go-omni/i18n/pattern.go

package i18n

import (
	"encoding/json"
	"math"
	"reflect"
	"regexp"
	"strconv"
	"strings"
	"unicode/utf8"

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

var paramName = regexp.MustCompile(`^[A-Za-z_][A-Za-z0-9_.-]*$`)
var integerText = regexp.MustCompile(`^-?(0|[1-9][0-9]*)$`)

type part struct{ text, param string }
type compiled struct {
	parts                []part
	selectKey, pluralKey string
	cases                map[string]*compiled
}
type budget struct{ nodes, bytes int }

func compile(p Pattern, schema map[string]Kind, used map[string]bool, b *budget, depth int) (*compiled, error) {
	b.nodes++
	b.bytes += len(p.Text)
	if depth > 24 || b.nodes > 100000 || b.bytes > 4<<20 || len(p.Text) > 64<<10 || !utf8.ValidString(p.Text) {
		return nil, &Diagnostic{Kind: "resource", Problem: "pattern exceeds limits or contains invalid UTF-8"}
	}
	c := &compiled{selectKey: p.Select, pluralKey: p.Plural}
	if p.Select != "" || p.Plural != "" {
		if p.Select != "" && p.Plural != "" || p.Text != "" || p.Literal || len(p.Cases) == 0 || len(p.Cases) > 128 {
			return nil, &Diagnostic{Kind: "pattern", Problem: "expected one select/plural branch with cases"}
		}
		key := p.Select
		if key == "" {
			key = p.Plural
		}
		kind, ok := schema[key]
		if !ok || p.Plural != "" && kind != Integer && kind != Number || p.Select != "" && kind != String && kind != Boolean {
			return nil, &Diagnostic{Kind: "parameter", Param: key, Problem: "branch requires a declared parameter of the appropriate type"}
		}
		used[key] = true
		if _, ok := p.Cases["other"]; !ok {
			return nil, &Diagnostic{Kind: "pattern", Problem: "branch requires other"}
		}
		c.cases = make(map[string]*compiled, len(p.Cases))
		for _, name := range sortedKeys(p.Cases) {
			if !validID(name) {
				return nil, &Diagnostic{Kind: "pattern", Problem: "invalid case name"}
			}
			if p.Plural != "" && !pluralCase(name) {
				return nil, &Diagnostic{Kind: "pattern", Problem: "invalid plural category or exact integer case"}
			}
			child, err := compile(p.Cases[name], schema, used, b, depth+1)
			if err != nil {
				return nil, err
			}
			c.cases[name] = child
		}
		return c, nil
	}
	if p.Cases != nil {
		return nil, &Diagnostic{Kind: "pattern", Problem: "cases require select or plural"}
	}
	if p.Literal {
		c.parts = []part{{text: p.Text}}
		return c, nil
	}
	var literal strings.Builder
	flush := func() {
		if literal.Len() > 0 {
			c.parts = append(c.parts, part{text: literal.String()})
			literal.Reset()
		}
	}
	for i := 0; i < len(p.Text); {
		ch := p.Text[i]
		if ch != '{' && ch != '}' {
			literal.WriteByte(ch)
			i++
			continue
		}
		if i+1 < len(p.Text) && p.Text[i+1] == ch {
			literal.WriteByte(ch)
			i += 2
			continue
		}
		if ch == '}' {
			return nil, &Diagnostic{Kind: "pattern", Problem: "unescaped closing brace"}
		}
		end := strings.IndexByte(p.Text[i+1:], '}')
		if end < 0 {
			return nil, &Diagnostic{Kind: "pattern", Problem: "unclosed placeholder"}
		}
		name := p.Text[i+1 : i+1+end]
		if _, ok := schema[name]; !ok {
			return nil, &Diagnostic{Kind: "parameter", Param: name, Problem: "undeclared placeholder"}
		}
		used[name] = true
		flush()
		c.parts = append(c.parts, part{param: name})
		i += end + 2
	}
	flush()
	return c, nil
}

func pluralCase(s string) bool {
	switch s {
	case "zero", "one", "two", "few", "many", "other":
		return true
	}
	if !strings.HasPrefix(s, "=") {
		return false
	}
	n, err := strconv.ParseInt(s[1:], 10, 64)
	return err == nil && strconv.FormatInt(n, 10) == s[1:]
}

func scalar(kind Kind, value any) (string, error) {
	bad := func() (string, error) {
		return "", &Diagnostic{Kind: "parameter", Problem: "wrong type, range, or non-finite value"}
	}
	if value == nil {
		return bad()
	}
	v := reflect.ValueOf(value)
	switch kind {
	case String:
		if v.Kind() == reflect.String && v.Type() != reflect.TypeOf(json.Number("")) && utf8.ValidString(v.String()) && len(v.String()) <= 64<<10 {
			return v.String(), nil
		}
	case Boolean:
		if v.Kind() == reflect.Bool {
			return strconv.FormatBool(v.Bool()), nil
		}
	case Integer:
		if n, ok := value.(json.Number); ok {
			x, e := n.Int64()
			if e == nil {
				return strconv.FormatInt(x, 10), nil
			}
			return bad()
		}
		switch v.Kind() {
		case reflect.Int, reflect.Int8, reflect.Int16, reflect.Int32, reflect.Int64:
			return strconv.FormatInt(v.Int(), 10), nil
		case reflect.Uint, reflect.Uint8, reflect.Uint16, reflect.Uint32, reflect.Uint64:
			if v.Uint() <= math.MaxInt64 {
				return strconv.FormatUint(v.Uint(), 10), nil
			}
		}
	case Number:
		var n float64
		if j, ok := value.(json.Number); ok {
			var e error
			n, e = j.Float64()
			if e != nil {
				return bad()
			}
		} else {
			switch v.Kind() {
			case reflect.Float32, reflect.Float64:
				n = v.Float()
			default:
				return bad()
			}
		}
		if !math.IsNaN(n) && !math.IsInf(n, 0) {
			if n == 0 {
				return "0", nil
			}
			// ECMAScript's plain/scientific thresholds, shared with the browser runtime.
			format := byte('f')
			if math.Abs(n) < 1e-6 || math.Abs(n) >= 1e21 {
				format = 'e'
			}
			s := strconv.FormatFloat(n, format, -1, 64)
			s = strings.ReplaceAll(strings.ReplaceAll(s, "e-0", "e-"), "e+0", "e+")
			return s, nil
		}
	}
	return bad()
}

func (c *compiled) render(tag language.Tag, values map[string]string) (string, error) {
	if c.cases != nil {
		key := values[c.selectKey]
		if c.pluralKey != "" {
			n := values[c.pluralKey]
			key = "=" + n
			if _, ok := c.cases[key]; !ok {
				var err error
				key, err = category(tag, n)
				if err != nil {
					return "", err
				}
			}
		}
		next := c.cases[key]
		if next == nil {
			next = c.cases["other"]
		}
		return next.render(tag, values)
	}
	var out strings.Builder
	for _, p := range c.parts {
		value := p.text
		if p.param != "" {
			value = values[p.param]
		}
		if out.Len()+len(value) > 1<<20 {
			return "", &Diagnostic{Kind: "resource", Problem: "rendered message exceeds 1 MiB"}
		}
		out.WriteString(value)
	}
	return out.String(), nil
}