alct/git-bundle-bug

## git-bundle-bug
#!/usr/bin/env bash

usage() {
  echo "
SYNOPSIS:
  git-bundle-bug <command> [DIR1] [DIR2]

  Where <command> is one of:

  bundles   Generate 30 bundles for a given git repo and output bundle hash, name and delta.
  dummy     Provision a git repo containing dummy files and commits.
  github    Clones 5 popular Github projects and generates 30 bundles per repo.
  help      Print this help message.

  and DIR1, DIR2:

  DIR1  '/absolute/path/to/working/dir' (default: /tmp/git-dummy-env)
        Either a git repo when used with the 'bundles' command or a empty dir
        when used with 'dummy' or 'github'.

  DIR2  '/absolute/path/to/bundles/dir' (default: defaults to '/tmp/git-bundles')
        The dir where the bundles will be created.

DESCRIPTION:
  Quick and dirty PoC to reproduce a bug where git does not generate the exact
  same bundle for the exact same repo. Used with the 'bundle' command,
  it generates 30 (supposedly identical) bundles for a given repo and prints a
  hash, bundle name and delta.

USAGE:
  For a completely automated process (provisioning of a dummy git repo, creation
  of a series of bundle based on that repo), use the 'dummy' command.
  NOTE, to avoid messing with existing repos, 'dummy' will fail if the working dir
  already exists.

    git-bundle-bug dummy

  For a completely automated process with real projects (cloning of 5 Github
  popular repos, generating 30 bundles per repo) use the 'github' command.
  BEWARE, with the default repos, it will generate around 15GB of data.
  You can edit the content of the 'github_projects' var within the
  'clone_github_repos' to limit or change the repos.

    git-bundle-bug github

  If you want to generate bundles for an existing repo, use the 'bundles' command
  with the absolute path to your repo.

    git-bundle-bug bundles /absolute/path/to/repo"

  die
}

die() {
  (( $# > 0 )) && echo "${*}"
  exit 1
}

get_datetime() {
  date --utc '+%Y-%m-%d_%H-%M-%S'
}

create_dummy_repo() {
  local git_repo="${1:-/tmp/git-dummy-env}"
  local dummy_file
  local i

  mkdir "${git_repo}" || die

  pushd "${git_repo}" >/dev/null || die

  git init >/dev/null || die

  for i in {1..10}; do
    dummy_file="file-${i}.txt"

    echo "foo ${i}" > "${dummy_file}"

    git add . >/dev/null || die
    git commit --message "adding ${dummy_file}" >/dev/null || die
  done

  popd >/dev/null || die

  echo "${git_repo}"
}

generate_bundles() {
  local git_repo="${1:-/tmp/git-dummy-env}"
  local bundles_dir="${2:-/tmp/git-bundles}"
  local tmp_bundle_output="/tmp/bundle_output.txt"
  local bundle_subdir
  local bundle
  local delta
  local i

  bundle_subdir="${bundles_dir}/$(get_datetime)_${git_repo##*/}"

  [[ -d "${bundle_subdir}" ]] || mkdir --parents "${bundle_subdir}"

  pushd "${git_repo}" >/dev/null || die

  for i in {1..30}; do
    bundle="${bundle_subdir}/bundle-${i}"

    script --quiet --return --command "git bundle create ${bundle} --all" "${tmp_bundle_output}" >/dev/null
    delta="$(grep --perl-regexp --only-matching "\(delta \d+\)(?=,)" "${tmp_bundle_output}")"
    rm --force "${tmp_bundle_output}"

    printf "%s\t%s\n" "$(md5sum "${bundle}")" "${delta}"
  done

  popd >/dev/null || die
}

clone_github_repos() {
  local working_dir="${1:-/tmp/git-dummy-env}"
  local tmp_repo_list="/tmp/repo_list.txt"
  local github_projects="freeCodeCamp/freeCodeCamp 996icu/996.ICU ohmyzsh/ohmyzsh flutter/flutter electron/electron"
  local project
  local repo_path

  [[ -d "${working_dir}" ]] || mkdir "${working_dir}"

  pushd "${working_dir}" >/dev/null || die

  for project in ${github_projects}; do
    repo_path="${working_dir}/${project##*/}"

    echo "Cloning ${project} into '${repo_path}' ..."
    echo "${repo_path}" >> "${tmp_repo_list}"
    git clone "https://github.com/${project}.git" 2>/dev/null 1>&2 || die "An error occured while cloning ${project}. Aborting."
  done

  popd >/dev/null || die
}

automatic_dummy() {
  create_dummy_repo "${@}"
  generate_bundles "${@}"
}

automatic_github() {
  local repo
  local tmp_repo_list="/tmp/repo_list.txt"

  clone_github_repos "${@}"

  for repo in $(< "${tmp_repo_list}"); do
    printf "\n%s\n" "${repo}"
    generate_bundles "${repo}" "${2}"
  done

  rm -f "${tmp_repo_list}"
}

main() {
  local command="${1}"

  (( $# < 1 )) && echo "Missing argument." && usage

  case "${command}" in
    "bundles") generate_bundles "${@:2}";;
    "dummy")   automatic_dummy  "${@:2}";;
    "github")  automatic_github "${@:2}";;
    "help")    usage ;;
    *) echo "Wrong argument." && usage
  esac
}

main "${@}"
	#!/usr/bin/env bash

	usage() {
	echo "
	SYNOPSIS:
	git-bundle-bug <command> [DIR1] [DIR2]

	Where <command> is one of:

	bundles Generate 30 bundles for a given git repo and output bundle hash, name and delta.
	dummy Provision a git repo containing dummy files and commits.
	github Clones 5 popular Github projects and generates 30 bundles per repo.
	help Print this help message.

	and DIR1, DIR2:

	DIR1 '/absolute/path/to/working/dir' (default: /tmp/git-dummy-env)
	Either a git repo when used with the 'bundles' command or a empty dir
	when used with 'dummy' or 'github'.

	DIR2 '/absolute/path/to/bundles/dir' (default: defaults to '/tmp/git-bundles')
	The dir where the bundles will be created.

	DESCRIPTION:
	Quick and dirty PoC to reproduce a bug where git does not generate the exact
	same bundle for the exact same repo. Used with the 'bundle' command,
	it generates 30 (supposedly identical) bundles for a given repo and prints a
	hash, bundle name and delta.

	USAGE:
	For a completely automated process (provisioning of a dummy git repo, creation
	of a series of bundle based on that repo), use the 'dummy' command.
	NOTE, to avoid messing with existing repos, 'dummy' will fail if the working dir
	already exists.

	git-bundle-bug dummy

	For a completely automated process with real projects (cloning of 5 Github
	popular repos, generating 30 bundles per repo) use the 'github' command.
	BEWARE, with the default repos, it will generate around 15GB of data.
	You can edit the content of the 'github_projects' var within the
	'clone_github_repos' to limit or change the repos.

	git-bundle-bug github

	If you want to generate bundles for an existing repo, use the 'bundles' command
	with the absolute path to your repo.

	git-bundle-bug bundles /absolute/path/to/repo"

	die
	}

	die() {
	(( $# > 0 )) && echo "${*}"
	exit 1
	}

	get_datetime() {
	date --utc '+%Y-%m-%d_%H-%M-%S'
	}

	create_dummy_repo() {
	local git_repo="${1:-/tmp/git-dummy-env}"
	local dummy_file
	local i

	mkdir "${git_repo}" \|\| die

	pushd "${git_repo}" >/dev/null \|\| die

	git init >/dev/null \|\| die

	for i in {1..10}; do
	dummy_file="file-${i}.txt"

	echo "foo ${i}" > "${dummy_file}"

	git add . >/dev/null \|\| die
	git commit --message "adding ${dummy_file}" >/dev/null \|\| die
	done

	popd >/dev/null \|\| die

	echo "${git_repo}"
	}

	generate_bundles() {
	local git_repo="${1:-/tmp/git-dummy-env}"
	local bundles_dir="${2:-/tmp/git-bundles}"
	local tmp_bundle_output="/tmp/bundle_output.txt"
	local bundle_subdir
	local bundle
	local delta
	local i

	bundle_subdir="${bundles_dir}/$(get_datetime)_${git_repo##*/}"

	[[ -d "${bundle_subdir}" ]] \|\| mkdir --parents "${bundle_subdir}"

	pushd "${git_repo}" >/dev/null \|\| die

	for i in {1..30}; do
	bundle="${bundle_subdir}/bundle-${i}"

	script --quiet --return --command "git bundle create ${bundle} --all" "${tmp_bundle_output}" >/dev/null
	delta="$(grep --perl-regexp --only-matching "\(delta \d+\)(?=,)" "${tmp_bundle_output}")"
	rm --force "${tmp_bundle_output}"

	printf "%s\t%s\n" "$(md5sum "${bundle}")" "${delta}"
	done

	popd >/dev/null \|\| die
	}

	clone_github_repos() {
	local working_dir="${1:-/tmp/git-dummy-env}"
	local tmp_repo_list="/tmp/repo_list.txt"
	local github_projects="freeCodeCamp/freeCodeCamp 996icu/996.ICU ohmyzsh/ohmyzsh flutter/flutter electron/electron"
	local project
	local repo_path

	[[ -d "${working_dir}" ]] \|\| mkdir "${working_dir}"

	pushd "${working_dir}" >/dev/null \|\| die

	for project in ${github_projects}; do
	repo_path="${working_dir}/${project##*/}"

	echo "Cloning ${project} into '${repo_path}' ..."
	echo "${repo_path}" >> "${tmp_repo_list}"
	git clone "https://github.com/${project}.git" 2>/dev/null 1>&2 \|\| die "An error occured while cloning ${project}. Aborting."
	done

	popd >/dev/null \|\| die
	}

	automatic_dummy() {
	create_dummy_repo "${@}"
	generate_bundles "${@}"
	}

	automatic_github() {
	local repo
	local tmp_repo_list="/tmp/repo_list.txt"

	clone_github_repos "${@}"

	for repo in $(< "${tmp_repo_list}"); do
	printf "\n%s\n" "${repo}"
	generate_bundles "${repo}" "${2}"
	done

	rm -f "${tmp_repo_list}"
	}

	main() {
	local command="${1}"

	(( $# < 1 )) && echo "Missing argument." && usage

	case "${command}" in
	"bundles") generate_bundles "${@:2}";;
	"dummy") automatic_dummy "${@:2}";;
	"github") automatic_github "${@:2}";;
	"help") usage ;;
	*) echo "Wrong argument." && usage
	esac
	}

	main "${@}"