#/usr/bin/perl -w
#Stringscan by Jeremy Price
#Call with perl stringscan.pl "/full/directory/path/from/the/top"

#Run through a directory structure, pick the strings out of files, hash them,
#sort them and then print them out with references to the files they came from.

#Bugs:  Will fail badly on files which include a ";" ie a ; in quotes.


use strict;
use diagnostics;

#Download this from CPAN with 'perl -MCPAN -e "install File::Recurse"
use File::Recurse;


my $cd = $ARGV[0];		#The directory to be scanned, supplied on the command line
my @dc;				#Directory contents.  A list of files in dir
my %fl;				#String locations (which files are they in?)
my $ft;				#The type of file we are examining
#ft is filetype, a hash of the file languages and the extension to identify them
my %ft=(	'c'=> '\.h$|\.c$', 
			'perl'=> '\.pl$'
		);
#An associative array, keyed by filetype, of regular expressions to find the strings.
#To add more, add another 'filetype'=>'regexp'
my %search=(	'c'=>'(\"[^;]*)',
				'perl'=>'(\"[^;]*)'
			);
my %vars;			#A cheap way of maintaining a list so that each item is unique



print "\nNow scanning for all files in directories below:  $cd";
my %files = &Recurse([$cd], {match => join('|',keys %ft)});
foreach my $dir (sort keys %files) { foreach my $file (@{ $files{$dir} }) {
    push @dc, "$dir/$file";   }}
    

FILE: foreach my $i (@dc){
	#WindBlows hack, comment out if you are on a UFS (Unified File System).
	$i=~ s/\//\\/g;
	#Try to detect the type of file we are examining
	$ft='none';	#Default (fall through) option
        foreach my $j ( keys %ft ) {$ft=$j if $i=~ /$ft{$j}/;}
			
	#Bail if we couldn't figure it out.
	next FILE if $ft eq 'none';

	open(FH, "<$i") || die "can't open file $i: $!";
	print "\n\nThe strings in $i\nvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvv\n";
	{ local($/)=undef;							#Change the record separator to nothing for a short while
		my $filedata=<FH>;						#read the whole file into filedata
		while ($filedata=~ /$search{$ft}/g){	#Apply the correct regex based on the file type ($ft)
			add_func($1,$i);					#Add it to the hash (see below)
		}
	}
	close FH;
}

#Output the hash
print "\n\nBelow is a list of all the strings found, and then the files they were found in\n";
foreach my $i ( keys %fl ) { 
	print "\n\n\n\nThis is the string: \n$i";
	print "\n\nIt was found in these files:\n ";
	print join(',\n', @{ $fl{$i} });
}



sub add_func(){
	my $string=$_[0];
	my $file=$_[1];
#Push the filename onto a list which is referenced by the string
	push @{ $fl{$string} },$file;
}