`timescale 1ns / 1ps
module Windows #(
    parameter DATA_WIDTH = 16,
    parameter IMAGE_WIDTH = 1936,
    parameter IMAGE_HEIGHT = 1088,
    parameter WINDOWS_WIDTH = 3,
    parameter WINDOWS_ANCHOR_X = 1,  //禁止大于WINDOWS_WIDTH-1
    parameter WINDOWS_ANCHOR_Y = 1  //禁止大于WINDOWS_WIDTH-1
) (
    // 基本信号
    input wire clk,
    input wire reset,
    //  数据线
    input wire [DATA_WIDTH - 1:0] in_data,
    output reg [DATA_WIDTH - 1:0] out_data[WINDOWS_WIDTH*WINDOWS_WIDTH],  // 数据输出线
    //  有效信号
    input wire in_valid,  // 上一模块输出数据有效
    output reg out_valid,  // 当前模块输出数据有效
    //  准备信号 Windows模块无法停止，因此默认不处理准备信号
    input wire in_ready,
    output wire out_ready
);

  assign out_ready = 1'b1;

  reg [DATA_WIDTH - 1:0] regx_in_data [WINDOWS_WIDTH-1];
  reg [DATA_WIDTH - 1:0] regx_out_data[WINDOWS_WIDTH-1];
  reg [WINDOWS_WIDTH - 2:0] regx_in_valid, regx_out_valid;

  reg [DATA_WIDTH - 1:0] data_out_shift[WINDOWS_WIDTH-1][2*(WINDOWS_WIDTH-1)];


  /*              outdata[x]:
SHIFT_REG1  ->  0 3 6 . .
SHIFT_REG0  ->  1 4 7 . .
   in_data  ->  2 5 8 . .
                . . .
                . . .
*/

  reg [15:0] pos_x, pos_y;
  always @(posedge clk) begin
    if (reset) begin
      pos_x <= 0;
      pos_y <= 0;
    end else if (regx_out_valid[WINDOWS_WIDTH-2]) begin
      pos_x <= (pos_x >= IMAGE_WIDTH - 1) ? (0) : (pos_x + 1);
      pos_y <= (pos_x >= IMAGE_WIDTH - 1)?((pos_y >= IMAGE_HEIGHT - 1)?(0):(pos_y + 1)):(pos_y);
    end else begin
      pos_x <= pos_x;
      pos_y <= pos_y;
    end
  end

  integer i, j;
  always @(posedge clk) begin
    if (reset) begin
      for (i = 0; i < WINDOWS_WIDTH * WINDOWS_WIDTH; i = i + 1) out_data[i] <= 0;
      out_valid <= 0;
    end else if (regx_out_valid[WINDOWS_WIDTH-2]) begin
      for (i = 0; i < WINDOWS_WIDTH; i = i + 1) begin
        for (j = 0; j < WINDOWS_WIDTH; j = j + 1) begin
          if (i == WINDOWS_WIDTH - 1) begin
            if (j == 0) out_data[(WINDOWS_WIDTH*i)+j] <= regx_out_data[WINDOWS_WIDTH-2];
            else out_data[(WINDOWS_WIDTH*i)+j] <= data_out_shift[j-1][2*i-1];
          end else out_data[(WINDOWS_WIDTH*i)+j] <= out_data[(WINDOWS_WIDTH*(i+1))+j];
        end
      end
      out_valid <= ~((pos_y <= WINDOWS_WIDTH-WINDOWS_ANCHOR_Y-1 && pos_x < WINDOWS_WIDTH-WINDOWS_ANCHOR_X-1) || (pos_y < WINDOWS_WIDTH-WINDOWS_ANCHOR_Y-1));
    end else begin
      for (i = 0; i < WINDOWS_WIDTH * WINDOWS_WIDTH - 1; i = i + 1) out_data[i] <= out_data[i];
      out_valid <= 0;
    end
  end

  always @(posedge clk) begin
    if (reset)
      for (i = 0; i < WINDOWS_WIDTH - 1; i = i + 1)
      for (j = 0; j < WINDOWS_WIDTH - 1; j = j + 1) data_out_shift[i][j] <= 0;
    else
      for (i = 0; i < WINDOWS_WIDTH - 1; i = i + 1) begin
        for (j = 0; j < 2 * (WINDOWS_WIDTH - 1); j = j + 1) begin
          if (i == WINDOWS_WIDTH - 2 && j == 0) data_out_shift[i][j] <= in_data;
          else if (j == 0) data_out_shift[i][j] <= regx_out_data[(WINDOWS_WIDTH-2-i)-1];
          else data_out_shift[i][j] <= data_out_shift[i][j-1];
        end
      end
  end

  always @(*) begin
    for (i = 0; i < WINDOWS_WIDTH - 1; i = i + 1) begin
      if (i == 0) regx_in_data[i] = in_data;
      else regx_in_data[i] = regx_out_data[i-1];
    end
    for (i = 0; i < WINDOWS_WIDTH - 1; i = i + 1) begin
      if (i == 0) regx_in_valid[i] = in_valid;
      else regx_in_valid[i] = regx_out_valid[i-1];
    end
  end

  generate
    genvar o;
    for (o = 0; o < WINDOWS_WIDTH - 1; o = o + 1'b1) begin : shift_register
      SHIFT_REGISTER #(
          .DATA_WIDTH (DATA_WIDTH),
          .IMAGE_WIDTH(IMAGE_WIDTH),
          .IFOUTIMME  (1'b1)
      ) shift_registerx (
          .clk      (clk),
          .reset    (reset),
          .in_data  (regx_in_data[o]),
          .out_data (regx_out_data[o]),
          .in_valid (regx_in_valid[o]),
          .out_valid(regx_out_valid[o])
      );
    end
  endgenerate

endmodule