scripts/sed-sources

#!/usr/bin/perl

use strict;
use File::Find;
use File::Temp qw/ tempfile tempdir /;
use Getopt::Std;
use Pod::Usage;
use Text::Tabs;

=head1 NAME

B<sed-sources> -- Performs multiple sed commands on files with the ability to expand or unexpand tabs.

=head1 SYNOPSIS

B<sed-sources> [options] [file dir ...]

=head1 DESCRIPTION

Performs multiple sed commands (modify builtin %seds hash) on source files
or any sources in directories. If no arguments are given, STDIN will be used
as the source. If source files or directories are specified as arguments,
all files will be transformed and overwritten with new versions. Use the B<-p>
option to preview changes to STDOUT, or use the B<-b> option to make a backup
or the original files.

=head1 OPTIONS

=over

=item B<-b>

Backup original source file by appending ".bak" before overwriting with the
newly transformed file.

=item B<-g>

Display verbose debug logging.

=item B<-e>

Expand tabs to spaces (in addition to doing sed substitutions).

=item B<-u>

Unexpand spaces to tabs (in addition to doing sed substitutions).

=item B<-p>

Preview changes to STDOUT without modifying original source files.

=item B<-r>

Skip variants when doing multiple files (no _profile or _debug variants). 

=item B<-t N>

Set the number of spaces per tab (default is 4) to use when expanding or
unexpanding.

=back

=head1 EXAMPLES
 
# Recursively process all source files in the current working directory 
# and and subdirectories and also expand tabs to spaces. All source files 
# will be overwritten with the newly transformed source files.

% sed-sources -e $cwd

# Recursively process all source files in the current working directory 
# and and subdirectories and also unexpand spaces to tabs and preview the
# results to STDOUT

% sed-sources -p -u $cwd

# Same as above except use 8 spaces per tab. 

% sed-sources -p -u -t8 $cwd

=cut


our $opt_b = 0;	# Backup original file?
our $opt_g = 0;	# Verbose debug output?
our $opt_e = 0;	# Expand tabs to spaces?
our $opt_h = 0; # Show help?
our $opt_m = 0;	# Show help manpage style?
our $opt_p = 0;	# Preview changes to STDOUT?
our $opt_t = 4;	# Number of spaces per tab?
our $opt_u = 0;	# Unexpand spaces to tabs?
getopts('eghmpt:u'); 

$opt_m and show_manpage();
$opt_h and help();

our %seds = (
	'\s+$' => "\n",		# Get rid of spaces at the end of a line
	'^\s+$' => "\n",	# Get rid spaces on lines that are all spaces
	'^\s*//\s*[Cc]opyright.*$' => '//', # Get rid of all copyright lines
);


sub show_manpage { exit pod2usage( verbose => 2 ); };
sub help { exit pod2usage( verbose => 3, noperldoc => 1 ); };


#----------------------------------------------------------------------
# process_opened_file_handle
#---------------------------------------------------------------------- 
sub process_opened_file_handle
{
	my $in_fh = shift;
	my $out_fh = shift;

	# Set the number of spaces per tab for expand/unexpand
	$tabstop = $opt_t; 
	
	while (my $line = <$in_fh>) 
	{
		foreach my $key (keys %seds)
		{
			my $value = $seds{"$key"};
			$line =~ s/$key/$value/g;
		}	
		if ($opt_e) {
			print $out_fh expand $line;
		} elsif ($opt_u) {
			print $out_fh unexpand $line;
		} else {
			print $out_fh $line;
		}
	}
}

#----------------------------------------------------------------------
# process_file
#---------------------------------------------------------------------- 
sub process_file
{
	my $in_path = shift;
	if (-T $in_path) 
	{ 
		my $out_fh;
		my $out_path;
		if ($opt_p)
		{
			# Preview to STDOUT
			$out_fh = *STDOUT;
			print "#---------------------------------------------------------------------- \n";
			print "# BEGIN: '$in_path'\n";
			print "#---------------------------------------------------------------------- \n";
		}
		else
		{
			($out_fh, $out_path) = tempfile();			
			$opt_g and print "temporary for '$in_path' is '$out_path'\n";
		}
		open (IN, "<$in_path") or die "error: can't open '$in_path' for reading: $!";
		process_opened_file_handle (*IN, $out_fh);
		

		# Close our input file
		close (IN);

		if ($opt_p)
		{
			print "#---------------------------------------------------------------------- \n";
			print "# END: '$in_path'\n";
			print "#---------------------------------------------------------------------- \n";
			print "\n\n";
		}
		else
		{
			# Close the output file if it wasn't STDOUT
			close ($out_fh);
		
			# Backup file if requested
			if ($opt_b)
			{
				my $backup_command = "cp '$in_path' '$in_path.bak'";
				$opt_g and print "\% $backup_command\n";
				system ($backup_command);
			}
		
			# Copy temp file over original
			my $copy_command = "cp '$out_path' '$in_path'";
			$opt_g and print "\% $copy_command\n";
			system ($copy_command);
		}
	}
}

our @valid_extensions = ( "h", "cpp", "c", "m", "mm" );

#----------------------------------------------------------------------
# find_callback
#---------------------------------------------------------------------- 
sub find_callback
{
	my $file = $_;
	my $fullpath = $File::Find::name;

	foreach my $ext (@valid_extensions)
	{
		my $ext_regex = "\\.$ext\$";
		if ($fullpath =~ /$ext_regex/i)
		{
			print "processing: '$fullpath'\n";
			process_file ($fullpath);
			return;
		}
	}
	print "  ignoring: '$fullpath'\n";
}


#----------------------------------------------------------------------
# main
#---------------------------------------------------------------------- 
sub main
{
	if (@ARGV == 0)
	{
		# no args, take from STDIN and put to STDOUT
		process_opened_file_handle (*STDIN, *STDOUT);
	}
	else
	{
		# Got args, any files we run into parse them, any directories
		# we run into, search them for files
		my $path;
		foreach $path (@ARGV)
		{
			if (-f $path)
			{
				print "processing: '$path'\n";
				process_file ($path);
			}
			else
			{
				print " searching: '$path'\n";
				find(\&find_callback, $path);					
			}
		}
	}
}



# call the main function
main();