#!/usr/bin/perl -I/usr/local/perl
use parseHTML;
sub usage {
my $nameFile="auditHTML";
print "
usage:\n
$nameFile takes an HTML file and give datas about the code.
$nameFile
\t -f OR -fr
\t -p [--print] prints tags
\t -po [--print_opened] prints only opened tags
\t -ps [--print_stats] prints stats on opened tags
\t --pattern text prefix for output
\t -e ' print only specified tags
\t --notext [-nt] do not print CDATAs
\t --nooption [-np] do not print tag options
\t --compact [-c] do not print End Of Line
";
}
sub retrieveRemoteFile {
my $url=shift;
my $nameFile=shift;
my $success=`wget $url -O $nameFile 2>&1 1>/dev/null`;
return 0 if (!$success) ;
return 1;
}
#===== mieux que getopts
sub getArgs {
my @args=@_;
my $oldKey="";
my %opts=();
for $el (@args) {
if ($el=~/^-/) {
$el=~s/-+//g;
$opts{$el}="";
$oldKey=$el;
}
else {
$opts{$oldKey}=$el;
}
}
return %opts;
}
#==========================
# MAIN
#_____________________________
# |
# Gestion des entrées :
my %opts=();
my $output;
my $eol = "\n";
%opts=getArgs(@ARGV);
if ((exists $opts{h}) || (exists $opts{help})) {
&usage();
exit 0;
}
if ((!exists $opts{f}) && (!exists $opts{fr})) {
print STDERR "I need a file !\n";
&usage();
exit 0;
}
if (exists $opts{f}) {
$file=$opts{f};
if (! -e $file) {
print STDERR "Error : file does not exist : $file\n";
exit;
}
}
if (exists $opts{fr}) {
$file=$opts{fr};
&retrieveRemoteFile($file,"/tmp/auditHTMLtmp.html");
$file="/tmp/auditHTMLtmp.html";
if (! -e $file) {
print STDERR "Error : file does not exist : $file\n";
exit;
}
}
if ((exists $opts{p}) || (exists $opts{print})) {
$printing=1;
} else {
$printing=0;
}
if ((exists $opts{po}) || (exists $opts{print_opened})) {
$printingOpened=1-$printing;
} else {
$printingOpened=0;
}
if ((exists $opts{ps}) || (exists $opts{print_stats})) {
$printStats=1;
} else {
$printStats=0;
}
if ((exists $opts{e}) || (exists $opts{extract})) {
my $r;
$r=$opts{e} if (exists $opts{e});
$r=$opts{extract} if (exists $opts{extract});
@extract=split(' ',uc($r));
}
if ((exists $opts{c}) || (exists $opts{compact})) {
$compact=1;
}
if (exists $opts{pattern}) {
$pattern=$opts{pattern};
} else {
$pattern="---";
}
if (exists $opts{o}) {
$output=$opts{o};
} else {
$output="";
}
if ($output) {
open(OUT,">$output");
*STDOUT=*OUT;
}
my ($notext,$nooption);
$notext=0;$nooption=0;
$notext=1 if ((exists $opts{notext}) ||(exists $opts{nt})) ;
$nooption=1 if ((exists $opts{nooption}) ||(exists $opts{no})) ;
# Fin Gestion des entrées
#_____________________________|
$n=parseHTML->new();
$n->setFile($file);
$n->parseFile();
my %opened;
for $r (@{$n->{listeTag}}) {
if ($printing )
{
if (($r->{name} eq 'TD') && ($r->{type} eq 'start') && ($compact==1)) {
$iaminatd=1;
$oldpattern=$pattern;
}
$n->printOneBal($r,"$pattern" x $r->{depth},{notexte=>$notext,nooption=>$nooption});
if ((! $iaminatd) && ($compact==1)) {
print "\n";
} else {
if ($compact==1) {$pattern=""};
}
if (($r->{name} eq 'TD') && ($r->{type} eq 'end') && ($compact==1)) {
$iaminatd=0;
$pattern=$oldpattern;
print "\n" ;
}
if (!$compact) { print "\n";}
}
if ($r->{statut} eq 'open') {
if ($printingOpened)
{
$n->printOneBal($r,"$pattern" x $r->{depth},{notexte=>$notext,nooption=>$nooption});
print "$eol";
}
$opened{$r->{name}}+=1;
}
}
if ($printStats) {
for $op (keys(%opened)) {
print $opened{$op}." tags $op ouvert(s)\n";
}
}
if (scalar(@extract)) {
my %tags;
my @lref;
for $tag (@extract) {
$tags{$tag}=1;
}
$lref=$n->extractBalBlock(\%tags);
for $r (@{$lref}) {
$n->printOneBal($r,"$pattern" x ($r->{depth}-1),{notexte=>$notext,nooption=>$nooption});
print "$eol";
}
}
#=================================
#
# End Main
#
#=================================