const std = @import("std");

const util = @import("util.zig");
const threads = @import("threads.zig");

inline fn lum_idxs(i: usize, j: usize) struct { usize, usize, usize } {
    return .{
        i / 2,
        j / 2,
        2 * (i % 2) + (j % 2),
    };
}

fn read_lum(f: std.fs.File, source_buff: [][][4]util.Block, target_buff: [][][4]util.BlockQuantized, io_buff: []u8, queue: *util.JobQueue) !void {
    const block_h = source_buff.len;
    const block_w = source_buff[0].len;

    for (0..block_h * 2) |i| {
        _ = try f.read(io_buff);
        var io_idx: usize = 0;
        for (0..8) |I| {
            for (0..block_w * 2) |j| {
                const idxs = lum_idxs(i, j);
                @memcpy(source_buff[idxs.@"0"][idxs.@"1"][idxs.@"2"][I * 8 .. (I + 1) * 8], io_buff[io_idx .. io_idx + 8]);
                io_idx += 8;
                if (I == 7) {
                    try queue.prepend(util.Job{
                        .source = &source_buff[idxs.@"0"][idxs.@"1"][idxs.@"2"],
                        .target = &target_buff[idxs.@"0"][idxs.@"1"][idxs.@"2"],
                    });
                }
            }
        }
    }
}

fn read_chrom(f: std.fs.File, source_buff: [][]util.Block, target_buff: [][]util.BlockQuantized, io_buff: []u8, queue: *util.JobQueue) !void {
    const block_h = source_buff.len;
    const block_w = source_buff[0].len;

    for (0..block_h) |i| {
        _ = try f.read(io_buff);
        var io_idx: usize = 0;
        for (0..8) |I| {
            for (0..block_w) |j| {
                @memcpy(source_buff[i][j][I * 8 .. (I + 1) * 8], io_buff[io_idx .. io_idx + 8]);
                io_idx += 8;
                if (I == 7) {
                    try queue.prepend(util.Job{
                        .source = &source_buff[i][j],
                        .target = &target_buff[i][j],
                    });
                }
            }
        }
    }
}

pub fn main_loop(f: std.fs.File, buffs: util.Buffers, thread_mgr: threads.ThreadManager) !void {
    try read_lum(f, buffs.Y, buffs.Y_quant, buffs.input_buff, thread_mgr.queue_wrp.queue);
    try read_chrom(f, buffs.U, buffs.U_quant, buffs.input_buff[0 .. buffs.input_buff.len / 2], thread_mgr.queue_wrp.queue);
    try read_chrom(f, buffs.V, buffs.V_quant, buffs.input_buff[0 .. buffs.input_buff.len / 2], thread_mgr.queue_wrp.queue);
    while (thread_mgr.signals.processed.load(.Acquire) != buffs.num_blocks) {}
}