From 76109c5c6b2ff7746601fd0c6a02a159b7de5356 Mon Sep 17 00:00:00 2001
From: Ryan Gregg <github@ryangregg.com>
Date: Wed, 8 Oct 2025 21:39:18 -0700
Subject: [PATCH] Add additional dedupe logic

---
 web/src/lib/utils/duplicate-utils.spec.ts | 16 ++++
 web/src/lib/utils/duplicate-utils.ts      | 94 ++++++++++++++++++++---
 2 files changed, 98 insertions(+), 12 deletions(-)

diff --git a/web/src/lib/utils/duplicate-utils.spec.ts b/web/src/lib/utils/duplicate-utils.spec.ts
index 4fa427989a..5937cd6dfb 100644
--- a/web/src/lib/utils/duplicate-utils.spec.ts
+++ b/web/src/lib/utils/duplicate-utils.spec.ts
@@ -11,6 +11,22 @@ describe('choosing a duplicate', () => {
     expect(suggestDuplicate(assets as AssetResponseDto[])).toEqual(assets[0]);
   });
 
+  it('prefers DNG over CR2 even when the DNG is smaller', () => {
+    const assets = [
+      { originalMimeType: 'image/x-canon-cr2', exifInfo: { fileSizeInByte: 500 } },
+      { originalMimeType: 'image/x-adobe-dng', exifInfo: { fileSizeInByte: 200 } },
+    ];
+    expect(suggestDuplicate(assets as AssetResponseDto[])).toEqual(assets[1]);
+  });
+
+  it('prefers HEIC over JPEG even when the HEIC is smaller', () => {
+    const assets = [
+      { originalMimeType: 'image/jpeg', exifInfo: { fileSizeInByte: 400 } },
+      { originalMimeType: 'image/heic', exifInfo: { fileSizeInByte: 150 } },
+    ];
+    expect(suggestDuplicate(assets as AssetResponseDto[])).toEqual(assets[1]);
+  });
+
   it('picks the asset with the most exif data if multiple assets have the same file size', () => {
     const assets = [
       { exifInfo: { fileSizeInByte: 200, rating: 5, fNumber: 1 } },
diff --git a/web/src/lib/utils/duplicate-utils.ts b/web/src/lib/utils/duplicate-utils.ts
index 1c783a3667..1279742c4c 100644
--- a/web/src/lib/utils/duplicate-utils.ts
+++ b/web/src/lib/utils/duplicate-utils.ts
@@ -2,10 +2,83 @@ import { getExifCount } from '$lib/utils/exif-utils';
 import type { AssetResponseDto } from '@immich/sdk';
 import { sortBy } from 'lodash-es';
 
+const formatPreferenceGroups: string[][] = [
+  ['image/x-adobe-dng', 'image/dng', 'dng'],
+  [
+    'image/x-canon-cr3',
+    'image/x-canon-cr2',
+    'image/x-nikon-nef',
+    'image/x-sony-arw',
+    'image/x-olympus-orf',
+    'image/x-fuji-raf',
+    'image/x-panasonic-rw2',
+    'image/x-panasonic-raw',
+    'image/x-pentax-pef',
+    'image/x-samsung-srw',
+    'cr3',
+    'cr2',
+    'nef',
+    'arw',
+    'orf',
+    'raf',
+    'rw2',
+    'raw',
+    'pef',
+    'srw',
+  ],
+  ['image/heic', 'image/heif', 'heic', 'heif'],
+  ['image/avif', 'avif'],
+  ['image/jpeg', 'image/jpg', 'jpeg', 'jpg'],
+];
+
+const DEFAULT_FORMAT_PRIORITY = formatPreferenceGroups.length;
+
+const formatPriorityLookup = formatPreferenceGroups.reduce<Map<string, number>>((lookup, group, index) => {
+  for (const format of group) {
+    lookup.set(format, index);
+  }
+  return lookup;
+}, new Map());
+
+const getExtension = (path?: string) => {
+  if (!path) {
+    return undefined;
+  }
+
+  const index = path.lastIndexOf('.');
+  if (index === -1 || index === path.length - 1) {
+    return undefined;
+  }
+
+  return path.slice(index + 1).toLowerCase();
+};
+
+const getAssetFormatPriority = (asset: AssetResponseDto) => {
+  const candidates = [
+    asset.originalMimeType?.toLowerCase(),
+    getExtension(asset.originalFileName),
+    getExtension(asset.originalPath),
+  ];
+
+  for (const candidate of candidates) {
+    if (!candidate) {
+      continue;
+    }
+
+    const priority = formatPriorityLookup.get(candidate);
+    if (priority !== undefined) {
+      return priority;
+    }
+  }
+
+  return DEFAULT_FORMAT_PRIORITY;
+};
+
 /**
  * Suggests the best duplicate asset to keep from a list of duplicates.
  *
  * The best asset is determined by the following criteria:
+ *  - Preferred original file format (based on mime type or extension)
  *  - Largest image file size in bytes
  *  - Largest count of exif data
  *
@@ -13,18 +86,15 @@ import { sortBy } from 'lodash-es';
  * @returns The best asset to keep
  */
 export const suggestDuplicate = (assets: AssetResponseDto[]): AssetResponseDto | undefined => {
-  let duplicateAssets = sortBy(assets, (asset) => asset.exifInfo?.fileSizeInByte ?? 0);
-
-  // Update the list to only include assets with the largest file size
-  duplicateAssets = duplicateAssets.filter(
-    (asset) => asset.exifInfo?.fileSizeInByte === duplicateAssets.at(-1)?.exifInfo?.fileSizeInByte,
-  );
-
-  // If there are multiple assets with the same file size, sort the list by the count of exif data
-  if (duplicateAssets.length >= 2) {
-    duplicateAssets = sortBy(duplicateAssets, getExifCount);
+  if (assets.length === 0) {
+    return undefined;
   }
 
-  // Return the last asset in the list
-  return duplicateAssets.pop();
+  const sorted = sortBy(assets, [
+    (asset) => getAssetFormatPriority(asset),
+    (asset) => -(asset.exifInfo?.fileSizeInByte ?? 0),
+    (asset) => -getExifCount(asset),
+  ]);
+
+  return sorted[0];
 };