feat: use pgvecto.rs (#3605)

2025-12-25 01:11:43 +03:00 · 2023-12-08 11:15:46 -05:00
parent 429ad28810
commit 1e99ba8167
99 changed files with 1935 additions and 2583 deletions
--- a/server/src/infra/sql/asset.repository.sql
+++ b/server/src/infra/sql/asset.repository.sql
@@ -57,8 +57,7 @@ SELECT
  "AssetEntity__AssetEntity_exifInfo"."profileDescription" AS "AssetEntity__AssetEntity_exifInfo_profileDescription",
  "AssetEntity__AssetEntity_exifInfo"."colorspace" AS "AssetEntity__AssetEntity_exifInfo_colorspace",
  "AssetEntity__AssetEntity_exifInfo"."bitsPerSample" AS "AssetEntity__AssetEntity_exifInfo_bitsPerSample",
-  "AssetEntity__AssetEntity_exifInfo"."fps" AS "AssetEntity__AssetEntity_exifInfo_fps",
-  "AssetEntity__AssetEntity_exifInfo"."exifTextSearchableColumn" AS "AssetEntity__AssetEntity_exifInfo_exifTextSearchableColumn"
+  "AssetEntity__AssetEntity_exifInfo"."fps" AS "AssetEntity__AssetEntity_exifInfo_fps"
 FROM
  "assets" "AssetEntity"
  LEFT JOIN "exif" "AssetEntity__AssetEntity_exifInfo" ON "AssetEntity__AssetEntity_exifInfo"."assetId" = "AssetEntity"."id"
@@ -133,8 +132,7 @@ SELECT
  "exifInfo"."profileDescription" AS "exifInfo_profileDescription",
  "exifInfo"."colorspace" AS "exifInfo_colorspace",
  "exifInfo"."bitsPerSample" AS "exifInfo_bitsPerSample",
-  "exifInfo"."fps" AS "exifInfo_fps",
-  "exifInfo"."exifTextSearchableColumn" AS "exifInfo_exifTextSearchableColumn"
+  "exifInfo"."fps" AS "exifInfo_fps"
 FROM
  "assets" "entity"
  LEFT JOIN "exif" "exifInfo" ON "exifInfo"."assetId" = "entity"."id"
@@ -217,11 +215,9 @@ SELECT
  "AssetEntity__AssetEntity_exifInfo"."colorspace" AS "AssetEntity__AssetEntity_exifInfo_colorspace",
  "AssetEntity__AssetEntity_exifInfo"."bitsPerSample" AS "AssetEntity__AssetEntity_exifInfo_bitsPerSample",
  "AssetEntity__AssetEntity_exifInfo"."fps" AS "AssetEntity__AssetEntity_exifInfo_fps",
-  "AssetEntity__AssetEntity_exifInfo"."exifTextSearchableColumn" AS "AssetEntity__AssetEntity_exifInfo_exifTextSearchableColumn",
  "AssetEntity__AssetEntity_smartInfo"."assetId" AS "AssetEntity__AssetEntity_smartInfo_assetId",
  "AssetEntity__AssetEntity_smartInfo"."tags" AS "AssetEntity__AssetEntity_smartInfo_tags",
  "AssetEntity__AssetEntity_smartInfo"."objects" AS "AssetEntity__AssetEntity_smartInfo_objects",
-  "AssetEntity__AssetEntity_smartInfo"."clipEmbedding" AS "AssetEntity__AssetEntity_smartInfo_clipEmbedding",
  "AssetEntity__AssetEntity_tags"."id" AS "AssetEntity__AssetEntity_tags_id",
  "AssetEntity__AssetEntity_tags"."type" AS "AssetEntity__AssetEntity_tags_type",
  "AssetEntity__AssetEntity_tags"."name" AS "AssetEntity__AssetEntity_tags_name",
@@ -230,7 +226,6 @@ SELECT
  "AssetEntity__AssetEntity_faces"."id" AS "AssetEntity__AssetEntity_faces_id",
  "AssetEntity__AssetEntity_faces"."assetId" AS "AssetEntity__AssetEntity_faces_assetId",
  "AssetEntity__AssetEntity_faces"."personId" AS "AssetEntity__AssetEntity_faces_personId",
-  "AssetEntity__AssetEntity_faces"."embedding" AS "AssetEntity__AssetEntity_faces_embedding",
  "AssetEntity__AssetEntity_faces"."imageWidth" AS "AssetEntity__AssetEntity_faces_imageWidth",
  "AssetEntity__AssetEntity_faces"."imageHeight" AS "AssetEntity__AssetEntity_faces_imageHeight",
  "AssetEntity__AssetEntity_faces"."boundingBoxX1" AS "AssetEntity__AssetEntity_faces_boundingBoxX1",
@@ -439,7 +434,6 @@ FROM
      "AssetEntity__AssetEntity_faces"."id" AS "AssetEntity__AssetEntity_faces_id",
      "AssetEntity__AssetEntity_faces"."assetId" AS "AssetEntity__AssetEntity_faces_assetId",
      "AssetEntity__AssetEntity_faces"."personId" AS "AssetEntity__AssetEntity_faces_personId",
-      "AssetEntity__AssetEntity_faces"."embedding" AS "AssetEntity__AssetEntity_faces_embedding",
      "AssetEntity__AssetEntity_faces"."imageWidth" AS "AssetEntity__AssetEntity_faces_imageWidth",
      "AssetEntity__AssetEntity_faces"."imageHeight" AS "AssetEntity__AssetEntity_faces_imageHeight",
      "AssetEntity__AssetEntity_faces"."boundingBoxX1" AS "AssetEntity__AssetEntity_faces_boundingBoxX1",
@@ -612,3 +606,73 @@ ORDER BY
  "AssetEntity"."createdAt" ASC
 LIMIT
  11
+
+-- AssetRepository.getAssetIdByCity
+WITH
+  "cities" AS (
+    SELECT
+      city
+    FROM
+      "exif" "e"
+    GROUP BY
+      city
+    HAVING
+      count(city) >= $1
+    ORDER BY
+      random() ASC
+    LIMIT
+      12
+  )
+SELECT DISTINCT
+  ON (c.city) "asset"."id" AS "data",
+  c.city AS "value"
+FROM
+  "assets" "asset"
+  INNER JOIN "exif" "e" ON "asset"."id" = e."assetId"
+  INNER JOIN "cities" "c" ON c.city = "e"."city"
+WHERE
+  (
+    "asset"."isVisible" = true
+    AND "asset"."fileCreatedAt" < NOW()
+    AND "asset"."type" = $2
+    AND "asset"."ownerId" IN ($3)
+    AND "asset"."isArchived" = $4
+  )
+  AND ("asset"."deletedAt" IS NULL)
+LIMIT
+  12
+
+-- AssetRepository.getAssetIdByTag
+WITH
+  "random_tags" AS (
+    SELECT
+      unnest(tags) AS "tag"
+    FROM
+      "smart_info" "si"
+    GROUP BY
+      tag
+    HAVING
+      count(*) >= $1
+    ORDER BY
+      random() ASC
+    LIMIT
+      12
+  )
+SELECT DISTINCT
+  ON (unnest("si"."tags")) "asset"."id" AS "data",
+  unnest("si"."tags") AS "value"
+FROM
+  "assets" "asset"
+  INNER JOIN "smart_info" "si" ON "asset"."id" = si."assetId"
+  INNER JOIN "random_tags" "t" ON "si"."tags" @> ARRAY[t.tag]
+WHERE
+  (
+    "asset"."isVisible" = true
+    AND "asset"."fileCreatedAt" < NOW()
+    AND "asset"."type" = $2
+    AND "asset"."ownerId" IN ($3)
+    AND "asset"."isArchived" = $4
+  )
+  AND ("asset"."deletedAt" IS NULL)
+LIMIT
+  12
--- a/server/src/infra/sql/person.repository.sql
+++ b/server/src/infra/sql/person.repository.sql
@@ -12,7 +12,6 @@ SELECT
  "AssetFaceEntity"."id" AS "AssetFaceEntity_id",
  "AssetFaceEntity"."assetId" AS "AssetFaceEntity_assetId",
  "AssetFaceEntity"."personId" AS "AssetFaceEntity_personId",
-  "AssetFaceEntity"."embedding" AS "AssetFaceEntity_embedding",
  "AssetFaceEntity"."imageWidth" AS "AssetFaceEntity_imageWidth",
  "AssetFaceEntity"."imageHeight" AS "AssetFaceEntity_imageHeight",
  "AssetFaceEntity"."boundingBoxX1" AS "AssetFaceEntity_boundingBoxX1",
@@ -138,7 +137,6 @@ SELECT
  "AssetFaceEntity"."id" AS "AssetFaceEntity_id",
  "AssetFaceEntity"."assetId" AS "AssetFaceEntity_assetId",
  "AssetFaceEntity"."personId" AS "AssetFaceEntity_personId",
-  "AssetFaceEntity"."embedding" AS "AssetFaceEntity_embedding",
  "AssetFaceEntity"."imageWidth" AS "AssetFaceEntity_imageWidth",
  "AssetFaceEntity"."imageHeight" AS "AssetFaceEntity_imageHeight",
  "AssetFaceEntity"."boundingBoxX1" AS "AssetFaceEntity_boundingBoxX1",
@@ -169,7 +167,6 @@ FROM
      "AssetFaceEntity"."id" AS "AssetFaceEntity_id",
      "AssetFaceEntity"."assetId" AS "AssetFaceEntity_assetId",
      "AssetFaceEntity"."personId" AS "AssetFaceEntity_personId",
-      "AssetFaceEntity"."embedding" AS "AssetFaceEntity_embedding",
      "AssetFaceEntity"."imageWidth" AS "AssetFaceEntity_imageWidth",
      "AssetFaceEntity"."imageHeight" AS "AssetFaceEntity_imageHeight",
      "AssetFaceEntity"."boundingBoxX1" AS "AssetFaceEntity_boundingBoxX1",
@@ -205,7 +202,6 @@ FROM
      "AssetFaceEntity"."id" AS "AssetFaceEntity_id",
      "AssetFaceEntity"."assetId" AS "AssetFaceEntity_assetId",
      "AssetFaceEntity"."personId" AS "AssetFaceEntity_personId",
-      "AssetFaceEntity"."embedding" AS "AssetFaceEntity_embedding",
      "AssetFaceEntity"."imageWidth" AS "AssetFaceEntity_imageWidth",
      "AssetFaceEntity"."imageHeight" AS "AssetFaceEntity_imageHeight",
      "AssetFaceEntity"."boundingBoxX1" AS "AssetFaceEntity_boundingBoxX1",
@@ -351,7 +347,6 @@ FROM
      "AssetEntity__AssetEntity_faces"."id" AS "AssetEntity__AssetEntity_faces_id",
      "AssetEntity__AssetEntity_faces"."assetId" AS "AssetEntity__AssetEntity_faces_assetId",
      "AssetEntity__AssetEntity_faces"."personId" AS "AssetEntity__AssetEntity_faces_personId",
-      "AssetEntity__AssetEntity_faces"."embedding" AS "AssetEntity__AssetEntity_faces_embedding",
      "AssetEntity__AssetEntity_faces"."imageWidth" AS "AssetEntity__AssetEntity_faces_imageWidth",
      "AssetEntity__AssetEntity_faces"."imageHeight" AS "AssetEntity__AssetEntity_faces_imageHeight",
      "AssetEntity__AssetEntity_faces"."boundingBoxX1" AS "AssetEntity__AssetEntity_faces_boundingBoxX1",
@@ -393,8 +388,7 @@ FROM
      "AssetEntity__AssetEntity_exifInfo"."profileDescription" AS "AssetEntity__AssetEntity_exifInfo_profileDescription",
      "AssetEntity__AssetEntity_exifInfo"."colorspace" AS "AssetEntity__AssetEntity_exifInfo_colorspace",
      "AssetEntity__AssetEntity_exifInfo"."bitsPerSample" AS "AssetEntity__AssetEntity_exifInfo_bitsPerSample",
-      "AssetEntity__AssetEntity_exifInfo"."fps" AS "AssetEntity__AssetEntity_exifInfo_fps",
-      "AssetEntity__AssetEntity_exifInfo"."exifTextSearchableColumn" AS "AssetEntity__AssetEntity_exifInfo_exifTextSearchableColumn"
+      "AssetEntity__AssetEntity_exifInfo"."fps" AS "AssetEntity__AssetEntity_exifInfo_fps"
    FROM
      "assets" "AssetEntity"
      LEFT JOIN "asset_faces" "AssetEntity__AssetEntity_faces" ON "AssetEntity__AssetEntity_faces"."assetId" = "AssetEntity"."id"
@@ -421,7 +415,6 @@ SELECT
  "AssetFaceEntity"."id" AS "AssetFaceEntity_id",
  "AssetFaceEntity"."assetId" AS "AssetFaceEntity_assetId",
  "AssetFaceEntity"."personId" AS "AssetFaceEntity_personId",
-  "AssetFaceEntity"."embedding" AS "AssetFaceEntity_embedding",
  "AssetFaceEntity"."imageWidth" AS "AssetFaceEntity_imageWidth",
  "AssetFaceEntity"."imageHeight" AS "AssetFaceEntity_imageHeight",
  "AssetFaceEntity"."boundingBoxX1" AS "AssetFaceEntity_boundingBoxX1",
@@ -473,7 +466,6 @@ SELECT
  "AssetFaceEntity"."id" AS "AssetFaceEntity_id",
  "AssetFaceEntity"."assetId" AS "AssetFaceEntity_assetId",
  "AssetFaceEntity"."personId" AS "AssetFaceEntity_personId",
-  "AssetFaceEntity"."embedding" AS "AssetFaceEntity_embedding",
  "AssetFaceEntity"."imageWidth" AS "AssetFaceEntity_imageWidth",
  "AssetFaceEntity"."imageHeight" AS "AssetFaceEntity_imageHeight",
  "AssetFaceEntity"."boundingBoxX1" AS "AssetFaceEntity_boundingBoxX1",
--- a/server/src/infra/sql/shared.link.repository.sql
+++ b/server/src/infra/sql/shared.link.repository.sql
@@ -77,7 +77,6 @@ FROM
      "9b1d35b344d838023994a3233afd6ffe098be6d8"."colorspace" AS "9b1d35b344d838023994a3233afd6ffe098be6d8_colorspace",
      "9b1d35b344d838023994a3233afd6ffe098be6d8"."bitsPerSample" AS "9b1d35b344d838023994a3233afd6ffe098be6d8_bitsPerSample",
      "9b1d35b344d838023994a3233afd6ffe098be6d8"."fps" AS "9b1d35b344d838023994a3233afd6ffe098be6d8_fps",
-      "9b1d35b344d838023994a3233afd6ffe098be6d8"."exifTextSearchableColumn" AS "e18de9deffa83f81ac3c43b5e8c2f08dba727bf8",
      "SharedLinkEntity__SharedLinkEntity_album"."id" AS "SharedLinkEntity__SharedLinkEntity_album_id",
      "SharedLinkEntity__SharedLinkEntity_album"."ownerId" AS "SharedLinkEntity__SharedLinkEntity_album_ownerId",
      "SharedLinkEntity__SharedLinkEntity_album"."albumName" AS "SharedLinkEntity__SharedLinkEntity_album_albumName",
@@ -143,7 +142,6 @@ FROM
      "d9f2f4dd8920bad1d6907cdb1d699732daff3c2f"."colorspace" AS "d9f2f4dd8920bad1d6907cdb1d699732daff3c2f_colorspace",
      "d9f2f4dd8920bad1d6907cdb1d699732daff3c2f"."bitsPerSample" AS "d9f2f4dd8920bad1d6907cdb1d699732daff3c2f_bitsPerSample",
      "d9f2f4dd8920bad1d6907cdb1d699732daff3c2f"."fps" AS "d9f2f4dd8920bad1d6907cdb1d699732daff3c2f_fps",
-      "d9f2f4dd8920bad1d6907cdb1d699732daff3c2f"."exifTextSearchableColumn" AS "96535c8046de591cca9b8c5825e6c5db502b0e6a",
      "6d7fd45329a05fd86b3dbcacde87fe76e33a422d"."id" AS "6d7fd45329a05fd86b3dbcacde87fe76e33a422d_id",
      "6d7fd45329a05fd86b3dbcacde87fe76e33a422d"."name" AS "6d7fd45329a05fd86b3dbcacde87fe76e33a422d_name",
      "6d7fd45329a05fd86b3dbcacde87fe76e33a422d"."avatarColor" AS "6d7fd45329a05fd86b3dbcacde87fe76e33a422d_avatarColor",
--- a/server/src/infra/sql/smart.info.repository.sql
+++ b/server/src/infra/sql/smart.info.repository.sql
@@ -0,0 +1,111 @@
+-- NOTE: This file is auto generated by ./sql-generator
+
+-- SmartInfoRepository.searchCLIP
+START TRANSACTION
+SET
+  LOCAL vectors.k = '100'
+SELECT
+  "a"."id" AS "a_id",
+  "a"."deviceAssetId" AS "a_deviceAssetId",
+  "a"."ownerId" AS "a_ownerId",
+  "a"."libraryId" AS "a_libraryId",
+  "a"."deviceId" AS "a_deviceId",
+  "a"."type" AS "a_type",
+  "a"."originalPath" AS "a_originalPath",
+  "a"."resizePath" AS "a_resizePath",
+  "a"."webpPath" AS "a_webpPath",
+  "a"."thumbhash" AS "a_thumbhash",
+  "a"."encodedVideoPath" AS "a_encodedVideoPath",
+  "a"."createdAt" AS "a_createdAt",
+  "a"."updatedAt" AS "a_updatedAt",
+  "a"."deletedAt" AS "a_deletedAt",
+  "a"."fileCreatedAt" AS "a_fileCreatedAt",
+  "a"."localDateTime" AS "a_localDateTime",
+  "a"."fileModifiedAt" AS "a_fileModifiedAt",
+  "a"."isFavorite" AS "a_isFavorite",
+  "a"."isArchived" AS "a_isArchived",
+  "a"."isExternal" AS "a_isExternal",
+  "a"."isReadOnly" AS "a_isReadOnly",
+  "a"."isOffline" AS "a_isOffline",
+  "a"."checksum" AS "a_checksum",
+  "a"."duration" AS "a_duration",
+  "a"."isVisible" AS "a_isVisible",
+  "a"."livePhotoVideoId" AS "a_livePhotoVideoId",
+  "a"."originalFileName" AS "a_originalFileName",
+  "a"."sidecarPath" AS "a_sidecarPath",
+  "a"."stackParentId" AS "a_stackParentId",
+  "e"."assetId" AS "e_assetId",
+  "e"."description" AS "e_description",
+  "e"."exifImageWidth" AS "e_exifImageWidth",
+  "e"."exifImageHeight" AS "e_exifImageHeight",
+  "e"."fileSizeInByte" AS "e_fileSizeInByte",
+  "e"."orientation" AS "e_orientation",
+  "e"."dateTimeOriginal" AS "e_dateTimeOriginal",
+  "e"."modifyDate" AS "e_modifyDate",
+  "e"."timeZone" AS "e_timeZone",
+  "e"."latitude" AS "e_latitude",
+  "e"."longitude" AS "e_longitude",
+  "e"."projectionType" AS "e_projectionType",
+  "e"."city" AS "e_city",
+  "e"."livePhotoCID" AS "e_livePhotoCID",
+  "e"."state" AS "e_state",
+  "e"."country" AS "e_country",
+  "e"."make" AS "e_make",
+  "e"."model" AS "e_model",
+  "e"."lensModel" AS "e_lensModel",
+  "e"."fNumber" AS "e_fNumber",
+  "e"."focalLength" AS "e_focalLength",
+  "e"."iso" AS "e_iso",
+  "e"."exposureTime" AS "e_exposureTime",
+  "e"."profileDescription" AS "e_profileDescription",
+  "e"."colorspace" AS "e_colorspace",
+  "e"."bitsPerSample" AS "e_bitsPerSample",
+  "e"."fps" AS "e_fps"
+FROM
+  "assets" "a"
+  INNER JOIN "smart_search" "s" ON "s"."assetId" = "a"."id"
+  LEFT JOIN "exif" "e" ON "e"."assetId" = "a"."id"
+WHERE
+  ("a"."ownerId" = $1)
+  AND ("a"."deletedAt" IS NULL)
+ORDER BY
+  "s"."embedding" <= > $2 ASC
+LIMIT
+  100
+COMMIT
+
+-- SmartInfoRepository.searchFaces
+START TRANSACTION
+SET
+  LOCAL vectors.k = '100'
+WITH
+  "cte" AS (
+    SELECT
+      "faces"."id" AS "faces_id",
+      "faces"."assetId" AS "faces_assetId",
+      "faces"."personId" AS "faces_personId",
+      "faces"."imageWidth" AS "faces_imageWidth",
+      "faces"."imageHeight" AS "faces_imageHeight",
+      "faces"."boundingBoxX1" AS "faces_boundingBoxX1",
+      "faces"."boundingBoxY1" AS "faces_boundingBoxY1",
+      "faces"."boundingBoxX2" AS "faces_boundingBoxX2",
+      "faces"."boundingBoxY2" AS "faces_boundingBoxY2",
+      1 + ("faces"."embedding" <= > $1) AS "distance"
+    FROM
+      "asset_faces" "faces"
+      INNER JOIN "assets" "asset" ON "asset"."id" = "faces"."assetId"
+      AND ("asset"."deletedAt" IS NULL)
+    WHERE
+      "asset"."ownerId" = $2
+    ORDER BY
+      "faces"."embedding" <= > $3 ASC
+    LIMIT
+      100
+  )
+SELECT
+  res.*
+FROM
+  "cte" "res"
+WHERE
+  res.distance <= $4
+COMMIT