From f1204d2749bc5b783edaebb722eeacacfdedb9b8 Mon Sep 17 00:00:00 2001
From: Trenton H <797416+stumpylog@users.noreply.github.com>
Date: Wed, 25 Jan 2023 08:09:35 -0800
Subject: [PATCH] Updates the installer library to be static in the final
 image, saving the installers into Git and curl-ing the correct revision

---
 .github/workflows/ci.yml                |   6 -
 .github/workflows/installer-library.yml | 139 ++++++++++++++++++++++++
 Dockerfile                              |  53 ++++-----
 docker-builders/Dockerfile.jbig2enc     |  15 ++-
 docker-builders/Dockerfile.pikepdf      |  22 +++-
 docker-builders/Dockerfile.psycopg2     |  18 ++-
 docker-builders/README.md               |  51 +++++++++
 7 files changed, 263 insertions(+), 41 deletions(-)
 create mode 100644 docker-builders/README.md

diff --git a/.github/workflows/ci.yml b/.github/workflows/ci.yml
index ac0b89611..adf03d4bb 100644
--- a/.github/workflows/ci.yml
+++ b/.github/workflows/ci.yml
@@ -212,12 +212,6 @@ jobs:
     name: Prepare Docker Pipeline Data
     if: github.event_name == 'push' && (startsWith(github.ref, 'refs/heads/feature-') || github.ref == 'refs/heads/dev' || github.ref == 'refs/heads/beta' || contains(github.ref, 'beta.rc') || startsWith(github.ref, 'refs/tags/v'))
     runs-on: ubuntu-22.04
-    # If the push triggered the installer library workflow, wait for it to
-    # complete here.  This ensures the required versions for the final
-    # image have been built, while not waiting at all if the versions haven't changed
-    concurrency:
-      group: build-installer-library
-      cancel-in-progress: false
     needs:
       - documentation
       - tests-backend
diff --git a/.github/workflows/installer-library.yml b/.github/workflows/installer-library.yml
index 32aaf85ee..56064ad86 100644
--- a/.github/workflows/installer-library.yml
+++ b/.github/workflows/installer-library.yml
@@ -169,3 +169,142 @@ jobs:
         PIKEPDF_VERSION=${{ fromJSON(needs.prepare-docker-build.outputs.pikepdf-json).version }}
         PILLOW_VERSION=${{ needs.prepare-docker-build.outputs.pillow-version }}
         LXML_VERSION=${{ needs.prepare-docker-build.outputs.lxml-version }}
+
+  commit-binary-files:
+    name: Store installers
+    needs:
+      - prepare-docker-build
+      - build-qpdf-debs
+      - build-jbig2enc
+      - build-psycopg2-wheel
+      - build-pikepdf-wheel
+    runs-on: ubuntu-22.04
+    steps:
+      -
+        name: Checkout
+        uses: actions/checkout@v3
+        with:
+          ref: binary-library
+      -
+        name: Set up Python
+        uses: actions/setup-python@v4
+        with:
+          python-version: "3.9"
+      -
+        name: Install system dependencies
+        run: |
+          sudo apt-get update -qq
+          sudo apt-get install -qq --no-install-recommends tree
+      -
+        name: Extract qpdf files
+        run: |
+          version=${{ fromJSON(needs.prepare-docker-build.outputs.qpdf-json).version }}
+          tag=${{ fromJSON(needs.prepare-docker-build.outputs.qpdf-json).image_tag }}
+
+          docker pull --quiet ${tag}
+          docker create --name qpdf-extract ${tag}
+
+          mkdir --parents qpdf/${version}/amd64
+          docker cp qpdf-extract:/usr/src/qpdf/${version}/amd64 qpdf/${version}
+
+          mkdir --parents qpdf/${version}/arm64
+          docker cp qpdf-extract:/usr/src/qpdf/${version}/arm64 qpdf/${version}
+
+          mkdir --parents qpdf/${version}/armv7
+          docker cp qpdf-extract:/usr/src/qpdf/${version}/armv7 qpdf/${version}
+      -
+        name: Extract psycopg2 files
+        run: |
+          version=${{ fromJSON(needs.prepare-docker-build.outputs.psycopg2-json).version }}
+          tag=${{ fromJSON(needs.prepare-docker-build.outputs.psycopg2-json).image_tag }}
+
+          docker pull --quiet --platform linux/amd64 ${tag}
+          docker create --platform linux/amd64 --name psycopg2-extract ${tag}
+          mkdir --parents psycopg2/${version}/amd64
+          docker cp psycopg2-extract:/usr/src/wheels/ psycopg2/${version}/amd64
+          mv psycopg2/${version}/amd64/wheels/* psycopg2/${version}/amd64
+          rm -r psycopg2/${version}/amd64/wheels/
+          docker rm psycopg2-extract
+
+          docker pull --quiet --platform linux/arm64 ${tag}
+          docker create --platform linux/arm64 --name psycopg2-extract ${tag}
+          mkdir --parents psycopg2/${version}/arm64
+          docker cp psycopg2-extract:/usr/src/wheels/ psycopg2/${version}/arm64
+          mv psycopg2/${version}/arm64/wheels/* psycopg2/${version}/arm64
+          rm -r psycopg2/${version}/arm64/wheels/
+          docker rm psycopg2-extract
+
+          docker pull --quiet --platform linux/arm/v7 ${tag}
+          docker create --platform linux/arm/v7 --name psycopg2-extract ${tag}
+          mkdir --parents psycopg2/${version}/armv7
+          docker cp psycopg2-extract:/usr/src/wheels/ psycopg2/${version}/armv7
+          mv psycopg2/${version}/armv7/wheels/* psycopg2/${version}/armv7
+          rm -r psycopg2/${version}/armv7/wheels/
+          docker rm psycopg2-extract
+      -
+        name: Extract pikepdf files
+        run: |
+          version=${{ fromJSON(needs.prepare-docker-build.outputs.pikepdf-json).version }}
+          tag=${{ fromJSON(needs.prepare-docker-build.outputs.pikepdf-json).image_tag }}
+
+          docker pull --quiet --platform linux/amd64 ${tag}
+          docker create --platform linux/amd64 --name pikepdf-extract ${tag}
+          mkdir --parents pikepdf/${version}/amd64
+          docker cp pikepdf-extract:/usr/src/wheels/ pikepdf/${version}/amd64
+          mv pikepdf/${version}/amd64/wheels/* pikepdf/${version}/amd64
+          rm -r pikepdf/${version}/amd64/wheels/
+          docker rm pikepdf-extract
+
+          docker pull --quiet --platform linux/arm64 ${tag}
+          docker create --platform linux/arm64 --name pikepdf-extract ${tag}
+          mkdir --parents pikepdf/${version}/arm64
+          docker cp pikepdf-extract:/usr/src/wheels/ pikepdf/${version}/arm64
+          mv pikepdf/${version}/arm64/wheels/* pikepdf/${version}/arm64
+          rm -r pikepdf/${version}/arm64/wheels/
+          docker rm pikepdf-extract
+
+          docker pull --quiet --platform linux/arm/v7 ${tag}
+          docker create --platform linux/arm/v7 --name pikepdf-extract ${tag}
+          mkdir --parents pikepdf/${version}/armv7
+          docker cp pikepdf-extract:/usr/src/wheels/ pikepdf/${version}/armv7
+          mv pikepdf/${version}/armv7/wheels/* pikepdf/${version}/armv7
+          rm -r pikepdf/${version}/armv7/wheels/
+          docker rm pikepdf-extract
+      -
+        name: Extract jbig2enc files
+        run: |
+          version=${{ fromJSON(needs.prepare-docker-build.outputs.jbig2enc-json).version }}
+          tag=${{ fromJSON(needs.prepare-docker-build.outputs.jbig2enc-json).image_tag }}
+
+          docker pull --quiet --platform linux/amd64 ${tag}
+          docker create --platform linux/amd64 --name jbig2enc-extract ${tag}
+          mkdir --parents jbig2enc/${version}/amd64
+          docker cp jbig2enc-extract:/usr/src/jbig2enc/build jbig2enc/${version}/amd64/
+          mv jbig2enc/${version}/amd64/build/* jbig2enc/${version}/amd64/
+          docker rm jbig2enc-extract
+
+          docker pull --quiet --platform linux/arm64 ${tag}
+          docker create --platform linux/arm64 --name jbig2enc-extract ${tag}
+          mkdir --parents jbig2enc/${version}/arm64
+          docker cp jbig2enc-extract:/usr/src/jbig2enc/build jbig2enc/${version}/arm64
+          mv jbig2enc/${version}/arm64/build/* jbig2enc/${version}/arm64/
+          docker rm jbig2enc-extract
+
+          docker pull --quiet --platform linux/arm/v7 ${tag}
+          docker create --platform linux/arm/v7 --name jbig2enc-extract ${tag}
+          mkdir --parents jbig2enc/${version}/armv7
+          docker cp jbig2enc-extract:/usr/src/jbig2enc/build jbig2enc/${version}/armv7
+          mv jbig2enc/${version}/armv7/build/* jbig2enc/${version}/armv7/
+          docker rm jbig2enc-extract
+      -
+        name: Show file structure
+        run: |
+          tree .
+      -
+        name: Commit files
+        run: |
+          git config --global user.name "github-actions"
+          git config --global user.email "41898282+github-actions[bot]@users.noreply.github.com"
+          git add pikepdf/ qpdf/ psycopg2/ jbig2enc/
+          git commit -m "Updating installer packages" || true
+          git push origin || true
diff --git a/Dockerfile b/Dockerfile
index 9522728d9..6588802bb 100644
--- a/Dockerfile
+++ b/Dockerfile
@@ -1,19 +1,5 @@
 # syntax=docker/dockerfile:1.4
 
-# Pull the installer images from the library
-# These are all built previously
-# They provide either a .deb or .whl
-
-ARG JBIG2ENC_VERSION
-ARG QPDF_VERSION
-ARG PIKEPDF_VERSION
-ARG PSYCOPG2_VERSION
-
-FROM ghcr.io/paperless-ngx/paperless-ngx/builder/jbig2enc:${JBIG2ENC_VERSION} as jbig2enc-builder
-FROM --platform=$BUILDPLATFORM ghcr.io/paperless-ngx/paperless-ngx/builder/qpdf:${QPDF_VERSION} as qpdf-builder
-FROM ghcr.io/paperless-ngx/paperless-ngx/builder/pikepdf:${PIKEPDF_VERSION} as pikepdf-builder
-FROM ghcr.io/paperless-ngx/paperless-ngx/builder/psycopg2:${PSYCOPG2_VERSION} as psycopg2-builder
-
 FROM --platform=$BUILDPLATFORM node:16-bullseye-slim AS compile-frontend
 
 # This stage compiles the frontend
@@ -58,24 +44,21 @@ LABEL org.opencontainers.image.url="https://github.com/paperless-ngx/paperless-n
 LABEL org.opencontainers.image.licenses="GPL-3.0-only"
 
 ARG DEBIAN_FRONTEND=noninteractive
-# Buildx provided
+# Buildx provided, must be defined to use though
 ARG TARGETARCH
 ARG TARGETVARIANT
 
 # Workflow provided
+ARG JBIG2ENC_VERSION
 ARG QPDF_VERSION
+ARG PIKEPDF_VERSION
+ARG PSYCOPG2_VERSION
 
 #
 # Begin installation and configuration
 # Order the steps below from least often changed to most
 #
 
-# copy jbig2enc
-# Basically will never change again
-COPY --from=jbig2enc-builder /usr/src/jbig2enc/src/.libs/libjbig2enc* /usr/local/lib/
-COPY --from=jbig2enc-builder /usr/src/jbig2enc/src/jbig2 /usr/local/bin/
-COPY --from=jbig2enc-builder /usr/src/jbig2enc/src/*.h /usr/local/include/
-
 # Packages need for running
 ARG RUNTIME_PACKAGES="\
   # Python
@@ -198,19 +181,29 @@ RUN set -eux \
 # Install the built packages from the installer library images
 # Use mounts to avoid copying installer files into the image
 # These change sometimes
-RUN --mount=type=bind,from=qpdf-builder,target=/qpdf \
-    --mount=type=bind,from=psycopg2-builder,target=/psycopg2 \
-    --mount=type=bind,from=pikepdf-builder,target=/pikepdf \
-  set -eux \
+RUN set -eux \
+  && echo "Getting binaries" \
+    && mkdir paperless-ngx \
+    && curl --fail --silent --show-error --output paperless-ngx.tar.gz --location https://github.com/paperless-ngx/paperless-ngx/archive/41d6e7e407af09a0882736d50c89b6e015997bff.tar.gz \
+    && tar -xf paperless-ngx.tar.gz --directory paperless-ngx --strip-components=1 \
+    && cd paperless-ngx \
+    # Setting a specific revision ensures we know what this installed
+    # and ensures cache breaking on changes
+  && echo "Installing jbig2enc" \
+    && cp ./jbig2enc/${JBIG2ENC_VERSION}/${TARGETARCH}${TARGETVARIANT}/jbig2 /usr/local/bin/ \
+    && cp ./jbig2enc/${JBIG2ENC_VERSION}/${TARGETARCH}${TARGETVARIANT}/libjbig2enc* /usr/local/lib/ \
   && echo "Installing qpdf" \
-    && apt-get install --yes --no-install-recommends /qpdf/usr/src/qpdf/${QPDF_VERSION}/${TARGETARCH}${TARGETVARIANT}/libqpdf29_*.deb \
-    && apt-get install --yes --no-install-recommends /qpdf/usr/src/qpdf/${QPDF_VERSION}/${TARGETARCH}${TARGETVARIANT}/qpdf_*.deb \
+    && apt-get install --yes --no-install-recommends ./qpdf/${QPDF_VERSION}/${TARGETARCH}${TARGETVARIANT}/libqpdf29_*.deb \
+    && apt-get install --yes --no-install-recommends ./qpdf/${QPDF_VERSION}/${TARGETARCH}${TARGETVARIANT}/qpdf_*.deb \
   && echo "Installing pikepdf and dependencies" \
-    && python3 -m pip install --no-cache-dir /pikepdf/usr/src/wheels/*.whl \
+    && python3 -m pip install --no-cache-dir ./pikepdf/${PIKEPDF_VERSION}/${TARGETARCH}${TARGETVARIANT}/*.whl \
     && python3 -m pip list \
   && echo "Installing psycopg2" \
-    && python3 -m pip install --no-cache-dir /psycopg2/usr/src/wheels/psycopg2*.whl \
-    && python3 -m pip list
+    && python3 -m pip install --no-cache-dir ./psycopg2/${PSYCOPG2_VERSION}/${TARGETARCH}${TARGETVARIANT}/psycopg2*.whl \
+    && python3 -m pip list \
+  && echo "Cleaning up image layer" \
+    && cd ../ \
+    && rm -rf paperless-ngx
 
 WORKDIR /usr/src/paperless/src/
 
diff --git a/docker-builders/Dockerfile.jbig2enc b/docker-builders/Dockerfile.jbig2enc
index 90318084f..388bdd1f7 100644
--- a/docker-builders/Dockerfile.jbig2enc
+++ b/docker-builders/Dockerfile.jbig2enc
@@ -29,7 +29,20 @@ RUN set -eux \
     && ./autogen.sh \
     && ./configure \
     && make \
+  && echo "Gathering package data" \
+    && dpkg-query -f '${Package;-40}${Version}\n' -W > ./pkg-list.txt \
   && echo "Cleaning up image" \
     && apt-get -y purge ${BUILD_PACKAGES} \
     && apt-get -y autoremove --purge \
-    && rm -rf /var/lib/apt/lists/*
+    && rm -rf /var/lib/apt/lists/* \
+  && echo "Moving files around" \
+    && mkdir build \
+    # Unlink a symlink that causes problems
+    && unlink ./src/.libs/libjbig2enc.la \
+    # Move what the link pointed to
+    && mv ./src/libjbig2enc.la ./build/ \
+    # Move the shared library .so files
+    && mv ./src/.libs/libjbig2enc* ./build/ \
+    # And move the cli binary
+    && mv ./src/jbig2 ./build/ \
+    && mv ./pkg-list.txt ./build/
diff --git a/docker-builders/Dockerfile.pikepdf b/docker-builders/Dockerfile.pikepdf
index c4d1ee1dc..e4181c538 100644
--- a/docker-builders/Dockerfile.pikepdf
+++ b/docker-builders/Dockerfile.pikepdf
@@ -7,12 +7,17 @@
 # Default to pulling from the main repo registry when manually building
 ARG REPO="paperless-ngx/paperless-ngx"
 
+# This does nothing, except provide a name for a copy below
 ARG QPDF_VERSION
 FROM --platform=$BUILDPLATFORM ghcr.io/${REPO}/builder/qpdf:${QPDF_VERSION} as qpdf-builder
 
-# This does nothing, except provide a name for a copy below
-
-FROM python:3.9-slim-bullseye as main
+#
+# Stage: builder
+# Purpose:
+#  - Build the pikepdf wheel
+#  - Build any dependent wheels which can't be found
+#
+FROM python:3.9-slim-bullseye as builder
 
 LABEL org.opencontainers.image.description="A intermediate image with pikepdf wheel built"
 
@@ -100,3 +105,14 @@ RUN set -eux \
     && apt-get -y purge ${BUILD_PACKAGES} \
     && apt-get -y autoremove --purge \
     && rm -rf /var/lib/apt/lists/*
+
+#
+# Stage: package
+# Purpose: Holds the compiled .whl files in a tiny image to pull
+#
+FROM alpine:3.17 as package
+
+WORKDIR /usr/src/wheels/
+
+COPY --from=builder /usr/src/wheels/*.whl ./
+COPY --from=builder /usr/src/wheels/pkg-list.txt ./
diff --git a/docker-builders/Dockerfile.psycopg2 b/docker-builders/Dockerfile.psycopg2
index 8fcf5264b..e3f182435 100644
--- a/docker-builders/Dockerfile.psycopg2
+++ b/docker-builders/Dockerfile.psycopg2
@@ -2,7 +2,12 @@
 # Inputs:
 #    - PSYCOPG2_VERSION - Version to build
 
-FROM python:3.9-slim-bullseye as main
+#
+# Stage: builder
+# Purpose:
+#  - Build the psycopg2 wheel
+#
+FROM python:3.9-slim-bullseye as builder
 
 LABEL org.opencontainers.image.description="A intermediate image with psycopg2 wheel built"
 
@@ -48,3 +53,14 @@ RUN set -eux \
     && apt-get -y purge ${BUILD_PACKAGES} \
     && apt-get -y autoremove --purge \
     && rm -rf /var/lib/apt/lists/*
+
+#
+# Stage: package
+# Purpose: Holds the compiled .whl files in a tiny image to pull
+#
+FROM alpine:3.17 as package
+
+WORKDIR /usr/src/wheels/
+
+COPY --from=builder /usr/src/wheels/*.whl ./
+COPY --from=builder /usr/src/wheels/pkg-list.txt ./
diff --git a/docker-builders/README.md b/docker-builders/README.md
new file mode 100644
index 000000000..14e684ccf
--- /dev/null
+++ b/docker-builders/README.md
@@ -0,0 +1,51 @@
+# Installer Library
+
+This folder contains the Dockerfiles for building certain installers or libraries, which are then pulled into the main image.
+
+## [jbig2enc](https://github.com/agl/jbig2enc)
+
+### Why
+
+JBIG is an image coding which can achieve better compression of images for PDFs.
+
+### What
+
+The Docker image builds a shared library file and utility, which is copied into the correct location in the final image.
+
+See Also:
+
+- [OCRMyPDF Documentation](https://ocrmypdf.readthedocs.io/en/latest/jbig2.html)
+
+## [psycopg2](https://www.psycopg.org/)
+
+### Why
+
+The pre-built wheels of psycopg2 are built on Debian 9, which provides a quite old version of libpq-dev. This causes issue with authentication methods.
+
+### What
+
+The image builds psycopg2 wheels on Debian 10 and places the produced wheels into `/usr/src/wheels/`.
+
+See Also:
+
+- [Issue 266](https://github.com/paperless-ngx/paperless-ngx/issues/266)
+
+## [qpdf](https://qpdf.readthedocs.io/en/stable/index.html)
+
+### Why
+
+qpdf and it's library provide tools to read, manipulate and fix up PDFs. Version 11 is also required by `pikepdf` 6+ and Debian 9 does not provide above version 10.
+
+### What
+
+The Docker image cross compiles .deb installers for each supported architecture of the main image. The installers are placed in `/usr/src/qpdf/${QPDF_VERSION}/${TARGETARCH}${TARGETVARIANT}/`
+
+## [pikepdf](https://pikepdf.readthedocs.io/en/latest/)
+
+### Why
+
+Required by OCRMyPdf, this is a general purpose library for PDF manipulation in Python via the qpdf libraries.
+
+### What
+
+The built wheels are placed into `/usr/src/wheels/`