dateKeys-dart/tool/go_unicode_tables.go

//go:build ignore

// Writes lib/src/go_unicode.dart: 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
// of authorkey (stage 7b of docs/PLAN_dart.md):
//
//   - 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.
//
// Run it with the toolchain of datekeys-go, Go 1.26.8, Unicode 15.0.0,
// without changing anything there:
//
//	cd ../datekeys-go && go run ../datekeys-dart/tool/go_unicode_tables.go \
//		-out ../datekeys-dart/lib/src/go_unicode.dart
package main

import (
	"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++
		}
	}
	// Check the runs against the function.
	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
}

func list(b *strings.Builder, name, doc string, rs [][3]rune) {
	fmt.Fprintf(b, "// %s\nconst List<int> %s = [\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 Dart file to write")
	flag.Parse()
	if *out == "" {
		log.Fatal("usage: go run tool/go_unicode_tables.go -out lib/src/go_unicode.dart")
	}
	// 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")
	}

	var b strings.Builder
	fmt.Fprintf(&b, `// Code generated by tool/go_unicode_tables.go with Go %s, Unicode %s.
// DO NOT EDIT: regenerate it.
//
// The lists run to hundreds of lines: the formatter leaves them as the
// generator writes them.
// dart format off

/// The code points that Go's strings.ToLower, strings.ToUpper and
/// strings.TrimSpace depend on, as the package unicode of Go %s defines
/// them, Unicode %s: authorkey.dart 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 platform, nor
/// the tables of the path rules (Unicode 18.0.0, spec §29.5.1).
///
/// Internal: lib/datekeys.dart does not export it.
library;

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

`, runtime.Version(), unicode.Version, runtime.Version(), unicode.Version, runtime.Version(), unicode.Version)
	list(&b, "_lowerChanges", "The runs [first, last, stride] of r with unicode.ToLower(r) != r.", runs(func(r rune) bool { return unicode.ToLower(r) != r }))
	list(&b, "_upperChanges", "The runs [first, last, stride] of r with unicode.ToUpper(r) != r.", runs(func(r rune) bool { return unicode.ToUpper(r) != r }))
	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 first points.
bool _inRuns(List<int> table, int r) {
  var lo = 0;
  var hi = table.length ~/ 3;
  while (lo < hi) {
    final m = (lo + hi) ~/ 2;
    if (table[3 * m + 1] < r) {
      lo = m + 1;
    } else {
      hi = m;
    }
  }
  if (lo == table.length ~/ 3) return false;
  final first = table[3 * lo];
  return r >= first && (r - first) % table[3 * lo + 2] == 0;
}

/// Whether Go's unicode.ToLower changes the code point [r].
bool goLowerChanges(int r) => _inRuns(_lowerChanges, r);

/// Whether Go's unicode.ToUpper changes the code point [r].
bool goUpperChanges(int r) => _inRuns(_upperChanges, r);

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