DateKeys-App/scripts/go-unicode-tables.go

//go:build ignore

// Writes src/lib/dkc/gounicode.ts: the three sets of code points that Go's
// strings.ToLower, strings.ToUpper and strings.TrimSpace depend on, as the
// package unicode of the Go that runs it defines them, for the key strings
// and the key files of authorkey.ts:
//
//   - the code points r with unicode.ToLower(r) != r, which strings.ToLower
//     changes, so that parsePublic tells "written in lower case" as Go does;
//   - those with unicode.ToUpper(r) != r, for parseSecret;
//   - those with unicode.IsSpace(r), which strings.TrimSpace trims from a
//     line of a key file.
//
// Each set is written as runs [first, last, stride], stride 1 or 2, in
// increasing order. No value is written by hand: the generator scans every
// code point and checks the runs it writes against the functions of Go, and
// the premises of authorkey.ts against strings.ToLower, strings.ToUpper and
// strings.TrimSpace. It also writes the SHA-256 of the three bit sets, which
// gounicode.test.ts recomputes from the runs.
//
// Run it with the Go toolchain of the reference implementation (the same
// tables as the Dart port's, Unicode 15.0.0 of Go 1.26):
//
//	go run scripts/go-unicode-tables.go -out src/lib/dkc/gounicode.ts
package main

import (
	"crypto/sha256"
	"flag"
	"fmt"
	"log"
	"os"
	"runtime"
	"strings"
	"unicode"
	"unicode/utf8"
)

// runs returns the code points of in as runs [first, last, stride]: a run
// of stride 2 is taken when it covers at least three points, so that
// alternating cases stay short.
func runs(in func(rune) bool) [][3]rune {
	var pts []rune
	for r := rune(0); r <= unicode.MaxRune; r++ {
		if r >= 0xd800 && r <= 0xdfff {
			continue // not in valid UTF-8: a decoder gives U+FFFD instead
		}
		if in(r) {
			pts = append(pts, r)
		}
	}
	var out [][3]rune
	for i := 0; i < len(pts); {
		j := i
		for j+1 < len(pts) && pts[j+1] == pts[j]+1 {
			j++
		}
		k := i
		for k+1 < len(pts) && pts[k+1] == pts[k]+2 {
			k++
		}
		switch {
		case j > i:
			out = append(out, [3]rune{pts[i], pts[j], 1})
			i = j + 1
		case k-i >= 2:
			out = append(out, [3]rune{pts[i], pts[k], 2})
			i = k + 1
		default:
			out = append(out, [3]rune{pts[i], pts[i], 1})
			i++
		}
	}
	member := map[rune]bool{}
	for _, r := range out {
		for c := r[0]; c <= r[1]; c += r[2] {
			if member[c] {
				log.Fatalf("U+%04X twice", c)
			}
			member[c] = true
		}
	}
	for r := rune(0); r <= unicode.MaxRune; r++ {
		if r >= 0xd800 && r <= 0xdfff {
			continue
		}
		if member[r] != in(r) {
			log.Fatalf("U+%04X: the runs disagree", r)
		}
	}
	return out
}

// bitSum is the SHA-256 of the set in: bit r at byte r>>3, mask 1<<(r&7).
func bitSum(in func(rune) bool) string {
	bits := make([]byte, (unicode.MaxRune+1)/8)
	for r := rune(0); r <= unicode.MaxRune; r++ {
		if r >= 0xd800 && r <= 0xdfff {
			continue
		}
		if in(r) {
			bits[r>>3] |= 1 << (r & 7)
		}
	}
	return fmt.Sprintf("%x", sha256.Sum256(bits))
}

func list(b *strings.Builder, name, doc string, rs [][3]rune) {
	fmt.Fprintf(b, "// %s\nconst %s: readonly number[] = [\n", doc, name)
	for _, r := range rs {
		fmt.Fprintf(b, "  0x%04x, 0x%04x, %d,\n", r[0], r[1], r[2])
	}
	b.WriteString("];\n\n")
}

func main() {
	out := flag.String("out", "", "the TypeScript file to write")
	flag.Parse()
	if *out == "" {
		log.Fatal("usage: go run scripts/go-unicode-tables.go -out src/lib/dkc/gounicode.ts")
	}
	// strings.ToLower and ToUpper change a string exactly when one of its
	// runes changes, or when it is not valid UTF-8: check that premise on
	// every code point.
	for r := rune(0); r <= unicode.MaxRune; r++ {
		if r >= 0xd800 && r <= 0xdfff {
			continue
		}
		s := string(r)
		if (strings.ToLower(s) != s) != (unicode.ToLower(r) != r) {
			log.Fatalf("strings.ToLower and unicode.ToLower disagree on U+%04X", r)
		}
		if (strings.ToUpper(s) != s) != (unicode.ToUpper(r) != r) {
			log.Fatalf("strings.ToUpper and unicode.ToUpper disagree on U+%04X", r)
		}
		if (strings.TrimSpace(s) == "") != unicode.IsSpace(r) {
			log.Fatalf("strings.TrimSpace and unicode.IsSpace disagree on U+%04X", r)
		}
	}
	if strings.ToLower("\xff") == "\xff" || strings.ToUpper("\xff") == "\xff" {
		log.Fatal("an invalid byte no longer changes")
	}
	if strings.TrimSpace(" \xff ") != "\xff" || utf8.RuneError != 0xfffd {
		log.Fatal("TrimSpace no longer stops at an invalid byte")
	}

	lower := func(r rune) bool { return unicode.ToLower(r) != r }
	upper := func(r rune) bool { return unicode.ToUpper(r) != r }
	var b strings.Builder
	fmt.Fprintf(&b, `// Code generated by scripts/go-unicode-tables.go with Go %s, Unicode %s.
// DO NOT EDIT: regenerate it.
//
// The code points that Go's strings.ToLower, strings.ToUpper and
// strings.TrimSpace depend on, as the package unicode of Go defines them:
// authorkey.ts reads the strings of a key and the lines of a key file with
// them, so that its errors are those of Go whatever the bytes. They are not
// the case mapping of the JavaScript engine (toLowerCase, \p{…}), whose
// version of Unicode changes with the engine, nor the tables of the path
// rules (Unicode 18.0.0, spec §29.5.1). Internal: index.ts does not
// re-export it.

/** The version of Unicode of the package unicode of Go %s. */
export const GO_UNICODE_VERSION = '%s';

/** The SHA-256 of the bit sets of the three tables (bit r at byte r >> 3), which the tests recompute. */
export const GO_UNICODE_SUMS = { lower: '%s', upper: '%s', space: '%s' } as const;

`, runtime.Version(), unicode.Version, runtime.Version(), unicode.Version, bitSum(lower), bitSum(upper), bitSum(unicode.IsSpace))
	list(&b, "LOWER_CHANGES", "The runs [first, last, stride] of r with unicode.ToLower(r) != r.", runs(lower))
	list(&b, "UPPER_CHANGES", "The runs [first, last, stride] of r with unicode.ToUpper(r) != r.", runs(upper))
	list(&b, "SPACES", "The runs [first, last, stride] of r with unicode.IsSpace(r).", runs(unicode.IsSpace))
	b.WriteString(`// Whether r is in the runs of table, by binary search on the last points.
function inRuns(table: readonly number[], r: number): boolean {
  let lo = 0;
  let hi = table.length / 3;
  while (lo < hi) {
    const m = (lo + hi) >>> 1;
    if (table[3 * m + 1]! < r) lo = m + 1;
    else hi = m;
  }
  if (lo === table.length / 3) return false;
  const first = table[3 * lo]!;
  return r >= first && (r - first) % table[3 * lo + 2]! === 0;
}

/** Whether Go's unicode.ToLower changes the code point r. */
export const goLowerChanges = (r: number): boolean => inRuns(LOWER_CHANGES, r);

/** Whether Go's unicode.ToUpper changes the code point r. */
export const goUpperChanges = (r: number): boolean => inRuns(UPPER_CHANGES, r);

/** Whether r is a space for Go's unicode.IsSpace. */
export const goIsSpace = (r: number): boolean => inRuns(SPACES, r);
`)
	if err := os.WriteFile(*out, []byte(b.String()), 0o644); err != nil {
		log.Fatal(err)
	}
	fmt.Printf("wrote %s\n", *out)
}