#!/usr/bin/env bash
#
# Usage:
#   -common/for-every-commit [--] COMMAND ARGS...
#
# Runs COMMAND ARGS... for every commit since the baseline
# (oldest first).  Fails as soon as any such command fails.
#
# Makes some assumptions:
#   - Your main branch is called `main`
#   - Your baseline repository is mentioned in your `Cargo.toml`
#   - When running under CI, CI_PROJECT_URL is set (as by gitlab)
#   - When not running under CI, your `origin` remote is sensible
#
# When run outside CI, will try to restore your branch to where you were.
#
# Requires TOML.pm, so you may need:
#     - maint/apt-install libtoml-perl
#
# *WARNING* this script can be dangerous.  It may `git clean`
# and `git reset` in order to try to check out differing versions.

set -e
set -o pipefail

BASE_BRANCHES='main'

# this include stanza is automatically maintained by update-shell-includes
common_dir=$(realpath "$0")
common_dir=$(dirname "$common_dir")
# shellcheck source=maint/common/bash-utils.sh
. "$common_dir"/bash-utils.sh

reject_options

repo=$(
    perl -MTOML -we '
        use strict;
        undef $/;
        my $toml = from_toml(<STDIN>);
        print $toml->{package}{repository} // die;
    ' <Cargo.toml
)
repo_host=$(
    perl -we '
        $ARGV[0] =~ m{^https://[-.0-9a-z]+/}
            or die "bad repository url";
        print "$&\n";
    ' "$repo"
)

echo "Cargo.toml package.repository: $repo"
echo "Upstream repository instance: $repo_host"

# Ideally we would somehow find which repo and branch this was going
# to be an MR for.  But, in fact, we can't even find out which
# repo this repo it was forked from, in a principled way, with env vars.
# This approach to finding the upstream will have to do for now.
case "$CI_PROJECT_URL" in
    "$repo_host"*)
	# `package.repository` in `Cargo.toml` is
	# on same instance as this CI job.
	#
	# Use $BASE_BRANCHES on `package.repository` as the baseline.
	git remote rm upstream.tmp >/dev/null 2>&1 ||:
	git remote rename upstream upstream.tmp >/dev/null 2>&1 ||:
	git remote add upstream "$repo"
	ORIGIN=upstream
	;;
    '')
	# Not running in CI, use a guess at the baseline:
	# $BASE_BRANCHES at whatever `origin` points to.
	ORIGIN=origin
	;;
    *)
	# We are apparently running in a different instance to upstream.
	# It's not clear what this testing would mean.
	# Should it use our repo URL?  That would be quite exciting.
	echo >&2 "CI_PROJECT_URL $CI_PROJECT_URL not recognised!"
	exit 4
	;;
esac

echo "Baseline repo remote $ORIGIN, at:"
git remote get-url "$ORIGIN"

tlref () {
    echo "refs/remotes/$ORIGIN/$tbranch"
}

x () {
    echo "+ $*"
    "$@"
}

git_checkout_maybe_clean () {
    x=$1; shift
    if
	$x git checkout -q "$1"
    then :
    else 
	echo "Checkout failed, trying with git clean and git reset"
	x git clean -xdff
	x git reset --hard
	x git checkout -q "$1"
    fi
}

old_branch=$(git symbolic-ref -q HEAD || test $? = 1)

restore_old_branch () {
    case "$old_branch" in
	refs/heads/*)
	    git_checkout_maybe_clean '' "${old_branch#refs/heads/}" ||
		echo '*** Failed to return to original branch ***'
	    ;;
	*)
	    echo "*** Was not originally on a branch, HEAD changed! ***"
	    ;;
    esac
}

trap 'restore_old_branch' 0

refspecs=()
for tbranch in $BASE_BRANCHES; do
    refspecs+=(+"refs/heads/$tbranch:$(tlref)")
done

# If you try to unshallow a repo that's already complete, you get
# an error!  Hence this.  Not all versions of git-rev-parse
# understand this option; we err on the side of trying the unshallow.
if [ "x$(git rev-parse --is-shallow-repository)" != xfalse ]; then
    # TODO really we want something like the converse of
    #  git fetch --shallow-exclude
    # but it doesn't seem to exist.
    git fetch --unshallow "$ORIGIN" "${refspecs[@]}"
fi

subtrees=$(
    git log -P  --grep '^git-subtree-split:' |
    sed -n 's/^    git-subtree-split: //p' |
    sed 's/$/../'
)
echo "Excluding subtree inputs: $subtrees"

for tbranch in $BASE_BRANCHES; do
    # shellcheck disable=SC2206
    trevlist=($subtrees "$(tlref)..HEAD")
    tcount=$(git rev-list --count "${trevlist[@]}")
    printf 'HEAD is %3d commits ahead of %s\n' "$tcount" "$tbranch"
    if [ "$count" ] && [ "$count" -le "$tcount" ]; then continue; fi
    count=$tcount
    branch=$tbranch
    revlist=("${trevlist[@]}")
done

echo "Testing every commit not already on $branch"

commits=$(git rev-list --reverse "${revlist[@]}")

i=0
for commit in $commits; do
    banner='##############################'
    printf '|\n%s %d/%d %s\n|\n' "$banner" "$i" "$count" "$banner"
    git log -n1 "$commit" | sed 's/^/| /'
    echo '|'

    git_checkout_maybe_clean x "$commit"

    if x "$@"; then :; else
	rc="$?"
	echo "
|
========================= Failed after $i/$count =========================
|
Earliest broken commit is  "
	git --no-pager log -n1 --pretty=oneline "$commit"
	exit "$rc"
    fi
    i=$(( i + 1 ))
done

echo "|
$banner ok $banner
|"

rc=0