Compare thumbnails, but allow a little difference in the perceptual hash

.github/actions/apt-install/action.yml

@@ -1,53 +0,0 @@
-name: Install apt packages
-description: >
-  Installs apt packages with a per-attempt timeout and retries. The hosted runner's Azure apt mirror intermittently serves archives at a crawl, which apt's own timeout does not catch. A stuck attempt is killed and retried, which usually lands on a healthy mirror node.
-
-inputs:
-  packages:
-    description: Space separated list of packages to install
-    required: true
-  attempts:
-    description: Maximum number of attempts
-    default: "3"
-  attempt-timeout:
-    description: Seconds before a single attempt is killed
-    default: "180"
-runs:
-  using: composite
-  steps:
-    - name: Install ${{ inputs.packages }}
-      shell: bash
-      env:
-        PACKAGES: ${{ inputs.packages }}
-        ATTEMPTS: ${{ inputs.attempts }}
-        ATTEMPT_TIMEOUT: ${{ inputs.attempt-timeout }}
-      run: |
-        SUDO=""
-        if [ "$(id -u)" -ne 0 ]; then
-          SUDO="sudo"
-        fi
-
-        attempt_install() {
-          $SUDO apt-get update -q
-          # Intentionally unquoted, this is a list of package names
-          $SUDO apt-get install -q --yes --no-install-recommends $PACKAGES
-        }
-        export -f attempt_install
-        export SUDO PACKAGES
-
-        for attempt in $(seq 1 "${ATTEMPTS}"); do
-          echo "::group::apt install attempt ${attempt}/${ATTEMPTS}"
-          start=$(date +%s)
-          if timeout --kill-after=10 "${ATTEMPT_TIMEOUT}" bash -c attempt_install; then
-            echo "::endgroup::"
-            echo "Installed in $(( $(date +%s) - start ))s"
-            exit 0
-          fi
-          echo "::endgroup::"
-          echo "::warning::apt install attempt ${attempt}/${ATTEMPTS} failed or exceeded ${ATTEMPT_TIMEOUT}s"
-          # A killed attempt can leave dpkg half configured
-          $SUDO dpkg --configure -a || true
-        done
-
-        echo "::error::apt install failed after ${ATTEMPTS} attempts"
-        exit 1

.github/workflows/ci-backend.yml

@@ -80,7 +80,7 @@ jobs:
     needs: changes
     if: needs.changes.outputs.backend_changed == 'true'
     name: "Python ${{ matrix.python-version }}"
-    runs-on: ubuntu-24.04
+    runs-on: ubuntu-26.04
     permissions:
       contents: read
     strategy:
@@ -108,13 +108,14 @@ jobs:
           enable-cache: true
           python-version: ${{ steps.setup-python.outputs.python-version }}
       - name: Install system dependencies
-        timeout-minutes: 12
-        uses: $/.github/actions/apt-install
-        with:
-          packages: unpaper tesseract-ocr imagemagick ghostscript poppler-utils
+        timeout-minutes: 10
+        run: |
+          sudo apt-get update -qq
+          sudo apt-get install -qq --no-install-recommends \
+            unpaper tesseract-ocr imagemagick ghostscript poppler-utils
       - name: Configure ImageMagick
         run: |
-          sudo cp docker/rootfs/etc/ImageMagick-6/paperless-policy.xml /etc/ImageMagick-6/policy.xml
+          sudo cp docker/rootfs/etc/ImageMagick-6/paperless-policy.xml /etc/ImageMagick-7/policy.xml
       - name: Install Python dependencies
         env:
           PYTHON_VERSION: ${{ steps.setup-python.outputs.python-version }}
@@ -158,7 +159,7 @@ jobs:
     needs: changes
     if: needs.changes.outputs.backend_changed == 'true'
     name: Check project typing
-    runs-on: ubuntu-24.04
+    runs-on: ubuntu-26.04
     permissions:
       contents: read
     env:

.github/workflows/ci-docker.yml

@@ -24,10 +24,10 @@ jobs:
       fail-fast: false
       matrix:
         include:
-          - runner: ubuntu-24.04
+          - runner: ubuntu-26.04
             arch: amd64
             platform: linux/amd64
-          - runner: ubuntu-24.04-arm
+          - runner: ubuntu-26.04-arm
             arch: arm64
             platform: linux/arm64
     runs-on: ${{ matrix.runner }}
@@ -163,7 +163,7 @@ jobs:
           archive: false
   merge-and-push:
     name: Merge and Push Manifest
-    runs-on: ubuntu-24.04
+    runs-on: ubuntu-26.04
     needs: build-arch
     if: needs.build-arch.outputs.should-push == 'true'
     environment: image-publishing

.github/workflows/ci-docs.yml

@@ -65,7 +65,7 @@ jobs:
     needs: changes
     if: needs.changes.outputs.docs_changed == 'true'
     name: Build Documentation
-    runs-on: ubuntu-24.04
+    runs-on: ubuntu-26.04
     steps:
       - uses: actions/configure-pages@45bfe0192ca1faeb007ade9deae92b16b8254a0d # v6.0.0
       - name: Checkout
@@ -102,7 +102,7 @@ jobs:
     name: Deploy Documentation
     needs: [changes, build]
     if: github.event_name == 'push' && github.ref == 'refs/heads/main' && needs.changes.outputs.docs_changed == 'true'
-    runs-on: ubuntu-24.04
+    runs-on: ubuntu-26.04
     permissions:
       pages: write
       id-token: write

.github/workflows/ci-frontend.yml

@@ -72,7 +72,7 @@ jobs:
     needs: changes
     if: needs.changes.outputs.frontend_changed == 'true'
     name: Install Dependencies
-    runs-on: ubuntu-24.04
+    runs-on: ubuntu-26.04
     permissions:
       contents: read
     steps:
@@ -104,7 +104,7 @@ jobs:
     name: Lint
     needs: [changes, install-dependencies]
     if: needs.changes.outputs.frontend_changed == 'true'
-    runs-on: ubuntu-24.04
+    runs-on: ubuntu-26.04
     permissions:
       contents: read
     steps:
@@ -137,7 +137,7 @@ jobs:
     name: "Unit Tests (${{ matrix.shard-index }}/${{ matrix.shard-count }})"
     needs: [changes, install-dependencies]
     if: needs.changes.outputs.frontend_changed == 'true'
-    runs-on: ubuntu-24.04
+    runs-on: ubuntu-26.04
     permissions:
       contents: read
     strategy:
@@ -188,10 +188,10 @@ jobs:
     name: E2E Tests
     needs: [changes, install-dependencies]
     if: needs.changes.outputs.frontend_changed == 'true'
-    runs-on: ubuntu-24.04
+    runs-on: ubuntu-26.04
     permissions:
       contents: read
-    container: mcr.microsoft.com/playwright:v1.62.1-noble
+    container: mcr.microsoft.com/playwright:v1.62.1-resolute
     env:
       PLAYWRIGHT_BROWSERS_PATH: /ms-playwright
       PLAYWRIGHT_SKIP_BROWSER_DOWNLOAD: 1
@@ -231,10 +231,9 @@ jobs:
       - name: Install dependencies
         run: cd src-ui && pnpm install --frozen-lockfile
       - name: Install backend system dependencies
-        timeout-minutes: 12
-        uses: $/.github/actions/apt-install
-        with:
-          packages: libmagic1
+        run: |
+          apt-get update
+          apt-get install --yes --quiet --no-install-recommends libmagic1
       - name: Install backend dependencies
         env:
           PYTHON_VERSION: ${{ steps.setup-python.outputs.python-version }}
@@ -246,7 +245,7 @@ jobs:
     name: Frontend Build
     needs: [changes, unit-tests, e2e-tests]
     if: needs.changes.outputs.frontend_changed == 'true'
-    runs-on: ubuntu-24.04
+    runs-on: ubuntu-26.04
     permissions:
       contents: read
     steps:

.github/workflows/ci-release.yml

@@ -14,7 +14,7 @@ permissions: {}
 jobs:
   wait-for-docker:
     name: Wait for Docker Build
-    runs-on: ubuntu-24.04
+    runs-on: ubuntu-26.04
     permissions:
       checks: read
       statuses: read
@@ -30,7 +30,7 @@ jobs:
   build-release:
     name: Build Release
     needs: wait-for-docker
-    runs-on: ubuntu-24.04
+    runs-on: ubuntu-26.04
     permissions:
       contents: read
     steps:
@@ -70,10 +70,9 @@ jobs:
         run: |
           uv sync --python "${PYTHON_VERSION}" --dev --frozen
       - name: Install system dependencies
-        timeout-minutes: 12
-        uses: $/.github/actions/apt-install
-        with:
-          packages: gettext liblept5
+        run: |
+          sudo apt-get update -qq
+          sudo apt-get install -qq --no-install-recommends gettext libleptonica6
       # ---- Build Documentation ----
       - name: Build documentation
         env:
@@ -145,7 +144,7 @@ jobs:
   publish-release:
     name: Publish Release
     needs: build-release
-    runs-on: ubuntu-24.04
+    runs-on: ubuntu-26.04
     permissions:
       contents: write
       pull-requests: write
@@ -197,7 +196,7 @@ jobs:
     name: Append Changelog
     needs: publish-release
     if: needs.publish-release.outputs.prerelease == 'false'
-    runs-on: ubuntu-24.04
+    runs-on: ubuntu-26.04
     permissions:
       contents: write
       pull-requests: write

.github/workflows/ci-static-analysis.yml

@@ -15,7 +15,7 @@ permissions:
 jobs:
   zizmor:
     name: Run zizmor
-    runs-on: ubuntu-24.04
+    runs-on: ubuntu-26.04
     permissions:
       contents: read
       actions: read
@@ -29,7 +29,7 @@ jobs:
         uses: zizmorcore/zizmor-action@cc914d7f3750a2d13d75c7f184a1060aa0e9d482 # v0.6.4
   semgrep:
     name: Semgrep CE
-    runs-on: ubuntu-24.04
+    runs-on: ubuntu-26.04
     container:
       image: semgrep/semgrep:1.155.0@sha256:cc869c685dcc0fe497c86258da9f205397d8108e56d21a86082ea4886e52784d
     if: github.actor != 'dependabot[bot]'

.github/workflows/cleanup-tags.yml

@@ -17,7 +17,7 @@ jobs:
   cleanup-images:
     name: Cleanup Image Tags for ${{ matrix.primary-name }}
     if: github.repository_owner == 'paperless-ngx'
-    runs-on: ubuntu-24.04
+    runs-on: ubuntu-26.04
     environment: registry-maintenance
     strategy:
       fail-fast: false
@@ -42,7 +42,7 @@ jobs:
   cleanup-untagged-images:
     name: Cleanup Untagged Images Tags for ${{ matrix.primary-name }}
     if: github.repository_owner == 'paperless-ngx'
-    runs-on: ubuntu-24.04
+    runs-on: ubuntu-26.04
     needs:
       - cleanup-images
     environment: registry-maintenance

.github/workflows/codeql-analysis.yml

@@ -21,7 +21,7 @@ on:
 jobs:
   analyze:
     name: Analyze
-    runs-on: ubuntu-24.04
+    runs-on: ubuntu-26.04
     permissions:
       actions: read
       contents: read

.github/workflows/crowdin.yml

@@ -13,7 +13,7 @@ jobs:
   synchronize-with-crowdin:
     name: Crowdin Sync
     if: github.repository_owner == 'paperless-ngx'
-    runs-on: ubuntu-24.04
+    runs-on: ubuntu-26.04
     environment: translation-sync
     steps:
       - name: Checkout

.github/workflows/issue-bot.yml

@@ -7,7 +7,7 @@ jobs:
     # Note: peakoss/anti-slop does not support the `issues` event yet (all of its
     # issue inputs are still commented out upstream), so the checks that the PR Bot
     # workflow gets from the action are implemented manually here.
-    runs-on: ubuntu-latest
+    runs-on: ubuntu-slim
     permissions:
       issues: write
     steps:

.github/workflows/pr-bot.yml

@@ -4,7 +4,7 @@ on:
     types: [opened]
 jobs:
   Anti-slop:
-    runs-on: ubuntu-latest
+    runs-on: ubuntu-slim
     permissions:
       contents: read
       issues: read
@@ -24,7 +24,7 @@ jobs:
             ASLOP-PR-VERIFY
   pr-bot:
     name: Automated PR Bot
-    runs-on: ubuntu-latest
+    runs-on: ubuntu-slim
     # Runs after Anti-slop so the welcome comment can see whether the PR was closed
     # instead of racing it. Still runs if that job fails, so labeling is not lost.
     needs: Anti-slop

.github/workflows/project-actions.yml

@@ -12,7 +12,7 @@ permissions:
 jobs:
   pr_opened_or_reopened:
     name: pr_opened_or_reopened
-    runs-on: ubuntu-24.04
+    runs-on: ubuntu-slim
     permissions:
       # write permission is required for autolabeler
       pull-requests: write

.github/workflows/repo-maintenance.yml

@@ -9,7 +9,7 @@ jobs:
   stale:
     name: 'Stale'
     if: github.repository_owner == 'paperless-ngx'
-    runs-on: ubuntu-24.04
+    runs-on: ubuntu-26.04
     permissions:
       issues: write
       pull-requests: write
@@ -34,7 +34,7 @@ jobs:
   lock-threads:
     name: 'Lock Old Threads'
     if: github.repository_owner == 'paperless-ngx'
-    runs-on: ubuntu-24.04
+    runs-on: ubuntu-26.04
     permissions:
       issues: write
       pull-requests: write
@@ -58,7 +58,7 @@ jobs:
   close-answered-discussions:
     name: 'Close Answered Discussions'
     if: github.repository_owner == 'paperless-ngx'
-    runs-on: ubuntu-24.04
+    runs-on: ubuntu-slim
     permissions:
       discussions: write
     steps:
@@ -117,7 +117,7 @@ jobs:
   close-outdated-discussions:
     name: 'Close Outdated Discussions'
     if: github.repository_owner == 'paperless-ngx'
-    runs-on: ubuntu-24.04
+    runs-on: ubuntu-slim
     permissions:
       discussions: write
     steps:
@@ -211,7 +211,7 @@ jobs:
   close-unsupported-feature-requests:
     name: 'Close Unsupported Feature Requests'
     if: github.repository_owner == 'paperless-ngx'
-    runs-on: ubuntu-24.04
+    runs-on: ubuntu-slim
     permissions:
       discussions: write
     steps:

.github/workflows/translate-strings.yml

@@ -8,7 +8,7 @@ env:
 jobs:
   generate-translate-strings:
     name: Generate Translation Strings
-    runs-on: ubuntu-latest
+    runs-on: ubuntu-26.04
     environment: translation-sync
     permissions:
       contents: write
@@ -25,10 +25,9 @@ jobs:
         id: setup-python
         uses: actions/setup-python@5fda3b95a4ea91299a34e894583c3862153e4b97 # v7.0.0
       - name: Install system dependencies
-        timeout-minutes: 12
-        uses: $/.github/actions/apt-install
-        with:
-          packages: gettext
+        run: |
+          sudo apt-get update -qq
+          sudo apt-get install -qq --no-install-recommends gettext
       - name: Install uv
         uses: astral-sh/setup-uv@c18668ad3cf93ea998bef934396af7bb5c839dc7 # v10.2.0
         with:

src/paperless/tests/parsers/test_mail_parser_live.py

@@ -137,9 +137,20 @@ class TestNginxService:
     reason="No Gotenberg/Tika servers to test with",
 )
 class TestParserLive:
-    @staticmethod
-    def imagehash(file: Path, hash_size: int = 18) -> str:
-        return f"{average_hash(Image.open(file), hash_size)}"
+    # Rasterizer versions shift a few pixels, so compare perceptual hashes by
+    # Hamming distance (out of 18 * 18 = 324 bits) rather than for equality
+    MAX_HASH_DISTANCE = 8
+
+    @classmethod
+    def assert_thumbnails_similar(cls, generated: Path, expected: Path) -> None:
+        distance = average_hash(Image.open(generated), 18) - average_hash(
+            Image.open(expected),
+            18,
+        )
+        assert distance <= cls.MAX_HASH_DISTANCE, (
+            f"Thumbnail {generated} differs from {expected} by {distance} bits "
+            f"(max {cls.MAX_HASH_DISTANCE})"
+        )
 
     def test_get_thumbnail(
         self,
@@ -168,12 +179,7 @@ class TestParserLive:
         assert thumb.exists()
         assert thumb.is_file()
 
-        assert self.imagehash(thumb) == self.imagehash(
-            simple_txt_email_thumbnail_file,
-        ), (
-            f"Created thumbnail {thumb} differs from expected file "
-            f"{simple_txt_email_thumbnail_file}"
-        )
+        self.assert_thumbnails_similar(thumb, simple_txt_email_thumbnail_file)
 
     def test_tika_parse_successful(self, mail_parser: MailDocumentParser) -> None:
         """
@@ -255,7 +261,7 @@ class TestParserLive:
         THEN:
             - Gotenberg shall be called to generate the PDF
             - The archive PDF shall contain the expected content
-            - The generated thumbnail shall match the expected image hash
+            - The generated thumbnail shall be perceptually close to the expected image
         """
         util_call_with_backoff(mail_parser.parse, [html_email_file, "message/rfc822"])
 
@@ -272,14 +278,4 @@ class TestParserLive:
             html_email_file,
             "message/rfc822",
         )
-        generated_thumbnail_hash = self.imagehash(generated_thumbnail)
-
-        # The created PDF is not reproducible, but the converted image
-        # should always look the same
-        expected_hash = self.imagehash(html_email_thumbnail_file)
-
-        assert generated_thumbnail_hash == expected_hash, (
-            f"PDF thumbnail differs from expected. "
-            f"Generated: {generated_thumbnail}, "
-            f"Hash: {generated_thumbnail_hash} vs {expected_hash}"
-        )
+        self.assert_thumbnails_similar(generated_thumbnail, html_email_thumbnail_file)