find-missing-annex

#!/usr/bin/env bash
# Search all repositories under given path for annexed files with missing data

usage() {
    echo "USAGE"
    echo "  $0 <directory>"
    echo ""
    echo "Scan a path recursively for annexed files with missing data"
    echo
    echo "  <directory>    path to scan (recursively)"
    echo ""
    echo "  -h, --help      display this help and exit"
    exit 1
}

if (( $# != 1 )); then
    usage
fi

case $1 in
    "-h" | "--help")
        usage
        ;;
    *)
        repostore=$1
        ;;
esac

echo "Scanning ${repostore}"
repositories=$(find ${repostore} -type d -iname "*.git")

# filter out repos that have no annex
annexrepos=()
count=0
for repo in ${repositories}; do
    (( count++ ))
    echo -n "Checking for annex branch in ${repo}:  "
    if git -C ${repo} show-ref --verify --quiet refs/heads/git-annex 2> /dev/null; then
        annexrepos+=("${repo}")
        echo "yes"
    else
        echo "no"
    fi
done

echo "Found ${#annexrepos[@]} annex repositories (of ${count} total)"
echo "Checking files..."

isannex() {
    repo="$1"
    fname="$2"
    size=$(git -C ${repo} cat-file -s HEAD:"${fname}")
    # echo ${fname} ${size}
    if (( ${size} < 1024 )) && $( git -C ${repo} show HEAD:"${fname}" | grep -I -q -F "annex/objects" ); then
         true
     else
         false
    fi
}

finddata() {
    fname="$2"
    annexid=$(git -C "${repo}" show HEAD:"${fname}")
    filehash=$(basename "${annexid}")
    # echo -n "Searching for ${filehash}: "
    if ! find ${repo} -name ${filehash} | grep -q .; then
        echo "${repo}/${fname} is MISSING (${filehash})"
    fi
}

checkfiles() {
    repo="$*"
    # echo "::: ${repo} :::"
    while IFS= read -r -d '' fname ; do
        if isannex ${repo} "${fname}"; then
            finddata ${repo} "${fname}"
        fi
    done < <(git -C ${repo} ls-tree -z --name-only -r --full-tree HEAD)
}

# Check each repository for annexed files
for repo in ${annexrepos[@]}; do
    checkfiles ${repo}
done