#!/bin/perl

# filename_changer.pl
# Meryll Larkin
# 20200615
# Replace illegal characters in Filenames AND DIRECTORY NAMES with safe characters
# in preparation for tar backups
# 20200625 Added report file
# 20200625 Added processing for Directory names as well as filenames
# Added exclamation point to illegal characters
# Added testing for path and file name length

use strict;
use File::Basename;

chomp (my $datestring =`date +"%Y%m%d"`);
chomp (my $pid = $$);
my $scriptname = basename($0);
$scriptname =~ s/\.pl$//;

my $REPORT = "out_${scriptname}_${datestring}_${pid}.log";
open (LOG, ">>$REPORT") || die "Unable to open log file $REPORT " . $! . "\n";
print LOG "$datestring $scriptname OUTPUT\n\n";

my @MORE_DIRECTORIES;
my $MORE_DIR_BOO = 1;
my @ERRORS;
# option:  permit user to provide starting directory as command line argument 
chomp (my $dir = $ARGV[0] );
print "dir = $dir\n";
my $DIR;

if (! $dir ) {
    print "Please provide the FULL ABSOLUTE PATH of the directory you want recursively processed: ";
    chomp ($dir=(<STDIN>));
}

# remove final / on directory name if that is how it was provided
$dir =~ s/\/$//;


while ($MORE_DIR_BOO) {
    my $DIR_BAD_FILE_COUNT = 0;
    print LOG "\n\n$dir\n";
    my $good_dir_name = &fix_filename($dir);
    if ( "$good_dir_name" ne "$dir" ) {
        if ( -e $good_dir_name) {
            &error_handler("$dir not changed $good_dir_name already exists");
        } else {
            rename ($dir, $good_dir_name);
            $DIR_BAD_FILE_COUNT++;
            print LOG "$DIR_BAD_FILE_COUNT  $dir   $good_dir_name\n";
            $dir = $good_dir_name;
        }
    }
    opendir ($DIR, $dir) || die "Unable to open $dir: " . $!;
    my @dirContents = (grep !/^[_\.]/, readdir ($DIR));
    foreach my $filename (@dirContents) {
        $filename = ${dir} . '/' . $filename;
        print "filename = $filename\n";
        if (-d $filename ) {
            print "$filename is a directory\n";
            push (@MORE_DIRECTORIES, $filename);
        }
        if ( -f $filename ) {
            my $good_name = &fix_filename ($filename);
            if ( "$good_name" eq  "$filename" ) {
                # we don't need to do anything, filename is already good
                next;
            }
            ##### IMPORTANT ######
            # Before renaming a file, make sure another file does not already exist with that name
            #
            if ( -e $good_name) {
                &error_handler("$filename not changed $good_name already exists");
            } else {
                print LOG "$DIR_BAD_FILE_COUNT  $filename   $good_name\n";
                rename ($filename, $good_name);
                $DIR_BAD_FILE_COUNT++;
            }
        }
    }
    my $scalar = scalar @MORE_DIRECTORIES;
    print "scalar @MORE_DIRECTORIES = $scalar\n";
    if ($scalar < 1) {
        $MORE_DIR_BOO = 0;
    } else {
        $dir = pop @MORE_DIRECTORIES;
        print "pop = $dir\n";
    }
    print LOG "Total files changed for $dir = $DIR_BAD_FILE_COUNT\n";
}
closedir $DIR;
my $scalar = scalar @ERRORS;
if ($scalar > 0) {
    print "\n\nSome errors were encountered:\n";
    print LOG "\n\nSome errors were encountered:\n";
    foreach my $error (@ERRORS) {
        print "  $error\n";
        print LOG "$error\n";
    }
}
exit 0;

# =====================================

sub fix_filename {
    my $full_path_filename = shift;
    my $filename_only = basename($full_path_filename);
    my $dir = dirname($full_path_filename);
    # print "Basename = $filename_only\n";
    # print "DIRNAME = $dir\n";
    # change all spaces into underscores
    $filename_only =~ s/\s/_/g;
    # if filename begins with non-alpha numeric character, remove it.
    $filename_only =~ s/^[^a-zA-Z\d]+//g;
    # special case, change [_] to underscore    
    $filename_only =~ s/\[_\]/_/g;
    # change all asterisks into plus signs
    $filename_only =~ s/\*/\+/g;
    # change all square brackets into parenthesis
    $filename_only =~ s/\[/(/g;
    $filename_only =~ s/\]/)/g;
    # change all commas into underscores
    $filename_only =~ s/\,/_/g;
    # change all semicolons into underscores
    $filename_only =~ s/\;/_/g;
    # change all ticks (back grave) into underscores
    $filename_only =~ s/\`/_/g;
    # remove all single quotes
    $filename_only =~ s/\'//g;
    # remove all double quotes
    $filename_only =~ s/\"//g;
    # remove all exclamation points
    $filename_only =~ s/\!//g;
    # change all backslashes into underscores
    $filename_only =~ s/\\/_/g;
    # remove all question marks
    $filename_only =~ s/\?//g;
    # change all octothorpes into underscores
    $filename_only =~ s/\#/_/g;
    # change all dollar signs into underscores
    $filename_only =~ s/\$/_/g;
    # change all percent signs into underscores
    $filename_only =~ s/\%/_/g;
    # change all pipes into underscores
    $filename_only =~ s/\|/_/g;
    # change all right arrows into underscores
    $filename_only =~ s/\>/_/g;
    # change all left arrows into underscores
    $filename_only =~ s/\</_/g;
    # remove all carrots
    $filename_only =~ s/\^//g;
    # change all consecutive underscores into single underscores
    $filename_only =~ s/__+/_/g;
  
    $full_path_filename = $dir . '/' .  $filename_only;

    # test length of path and filename
    my $limit = substr($full_path_filename,254,1);
    if ( $limit eq "" ) {
       # $full_path_filename is less than 255 chars
    } else {
       &error_handler("EXCEEDS 250 chars:   $full_path_filename");
    }
    return $full_path_filename;
}

# =====================================
 
sub error_handler {
    my $errstring = shift;
    push (@ERRORS, $errstring);
}

# =====================================
#
#
# Search for definite tar breakers:
#
# find /<my_dir>   \( -name '*  *' -o -name '* .*'  -o -name '* -*'  -o -name '-*'
# -o -name '*\**' -o -name '*;*' -o -name '*]*' -o -name '*[*' -o -name '*`*'
# -o -name '*"*' -o -name "*'*" -o -name "*\!*" -o -name "*\\*"  \)  2>/dev/null
#
# Search for maybe tar breakers:  ?  # $ % | <>^
#
# find /<my_dir>  \(  -name '*\?*'  -o -name '*\#*' -o -name '*;*' -o -name '*\$*'
# -o -name '*\%*' -o -name '*\|*' -o -name '*\<*' -o -name '*>*' -o -name
# '*^*' \)  2> /dev/null
#
# Here are the characters that I am sure will not break tar nor anything else
# on Linux:
# Alphanumeric, hyphen, underscores, periods  (hyphen is same as minus sign -
# but don't use it as first character in filename)
# AZ az 1234567890-_.
#
# Here are more characters that I  am pretty sure are safe.  Just use them in
# the middle of the file or the end of the file
# I came to this conclusion because I found multiple  files in /usr or /bin or
# /lib or /etc with these characters in the filename
# Parenthesis, equal sign, plus sign, tilde, carrot, ampersand, curly bracket,
# colon, percent
# ()=+~^&{},@:%
#