parser refactored

This commit is contained in:
Timon Ringwald 2022-08-04 18:13:41 +02:00
parent b516177d7c
commit 2a3194c37e
6 changed files with 90 additions and 64 deletions

8
errors.go Normal file
View File

@ -0,0 +1,8 @@
package parsers
import "git.milar.in/milarin/adverr"
var (
ErrTorrentParserInsufficientData = adverr.NewErrTmpl("ErrTorrentParserInsufficientData", "regex '%s' must at least provide title and episode")
ErrTorrentParserInvalidGroupReference = adverr.NewErrTmpl("ErrTorrentParserInvalidGroupReference", "options references group %d but regex only has %d groups")
)

7
go.mod
View File

@ -2,4 +2,9 @@ module git.milar.in/animan/parsers
go 1.18
require git.milar.in/animan/model v0.0.0-20220803215306-905563410463
require (
git.milar.in/animan/model v0.0.0-20220804154959-f8b66e8c9eda
git.milar.in/milarin/adverr v0.2.1
)
require git.milar.in/milarin/anilist v1.5.0 // indirect

8
go.sum
View File

@ -1,2 +1,6 @@
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=
git.milar.in/animan/model v0.0.0-20220804154959-f8b66e8c9eda h1:9XG7Dn+UtyLwwNcuol10EdHcBQjYD+1WzmnJbRUJQfs=
git.milar.in/animan/model v0.0.0-20220804154959-f8b66e8c9eda/go.mod h1:oXDr3slnzXoccIrci2wotX0cWwMVuAQ9dSvmh1buE4c=
git.milar.in/milarin/adverr v0.2.1 h1:eyXFGC+Ui/kcNt2+NqP3HiAplwxzqeNr9DfitsUb3c4=
git.milar.in/milarin/adverr v0.2.1/go.mod h1:wwfglcey4R3vqjNL/d8mbnvFJGzETRXzAEolIHZY32w=
git.milar.in/milarin/anilist v1.5.0 h1:fSiAXY/topNk4ISEp2QtcG9HHKLJfMc8w05iqc+Paf0=
git.milar.in/milarin/anilist v1.5.0/go.mod h1:8PTHXFMA45JpfRFIpcdrKwDHue8fbT/wwV1BuHFn6c0=

View File

@ -16,9 +16,15 @@ console.log(s);
var Parsers = []model.Parser{
{
Identity: "Erai-Raws",
TorrentParser: torrentNameEpLangsRegex(
`^\[Erai-raws\] (.*) - (.*?) \[1080p\](?:\[HEVC\])?(?:\[Multiple Subtitle\])? (\[.*?\]+)?$`,
SquareBracketsLanguageParser,
TorrentParser: regexTorrentParser(
`^\[Erai-raws\] (.*) - (.*?) \[(.*?)p\](?:\[HEVC\])?(?:\[Multiple Subtitle\])? (\[.*?\]+)?$`,
TorrentParseOptions{
Name: 1,
Episode: 2,
Resolution: 3,
Languages: 4,
LanguageParser: SquareBracketsLanguageParser,
},
),
FileParser: nil,
},

View File

@ -1,58 +0,0 @@
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
}
}

61
torrent_parsing.go Normal file
View File

@ -0,0 +1,61 @@
package parsers
import (
"regexp"
"git.milar.in/animan/model"
)
// TorrentParseOptions holds the subgroup index in which information can be found in a given regex
// as well as some parser specific functions
type TorrentParseOptions struct {
Name int
Episode int
Languages int
Subtitles int
Resolution int
LanguageParser LanguageParserFunc
}
func regexTorrentParser(regex string, options TorrentParseOptions) model.TorrentParserFunc {
pattern := regexp.MustCompile(regex)
if options.Name == 0 || options.Episode == 0 {
panic(ErrTorrentParserInsufficientData.New(regex))
}
for _, g := range []int{options.Name, options.Episode, options.Languages, options.Subtitles, options.Resolution} {
if g > pattern.NumSubexp() {
panic(ErrTorrentParserInvalidGroupReference.New(g, pattern.NumSubexp()))
}
}
return func(parser *model.Parser, torrent *model.Torrent) (ParsedTorrent *model.ParsedTorrent, ok bool) {
matches := pattern.FindStringSubmatch(torrent.Title)
if matches == nil {
return nil, false
}
episode, ok := atoi(matches[options.Episode])
if !ok {
return nil, false
}
resolution, err := model.ParseResolution(matches[options.Resolution])
if err != nil {
return nil, false
}
return &model.ParsedTorrent{
Title: matches[options.Name],
Episode: episode,
Resolution: resolution,
Parser: parser,
Languages: options.LanguageParser(matches[options.Languages]),
Torrent: torrent,
}, true
}
}