#include <stdlib.h>
#include <string.h>
#include <stdint.h>
#include <dpmi.h>
#include <sys/nearptr.h>
#include <pc.h>
#include <dos.h>

#include "vga.h"

static uint8_t buffer[320 * 200];
static uint8_t *screen = NULL;
static uint8_t *target = NULL;

uint8_t open_framebuffer()
{
    if (__djgpp_nearptr_enable() == 0)
        return 0;

    target = screen = (uint8_t *)(0xa0000 + __djgpp_conventional_base);
    return 1;
}

void blit_target(uint8_t t)
{
    target = t ? buffer : screen;
}

void close_framebuffer()
{
    __djgpp_nearptr_disable();
}

uint8_t set_mode(uint8_t mode)
{
    __dpmi_regs regs = { 0 };

    /* detect VGA card */
    regs.x.ax = 0x1a00;
    __dpmi_int(0x10, &regs);
    if (regs.h.al != 0x1a)
        return 0;

    memset(&regs, 0, sizeof(regs));
    regs.x.ax = mode;
    __dpmi_int(0x10, &regs);

    if (mode == 0x13)
    {
        /* setup the VGA: 320x200 @ 60Hz */
        /* from: https://gist.github.com/juj/34306e6da02a8a7043e393f01e013f24 */
        disable();

        /* XXX: probably not neeed */
        wait_vsync();

        outportw(0x3d4, 0x0011); /* Turn off write protect to CRTC registers */
        outportw(0x3d4, 0x0b06); /* New vertical total=525 lines, bits 0-7 */
        outportw(0x3d4, 0x3e07); /* New vertical total=525 lines, bits 8-9 */

        outportw(0x3d4, 0xb910); /* Vsync start=scanline 185 */
        outportw(0x3d4, 0x8f12); /* Vertical display end=scanline 399, bits 0-7 */
        outportw(0x3d4, 0xb815); /* Vertical blanking start=scanline 440, bits 0-7 */
        outportw(0x3d4, 0xe216); /* Adjust vblank end position */
        outportw(0x3d4, 0x8511); /* Vsync length=2 lines + turn write protect back on */

        /* XXX: probably not neeed */
        wait_vsync();

        outportw(0x3d4, 0x2813); /* 8 pixel chars */
        outportw(0x3d4, 0x8e11); /* restore retrace */
        enable();
    }

    return 1;
}

void wait_vsync()
{
    while (inportb(0x3da) & 8);
    while (!(inportb(0x3da) & 8));
}

void wait_frames(uint16_t frames)
{
    /* wait some time */
    for (uint16_t i = 0; i < frames; i++)
        wait_vsync();
}

void set_palette(const uint8_t *palette)
{
    outportb(0x3c8, 0);
    for (int i = 0; i < 768; i++)
        outportb(0x3c9, palette[i] >> 2);
}

void blit(const uint8_t *sprite, const Rect *dst)
{
    uint8_t b;
    uint8_t *dtarget = target + dst->x + dst->y * 320;
    const uint16_t inc = 320 - dst->w;

    for (uint16_t j = dst->h; j > 0; j--)
    {
        for (uint16_t i = dst->w; i > 0; i--)
        {
            b = *sprite++;

            /* transparent */
            if (b == TRANSPARENT)
            {
                dtarget++;
                continue;
            }

            *dtarget++ = b;
        }
        dtarget += inc;
    }
}

void blit_c(const uint8_t *sprite, const Rect *dst, uint8_t c)
{
    uint8_t *dtarget = target + dst->x + dst->y * 320;

    asm volatile (
        "movl $320, %%ebp\n\t"
        "subl %%ecx, %%ebp\n\t"
        "blit_c_w:\n\t"
        "movl %%ecx, %%eax\n\t"
        "blit_c_w_loop:\n\t"
        "movb (%%esi), %%ah\n\t"
        "cmpb $%p5, %%ah\n\t"
        "je blit_c_tr\n\t"
        "orb %%ah, %%ah\n\t"
        "je blit_c_store\n\t"
        "movb %%dl, %%ah\n\t"
        "blit_c_store:\n\t"
        "movb %%ah, (%%edi)\n\t"
        "blit_c_tr:\n\t"
        "incl %%esi\n\t"
        "incl %%edi\n\t"
        "decb %%al\n\t"
        "jne blit_c_w_loop\n\t"
        "addl %%ebp, %%edi\n\t"
        "decl %%ebx\n\t"
        "jne blit_c_w\n\t"
        : /*no output */
        : "S" (sprite), "D" (dtarget),
        "b" ((uint32_t)dst->h), "d" ((uint32_t)c),
        "c" ((uint32_t)dst->w), "i" (TRANSPARENT)
        : "eax", "ebp"
    );
}

void blitrc(const uint8_t *sprite, const Rect *src, const Rect *dst)
{
    uint8_t *dtarget = target + dst->x + dst->y * 320;

    sprite += src->x + src->y * src->w;

    asm volatile (
        "movl $320, %%ebp\n\t"
        "subl %%ecx, %%ebp\n\t"
        "blitrc_w:\n\t"
        "movl %%ecx, %%eax\n\t"
        "blitrc_w_loop:\n\t"
        "movb (%%esi), %%ah\n\t"
        "cmpb $%p5, %%ah\n\t"
        "je blitrc_tr\n\t"
        "movb %%ah, (%%edi)\n\t"
        "blitrc_tr:\n\t"
        "incl %%esi\n\t"
        "incl %%edi\n\t"
        "decb %%al\n\t"
        "jne blitrc_w_loop\n\t"
        "addl %%edx, %%esi\n\t"
        "addl %%ebp, %%edi\n\t"
        "decl %%ebx\n\t"
        "jne blitrc_w\n\t"
        : /*no output */
        : "S" (sprite), "D" (dtarget),
        "b" ((uint32_t)dst->h), "d" ((uint32_t)(src->w - dst->w)),
        "c" ((uint32_t)dst->w), "i" (TRANSPARENT)
        : "eax", "ebp"
    );
}

void blit_erase(uint8_t c)
{
    memset(target, c, 320 * 200);
}

void blit_copy_all()
{
    memcpy(screen, buffer, 320 * 200);
}

void blit_copy(const Rect *dst)
{
    uint8_t *src = buffer + dst->x + dst->y * 320;
    uint8_t *dtarget = screen + dst->x + dst->y * 320;

    for (uint16_t j = dst->h; j > 0; j--)
    {
        memcpy(dtarget, src, dst->w);
        src += 320;
        dtarget += 320;
    }
}

void blit_copy16(const Rect *dst)
{
    uint8_t *src = buffer + dst->x + dst->y * 320;
    uint8_t *dtarget = screen + dst->x + dst->y * 320;

    asm volatile (
        "movl $0x130, %%ebx\n\t"
        "cld\n\t"
        "blit_copy16_loop:\n\t"
        "movsl\n\t"
        "movsl\n\t"
        "movsl\n\t"
        "movsl\n\t"
        "addl %%ebx, %%esi\n\t"
        "addl %%ebx, %%edi\n\t"
        "decl %%eax\n\t"
        "orl %%eax, %%eax\n\t"
        "jne blit_copy16_loop\n\t"
        : /* no output */
        : "S" (src), "D" (dtarget), "a" ((uint32_t)dst->h)
        : "ecx", "ebx"
    );
}