_inter_predictors_for_planes(xd, bsize, mi_row, mi_col, 0,
                                    MAX_MB_PLANE - 1);
}

void av1_setup_dst_planes(struct macroblockd_plane planes[MAX_MB_PLANE],
                           const YV12_BUFFER_CONFIG *src, int mi_row,
                           int mi_col) {
  uint8_t *const buffers[MAX_MB_PLANE] = { src->y_buffer, src->u_buffer,
                                           src->v_buffer };
  const int strides[MAX_MB_PLANE] = { src->y_stride, src->uv_stride,
                                      src->uv_stride };
  int i;

  for (i = 0; i < MAX_MB_PLANE; ++i) {
    struct macroblockd_plane *const pd = &planes[i];
    setup_pred_plane(&pd->dst, buffers[i], strides[i], mi_row, mi_col, NULL,
                     pd->subsampling_x, pd->subsampling_y);
  }
}

void av1_setup_pre_planes(MACROBLOCKD *xd, int idx,
                           const YV12_BUFFER_CONFIG *src, int mi_row,
                           int mi_col, const struct scale_factors *sf) {
  if (src != NULL) {
    int i;
    uint8_t *const buffers[MAX_MB_PLANE] = { src->y_buffer, src->u_buffer,
                                             src->v_buffer };
    const int strides[MAX_MB_PLANE] = { src->y_stride, src->uv_stride,
                                        src->uv_stride };
    for (i = 0; i < MAX_MB_PLANE; ++i) {
      struct macroblockd_plane *const pd = &xd->plane[i];
      setup_pred_plane(&pd->pre[idx], buffers[i], strides[i], mi_row, mi_col,
                       sf, pd->subsampling_x, pd->subsampling_y);
    }
  }
}