Clone a repository with Git LFS files and leverage a local cache.

clone.sh

#!/usr/bin/env bash
#
# Clone a repository with Git LFS files and leverage a local cache.
#
# Usage:
#
#   Clone and checkout any branch:
#   $ clone.sh <repo-url> <cache-dir> <working-copy-dir> <branch-name>
#
#   Clone and checkout a PR head:
#   $ clone.sh <repo-url> <cache-dir> <working-copy-dir> pull/<pr-id>/head
#
#   Clone and checkout a PR merge:
#   $ clone.sh <repo-url> <cache-dir> <working-copy-dir> pull/<pr-id>/merge
#

GIT_REPO_URL=$1
GIT_OBJECT_CACHE_DIR=$2
WORKING_COPY_DIR=$3
REF=$4

# Abort the script if a single command fails.
set -e

# Check if $DEBUG variable is defined (e.g. with "export DEBUG=1")
# If it is defined, then print all commands executed by this script
# and all internal commands executed by Git.
if [[ -n "$DEBUG" ]]; then
    set -x
    export GIT_TRACE=1
fi

# Helper function that retries a command if it failed.
# This is useful for Git commands that perform network requests as
# network requests can always fail.
function execute_with_retry {
    COMMAND=$1
    RETRIES=7   # longest continuous wait should be 64s (2**6)
    COUNT=1     # first try after 4s, if needed
    RET=1       # make command overwrite this
    while [ $COUNT -lt $RETRIES ]; do
        set +e
        $COMMAND
        RET=$?
        set -e
        if [ $RET -eq 0 ]; then
            break
        fi
        COUNT=$((COUNT+1))
        DELAY=$((2**COUNT))
        sleep $DELAY
    done
    if [ $RET -gt 0 ]; then
        echo "'$COMMAND' failed with exit code '$RET'"
        exit $RET
    fi
}

# Create cache if it does not exist yet
if [ -z "$(ls -A "$GIT_OBJECT_CACHE_DIR" 2>/dev/null)" ]; then
    printf "\n## Initializing repo cache...\n"
    execute_with_retry "git clone --bare $GIT_REPO_URL $GIT_OBJECT_CACHE_DIR"
fi

# Update the cache
printf "\n## Fetching latest changes into repo cache...\n"
pushd "$GIT_OBJECT_CACHE_DIR" >/dev/null
    CACHE_URL=$(git remote get-url origin)
    if [ $CACHE_URL != $GIT_REPO_URL ]; then
        printf "\nERROR: Repo cache ($CACHE_URL) does not match request repo ($GIT_REPO_URL)!\n"
        exit 1
    fi
    execute_with_retry "git fetch origin $REF:$REF"
    execute_with_retry "git lfs fetch origin $REF"
popd

# Create a fresh clone and leverage the cache
printf "\n## Setup working copy...\n"
rm -rf "$WORKING_COPY_DIR"
mkdir -p "$WORKING_COPY_DIR"
pushd "$WORKING_COPY_DIR" >/dev/null
    git init .
    mkdir -p .git/objects/info/
    echo "${GIT_OBJECT_CACHE_DIR}/objects" > .git/objects/info/alternates
    git remote add origin "$GIT_REPO_URL"

    execute_with_retry "git fetch origin $REF:local-ci-branch/$REF"
    execute_with_retry "git checkout -f local-ci-branch/$REF"
popd

printf "\n## DONE!\n"