chkhsrc.pl to HTML.

index -|- end

Generated: Tue Feb 2 17:54:23 2010 from chkhsrc.pl 2007/06/02 5.4 KB.

#!/perl -w
# NAME: chkhsrc.pl
# AIM: Given a input FOLDER, check all the HTML found for a <img src="...."
# reference and make sure that reference EXISTS ...
# 31/05/2007 - geoff mclane - geoffair.com/mperl/index.htm
use strict;
use warnings;
use File::Basename;
unshift(@INC, 'C:/GTools/perl');
require 'logfile.pl' or die "Unable to load logfile.pl ...\n";
# log file stuff
my ($LF);
my $pgmname = $0;
if ($pgmname =~ /\w{1}:\\.*/) {
   my @tmpsp = split(/\\/,$pgmname);
   $pgmname = $tmpsp[-1];
}
my $outfile = "temp.$pgmname.txt";
open_log($outfile);
prt( "$0 ... Hello, World ...\n" );
my $recurse = 1;   # recursive
my $ignfpd = 1;      # ignore FRONTPAGE folders
my @fpfolders = qw( _vti_cnf _vti_pvt _private _derived );
my $in_folder = "C:\\HOMEPAGE\\GeoffAir";
my @in_files = ();
my $cnt = 0;
my $file = '';
my $warnings = '';
my $imgcnt = 0;
# debug only bits
my $dbg1 = 0;   # show entering folder ...
my $dbg10 = 0;   # show diag for get_img_srcs() ...
my $dbg11 = 0;   # in image processing show entered/exits script 
my $dbg12 = 0;   # in image processing show processing count
my $dbg13 = 0;   # in image processing show ok - found file
my $dbg14 = 0;   # in image processing show image count found
my $dbg15 = 0;   # in image processing show image count when NONE found
parse_args(@ARGV);
process_folder( $in_folder );
$cnt = scalar @in_files;
prt( "Found $cnt HTML files to process ...\n" );
foreach $file (@in_files) {
   my ($nm,$dir) = fileparse($file);
   if (open INF, "<$file") {
      my @lines = <INF>;
      close INF;
      my @srcs = get_img_srcs($file, @lines);
      $imgcnt += check_images( $file, @srcs );
   }
}
prt( "Processed $imgcnt image sources ...\n" );
if (length($warnings)) {
   prt( "\nWARNINGS FOLLOW:\n$warnings\n" );
} else {
   prt( "No warnings ...\n" );
}
close_log($outfile,1);
exit(0);
##################################
sub check_images {
   my ($file, @srcs) = @_;
   my ($nm, $dir) = fileparse($file);
   my $scnt = scalar @srcs;
   if ($scnt) {
      prt( "Found $scnt imgs in $nm ...\n" ) if ($dbg14);
      for (my $i = 0; $i < $scnt; $i++) {
         my $src = $srcs[$i][0];
         my $lnn = $srcs[$i][1];
         if ($src =~ /^http:\/\//i) {
            # remote HREF
         } else {
            my $ff = $dir.$src;
            if ( -f $ff ) {
               prt( "$src - ok\n" ) if ($dbg13);
            } else {
               my $msg = "WARNING: [$src] $file:$lnn NOT FOUND!";
               $warnings .= "\n" if length($warnings);
               $warnings .= $msg;
               prt( "$msg\n" );
            }
         }
      }
   } else {
      prt( "Found NO imgs in [$file] ...\n" ) if ($dbg15);
   }
   return $scnt;
}
sub get_img_srcs {
   my ($fil, @lns) = @_;
   my $lc = scalar @lns;
   my $scnt = 0;
   my ($nm,$dir) = fileparse( $fil );
   prt( "Processing $lc lines from [$nm] dir=[$dir]...\n" ) if ($dbg12);
   my @isrc = ();
   my $ln = '';
   my $bal = '';
   my $inscript = 0;
   my $msg = '';
   my $bgnln = 0;
   my $lnnos = '';
   for (my $i = 0; $i < $lc; $i++) {
      $ln = $bal;
      $bal = '';
      $ln .= $lns[$i];
      chomp $ln;
      prt( "$i [$ln] ...\n" ) if ($dbg10);
      if ($inscript) {
         if ($ln =~ /<\/script>/i) {
            $inscript =0;
            prt( "EXIT a SCRIPT ...\n" ) if ($dbg11);
         }
         next;
      }
      if ( $ln =~ /<img\s+(.*)/i ) {
         my $iln = $1;
         prt( "Found [$iln] ...\n" ) if ($dbg10);
         $bgnln = $i;
         while ( !($iln =~ />/) && ($i < $lc)) {
            $i++;
            my $nxln = $lns[$i];
            chomp $nxln;
            prt( "Adding [$nxln] ...\n" ) if ($dbg10);
            $iln .= ' '.$nxln;
         }
         $lnnos = "$bgnln:$i";
         my $ind = index($iln, '>');
         if ($ind != -1) {
            $bal = substr($iln, $ind+1);
            $iln = substr($iln, 0, $ind+1);
         }
         #if ($iln =~ /src=\"(.+)\"/i) {
         if ($iln =~ /src=\"(\S+)\"/i) {
            prt( "SRC = $1 In line [$iln]$lnnos...\n" ) if ($dbg10);
            push(@isrc, [$1, $lnnos]);
            $scnt++;
         } else {
            $msg = "WARNING: SRC NOT FOUND in [$iln]$lnnos...";
            $warnings .= "\n" if length($warnings);
            $warnings .= $msg;
            prt( "$msg\n" );
         }
      } elsif ( $ln =~ /<script.*>/i ) {
         $inscript = 1;
         prt( "Entered a SCRIPT ...\n" ) if ($dbg11);
         if ($ln =~ /<\/script>/i) {
            $inscript =0;
            prt( "EXIT a SCRIPT ...\n" ) if ($dbg11);
         }
      }
   }
   prt( "Returning $scnt img sources ...\n") if ($dbg10);
   return @isrc;
}
sub parse_args {
   my (@av) = @_;
   while (@av) {
      $in_folder = $av[0];
      shift @av;
   }
}
sub is_my_ext {
   my ($fil) = shift;
   my ($nm,$dir,$ext) = fileparse( $fil, qr/\.[^.]*/ );
   if ((lc($ext) eq ".htm")||(lc($ext) eq ".html")) {
      return 1;
   }
   return 0;
}
# my $ignfpd = 1;   # ignore FRONTPAGE folders
sub is_fp_folder {
   my ($inf) = shift;
   foreach my $fil (@fpfolders) {
      if (lc($inf) eq lc($fil)) {
         return 1;
      }
   }
   return 0;
}
sub process_folder {
   my ($inf) = shift;
   my $fcnt = 0;
   prt( "Processing $inf folder ...\n" ) if ($dbg1);
   if ( opendir( DIR, $inf ) ) {
      my @files = readdir(DIR);
      closedir DIR;
      foreach my $fil (@files) {
         if (($fil eq ".")||($fil eq "..")) {
            next;
         }
         my $ff = $inf."\\".$fil;
         if ( -d $ff ) {
            if ($recurse) {
               if ($ignfpd && is_fp_folder($fil)) {   # ignore FRONTPAGE folders
                  next;
               }
               process_folder( $ff );
            }
         } else {
            if (is_my_ext($fil)) {
               push(@in_files, $ff);
               $fcnt++;
            }
         }
      }
      prt( "Processed $inf folder finding $fcnt HTML files ...\n" );
   } else {
      prt( "ERROR: Failed to open folder $inf ...\n" );
   }
}
# eof - chkhsrc.pl

index -|- top

checked by tidy  Valid HTML 4.01 Transitional