initial commit

go.mod

@@ -0,0 +1,5 @@
+module git.milar.in/animan/parsers
+
+go 1.18
+
+require git.milar.in/animan/model v0.0.0-20220803215306-905563410463

go.sum

@@ -0,0 +1,2 @@
+git.milar.in/animan/model v0.0.0-20220803215306-905563410463 h1:M7Xxqbv7WQ3xwW66APwI7KQ8wD8IN/v6ZZbq1JB1T2M=
+git.milar.in/animan/model v0.0.0-20220803215306-905563410463/go.mod h1:FmgisUVGJd8EEvGP4uk9pEv6Ic9aiEhAy5eZF8zZsgU=

lang_parser_funcs.go

@@ -0,0 +1,11 @@
+package parsers
+
+import "regexp"
+
+type LanguageParserFunc func(str string) []string
+
+var SquareBracketsLanguageParserRegex = regexp.MustCompile(`\[.+?\]`)
+
+func SquareBracketsLanguageParser(str string) []string {
+	return SquareBracketsLanguageParserRegex.FindAllString(str, -1)
+}

parsers.go

@@ -0,0 +1,25 @@
+package parsers
+
+import "git.milar.in/animan/model"
+
+/*
+how to get all torrent names on a nyaa page:
+
+let s = "";
+document.querySelectorAll("tr > td:nth-child(2) > a:not(.comments)").forEach((element, index) => {
+  s += element.textContent + "\n";
+})
+console.log(s);
+
+*/
+
+var Parsers = []model.Parser{
+	{
+		Identity: "Erai-Raws",
+		TorrentParser: torrentNameEpLangsRegex(
+			`^\[Erai-raws\] (.*) - (.*?) \[1080p\](?:\[HEVC\])?(?:\[Multiple Subtitle\])? (\[.*?\]+)?$`,
+			SquareBracketsLanguageParser,
+		),
+		FileParser: nil,
+	},
+}

torrent_parser_funcs.go

@@ -0,0 +1,58 @@
+package parsers
+
+import (
+	"regexp"
+
+	"git.milar.in/animan/model"
+)
+
+// torrentNameEpRegex returns a TorrentParserFunc with the following regex groups:
+// 1: name
+// 2: episode
+func torrentNameEpRegex(regex string) model.TorrentParserFunc {
+	pattern := regexp.MustCompile(regex)
+
+	return func(parser *model.Parser, torrent *model.Torrent) (ParsedTorrent *model.ParsedTorrent, ok bool) {
+		matches := pattern.FindStringSubmatch(torrent.Title)
+		if len(matches) <= 2 {
+			return nil, false
+		}
+
+		name := matches[1]
+		episode, ok := atoi(matches[2])
+
+		return &model.ParsedTorrent{
+			Torrent: torrent,
+			Name:    name,
+			Episode: episode,
+			Parser:  parser,
+		}, ok
+	}
+}
+
+// torrentNameEpRegex returns a TorrentParserFunc with the following regex groups:
+// 1: name
+// 2: episode
+// 3: languages
+func torrentNameEpLangsRegex(regex string, languageParser LanguageParserFunc) model.TorrentParserFunc {
+	pattern := regexp.MustCompile(regex)
+
+	return func(parser *model.Parser, torrent *model.Torrent) (ParsedTorrent *model.ParsedTorrent, ok bool) {
+		matches := pattern.FindStringSubmatch(torrent.Title)
+		if len(matches) <= 2 {
+			return nil, false
+		}
+
+		name := matches[1]
+		episode, ok := atoi(matches[2])
+		languages := languageParser(matches[3])
+
+		return &model.ParsedTorrent{
+			Torrent:   torrent,
+			Name:      name,
+			Episode:   episode,
+			Languages: languages,
+			Parser:    parser,
+		}, ok
+	}
+}

utils.go

@@ -0,0 +1,11 @@
+package parsers
+
+import "strconv"
+
+func atoi(s string) (int, bool) {
+	v, err := strconv.Atoi(s)
+	if err != nil {
+		return 0, false
+	}
+	return v, true
+}