annexstuff

#!/usr/bin/bash

set -e

gitconfig=(
"user.name annextester"
"user.email [email protected]"
"annex.addunlocked true"
"annex.largefiles largerthan=10M"
"annex.backends WORM"
"annex.thin true"
)

annexver="6"

printstatus() {
    echo "------------------------------"
    echo "Current status after $1"
    echo "------------------------------"
    nfiles=$(ls | wc -l)
    echo -e "\nFile types"
    if [ "$nfiles" -gt "0" ]
    then
        file *
    else
        echo "No files yet"
    fi

    echo -e "\nFile sizes"
    if [ "$nfiles" -gt "0" ]
    then
        du -hd0 * .git/*
    else
        du -hd0 .git/*
    fi


    echo -e "\nLast recorded sizes"
    if [ -f *_sizes.txt ]
    then
        cat *_sizes.txt
    else
        echo "Not recorded yet"
    fi

    echo -e "\nHashes"
    if [ "$nfiles" -gt "0" ]
    then
        md5sum randfile*
    else
        echo "No files yet"
    fi

    echo -e "\nLast recorded hashes"
    if [ -f *.md5 ]
    then
        cat *.md5
        echo "Checking"
        md5sum -c *.md5 || echo "HASHES CHANGED"
    else
        echo "Not recorded yet"
    fi

    if [ "$nfiles" -gt "0" ]
    then
        echo -e "\nChecking for hard links"
        for rf in randfile_*
        do
            echo "${rf} ->"
            find .git -samefile $rf
        done
    fi

    echo
    read -p "Press enter to continue"
    echo "--------------------------"
}

# create repo on gin
reponame=annextest-${RANDOM}
gin keys  # should have non-temporary key
echo "Creating repository ${reponame} on GIN"
gin create ${reponame}

# clone it
git clone [email protected]:achilleas/${reponame} && cd ${reponame}
printstatus "Clone"

echo "Setting git config options"
for conf in "${gitconfig[@]}"
do
    echo ${conf}
    git config ${conf}
done

echo "Initialising annex with version ${annexver}"
git annex init --version=${annexver}

# create files
echo "Generating test files using /dev/urandom"
dd if=/dev/urandom of=randfile_1 bs=1M count=15
dd if=/dev/urandom of=randfile_2 bs=1M count=30
dd if=/dev/urandom of=randfile_3 bs=1M count=32
printstatus "File creation"

# create md5sums and check them into git
echo "Calculating md5 sums for tracking file state manually"
md5sum randfile* > ${reponame}.md5
git add ${reponame}.md5
git commit -m "File hashes"
printstatus "Hash calculation"

# annex large files and push thewm
echo "Doing git annex add ."
git annex add .
# echo "Commit"
# git commit -a -m "Data files"
printstatus "git annex add"

echo "Running sync"
git annex sync --content &> /dev/null
printstatus "git annex sync"

# compute sizes
echo "Computing sizes"
du -hd0 * .git/* > ${reponame}_sizes.txt
printstatus "size calculation"

echo "Adding size file to git"
git add ${reponame}_sizes.txt
git commit -m "Sizes"
git push
printstatus "git add sizes, commit, push"

# tag current commit to test reverting
tagname="InitData"
echo "Tagging current commit as ${tagname}"
git tag "${tagname}"
git push --tags

# change files
echo "Appending data to file 1"
dd if=/dev/urandom of=randfile_1 bs=1M count=1 conv=notrunc oflag=append
printstatus "data append"
echo "Truncating file 3"
dd if=/dev/urandom of=randfile_3 bs=1M count=11
printstatus "data replace"

# recompute hashes
echo "Calculating new md5 sums"
md5sum randfile* > ${reponame}.md5
git add ${reponame}.md5
git commit -m "File hashes"
printstatus "Hash calculation"

# update git and annex
echo "Doing git annex add ."
git annex add .
printstatus "git annex add"

echo "Running sync"
git annex sync --content &> /dev/null
printstatus "git annex sync"

# recompute sizes
echo "Computing sizes"
du -hd0 * .git/* > ${reponame}_sizes.txt
printstatus "size calculation"

echo "Adding size file to git"
git add ${reponame}_sizes.txt
git commit -m "Sizes"
git push
printstatus "git add sizes, commit, push"

# revert to previous version
echo "Checking out tagged commit"
git checkout "${tagname}"
printstatus "git checkout tag"

# check sizes and hashes

echo "FINISHED WITH TEST REPOSITORY ${reponame}"