FileCircus.pl

#!perl -w

use utf8;
use Encode qw/decode from_to decode_utf8/;
use File::Stat qw/stat/;
use File::Find;
use File::Basename;
use File::Copy qw/cp mv/;
use Time::localtime;

use Data::Dumper qw/Dumper/;


%months = (
    'Jan' => 'janeiro', 'Feb' => 'fevereiro', 'Mar' => 'marco',
    'Apr' => 'abril', 'May' => 'maio', 'Jun' => 'junho',
    'Jul' => 'julho', 'Aug' => 'agosto', 'Sep' => 'setembro',
    'Oct' => 'outubro', 'Nov' => 'novembro', 'Dec' => 'dezembro',
);

find({
    wanted => \&organizeFiles,
}, @ARGV);


##
##
## Sanitize filenames, fx enconding and removing spaces.
sub sanitizeFilename {
    return  if -d;
    fileparse_set_fstype('MSWin32');

    my $filename = basename($_);
    my $sanitizedFilename = $filename;

    $sanitizedFilename =~ s{[\_\-\+]}{ }ig; # creating spaces
    $sanitizedFilename =~ s{\s+?}{_}ig; # creating spaces
    $sanitizedFilename =~ s{([0-9])([a-z])}{$1 $2}ig; # spacing numbers and words
    $sanitizedFilename =~ s{\d{8}}{}ig; # spacing numbers and words
    $sanitizedFilename =~ s{^\s*|\s*$}{}ig;
    $sanitizedFilename =~ s{[^a-z0-9\s\.]+?}{}ig;

    mv $filename, $sanitizedFilename;

    print 'Sanitization complete: ', $sanitizedFilename, "\n"  if !$!;
    print 'Error: ', $filename,' -> ', $sanitizedFilename,' ', $!, "\n"  if $!;
}

##
##
## Copy files to its respective folder with year and month
sub organizeFiles {
    return  if -d;

    fileparse_set_fstype('MSWin32');
    # print 'Processing file: ', $_, "\n";

    my $filename  = basename($_);
    my $filedir = $File::Find::dir;
    my $fullfilename = $File::Find::name;
    my $basePathToFiles = '/var/www/html/oi/oi_repositorio_ofertas/organized';
    my $creationDate = ctime( stat($filename)->mtime );

    my ($month, $year) = ($1, $2)  if $creationDate =~ m/^\w{3,4}\s(\w{3,4}).+(\w{4})$/i;
    my $pathToCopy = "$basePathToFiles/$year/$months{$month}";

    mkdir $pathToCopy  if !-d $pathToCopy;

    cp $filename, $pathToCopy  if -d $pathToCopy;

    print 'Copied to', $pathToCopy, "\n"  if !$!;
    print 'Error: ', $!  if $!;
}

##
##
## Convert all files to PDF
sub proccess_file {
    return  if -d;
    return  if !m/\.docx?$/;

    my $filename  = $_;
    my $filedir = $File::Find::dir;
    my $fullfilename = $File::Find::name;

    $fullfilename =~ s{(\s)}{\\$1}g;
    $filedir =~ s{(\s)}{\\$1}g;
    $filename =~ s{(\s)}{\\$1}g;

    print 'Starting conversion of: ', $filename, "\n";
    system("libreoffice --invisible --convert-to pdf $filename");
    print 'Converted!', "\n";
}


__DATA__


Steps:

[X] cast lockfiles into the void
[X] Sanitize files, remove "_" and decode broken enconding filenames
[X] Transform all files to PDF
[X] Separate files by Year and Month
    [X]Parse file meta data
[?] Organize files by Offers
    Open file and search for known names
    Save files inside the year and month

-----------------------------------------------
-----------------------------------------------
unzip all:
    find . -name *.zip  -exec unzip "{}" \;

remove all zips:
    find . -name *.zip  -exec rm -f "{}" \;