From d4e5e659350ab1b5d1005f0aae51ac25bb39c6ac Mon Sep 17 00:00:00 2001
From: Timon Ringwald <timon.ringwald@mequen.de>
Date: Mon, 15 Aug 2022 13:40:13 +0200
Subject: [PATCH] lang code parsing

---
 lang_codes.go      | 144 +++++++++++++++++++++++++++++++++++++++++++++
 torrent_parsing.go |   4 +-
 2 files changed, 146 insertions(+), 2 deletions(-)
 create mode 100644 lang_codes.go

diff --git a/lang_codes.go b/lang_codes.go
new file mode 100644
index 0000000..4bf8814
--- /dev/null
+++ b/lang_codes.go
@@ -0,0 +1,144 @@
+package parsers
+
+import "strings"
+
+// langSynonyms converts all irregular lang codes to ISO 639-1
+var langSynonyms = map[string]string{
+	// english
+	"ENG": "en", // Erai-Raws
+	"US":  "en", // Erai-Raws
+
+	// portuguese
+	"POR-BR": "pt", // Erai-Raws
+	"POR":    "pt", // Erai-Raws
+	"BR":     "pt", // Erai-Raws
+
+	// spanish
+	"SPA-LA": "es", // Erai-Raws
+	"SPA":    "es", // Erai-Raws
+	"MX":     "es", // Erai-Raws
+
+	// arabic
+	"ARA": "ar", // Erai-Raws
+	"SA":  "ar", // Erai-Raws
+
+	// french
+	"FRE": "fr", // Erai-Raws
+
+	// german
+	"GER": "de", // Erai-Raws
+
+	// italian
+	"ITA": "it", // Erai-Raws
+
+	// russian
+	"RUS": "ru", // Erai-Raws
+
+	// japanese
+	"JPN": "ja", // Erai-Raws
+	"JP":  "ja", // Erai-Raws
+
+	// polish
+	"POL": "pl", // Erai-Raws
+
+	// dutch
+	"DUT": "nl", // Erai-Raws
+
+	// norwegian
+	"NOB": "no", // Erai-Raws
+
+	// turkish
+	"TUR": "tr", // Erai-Raws
+
+	// swedish
+	"SWE": "sv", // Erai-Raws
+	"SE":  "sv", // Erai-Raws
+
+	// greek
+	"GRE": "el", // Erai-Raws
+	"GR":  "el", // Erai-Raws
+
+	// hebrew
+	"HEB": "he", // Erai-Raws
+	"IL":  "he", // Erai-Raws
+
+	// romanian
+	"RUM": "ro", // Erai-Raws
+
+	// indonesian
+	"IND": "id", // Erai-Raws
+
+	// thai
+	"THA": "th", // Erai-Raws
+
+	// korean
+	"KOR": "ko", // Erai-Raws
+	"KR":  "ko", // Erai-Raws
+
+	// danish
+	"DAN": "da", // Erai-Raws
+	"DK":  "da", // Erai-Raws
+
+	// chinese (simplified & traditional)
+	"CHI": "zh", // Erai-Raws
+	"CN":  "zh", // Erai-Raws
+
+	// bulgarian
+	"BUL": "bg", // Erai-Raws
+
+	// vietnamese
+	"VIE": "vi", // Erai-Raws
+	"VN":  "vi", // Erai-Raws
+
+	// hindi
+	"HIN": "hi", // Erai-Raws
+	"IN":  "hi", // Erai-Raws
+
+	// tamil
+	"TEL": "ta", // Erai-Raws
+	"LK":  "ta", // Erai-Raws
+
+	// ukrainian
+	"UKR": "uk", // Erai-Raws
+	"UA":  "uk", // Erai-Raws
+
+	// hungarian
+	"HUN": "hu", // Erai-Raws
+
+	// czech
+	"CES": "cs", // Erai-Raws
+	"CZ":  "cs", // Erai-Raws
+
+	// croatian
+	"HRV": "hr", // Erai-Raws
+
+	// malaysian
+	"MAY": "ms", // Erai-Raws
+	"MY":  "ms", // Erai-Raws
+
+	// slovakian
+	"SLK": "sk", // Erai-Raws
+
+	// filipino
+	"FIL": "tl", // Erai-Raws
+	"PH":  "tl", // Erai-Raws
+}
+
+// ParseLanguage converts irregular language codes to ISO 639-1
+func ParseLanguage(str string) string {
+	if code, ok := langSynonyms[str]; ok {
+		return code
+	}
+
+	return strings.ToLower(str)
+}
+
+// ParseLanguages converts multiple irregular language codes to ISO 639-1.
+// It simply calls ParseLanguage for each language code
+func ParseLanguages(langCodes []string) []string {
+	codes := make([]string, 0, len(langCodes))
+	for _, irregularLangCode := range langCodes {
+		codes = append(codes, ParseLanguage(irregularLangCode))
+	}
+	return codes
+}
diff --git a/torrent_parsing.go b/torrent_parsing.go
index b954149..431f274 100644
--- a/torrent_parsing.go
+++ b/torrent_parsing.go
@@ -84,8 +84,8 @@ func regexTorrentParser(regex string, options TorrentParseOptions) model.Torrent
 			Episode:            episode,
 			Resolution:         resolution,
 			Parser:             parser,
-			Languages:          languages,
-			Subtitles:          subtitles,
+			Languages:          ParseLanguages(languages),
+			Subtitles:          ParseLanguages(subtitles),
 
 			Torrent: torrent,
 		}, true