convy/internal/lang/vocabulary_test.go

package lang_test

import (
	"strings"
	"testing"

	"git.vakhrushev.me/av/convy/internal/lang"
)

func TestLeadTakesLongestMatch(t *testing.T) {
	v, err := lang.Lookup(1, "ru")
	if err != nil {
		t.Fatal(err)
	}

	cases := []struct {
		text string
		want string
	}{
		{"ДОЛЖЕН.", "ДОЛЖЕН"},
		{"НЕ ДОЛЖЕН.", "НЕ ДОЛЖЕН"},
		{"НЕ СЛЕДУЕТ.", "НЕ СЛЕДУЕТ"},
		{"СЛЕДУЕТ.", "СЛЕДУЕТ"},
		{"СНЯТО 2026-07-26.", "СНЯТО"},
		{"ПОЧЕМУ.", "ПОЧЕМУ"},
		{"", ""},
		{"Заголовок правила", ""},
		{"ДОЛЖЕНСТВОВАНИЕ", ""},
	}

	for _, tc := range cases {
		got, ok := v.Lead(tc.text)
		if tc.want == "" {
			if ok {
				t.Errorf("Lead(%q) = %q, wanted no word to be found", tc.text, got)
			}
			continue
		}
		if !ok || got != tc.want {
			t.Errorf("Lead(%q) = %q, %v; wanted %q", tc.text, got, ok, tc.want)
		}
	}
}

func TestForeignExcludesOwnWords(t *testing.T) {
	foreign := lang.Foreign(1, "ru")
	if len(foreign) == 0 {
		t.Fatal("the Russian vocabulary yielded no foreign word at all")
	}
	if code, ok := foreign["MUST"]; !ok || code != "en" {
		t.Errorf("MUST should be recognized as a word of the en vocabulary, got %q, %v", code, ok)
	}

	v, err := lang.Lookup(1, "ru")
	if err != nil {
		t.Fatal(err)
	}
	for word := range foreign {
		if _, own := v.Modal(word); own {
			t.Errorf("the word %q is the suite's own, yet landed among the foreign ones", word)
		}
		if _, own := v.Mark(word); own {
			t.Errorf("the mark %q is the suite's own, yet landed among the foreign ones", word)
		}
	}
}

func TestUnknownVersionAndCode(t *testing.T) {
	if _, err := lang.Lookup(99, "ru"); err == nil {
		t.Error("an unknown language version was accepted without an error")
	}
	if _, err := lang.Lookup(1, "xx"); err == nil {
		t.Error("an unknown vocabulary was accepted without an error")
	}
}

// A copy in a consuming repository carries no manifest, so the only thing that
// says which words of it are normative is the line it names the language by.
func TestRecognizeReadsTheLanguageOffTheVersionLine(t *testing.T) {
	ru, err := lang.Lookup(1, "ru")
	if err != nil {
		t.Fatal(err)
	}
	en, err := lang.Lookup(1, "en")
	if err != nil {
		t.Fatal(err)
	}

	for _, want := range []lang.Vocabulary{ru, en} {
		got, ok := lang.Recognize(want.VersionLine())
		if !ok {
			t.Fatalf("the version line of %q was not recognized", want.Code)
		}
		if got.Code != want.Code || got.Version != want.Version {
			t.Errorf("read as %q version %d, expected %q version %d",
				got.Code, got.Version, want.Code, want.Version)
		}
	}

	if _, ok := lang.Recognize("Обычная проза, ничего не объявляющая."); ok {
		t.Errorf("prose naming no words passed for a version line")
	}
}

// Words say which natural language, the number says which version. One without
// the other names no vocabulary: two versions of the same language would be
// told apart by nothing at all.
func TestRecognizeNeedsTheVersionToAgree(t *testing.T) {
	ru, err := lang.Lookup(1, "ru")
	if err != nil {
		t.Fatal(err)
	}
	line := ru.VersionLine()

	if _, ok := lang.Recognize(line); !ok {
		t.Fatalf("the line the tool itself writes was not recognized")
	}
	wrong := strings.Replace(line, "версии 1", "версии 7", 1)
	if wrong == line {
		t.Fatal("the fixture did not change the version")
	}
	if v, ok := lang.Recognize(wrong); ok {
		t.Errorf("a line naming version 7 was read as version %d (%s)", v.Version, v.Code)
	}
}