go-omni/i18n/plural.go

package i18n

import (
	_ "embed"
	"encoding/json"
	"fmt"
	"math/big"
	"regexp"
	"strconv"
	"strings"

	"golang.org/x/text/language"
)

// PluralRulesVersion pins both runtimes to the same Unicode data. Changing it
// is a presentation compatibility change, independent of message format Version.
const PluralRulesVersion = "cldr-46.1"

//go:embed cldr/plurals.json
var pluralData []byte

//go:embed cldr/LICENSE
var unicodeLicense string

type pluralRange struct{ lo, hi *big.Int }
type pluralCondition struct {
	operand byte
	modulus *big.Int
	negate  bool
	ranges  []pluralRange
}
type pluralRule [][]pluralCondition // OR of AND groups, per the CLDR grammar.
type pluralSet struct {
	text  map[string]string
	rules map[string]pluralRule
}

var pluralConditionSyntax = regexp.MustCompile(`^([nivwftec])(?: % ([1-9][0-9]*))? (!=|=) ([0-9]+(?:\.\.[0-9]+)?(?:,\s*[0-9]+(?:\.\.[0-9]+)?)*)$`)
var cardinalRules = loadCardinalRules()

func bigDecimal(s string) *big.Int {
	if s == "" {
		return new(big.Int)
	}
	n, ok := new(big.Int).SetString(s, 10)
	if !ok {
		panic("i18n: invalid embedded plural integer")
	}
	return n
}
func parsePluralRule(text string) (pluralRule, error) {
	var result pluralRule
	if len(text) > 2048 {
		return nil, fmt.Errorf("plural rule too large")
	}
	for _, or := range strings.Split(text, " or ") {
		var group []pluralCondition
		for _, and := range strings.Split(or, " and ") {
			parts := pluralConditionSyntax.FindStringSubmatch(strings.TrimSpace(and))
			if parts == nil {
				return nil, fmt.Errorf("invalid CLDR plural condition")
			}
			c := pluralCondition{operand: parts[1][0], negate: parts[3] == "!="}
			if parts[2] != "" {
				c.modulus = bigDecimal(parts[2])
			}
			for _, r := range strings.Split(parts[4], ",") {
				lo, hi, has := strings.Cut(strings.TrimSpace(r), "..")
				if !has {
					hi = lo
				}
				low, high := bigDecimal(lo), bigDecimal(hi)
				if low.Cmp(high) > 0 {
					return nil, fmt.Errorf("reversed plural range")
				}
				c.ranges = append(c.ranges, pluralRange{low, high})
			}
			group = append(group, c)
		}
		result = append(result, group)
	}
	return result, nil
}

func loadCardinalRules() map[string]pluralSet {
	var data struct {
		Supplemental struct {
			Rules map[string]map[string]string `json:"plurals-type-cardinal"`
		} `json:"supplemental"`
	}
	if err := json.Unmarshal(pluralData, &data); err != nil {
		panic("i18n: invalid embedded CLDR data")
	}
	out := map[string]pluralSet{}
	for locale, entries := range data.Supplemental.Rules {
		set := pluralSet{text: map[string]string{}, rules: map[string]pluralRule{}}
		for name, raw := range entries {
			category := strings.TrimPrefix(name, "pluralRule-count-")
			text, _, _ := strings.Cut(raw, "@")
			text = strings.TrimSpace(text)
			if category == "other" {
				continue
			}
			rule, err := parsePluralRule(text)
			if err != nil {
				panic("i18n: unsupported embedded CLDR rule for " + locale)
			}
			set.text[category] = text
			set.rules[category] = rule
		}
		out[strings.ReplaceAll(locale, "_", "-")] = set
	}
	return out
}

func rulesFor(tag language.Tag) pluralSet {
	for {
		if rules, ok := cardinalRules[tag.String()]; ok {
			return rules
		}
		if tag == language.Und {
			break
		}
		tag = tag.Parent()
	}
	return pluralSet{text: map[string]string{}, rules: map[string]pluralRule{}}
}

// operands expands the canonical numeric string exactly, avoiding any float
// conversion for int64 and any browser rounding of tiny fractional operands.
func operands(text string) (map[byte]*big.Int, bool) {
	text = strings.TrimPrefix(text, "-")
	mantissa, exponent, has := strings.Cut(text, "e")
	shift := 0
	if has {
		shift, _ = strconv.Atoi(exponent)
	}
	whole, fraction, _ := strings.Cut(mantissa, ".")
	digits := whole + fraction
	point := len(whole) + shift
	switch {
	case point <= 0:
		whole = "0"
		fraction = strings.Repeat("0", -point) + digits
	case point >= len(digits):
		whole = digits + strings.Repeat("0", point-len(digits))
		fraction = ""
	default:
		whole = digits[:point]
		fraction = digits[point:]
	}
	tail := strings.TrimRight(fraction, "0")
	i, f := bigDecimal(whole), bigDecimal(fraction)
	return map[byte]*big.Int{'n': i, 'i': i, 'v': big.NewInt(int64(len(fraction))), 'w': big.NewInt(int64(len(tail))), 'f': f, 't': bigDecimal(tail), 'c': new(big.Int), 'e': new(big.Int)}, f.Sign() != 0
}
func (rule pluralRule) matches(values map[byte]*big.Int, fractional bool) bool {
	for _, group := range rule {
		all := true
		for _, condition := range group {
			value := values[condition.operand]
			if condition.modulus != nil {
				value = new(big.Int).Mod(value, condition.modulus)
			}
			found := false
			if condition.operand != 'n' || !fractional {
				for _, r := range condition.ranges {
					if value.Cmp(r.lo) >= 0 && value.Cmp(r.hi) <= 0 {
						found = true
						break
					}
				}
			}
			if condition.negate {
				found = !found
			}
			if !found {
				all = false
				break
			}
		}
		if all {
			return true
		}
	}
	return false
}
func category(tag language.Tag, text string) (string, error) {
	values, fractional := operands(text)
	rules := rulesFor(tag).rules
	for _, category := range []string{"zero", "one", "two", "few", "many"} {
		if rule, ok := rules[category]; ok && rule.matches(values, fractional) {
			return category, nil
		}
	}
	return "other", nil
}