subber: further refine cleaner & srt parser

martinlindhe · Oct 11, 2024 · 3d9988f · 3d9988f
1 parent 57da01c
commit 3d9988f
Show file tree

Hide file tree

Showing 4 changed files with 15 additions and 11 deletions.
diff --git a/cleaner.go b/cleaner.go
@@ -42,7 +42,8 @@ var (
 		"encoded by",
 		"downloaded from",
 		"web-dl",
-		"subscene",
+		"subscene", "podnapisi", "bokutox", "team nanban",
+		"fury_don@hotmail.com",
 		"broadcasttext",
 		"seriessub", "subtitlesource",
 		"addic7ed", "addicted.com", "vaioholics",
@@ -51,18 +52,18 @@ var (
 		"cssubs", "tvsub", "uksubtitles",
 		"ragbear.com", "ydy.com", "yyets.net", "indivx.net", "sub-way.fr",
 		"forom.com", "forom. com", "facebook.com", "hdvietnam.com", "sapo.pt", "softhome.net",
-		"americascardroom.com",
-		"@gmail.com", "@hotmail.com", "@hotmail.fr",
-		"napisy.org", "1000fr.com", "yts.mx",
+		"americascardroom.com", "subti.com", "tugazx", "pirata-tuga",
+		"napisy.org", "1000fr.com", "yts.mx", "yts.am", "yts.ag", "yts.lt",
 		"opensubtitles", "open subtitles", "s u b t i t l e",
 		"sous-titres.eu", "300mbfilms.com", "put.io", "subtitulos.es", "osdb.link", "300mbunited",
 		"simail.si", "sf-film.dk", "sf.net", "vitac.com", "rapidpremium", "psarips",
 		"yify-torrents", "yify torrents", "yify movies",
 		"thepiratebay", "anoxmous", "verdikt", "la fisher team", "red bee media",
-		"mkv player", "best watched using", "advertise your product", "remove all ads",
+		"mkv player", "mkv-potplayer-vlc", "best watched using", "best play with",
+		"advertise your product", "remove all ads",
 		"memoryonsmells", "1st-booking",
-		":[gwc]:", "ripped by", "ripped with subrip", "titra film",
-		"hiqve",
+		":[gwc]:", "ripped with subrip", "titra film",
+		"hiqve", "kentir.bb", "w-bb.org", "sub download",
 		"trimark home video",
 
 		// swedish:

diff --git a/cmd/subber/subber.go b/cmd/subber/subber.go
@@ -64,9 +64,6 @@ func parseAndWriteSubFile(inFileName, outFileName string, filterName string, kee
 	if err != nil {
 		return err
 	}
-	if len(data) == 0 {
-		return fmt.Errorf("input file is empty")
-	}
 
 	out, err := cleanupSub(data, filterName, keepAds, sync, inFileName)
 	if err != nil {

diff --git a/parser.go b/parser.go
@@ -8,6 +8,9 @@ import (
 
 // Parse tries to parse a subtitle
 func Parse(b []byte) (Subtitle, error) {
+	if len(b) <= 10 {
+		return Subtitle{}, fmt.Errorf("parse: empty input")
+	}
 	s := ConvertToUTF8(b)
 	if looksLikeCCDBCapture(s) {
 		return NewFromCCDBCapture(s)
@@ -29,6 +32,9 @@ func LooksLikeTextSubtitle(filename string) bool {
 	if err != nil {
 		log.Fatal(err)
 	}
+	if len(data) <= 10 {
+		log.Fatal(fmt.Errorf("parse: empty input in '%s'", filename))
+	}
 	s := ConvertToUTF8(data)
 	return looksLikeCCDBCapture(s) || looksLikeSSA(s) || looksLikeDCSub(s) || looksLikeSRT(s) || looksLikeVTT(s)
 }
diff --git a/srt.go b/srt.go
@@ -22,7 +22,7 @@ func init() {
 
 func looksLikeSRT(s string) bool {
 	s = strings.TrimSpace(s)
-	return strings.HasPrefix(s, "0\n") || strings.HasPrefix(s, "0\r\n") || strings.HasPrefix(s, "1\n") || strings.HasPrefix(s, "1\r\n") || strings.HasPrefix(s, "2\n") || strings.HasPrefix(s, "2\r\n") || strings.HasPrefix(s, "3\n") || strings.HasPrefix(s, "3\r\n")
+	return strings.HasPrefix(s, "0\n") || strings.HasPrefix(s, "0\r\n") || strings.HasPrefix(s, "1\n") || strings.HasPrefix(s, "1\r\n") || strings.HasPrefix(s, "2\n") || strings.HasPrefix(s, "2\r\n") || strings.HasPrefix(s, "3\n") || strings.HasPrefix(s, "3\r\n") || strings.HasPrefix(s, "4\n") || strings.HasPrefix(s, "4\r\n")
 }
 
 // NewFromSRT parses a .srt text into Subtitle, assumes s is a clean utf8 string