refactor(processor): improve release comparing (#67)

* refactor(processor): improve release comparing

* chore(config): add compareRepackStatus to config

* refactor(processor): rewrote release comparer & added more logging

* chore(tests): adjust formatting tests

* fix(processor): log formatting

* fix(release): copy paste error

* fix(release): continue after condition not matching

* fix(processor): incorrect response codes

* chore(processor): clearer logging

* chore: add missing license header

* chore(processor): improve logging again

* feat(config): fuzzyMatching options

* chore: remove fuzzy matching debug logging

* chore(config): added default value info

* docs(readme): added info about fuzzy matching

* chore(release): update comments to reflect name changes

* fix(release): don't modify slice when comparing

* fix(slices): check for empty hdr slice

README.md

@@ -124,6 +124,22 @@ using the parsed folder name the files will be hardlinked into the exact folder
 You can take a look at the [Webhook](#webhook) section to see what you would need to add in your autobrr filter to
 make use of this feature.
 
+### Fuzzy Matching
+
+In this section, you can toggle comparing rules. I will explain each of them in more detail here.
+
+1. **skipRepackCompare**: When set to `true`, the comparer skips checking the repack status of the season pack release
+   against the episodes in your client. The episode in the example will only be accepted as a match by seasonpackarr if
+   you enable this option:
+   - Announce name: `Show.S01.1080p.WEB-DL.DDPA5.1.H.264-RlsGrp`
+   - Episode name: `Show.S01E01.1080p.WEB-DL.REPACK.DDPA5.1.H.264-RlsGrp`
+
+2. **simplifyHdrCompare**: If set to `true`, this option simplifies the HDR formats `HDR10`, `HDR10+`, and `HDR+` to
+   just `HDR`. This increases the likelihood of matching renamed releases that specify a more advanced HDR format in the
+   announce name than in the episode title:
+   - Announce name: `Show.S01.2160p.WEB-DL.DDPA5.1.DV.HDR10+.H.265-RlsGrp`
+   - Episode name: `Show.S01E01.2160p.WEB-DL.DDPA5.1.DV.HDR.H.265-RlsGrp`
+
 ## autobrr Filter setup
 
 Support for multiple Sonarr and qBittorrent instances with different pre import directories was added with v0.4.0, so

config.yaml

@@ -116,6 +116,24 @@ logLevel: "DEBUG"
 #
 # parseTorrentFile: false
 
+# Fuzzy Matching
+# You can decide for which criteria the matching should be less strict, e.g. repack status and HDR format
+#
+fuzzyMatching:
+  # Skip Repack Compare
+  # Toggle comparing of the repack status of a release, e.g. repacked episodes will be treated the same as a non-repacked ones
+  #
+  # Default: false
+  #
+  skipRepackCompare: false
+
+  # Simplify HDR Compare
+  # Toggle simplification of HDR formats for comparing, e.g. HDR10+ will be treated the same as HDR
+  #
+  # Default: false
+  #
+  simplifyHdrCompare: false
+
 # API Token
 # If not defined, removes api authentication
 #

internal/config/config.go

@@ -143,6 +143,24 @@ logLevel: "DEBUG"
 #
 # parseTorrentFile: false
 
+# Fuzzy Matching
+# You can decide for which criteria the matching should be less strict, e.g. repack status and HDR format
+#
+fuzzyMatching:
+  # Skip Repack Compare
+  # Toggle comparing of the repack status of a release, e.g. repacked episodes will be treated the same as a non-repacked ones
+  #
+  # Default: false
+  #
+  skipRepackCompare: false
+
+  # Simplify HDR Compare
+  # Toggle simplification of HDR formats for comparing, e.g. HDR10+ will be treated the same as HDR
+  #
+  # Default: false
+  #
+  simplifyHdrCompare: false
+
 # API Token
 # If not defined, removes api authentication
 #
@@ -277,7 +295,11 @@ func (c *AppConfig) defaults() {
 		SmartMode:          false,
 		SmartModeThreshold: 0.75,
 		ParseTorrentFile:   false,
-		APIToken:           "",
+		FuzzyMatching: domain.FuzzyMatching{
+			SkipRepackCompare:  false,
+			SimplifyHdrCompare: false,
+		},
+		APIToken: "",
 	}
 }
 

internal/domain/config.go

@@ -12,6 +12,11 @@ type Client struct {
 	PreImportPath string `yaml:"preImportPath"`
 }
 
+type FuzzyMatching struct {
+	SkipRepackCompare  bool `yaml:"skipRepackCompare"`
+	SimplifyHdrCompare bool `yaml:"simplifyHdrCompare"`
+}
+
 type Config struct {
 	Version            string
 	ConfigPath         string
@@ -25,5 +30,6 @@ type Config struct {
 	SmartMode          bool               `yaml:"smartMode"`
 	SmartModeThreshold float32            `yaml:"smartModeThreshold"`
 	ParseTorrentFile   bool               `yaml:"parseTorrentFile"`
+	FuzzyMatching      FuzzyMatching      `yaml:"fuzzyMatching"`
 	APIToken           string             `yaml:"apiToken"`
 }

internal/domain/release.go

@@ -0,0 +1,14 @@
+// Copyright (c) 2023 - 2024, nuxen and the seasonpackarr contributors.
+// SPDX-License-Identifier: GPL-2.0-or-later
+
+package domain
+
+import (
+	"github.com/autobrr/go-qbittorrent"
+	"github.com/moistari/rls"
+)
+
+type Entry struct {
+	T qbittorrent.Torrent
+	R rls.Release
+}

internal/http/processor.go

@@ -14,11 +14,14 @@ import (
 	"seasonpackarr/internal/config"
 	"seasonpackarr/internal/domain"
 	"seasonpackarr/internal/logger"
+	"seasonpackarr/internal/release"
 	"seasonpackarr/internal/utils"
 
 	"github.com/autobrr/go-qbittorrent"
 	"github.com/moistari/rls"
 	"github.com/rs/zerolog"
+
+	"golang.org/x/exp/slices"
 )
 
 type processor struct {
@@ -27,11 +30,6 @@ type processor struct {
 	req *request
 }
 
-type entry struct {
-	t qbittorrent.Torrent
-	r rls.Release
-}
-
 type request struct {
 	Name       string
 	Torrent    json.RawMessage
@@ -40,7 +38,7 @@ type request struct {
 }
 
 type entryTime struct {
-	e   map[string][]entry
+	e   map[string][]domain.Entry
 	d   map[string]rls.Release
 	t   time.Time
 	err error
@@ -125,7 +123,7 @@ func (p processor) getAllTorrents(client *domain.Client) entryTime {
 	}
 
 	nt := time.Now()
-	res = &entryTime{e: make(map[string][]entry), t: nt.Add(nt.Sub(cur)), d: res.d}
+	res = &entryTime{e: make(map[string][]domain.Entry), t: nt.Add(nt.Sub(cur)), d: res.d}
 
 	for _, t := range torrents {
 		r, ok := res.d[t.Name]
@@ -135,7 +133,7 @@ func (p processor) getAllTorrents(client *domain.Client) entryTime {
 		}
 
 		s := utils.GetFormattedTitle(r)
-		res.e[s] = append(res.e[s], entry{t: t, r: r})
+		res.e[s] = append(res.e[s], domain.Entry{T: t, R: r})
 	}
 
 	torrentMap.Store(set, res)
@@ -187,8 +185,8 @@ func (p processor) ProcessSeasonPack(w netHTTP.ResponseWriter, r *netHTTP.Reques
 		return
 	}
 
-	requestrls := entry{r: rls.ParseString(p.req.Name)}
-	v, ok := mp.e[utils.GetFormattedTitle(requestrls.r)]
+	requestrls := domain.Entry{R: rls.ParseString(p.req.Name)}
+	v, ok := mp.e[utils.GetFormattedTitle(requestrls.R)]
 	if !ok {
 		p.log.Info().Msgf("no matching releases in client %q: %q", clientName, p.req.Name)
 		netHTTP.Error(w, fmt.Sprintf("no matching releases in client %q: %q", clientName, p.req.Name), 200)
@@ -199,17 +197,66 @@ func (p processor) ProcessSeasonPack(w netHTTP.ResponseWriter, r *netHTTP.Reques
 	p.log.Debug().Msgf("formatted season pack name: %q", packDirName)
 
 	for _, child := range v {
-		if checkCandidates(&requestrls, &child) == 210 {
+		if release.CheckCandidates(&requestrls, &child, p.cfg.Config.FuzzyMatching) == 210 {
 			p.log.Info().Msgf("release already in client %q: %q", clientName, p.req.Name)
 			netHTTP.Error(w, fmt.Sprintf("release already in client %q: %q", clientName, p.req.Name), 210)
 			return
 		}
 	}
 
 	var matchedEpisodes []int
+	var responseCodes []int
 
 	for _, child := range v {
-		switch res := checkCandidates(&requestrls, &child); res {
+		switch res := release.CheckCandidates(&requestrls, &child, p.cfg.Config.FuzzyMatching); res {
+		case 201:
+			p.log.Info().Msgf("resolution did not match: request(%s => %s), client(%s => %s)",
+				requestrls.R.String(), requestrls.R.Resolution, child.R.String(), child.R.Resolution)
+			responseCodes = append(responseCodes, res)
+			continue
+
+		case 202:
+			p.log.Info().Msgf("source did not match: request(%s => %s), client(%s => %s)",
+				requestrls.R.String(), requestrls.R.Source, child.R.String(), child.R.Source)
+			responseCodes = append(responseCodes, res)
+			continue
+
+		case 203:
+			p.log.Info().Msgf("release group did not match: request(%s => %s), client(%s => %s)",
+				requestrls.R.String(), requestrls.R.Group, child.R.String(), child.R.Group)
+			responseCodes = append(responseCodes, res)
+			continue
+
+		case 204:
+			p.log.Info().Msgf("cut did not match: request(%s => %s), client(%s => %s)",
+				requestrls.R.String(), requestrls.R.Cut, child.R.String(), child.R.Cut)
+			responseCodes = append(responseCodes, res)
+			continue
+
+		case 205:
+			p.log.Info().Msgf("edition did not match: request(%s => %s), client(%s => %s)",
+				requestrls.R.String(), requestrls.R.Edition, child.R.String(), child.R.Edition)
+			responseCodes = append(responseCodes, res)
+			continue
+
+		case 206:
+			p.log.Info().Msgf("repack status did not match: request(%s => %s), client(%s => %s)",
+				requestrls.R.String(), requestrls.R.Other, child.R.String(), child.R.Other)
+			responseCodes = append(responseCodes, res)
+			continue
+
+		case 207:
+			p.log.Info().Msgf("hdr metadata did not match: request(%s => %s), client(%s => %s)",
+				requestrls.R.String(), requestrls.R.HDR, child.R.String(), child.R.HDR)
+			responseCodes = append(responseCodes, res)
+			continue
+
+		case 208:
+			p.log.Info().Msgf("streaming service did not match: request(%s => %s), client(%s => %s)",
+				requestrls.R.String(), requestrls.R.Collection, child.R.String(), child.R.Collection)
+			responseCodes = append(responseCodes, res)
+			continue
+
 		case 210:
 			p.log.Info().Msgf("release already in client %q: %q", clientName, p.req.Name)
 			netHTTP.Error(w, fmt.Sprintf("release already in client %q: %q", clientName, p.req.Name), res)
@@ -221,9 +268,9 @@ func (p processor) ProcessSeasonPack(w netHTTP.ResponseWriter, r *netHTTP.Reques
 			return
 
 		case 250:
-			m, err := p.getFiles(child.t.Hash)
+			m, err := p.getFiles(child.T.Hash)
 			if err != nil {
-				p.log.Error().Err(err).Msgf("error getting files: %q", child.t.Name)
+				p.log.Error().Err(err).Msgf("error getting files: %q", child.T.Name)
 				continue
 			}
 
@@ -233,8 +280,8 @@ func (p processor) ProcessSeasonPack(w netHTTP.ResponseWriter, r *netHTTP.Reques
 				break
 			}
 
-			episodeRls := rls.ParseString(child.t.Name)
-			episodePath := filepath.Join(child.t.SavePath, fileName)
+			episodeRls := rls.ParseString(child.T.Name)
+			episodePath := filepath.Join(child.T.SavePath, fileName)
 			packPath := filepath.Join(client.PreImportPath, packDirName, filepath.Base(fileName))
 
 			matchedEpisodes = append(matchedEpisodes, episodeRls.Episode)
@@ -253,11 +300,18 @@ func (p processor) ProcessSeasonPack(w netHTTP.ResponseWriter, r *netHTTP.Reques
 
 			newMatches := append(oldMatches.([]matchPaths), currentMatch...)
 			matchesMap.Store(p.req.Name, newMatches)
-			p.log.Debug().Msgf("matched torrent from client %q: %q %q", clientName, child.t.Name, child.t.Hash)
+			p.log.Debug().Msgf("matched torrent from client %q: %q %q", clientName, child.T.Name, child.T.Hash)
+			responseCodes = append(responseCodes, res)
 			continue
 		}
 	}
 
+	if !slices.Contains(responseCodes, 250) {
+		p.log.Info().Msgf("no matching releases in client %q: %q", clientName, p.req.Name)
+		netHTTP.Error(w, fmt.Sprintf("no matching releases in client %q: %q", clientName, p.req.Name), 200)
+		return
+	}
+
 	if matchesSlice, ok := matchesMap.Load(p.req.Name); ok {
 		if p.cfg.Config.SmartMode {
 			reqRls := rls.ParseString(p.req.Name)
@@ -270,7 +324,7 @@ func (p processor) ProcessSeasonPack(w netHTTP.ResponseWriter, r *netHTTP.Reques
 			}
 			matchedEpisodes = utils.DedupeSlice(matchedEpisodes)
 
-			percentEpisodes := percentOfTotalEpisodes(totalEpisodes, matchedEpisodes)
+			percentEpisodes := release.PercentOfTotalEpisodes(totalEpisodes, matchedEpisodes)
 
 			if percentEpisodes < p.cfg.Config.SmartModeThreshold {
 				// delete match from matchesMap if threshold is not met
@@ -359,34 +413,3 @@ func (p processor) ParseTorrent(w netHTTP.ResponseWriter, r *netHTTP.Request) {
 		netHTTP.Error(w, fmt.Sprintf("created hardlink of %q into %q", match.episodePath, match.packPath), 250)
 	}
 }
-
-func checkCandidates(requestrls, child *entry) int {
-	rlsRelease := requestrls.r
-	rlsInClient := child.r
-
-	// check if season pack and no extension
-	if fmt.Sprint(rlsRelease.Type) == "series" && rlsRelease.Ext == "" {
-		// compare formatted titles
-		if utils.GetFormattedTitle(rlsInClient) == utils.GetFormattedTitle(rlsRelease) {
-			// check if same episode
-			if rlsInClient.Episode == rlsRelease.Episode {
-				// release is already in client
-				return 210
-			}
-			// season pack with matching episodes
-			return 250
-		}
-	}
-	// not a season pack
-	return 211
-}
-
-func percentOfTotalEpisodes(totalEpisodes int, matchedEpisodes []int) float32 {
-	if totalEpisodes == 0 {
-		return 0
-	}
-	count := len(matchedEpisodes)
-	percent := float32(count) / float32(totalEpisodes)
-
-	return percent
-}