package terminology

import (
	"reflect"
	"strings"
	"testing"
)

func TestParseTypesKeepsOnlyAskedAndVocabulary(t *testing.T) {
	id := func(s string) string { return s }
	reply := strings.Join([]string{
		"元石\tterm",
		"方源   name",   // spaces instead of a tab — tolerant split
		"青茅山 | place", // padded pipe
		"家老\tPERSON",  // off-vocabulary → refused and counted
		"陌生\tterm",    // never asked → refused and counted
		"元石\tname",    // duplicate key → first answer wins
		"мусор",       // one field → refused and counted
	}, "\n")
	got, st := ParseTypes(reply, []string{"元石", "方源", "青茅山", "家老"}, id)
	want := map[string]string{"元石": "term", "方源": "name", "青茅山": "place"}
	if !reflect.DeepEqual(got, want) {
		t.Fatalf("parse = %#v, want %#v", got, want)
	}
	// off-vocabulary PERSON + unasked 陌生 + malformed мусор = 3 bad lines (the duplicate is silently ignored,
	// not counted, exactly like ParseReply).
	if st.Bad != 3 {
		t.Fatalf("unusable/off-vocabulary lines must be COUNTED, got %d", st.Bad)
	}
}

func TestParseTypesNormalizesKey(t *testing.T) {
	got, _ := ParseTypes("FANG\tname", []string{"fang"}, strings.ToLower)
	if got["fang"] != "name" {
		t.Fatalf("the reply key must be normalized by the caller's function, got %#v", got)
	}
}

// TestTypeLabelMismatchesIsHonest pins the screen's DELIBERATE limit: it flags a name/place whose rendering
// was clearly translated (multi-word), but it is NOT a safety net — a mistyped name rendered as one token,
// lower-case (元石→юаньши) or capitalised (元海→Юаньхай), passes it clean. The classifier phase is what
// prevents that harm; asserting the miss keeps a future reader from mistaking this for the guard.
func TestTypeLabelMismatchesIsHonest(t *testing.T) {
	rows := []LabelRow{
		{Src: "花家", Type: "name", Dst: "Дом Хуа"},           // translated name → FLAG
		{Src: "青茅山", Type: "place", Dst: "гора Цинмао"},     // translated place → FLAG
		{Src: "元石", Type: "name", Dst: "юаньши"},            // the harm, single lower-case token → MISSED
		{Src: "元海", Type: "name", Dst: "Юаньхай"},           // the harm, capitalised token → MISSED
		{Src: "灵泉", Type: "term", Dst: "духовный источник"}, // term is not screened → not flagged
	}
	got := TypeLabelMismatches(rows)
	var srcs []string
	for _, r := range got {
		srcs = append(srcs, r.Src)
	}
	if !reflect.DeepEqual(srcs, []string{"花家", "青茅山"}) {
		t.Fatalf("only the translated name/place rows must flag, got %v", srcs)
	}
}