resource-scraper/internal/releaseparser/groups/moonblood/parser_test.go

package moonblood

import (
	"bufio"
	"os"
	"testing"
)

func TestParseMoonblood(t *testing.T) {
	file, err := os.Open("../../resources/moonblood.txt")
	if err != nil {
		t.Fatal(err)
	}
	defer file.Close()

	scanner := bufio.NewScanner(file)
	total, success := 0, 0

	for scanner.Scan() {
		line := scanner.Text()
		if line == "" {
			continue
		}

		result := Parse(line)
		total++

		if result.Group == "MOONBLOOD" && result.Year != 0 && result.Format == "FLAC" {
			success++
		}

		if result.Confidence < 0.7 {
			t.Logf("Low confidence: %s (%.2f)", line, result.Confidence)
		}
	}

	t.Logf("Parsed %d releases, %d successful (%.1f%%)", total, success, float64(success)/float64(total)*100)
}

func TestParseFullTail(t *testing.T) {
	r := Parse("Fane-Dragged_Through_the_Unending-24BIT-WEB-FLAC-2025-MOONBLOOD")
	if r.Artist != "Fane" {
		t.Errorf("Artist = %q, want %q", r.Artist, "Fane")
	}
	if r.Title != "Dragged_Through_the_Unending" {
		t.Errorf("Title = %q, want %q", r.Title, "Dragged_Through_the_Unending")
	}
	if r.Year != 2025 || r.Format != "FLAC" || r.Source != "WEB" || r.BitDepth != "24BIT" {
		t.Errorf("tail fields wrong: %+v", r)
	}
	if len(r.UnknownTokens) != 0 {
		t.Errorf("UnknownTokens = %v, want empty", r.UnknownTokens)
	}
	if r.Confidence < 0.99 {
		t.Errorf("Confidence = %.2f, want ~1.0", r.Confidence)
	}
}

func TestParseEditionAndDemo(t *testing.T) {
	r := Parse("Satans_Grind-Rotten_Demo-DEMO-16BIT-WEB-FLAC-2016-MOONBLOOD")
	if r.Edition != "DEMO" {
		t.Errorf("Edition = %q, want %q", r.Edition, "DEMO")
	}
	if r.Artist != "Satans_Grind" || r.Title != "Rotten_Demo" {
		t.Errorf("Artist/Title = %q/%q", r.Artist, r.Title)
	}
	if len(r.UnknownTokens) != 0 {
		t.Errorf("UnknownTokens = %v, want empty", r.UnknownTokens)
	}
}

func TestParseArtistWithHyphen(t *testing.T) {
	// Artist itself contains a hyphen; no part of it may leak into UnknownTokens.
	r := Parse("Ultra-Violence-All_My_Life-EP-16BIT-WEB-FLAC-2026-MOONBLOOD")
	if r.Artist != "Ultra" || r.Title != "Violence-All_My_Life" {
		t.Errorf("Artist/Title = %q/%q, want Ultra / Violence-All_My_Life", r.Artist, r.Title)
	}
	if r.Edition != "EP" || r.BitDepth != "16BIT" || r.Source != "WEB" || r.Year != 2026 {
		t.Errorf("tail fields wrong: %+v", r)
	}
	if len(r.UnknownTokens) != 0 {
		t.Errorf("UnknownTokens = %v, want empty", r.UnknownTokens)
	}
}

func TestParseLanguageAndCDSource(t *testing.T) {
	r := Parse("Burn_Thou_in_Effigy-Memorias_Postumas_do_Imperador-BR-CD-FLAC-2026-MOONBLOOD")
	if r.Language != "BR" || r.Source != "CD" || r.Year != 2026 || r.Format != "FLAC" {
		t.Errorf("tail fields wrong: %+v", r)
	}
	if len(r.UnknownTokens) != 0 {
		t.Errorf("UnknownTokens = %v, want empty", r.UnknownTokens)
	}
}

func TestParseNoYear(t *testing.T) {
	// Without a year the tail still parses; group stays the last token.
	r := Parse("Fane-Gallant_Shroud-EP-16BIT-WEB-FLAC-MOONBLOOD")
	if r.Group != "MOONBLOOD" || r.Year != 0 {
		t.Errorf("Group/Year = %q/%d", r.Group, r.Year)
	}
	if r.Edition != "EP" || r.Source != "WEB" || r.Format != "FLAC" {
		t.Errorf("tail fields wrong: %+v", r)
	}
	if r.Artist != "Fane" || r.Title != "Gallant_Shroud" {
		t.Errorf("Artist/Title = %q/%q", r.Artist, r.Title)
	}
}

func TestParseResolutionSlot(t *testing.T) {
	r := Parse("Ice_Nine_Kills-Last_Chance_To_Make_Amends-REMASTERED-24BIT-44KHZ-WEB-FLAC-2024-FLACCiD")
	if r.Group != "FLACCiD" {
		t.Errorf("Group = %q, want FLACCiD", r.Group)
	}
	if r.Resolution != "44KHZ" {
		t.Errorf("Resolution = %q, want 44KHZ", r.Resolution)
	}
	if r.BitDepth != "24BIT" || r.Edition != "REMASTERED" || r.Source != "WEB" || r.Year != 2024 {
		t.Errorf("tail fields wrong: %+v", r)
	}
	if r.Artist != "Ice_Nine_Kills" || r.Title != "Last_Chance_To_Make_Amends" {
		t.Errorf("Artist/Title = %q/%q", r.Artist, r.Title)
	}
	if len(r.UnknownTokens) != 0 {
		t.Errorf("UnknownTokens = %v, want empty", r.UnknownTokens)
	}
}

func TestParseVinylIsSource(t *testing.T) {
	r := Parse("ROTH-Alles_Endet-DE-VINYL-FLAC-2026-TOTENKVLT")
	if r.Language != "DE" {
		t.Errorf("Language = %q, want DE", r.Language)
	}
	if r.Source != "VINYL" {
		t.Errorf("Source = %q, want VINYL", r.Source)
	}
	if r.Edition != "" {
		t.Errorf("Edition = %q, want empty (VINYL is source, not edition)", r.Edition)
	}
	if len(r.UnknownTokens) != 0 {
		t.Errorf("UnknownTokens = %v, want empty", r.UnknownTokens)
	}
}

func TestParseCaseInsensitiveDictionaries(t *testing.T) {
	r := Parse("Artist-Some_Title-16bit-web-flac-2020-moonblood")
	if r.BitDepth != "16BIT" || r.Source != "WEB" || r.Format != "FLAC" {
		t.Errorf("case-insensitive lookup failed: %+v", r)
	}
	if r.Artist != "Artist" || r.Title != "Some_Title" {
		t.Errorf("Artist/Title = %q/%q", r.Artist, r.Title)
	}
}

func TestParseUnknownTailTokenReported(t *testing.T) {
	// Невідомий токен між двома відомими маркерами → у UnknownTokens.
	r := Parse("Artist-Some_Title-SPLIT-FOO-16BIT-WEB-FLAC-2025-MOONBLOOD")
	if len(r.UnknownTokens) != 1 || r.UnknownTokens[0] != "FOO" {
		t.Errorf("UnknownTokens = %v, want [FOO]", r.UnknownTokens)
	}
	if r.Edition != "SPLIT" || r.BitDepth != "16BIT" || r.Source != "WEB" || r.Year != 2025 {
		t.Errorf("tail fields wrong: %+v", r)
	}
	if r.Artist != "Artist" || r.Title != "Some_Title" {
		t.Errorf("Artist/Title = %q/%q", r.Artist, r.Title)
	}
}

func TestParseCoreTokensNeverReported(t *testing.T) {
	// Артист/тайтл ніколи не потрапляють в UnknownTokens.
	r := Parse("Fane-Dragged_Through_the_Unending-24BIT-WEB-FLAC-2025-MOONBLOOD")
	if len(r.UnknownTokens) != 0 {
		t.Errorf("UnknownTokens = %v, want empty", r.UnknownTokens)
	}
	if r.Artist != "Fane" || r.Title != "Dragged_Through_the_Unending" {
		t.Errorf("Artist/Title = %q/%q", r.Artist, r.Title)
	}
}

func TestParseAnyGroupNoWarning(t *testing.T) {
	cases := []string{
		"Jetsex-Down_Down_Down-24BIT-WEB-FLAC-2025-FLACCiD",
		"Sabaton-The_Last_Stand-24BIT-96KHZ-WEB-FLAC-2016-FLACCiD",
		"Wind_Walkers-What_If_I_Break-16BIT-WEB-FLAC-2023-FLACCiD",
	}
	for _, line := range cases {
		r := Parse(line)
		if r.Group != "FLACCiD" {
			t.Errorf("%s: Group = %q, want FLACCiD", line, r.Group)
		}
		if len(r.Warnings) != 0 {
			t.Errorf("%s: Warnings = %v, want empty", line, r.Warnings)
		}
		if len(r.UnknownTokens) != 0 {
			t.Errorf("%s: UnknownTokens = %v, want empty", line, r.UnknownTokens)
		}
		if r.Year == 0 || r.Format != "FLAC" || r.Source != "WEB" || r.BitDepth == "" {
			t.Errorf("%s: tail fields wrong: %+v", line, r)
		}
	}
}