From 89859369f9cb8b2656d713c4dbbf8babdaf7ddd8 Mon Sep 17 00:00:00 2001
From: InfiniteStash <117855276+InfiniteStash@users.noreply.github.com>
Date: Sun, 25 Feb 2024 22:56:44 +0000
Subject: [PATCH] Add phash duration cutoff option

---
 README.md                    | 1 +
 pkg/manager/config/config.go | 8 +++++++-
 pkg/scene/scene.go           | 8 ++++++++
 3 files changed, 16 insertions(+), 1 deletion(-)

diff --git a/README.md b/README.md
index 94e51d3a7..a0aeb9c8a 100644
--- a/README.md
+++ b/README.md
@@ -92,6 +92,7 @@ There are two ways to authenticate a user in Stash-box: a session or an API key.
 | `s3.max_dimension` | (none) | If set, a resized copy will be created for any image whose dimensions exceed this number. This copy will be served in place of the original. |
 | `s3.upload_headers` | (none) | A map of headers to send with each upload request. For example, DigitalOcean requires the `x-amz-acl` header to be set to `public-read` or it does not make the uploaded images available. |
 | `phash_distance` | 0 | Determines what binary distance is considered a match when querying with a pHash fingeprint. Using more than 8 is not recommended and may lead to large amounts of false positives. **Note**: The [pg-spgist_hamming extension](#phash-distance-matching) must be installed to use distance matching, otherwise you will get errors. |
+| `phash_duration_cutoff` | 0 | Optionally sets a minimum duration for pHash fingerprint submission, in seconds. Fingerprints below this cutoff will be quietly dropped. Set to 0 to disable cutoff. |
 | `favicon_path` | (none) | Location where favicons for linked sites should be stored. Leave empty to disable. |
 | `draft_time_limit` | (24h) | Time, in seconds, before a draft is deleted. |
 | `profiler_port` | 0 | Port on which to serve pprof output. Omit to disable entirely. |
diff --git a/pkg/manager/config/config.go b/pkg/manager/config/config.go
index 29bceb0b9..957a1bbc8 100644
--- a/pkg/manager/config/config.go
+++ b/pkg/manager/config/config.go
@@ -85,7 +85,8 @@ type config struct {
 		PostgresConfig `mapstructure:",squash"`
 	}
 
-	PHashDistance int `mapstructure:"phash_distance"`
+	PHashDistance       int `mapstructure:"phash_distance"`
+	PHashDurationCutoff int `mapstructure:"phash_duration_cutoff"`
 
 	Title string `mapstructure:"title"`
 
@@ -112,6 +113,7 @@ var C = &config{
 	EmailPort:                  25,
 	ImageBackend:               string(FileBackend),
 	PHashDistance:              0,
+	PHashDurationCutoff:        0,
 	VoteApplicationThreshold:   3,
 	VotePromotionThreshold:     10,
 	VoteCronInterval:           "5m",
@@ -280,6 +282,10 @@ func GetPHashDistance() int {
 	return C.PHashDistance
 }
 
+func GetPhashDurationCutoff() int {
+	return C.PHashDurationCutoff
+}
+
 func InitializeDefaults() error {
 	// generate some api keys
 	const apiKeyLength = 32
diff --git a/pkg/scene/scene.go b/pkg/scene/scene.go
index 040dabe36..6a1ec9b21 100644
--- a/pkg/scene/scene.go
+++ b/pkg/scene/scene.go
@@ -7,6 +7,7 @@ import (
 	"github.com/gofrs/uuid"
 	"github.com/pkg/errors"
 	"github.com/stashapp/stash-box/pkg/image"
+	"github.com/stashapp/stash-box/pkg/manager/config"
 	"github.com/stashapp/stash-box/pkg/models"
 	"github.com/stashapp/stash-box/pkg/user"
 )
@@ -235,6 +236,13 @@ func Destroy(fac models.Repo, input models.SceneDestroyInput) (bool, error) {
 func SubmitFingerprint(ctx context.Context, fac models.Repo, input models.FingerprintSubmission) (bool, error) {
 	qb := fac.Scene()
 
+	// Optionally filter out phashes below a certain duration.
+	// Phashes are inaccurate for short videos which can make them undesirable.
+	cutoff := config.GetPhashDurationCutoff()
+	if cutoff > 0 && input.Fingerprint.Algorithm == models.FingerprintAlgorithmPhash && cutoff >= input.Fingerprint.Duration {
+		return true, nil
+	}
+
 	// Find the scene
 	scene, err := qb.Find(input.SceneID)