// unix cmp util
// license: this code is released in the public domain. Do what thou wilt.
// @date 2023-03-15 09:41:27Z

const std = @import("std");
const debug = std.debug;
const assert = debug.assert;
const os = std.os;
const mem = std.mem;

fn usage() void{
  debug.print("usage: cmp file-a [file-b]\n  compares 2 files bytes\n", .{});
  os.exit(1);
}

pub fn main() !void{
  var av = std.process.args();
  _ = av.skip();
  const fna = av.next() orelse "";
  const fnb = av.next() orelse "";
  if(0==fna.len) usage();

  const fda = try os.open(fna, os.O.RDONLY, 0o444);
  defer os.close(fda);
  const fdb = if(0==fnb.len) os.STDIN_FILENO else try os.open(fnb, os.O.RDONLY, 0o444);
  defer os.close(fdb);

  const bsz = 64*1024;
  var bufa: [bsz]u8 = undefined;
  var bufb: [bsz]u8 = undefined;
  var offset: usize = 0;
  var lineno: usize = 1;
  while(true){
    const rsza = try os.read(fda, &bufa);
    const rszb = try os.read(fdb, &bufb);
    if(rsza != rszb){
      const sfn = if(rsza<rszb) fna else fnb;
      const o = if(rsza<rszb) rsza else rszb;
      debug.print("cmp: prefixes match but file {s} is shorter ({} bytes)\n", .{sfn, offset+o});
      os.exit(1);
    }
    if(0==rsza) break;
    if(mem.indexOfDiff(u8, bufa[0..rsza], bufb[0..rsza])) |o| {
      lineno += countScalarSIMD(u8, bufa[0..o], '\n');
      debug.print("cmp: mismatch at byte:{} line:{}\n", .{offset+o+1, lineno});
      os.exit(1);
    }
    offset += rsza;
    lineno += countScalarSIMD(u8, bufa[0..rsza], '\n');
  }
}

// my simd/vectorized countScalar version
pub fn countScalarSIMD(comptime T: type, haystack: []const T, needle: T) usize {
  var cnt: usize = 0;
  const vsz = 16;
  const lnub = (haystack.len % vsz);
  const l = haystack.len - lnub;
  const vneedle = @splat(vsz, needle);
  var i: usize = 0; while(i<l):(i+=vsz) {
    const slide: @Vector(vsz, u8) = haystack[i..][0..vsz].*;
    const mask = (vneedle == slide);
    comptime var j = 0; inline while(j<vsz):(j+=1) { cnt += @boolToInt(mask[j]); }
  }
  i=l; while(i<haystack.len):(i+=1) {
    if(needle == haystack[i]) cnt += 1;
  }
  return cnt;
}