git-site-gen/lib/Gsg/Gather.pm

package Gsg::Gather;
use strict;
use warnings;
use Log::Log4perl qw(:easy);
use Shellex::Shellex qw(shellex findBin);
use Exporter qw(import);
our @EXPORT_OK = qw(get_file_tree get_projects trim_project_paths get_diff_stat);

sub get_diff_stat($$$$) {

	my $project_dir = shift;
	my $newest_commit = shift;
	my $compare_commit = shift;
	my $logger = shift;

	# git --git-dir=/home/git/git-site-gen.git/ diff --stat 37f54811d49d41a4d794594e5bbaaee2271d82ad 1afd193eda9a6bc703011a72afa273e560355713
	my $gitCmd = findBin("git",$logger);
	my $diff_stat = shellex("$gitCmd --git-dir=$project_dir $newest_commit $compare_commit",$logger);

	return $diff_stat;

}

sub get_projects($$$) {

	my $git_dir = shift;
	my $ignored_projects_ref = shift;
	my $logger = shift;
	my $ls_cmd = findBin("ls",$logger);
	my @git_project_dirs;
	foreach my $dir ( split("\n", shellex("$ls_cmd -d $git_dir/*/",$logger)) ) {
		if ( $dir !~ m/\.git/ ) {
			next;
		}
		if ( grep( /^$dir$/, @$ignored_projects_ref ) ) {
			$logger->info("Found $dir in ignore list, skipping...");
			next;
		} else {
			push(@git_project_dirs,$dir);
		}
	}

	return \@git_project_dirs;

}

sub trim_project_paths($$) {

	my $projects_ref = shift;
	my $logger = shift;

	my @trimmed_projects;
	foreach my $project_path ( @$projects_ref ) {
		# Chop parts of the path we dont need for the web root
		# /some/path/project.git/ -> project.git/
		if ( $project_path =~ m/\/?([^\/]+\.[^\.]+$)/ ) {
			push(@trimmed_projects, $1);
		}
	}

	$logger->info("Returning trimmed project paths");
	return \@trimmed_projects;

}

sub get_file_tree($$) {

	my $projectDir = shift;
	my $logger = shift;
	my $gitCmd = findBin("git",$logger);

	# Get files
	my %file_tree;
	foreach my $file ( split("\n", shellex("$gitCmd --git-dir=\"$projectDir\" ls-tree --full-tree -r HEAD",$logger)) ) {
		chomp $file;
		$file =~ /([a-z0-9]{40})\t(.*)$/;
		# Name - object id
		$file_tree{$2} = $1;
	}

	# Get file content
	my %file_content;
	foreach my $filename ( keys %file_tree ) {
		my $content = shellex("$gitCmd --git-dir=\"$projectDir\" show $file_tree{$filename}",$logger);
		# - TODO - 
		# A hack -- interested in a better way to detect if git files are binary
		# Also dramatically increases run time (~3 seconds additional run time, will likely ballon on bigger git repos)
		my $file_cmd = findBin("file",$logger);
		my $rm_cmd = findBin("rm",$logger);
		my $test_write_path = "/tmp/test";
		my $bin_test = shellex("$gitCmd --git-dir=\"$projectDir\" show $file_tree{$filename} > $test_write_path && $file_cmd -i $test_write_path && $rm_cmd $test_write_path",$logger);

		if ( $bin_test !~ m/text/ ) {
			$content = "Binary file";
		}

		chomp $content;
		# Name - file content
		$file_content{$filename} = $content;
	}

	# Get logs
	my @commit_ids;
	foreach my $log_line ( split("\n",shellex("$gitCmd --git-dir=\"$projectDir\" log",$logger)) ) {
		if ( $log_line =~ m/commit\ ([a-z0-9]{40})/ ) {
			push(@commit_ids,$1);
		}
	}

	my %commits;
	foreach my $commit_id ( @commit_ids ) {
		my $commit_info = shellex("git --git-dir=\"$projectDir\" show $commit_id",$logger);
		chomp $commit_info;
		$commits{$commit_id} = $commit_info;
	}

	# We return commit_ids as well to preserve ordering
	return ( \%file_tree, \%file_content, \%commits, \@commit_ids );

}

1;