package checks

import (
	"testing"

	"textmachine/backend/internal/lang"
)

// TestTargetSeamInertness is the П1 acceptance/mutation check (D39.62): a target with NO readability data
// (→en) runs the whole of layer 7 inert BY ABSENCE of data — not by a language branch — while ru is active.
// This is what makes "a pair not in the repo passes layer 7 without a Go change" true.
func TestTargetSeamInertness(t *testing.T) {
	en := CompileCheckers(nil, lang.TargetChecksFor("en"))
	if en.TargetActive() {
		t.Error("a target with no data must be inactive (layer 7 inert by absence)")
	}
	if en.TargetScriptNonLatin() {
		t.Error("a target with no declared word script must not read as non-Latin")
	}
	if got := en.tokenizeWords("hello слово"); got != nil {
		t.Errorf("a target with no word script must not tokenize, got %v", got)
	}
	if n, _ := en.lintBrokenWord("войть slово"); n != 0 {
		t.Error("broken-word lint must be inert for a dataless target")
	}
	// The pattern-based sanitizer classes are inert for a dataless target (the pipeline also gates the whole
	// sanitizer on TargetActive() upstream); a would-be ru preamble is not flagged for →en.
	if r := en.SanitizeOutput("Вот отредактированный перевод: текст."); r.Preamble != 0 || r.TrailingNote != 0 {
		t.Errorf("dataless target must not fire the pattern sanitizer classes, got %+v", r)
	}

	ru := CompileCheckers(nil, lang.TargetChecksFor("ru"))
	if !ru.TargetActive() {
		t.Error("ru ships readability data → must be active")
	}
	if !ru.TargetScriptNonLatin() {
		t.Error("ru is a non-Latin word script")
	}
	if got := ru.tokenizeWords("сло́во"); len(got) != 1 || got[0] != "слово" {
		t.Errorf("ru tokenizer must fold the U+0301 stress mark into one word, got %v (#11)", got)
	}
}