repo-merge-tool

#!/usr/bin/env perl

use strict;
use warnings;

use File::Basename;

my $bfg_jar_url = "https://repo1.maven.org/maven2/com/madgag/bfg/1.13.0/bfg-1.13.0.jar";

sub trim {
    my $s = shift;
    $s =~ s/^\s+|\s+$//g;
    return $s
};

my $is_bare = trim(`git rev-parse --is-bare-repository`) eq "true";

if (!$is_bare) {
    print "You may only run this tool in a bare git repository\n";
    print "Please clone your repo with --mirror\n";
    print "and try again.\n";
    die;
}

my $dirname = dirname(__FILE__);
my $bfg_file = "${dirname}/bfg.jar";
if (!-e $bfg_file) {
    print "File missing!\n";
    print "${bfg_file}\n";
    system "curl -o ${bfg_file} ${bfg_jar_url}";
}

sub prompt {
  my ($query) = @_;
  local $| = 1;
  print $query;
  chomp(my $answer = <STDIN>);
  return $answer;
}

sub prompt_yn {
  my ($query) = @_;
  my $answer = prompt("$query [Y/n]: ");
  return lc($answer) eq 'y';
}

print "Which files would you like to purge?\n";

my @index_filters = ();

my $nuke_files = prompt_yn(" - remove large files (history only)");
my $nuke_min_file_size = "";
if ($nuke_files) {
    $nuke_min_file_size = prompt(" - how big? (10M/1M/128K/etc) ");
}
if (prompt_yn(" - 3rdparty/workspace.bzl")) {
    push @index_filters, 'git rm --cached --ignore-unmatch 3rdparty/workspace.bzl';
}

if (prompt_yn(" - 3rdparty/jvm/...")) {
    push @index_filters, 'git rm -r --cached --ignore-unmatch 3rdparty/jvm';
}

if (prompt_yn(" - shared_tools/...")) {
    push @index_filters, 'git rm -r --cached --ignore-unmatch shared_tools';
}

if (prompt_yn(" - WORKSPACE")) {
    push @index_filters, 'git rm --cached --ignore-unmatch WORKSPACE';
}

if (prompt_yn(" - stripe-build.yaml")) {
    push @index_filters, 'git rm --cached --ignore-unmatch stripe-build.yaml';
}

if (prompt_yn(" - Dockerfile")) {
    push @index_filters, 'git rm --cached --ignore-unmatch Dockerfile';
}

if (prompt_yn(" - jvm_dependencies.yaml")) {
    push @index_filters, 'git rm --cached --ignore-unmatch bazel_deps.yaml';
    push @index_filters, 'git rm --cached --ignore-unmatch jvm_dependencies.yaml';
}

if (prompt_yn(" - .bazelrc")) {
    push @index_filters, 'git rm --cached --ignore-unmatch .bazelrc';
    push @index_filters, 'git rm --cached --ignore-unmatch .bazelrc-ci-startup-options';
}

if (prompt_yn(" - .gitignore")) {
    push @index_filters, 'git rm --cached --ignore-unmatch .gitignore';
}

if (prompt_yn(" - bazel")) {
    push @index_filters, 'git rm --cached --ignore-unmatch bazel';
    push @index_filters, 'git rm --cached --ignore-unmatch tools/bazel';
}

if (prompt_yn(" - tools/coursier")) {
    push @index_filters, 'git rm --cached --ignore-unmatch tools/coursier';
}

print "\n";

my $move_to_subdir = prompt_yn("move to a subdir?");
my $subdir_name = "";
if ($move_to_subdir) {
    $subdir_name = prompt("name? ");
}

if ($nuke_files) {
    system "java -jar ${bfg_file} -b ${nuke_min_file_size}"
}

if (@index_filters) {
    my $filter_command = join ';', @index_filters;
    print "filter command:\n";
    print "${filter_command}\n";
    system "git filter-branch -f --index-filter '${filter_command}' HEAD"
}

if ($move_to_subdir) {
    system <<"EOF";
git worktree add -f tmp
cd tmp
git ls-tree -dr --name-only head | xargs -I {} mkdir -p ./${subdir_name}/{}
git ls-files | xargs -I {} git mv {} ./${subdir_name}/{}
git commit -m 'move all files into directory: ${subdir_name}'
git clean -df
git worktree remove tmp
EOF
}