CopyImages.pl

1. !/usr/bin/perl -w

use strict; use Getopt::Long; use File::Copy;

1. If there is no XML file the script should grab all the images and put them into
2. /tmp/wikiimages Template:PERL

1. 0) ------------------------------ Globals ----------------------------------- #

my $tmpPath = "/tmp/"; # Absolute path to tmp dir my $debug = 0; # debugging switch my @xmlImages; # List of images obtained from 'xmlFile' my @wantedPaths; # List of wanted image paths which exist from 'xmlFile'

my %options = ( # Set options using Getopt::Long

 wikiImagePath => "",
 xmlFile       => "",
 tmpDir        => "wikiImages", # Could use wiki/images path too
 rmTmpDir      => 0,
 debug         => 0,

);

1. 1) ------------------------------ Get options ------------------------------ #

GetOptions(

 \%options,
 'wikiImagePath=s',
 'xmlFile=s',
 'tmpDir=s',
 'rmTmpDir',
 'debug',

);

1. Append /tmp if missing

($options{'tmpDir'} =~ m/$tmpPath/) or

   ($options{'tmpDir'} = $tmpPath . $options{'tmpDir'});

unless( $options{'wikiImagePath'} ) {

   die("--wikiImagePath relative or absolute path required\n");

}

my $cmd = "find $options{'wikiImagePath'} " # find files in image path

 . q(| egrep '\..{3,4}$' )                  # 3-4 letter extensions only
 . q(| egrep -v '(thumb|archive)');         # Remove thumb archive directories

if( $options{'rmTmpDir'} ) {

 print("[ Removing temporary directory $options{'tmpDir'} ]\n");
 system("rm -rf $options{'tmpDir'}\n") and 
     die("Could not delete $options{'tmpDir'}: $!"); 

}

1. Create tmp subdirectory

opendir(DIR , $options{'tmpDir'}); my $rv = readdir(DIR); closedir(DIR); unless($rv) {

   mkdir $options{'tmpDir'} or die("Could not create directory"); 

}

if ($options{'debug'}) {

 print "== Input arguements ==\n";
 foreach my$values ( keys %options ) {
   print "$values => $options{$values}\n";
 }

}

if( $options{'xmlFile'} eq "") {

 print "[ Copying ALL images to: $options{'tmpDir'} ]\n";

 if ($options{'debug'} ) {
   print"== Executing command ==\n";
   print "$cmd\n";
 }
 # Copying images
 system($cmd . q(| while read i; do cp $i ) . $options{'tmpDir'} . q(; done));
 exit;

} else {

 # Saving list of all images
 system($cmd . " > /tmp/allImages"); #  print ($cmd . "> /tmp/allImages\n");

}

1. 2) ---------------- Open input XML file and list of images ------------------ #

open(XMLFILE, $options{'xmlFile'}) or die("Could not open file ($options{'xmlFile'}): $!\n"); open(ALLPATHS, "/tmp/allImages") or die("Could not open file /tmp/allImages: $!\n");

1. 3) ------------------------------ Grab images ------------------------------- #

{

 # Not scalable  
 local $/=undef;
 my $xmlImage = <XMLFILE>;

 # ($xmlImage =~ m/\<nowiki(.+?)\&nowiki\\gt;/) && (print $1);
 if ( $options{'debug'} ) {
   print "== XML nowiki sections ==\n";
   while ($xmlImage =~ m/\<nowiki\>(.+?)<\/nowiki\>/g) {print "$1\n"};
 }
 # Remove content in between <nowiki>'s
 $xmlImage =~ s/\<nowiki\>(.+?)<\/nowiki\>//g;
 # Grab image matches (don't forget |'s and ]]'s in image tags)
 @xmlImages = $xmlImage =~ m/\[{2}Image:(.+?)(?=\||\]{2})/g; 

}

if($options{'debug'}) {

 local $" = "\n";
 print "== \@xmlImages ==\n @xmlImages\n";

}

1. Not scalable

my @allImagePaths=<ALLPATHS>;

1. 4) ----------------- Test $imageToMatch against $eachImagePath -------------- #

foreach my $imageToMatch (@xmlImages) {

 chop $imageToMatch;
 # Convert space separated image names to '_'s
 $imageToMatch =~ s/ /_/g;
 # Matching xmlFile images against images in ALLPATHS 
 foreach my $eachImagePath (@allImagePaths) {
   ($eachImagePath =~ m/$imageToMatch/) and (push @wantedPaths, $eachImagePath);
 }

}

if ( $options{'debug'} ) {

 print "==Wanted image paths==\n@wantedPaths";

}

if ( scalar(@wantedPaths) == 0 ) {

 die("[ No images found in XML file $options{'xmlFile'} ]\n");

}

print "[ Copying images found in XML file $options{'xmlFile'} to: $options{'tmpDir'} ]\n";

1. 5) --------------------- Copy subset of wanted images ----------------------- #

foreach my $wantedPath ( @wantedPaths ) {

 chomp($wantedPath);
 print "cp $wantedPath $options{'tmpDir'}\n";
   

1. my $fullPath = substr($wantedPath, 1, length $wantedPath);
2. print "$fullPath\n";

 # Copy files
 copy($wantedPath, $options{'tmpDir'}) or die("Could not copy files to $options{'tmpDir'}: $!\n")

}