From 3dbb3ebe5e4e562832f6f91c8231ebbb1219a76e Mon Sep 17 00:00:00 2001
From: Andrew Gouin <andrew@gouin.io>
Date: Wed, 15 Nov 2023 14:30:12 -0700
Subject: [PATCH] Make imported builds more heighliner-y

---
 chains.yaml                    |   7 ++
 dockerfile/imported/Dockerfile | 213 +++++++++++++++++++++++++++++++--
 2 files changed, 211 insertions(+), 9 deletions(-)

diff --git a/chains.yaml b/chains.yaml
index 5547adc..5056ebb 100644
--- a/chains.yaml
+++ b/chains.yaml
@@ -1165,6 +1165,13 @@
   build-env:
     - BUILD_TAGS=muslc
 
+# Union
+- name: union
+  dockerfile: imported
+  base-image: ghcr.io/unionlabs/uniond
+  binaries:
+    - /nix/store/*-uniond-${ARCH}-unknown-linux-musl/bin/uniond
+
 # Vidulum
 - name: vidulum
   github-organization: vidulum
diff --git a/dockerfile/imported/Dockerfile b/dockerfile/imported/Dockerfile
index 2bafcbe..110b62c 100644
--- a/dockerfile/imported/Dockerfile
+++ b/dockerfile/imported/Dockerfile
@@ -1,17 +1,212 @@
 ARG BASE_IMAGE
 ARG VERSION
-FROM $BASE_IMAGE:$VERSION
+FROM $BASE_IMAGE:$VERSION AS imported
+
+# Use minimal busybox from infra-toolkit image for final scratch image
+FROM ghcr.io/strangelove-ventures/infra-toolkit:v0.1.0 AS infra-toolkit
+RUN addgroup --gid 1025 -S heighliner && adduser --uid 1025 -S heighliner -G heighliner
+
+# Use ln and rm from full featured busybox for assembling final image
+FROM busybox:1.34.1-musl AS busybox-full
+
+FROM debian:latest AS build-env
+
+COPY --from=imported / /imported
+
+RUN export ARCH=$(uname -m);\
+    BIN="$(eval "echo "/imported/nix/store/*-uniond-$(uname -m)-unknown-linux-musl/bin/uniond"")";\
+    echo $BIN
+
+# Copy all binaries to /root/bin, for a single place to copy into final image.
+# If a colon (:) delimiter is present, binary will be renamed to the text after the delimiter.
+RUN mkdir /root/bin
+ARG BINARIES
+ENV BINARIES_ENV ${BINARIES}
+RUN bash -c 'set -eux;\
+  export ARCH=$(uname -m);\
+  BINARIES_ARR=();\
+  IFS=, read -ra BINARIES_ARR <<< "$BINARIES_ENV";\
+  for BINARY in "${BINARIES_ARR[@]}"; do\
+    BINSPLIT=();\
+    IFS=: read -ra BINSPLIT <<< "$BINARY";\
+    BINPATH="${BINSPLIT[1]+"${BINSPLIT[1]}"}";\
+    BIN="$(eval "echo "/imported${BINSPLIT[0]+"${BINSPLIT[0]}"}"")";\
+    if [ ! -z "$BINPATH" ]; then\
+      if [[ $BINPATH == *"/"* ]]; then\
+        mkdir -p "$(dirname "${BINPATH}")";\
+        cp "$BIN" "${BINPATH}";\
+      else\
+        cp "$BIN" "/root/bin/${BINPATH}";\
+      fi;\
+    else\
+      cp "$BIN" /root/bin/;\
+    fi;\
+  done'
+
+RUN mkdir -p /root/lib
+ARG LIBRARIES
+ENV LIBRARIES_ENV ${LIBRARIES}
+RUN bash -c 'set -eux;\
+  export ARCH=$(uname -m);\
+  LIBRARIES_ARR=($LIBRARIES_ENV); for LIBRARY in "${LIBRARIES_ARR[@]}"; do LIB="$(eval "echo "$LIBRARY"")"; cp /imported$LIB /root/lib/; done'
+
+# Copy over directories
+RUN mkdir -p /root/dir_abs && touch /root/dir_abs.list
+ARG DIRECTORIES
+ENV DIRECTORIES_ENV ${DIRECTORIES}
+RUN bash -c 'set -eux;\
+  DIRECTORIES_ARR=($DIRECTORIES_ENV);\
+  i=0;\
+  for DIRECTORY in "${DIRECTORIES_ARR[@]}"; do \
+    cp -R /imported$DIRECTORY /root/dir_abs/$i;\
+    echo $DIRECTORY >> /root/dir_abs.list;\
+    ((i = i + 1));\
+  done'
+
+# Determine shared library dependencies for both bins and libs
+RUN cp -R /root/lib/* /root/bin/ || true
+RUN mkdir -p /root/lib_abs && touch /root/lib_abs.list
+RUN bash -c 'set -eux;\
+    export ARCH=$(uname -m);\
+    i=0; for BIN in /root/bin/*; do\
+    echo "Getting $(uname -m) libs for bin: $BIN";\
+    readarray -t LIBS < <(ldd "$BIN");\
+    for LIB in "${LIBS[@]}"; do\
+      PATH1=$(echo $LIB | awk "{print \$1}");\
+      if [ "$PATH1" = "linux-vdso.so.1" ]; then continue; fi;\
+      PATH2=$(echo $LIB | awk "{print \$3}");\
+      PATH3=$(echo $LIB | awk "{print \$4}");\
+      if [ "$PATH2" == "not" ] && [ "$PATH3" == "found" ]; then continue; fi;\
+      if [ ! -z "$PATH2" ]; then\
+        if cat /root/lib_abs.list | grep -x "$PATH2"; then\
+          echo "Skipping $PATH2, already accounted for";\
+          continue;\
+        else\
+          echo "Copying lib2: $PATH2";\
+          cp -L $PATH2 /root/lib_abs/$i;\
+          echo $PATH2 >> /root/lib_abs.list;\
+        fi;\
+      else\
+        if cat /root/lib_abs.list | grep -x "$PATH1"; then\
+          echo "Skipping $PATH1, already accounted for";\
+          continue;\
+        else\
+          echo "Copying lib1: $PATH1";\
+          cp -L $PATH1 /root/lib_abs/$i;\
+          echo $PATH1 >> /root/lib_abs.list;\
+        fi;\
+      fi;\
+      ((i = i + 1));\
+    done;\
+  done'
+
+ARG TARGET_LIBRARIES
+ENV TARGET_LIBRARIES_ENV ${TARGET_LIBRARIES}
+RUN bash -c 'set -eux;\
+  export ARCH=$(uname -m);\
+  i=$(wc -l < /root/lib_abs.list);\
+  LIBRARIES_ARR=($TARGET_LIBRARIES_ENV); for LIBRARY in "${LIBRARIES_ARR[@]}"; do LIB="$(eval "echo "$LIBRARY"")";\
+    if cat /root/lib_abs.list | grep -x "$LIB"; then\
+      echo "Skipping $LIB, already accounted for";\
+      continue;\
+    else\
+      echo "Copying lib2: $LIB";\
+      cp -L $LIB /root/lib_abs/$i;\
+      echo $LIB >> /root/lib_abs.list;\
+      ((i = i + 1));\
+    fi;\
+  done'
+
+# Build final image from scratch
+FROM scratch
 
 LABEL org.opencontainers.image.source="https://github.com/strangelove-ventures/heighliner"
 
-# supports either debian based or alpine based
-RUN (apt update && apt install -y ca-certificates jq curl git gcc nano lz4 wget unzip) || \
-  apk add --no-cache ca-certificates jq curl git gcc nano lz4 wget unzip
+WORKDIR /bin
+
+# Install ln (for making hard links), rm (for cleanup), mv, mkdir, and dirname from full busybox image (will be deleted, only needed for image assembly)
+COPY --from=busybox-full /bin/ln /bin/rm /bin/mv /bin/mkdir /bin/dirname ./
+
+# Install minimal busybox image as shell binary (will create hardlinks for the rest of the binaries to this data)
+COPY --from=infra-toolkit /busybox/busybox /bin/sh
+
+# Install jq
+COPY --from=infra-toolkit /usr/local/bin/jq /bin/
+
+# Add hard links for read-only utils
+# Will then only have one copy of the busybox minimal binary file with all utils pointing to the same underlying inode
+RUN for b in \
+  cat \
+  date \
+  df \
+  du \
+  env \
+  grep \
+  head \
+  less \
+  ls \
+  md5sum \
+  pwd \
+  sha1sum \
+  sha256sum \
+  sha3sum \
+  sha512sum \
+  sleep \
+  stty \
+  tail \
+  tar \
+  tee \
+  tr \
+  watch \
+  which \
+  ; do ln sh $b; done
+
+# Install chain binaries
+COPY --from=build-env /root/bin /bin
+
+# Install libraries that don't need absolute path
+COPY --from=build-env /root/lib /lib
+
+# Copy over absolute path libraries
+COPY --from=build-env /root/lib_abs /root/lib_abs
+COPY --from=build-env /root/lib_abs.list /root/lib_abs.list
+
+# Move absolute path libraries to their absolute locations.
+RUN sh -c 'i=0; while read FILE; do\
+      echo "$i: $FILE";\
+      DIR="$(dirname "$FILE")";\
+      mkdir -p "$DIR";\
+      mv /root/lib_abs/$i $FILE;\
+      i=$((i+1));\
+    done < /root/lib_abs.list'
+
+# Copy over absolute path directories
+COPY --from=build-env /root/dir_abs /root/dir_abs
+COPY --from=build-env /root/dir_abs.list /root/dir_abs.list
+
+# Move absolute path directories to their absolute locations.
+RUN sh -c 'i=0; while read DIR; do\
+      echo "$i: $DIR";\
+      PLACEDIR="$(dirname "$DIR")";\
+      mkdir -p "$PLACEDIR";\
+      mv /root/dir_abs/$i $DIR;\
+      i=$((i+1));\
+    done < /root/dir_abs.list'
+
+RUN mkdir -p /usr/bin && ln -s /bin/env /usr/bin/env
+
+ARG FINAL_IMAGE
+RUN if [ ! -z "$FINAL_IMAGE" ]; then sh -c "$FINAL_IMAGE"; fi
+
+# Remove write utils used to construct image and tmp dir/file for lib copy.
+RUN rm -rf ln rm mv mkdir dirname /root/lib_abs /root/lib_abs.list
+
+# Install trusted CA certificates
+COPY --from=infra-toolkit /etc/ssl/cert.pem /etc/ssl/cert.pem
+
+# Install heighliner user
+COPY --from=infra-toolkit /etc/passwd /etc/passwd
+COPY --from=infra-toolkit --chown=1025:1025 /home/heighliner /home/heighliner
 
-RUN addgroup --gid 1025 -S heighliner || groupadd -g 1025 -r heighliner
-RUN adduser --uid 1025 -S heighliner -G heighliner || useradd -u 1025 --no-log-init -r -g heighliner heighliner
 WORKDIR /home/heighliner
 USER heighliner
-
-ENTRYPOINT []
-CMD []