/*
 * gfx_256.c — 256-colour (mode 0x81) public drawing API.
 *
 * Single-shot wrappers do one W3-begin / one W3-end around their raw
 * call.  Composites (rect, fill_rect, line) wrap a single begin/end
 * around many raw calls so the W3 dance is paid once per operation,
 * not once per pixel/byte.
 */

#include <gfx.h>
#include <stdint.h>

/* From gfx_raw_common.c — DI + map _gfx_bank into W3, save previous. */
extern void _gfx_w3_video_begin(void);
extern void _gfx_w3_video_end(void);

/* From gfx_raw_256.c — W3-naive primitives. */
extern void _gfx_putpixel256_raw(int x, int y, uint8_t color);
extern void _gfx_hline256_raw   (int x, int y, int len, uint8_t color);
extern void _gfx_vline256_raw   (int x, int y, int len, uint8_t color);
extern void _gfx_clear256_raw   (uint8_t color);

void gfx_clear256(uint8_t color)
{
    _gfx_w3_video_begin();
    _gfx_clear256_raw(color);
    _gfx_w3_video_end();
}

void gfx_putpixel256(int x, int y, uint8_t color)
{
    if ((unsigned)x >= GFX_WIDTH || (unsigned)y >= GFX_HEIGHT) return;
    _gfx_w3_video_begin();
    _gfx_putpixel256_raw(x, y, color);
    _gfx_w3_video_end();
}

void gfx_hline256(int x, int y, int len, uint8_t color)
{
    _gfx_w3_video_begin();
    _gfx_hline256_raw(x, y, len, color);
    _gfx_w3_video_end();
}

void gfx_vline256(int x, int y, int len, uint8_t color)
{
    _gfx_w3_video_begin();
    _gfx_vline256_raw(x, y, len, color);
    _gfx_w3_video_end();
}

void gfx_rect256(int x, int y, int w, int h, uint8_t color)
{
    if (w <= 0 || h <= 0) return;
    _gfx_w3_video_begin();
    _gfx_hline256_raw(x, y,         w, color);
    _gfx_hline256_raw(x, y + h - 1, w, color);
    if (h > 2) {
        _gfx_vline256_raw(x,         y + 1, h - 2, color);
        _gfx_vline256_raw(x + w - 1, y + 1, h - 2, color);
    }
    _gfx_w3_video_end();
}

void gfx_fill_rect256(int x, int y, int w, int h, uint8_t color)
{
    if (w <= 0 || h <= 0) return;

    /* Pick the orientation with fewer accelerator bursts.  Each burst
     * paints up to 256 contiguous bytes (horizontal) or up to 256
     * vertical pixels in one column.
     *   row-major (hlines): h × ceil(w/256) bursts
     *   col-major (vlines): w × ceil(h/256) bursts — for h ≤ 256 = w
     * Vertical wins for tall-narrow rects; horizontal for short-wide. */
    int h_bursts = h * ((w + 255) >> 8);
    int v_bursts = w * ((h + 255) >> 8);
    _gfx_w3_video_begin();
    if (h_bursts <= v_bursts) {
        for (int yy = 0; yy < h; yy++)
            _gfx_hline256_raw(x, y + yy, w, color);
    } else {
        for (int xx = 0; xx < w; xx++)
            _gfx_vline256_raw(x + xx, y, h, color);
    }
    _gfx_w3_video_end();
}

void gfx_line256(int x0, int y0, int x1, int y1, uint8_t color)
{
    /* Orthogonal lines route to the accelerator. */
    if (y0 == y1) {
        int x = x0 <= x1 ? x0 : x1;
        int w = (x0 <= x1 ? x1 - x0 : x0 - x1) + 1;
        gfx_hline256(x, y0, w, color);
        return;
    }
    if (x0 == x1) {
        int y = y0 <= y1 ? y0 : y1;
        int h = (y0 <= y1 ? y1 - y0 : y0 - y1) + 1;
        gfx_vline256(x0, y, h, color);
        return;
    }

    /* Bresenham — single W3-setup around the whole loop. */
    int dx = x1 - x0; int sx = dx < 0 ? -1 : 1; if (dx < 0) dx = -dx;
    int dy = y1 - y0; int sy = dy < 0 ? -1 : 1; if (dy < 0) dy = -dy;
    int err = (dx > dy ? dx : -dy) / 2;
    int x = x0, y = y0;
    _gfx_w3_video_begin();
    for (;;) {
        _gfx_putpixel256_raw(x, y, color);
        if (x == x1 && y == y1) break;
        int e2 = err;
        if (e2 > -dx) { err -= dy; x += sx; }
        if (e2 <  dy) { err += dx; y += sy; }
    }
    _gfx_w3_video_end();
}