chetwyn
Novice
Apr 26, 2006, 6:10 AM
Post #4 of 6
(1202 views)
|
|
Re: [davorg] Pattern matching issue with HTML file, please help...
[In reply to]
|
Can't Post
|
|
OK. Here's what I've done... It works prtty well. I'm not sure how to format in this forum but here it is:
#!/usr/bin/perl
srand();
$progname = $0;
$progname =~ s@(.*?)(/|\\)@@ig;
$generate = '';
$credit = <<EOT;
HTML Formatter
EOT
$amount = 0;
$replace = 0;
$lower = 0;
$upper = 0;
$quiet = 0;
$convert = 0;
$usetabs = 0;
$tags = "<html|</html|<body|</body|<head|</head|<title";
$tags .= "|<isindex|<link|<meta|<!doctype";
$tags .= "|<table|<tr|<th|<td|</th|</tr|</table|<caption";
$tags .= "|<thead|</thead|<tbody|</tbody";
$tags .= "|<p|<br|<blockquote|<hr|<center|</center|<div|</div";
$tags .= "|<col|<colgroup";
$tags .= "|<marquee|</marquee";
$tags .= "|<style|</style";
$tags .= "|<h1|<h2|<h3|<h4|<h5|<h6";
$tags .= "|<ul|</ul|<ol|</ol|<dl|</dl|<li|<dt|<dd|<dir|</dir|<menu|</menu";
$tags .= "|<map|<area|</map";
$tags .= "|<base|<basefont|<bgsound";
$tags .= "|<object|<applet|<param|</object|</applet|<embed|</embed";
$tags .= "|<frameset|<frame|<noframes|</noframes|</frame|</frameset";
$tags .= "|<form|</form|<input|<select|<option|</select|<textarea";
$tagindent = "<table|<tr|<td";
$tagindent .= "|<select|<form";
$tagindent .= "|<frameset";
$tagindent .= "|<ul|<ol|<dl|<dir|<menu|<map";
$tagunindent = "</table|</tr|</td";
$tagunindent .= "|</select|</form";
$tagunindent .= "|</frameset";
$tagunindent .= "|</ul|</ol|</dl|</dir|</menu|</map";
NGetOpt('n:i','r','l','u','q','h','t','todo');
if ($opt_todo) {
print <<EOT;
$credit
To do list:
Everything's done? Everything's DONE! So bring it on
EOT
exit;
}
@files = @ARGV;
if ($#files+1 <= 0 || $opt_h) {
usage();
exit(0);
}
if (defined($opt_n) && $opt_n >= 0) { $amount = $opt_n; }
if (defined($opt_t)) { $usetabs = 1; }
if ($opt_r) { $replace = 1; }
if ($opt_l) { $lower = 1; }
if ($opt_u) { $upper = 1; }
if ($opt_c) { $convert = 1; }
if ($opt_q) { $quiet = 1; }
if (!$quiet) { print "$credit\n\nProcessing...\n"; }
chdir('.') ? print "+\n" : print "-\n";
opendir(DIR,'.');
@files = readdir(DIR);
foreach(@files) {
@lines = ();
@temp = ();
if($_ !~ m/\.(htm.*?)/gi){
next;
}
$filein = $_;
if ($replace) {
$fileout = $filein;
} else {
$fileout = "$filein.out";
}
open(i,"<$filein") || die "Can't open $_ ";
while (!eof(i)) {
$line = <i>;
push @temp, $line;
}
close(i);
splitlines();
if (!$quiet) { print "$_\n"; }
$SCRIPT = 0;
$COMMENT = 0;
$PRE = 0;
$temp = '';
foreach(@lines) {
$SCRIPT = 0 if ($line =~ m@(</script|%>)@ig);
$COMMENT = 0 if ($line =~ m@(-->|</comment)@ig);
$PRE = 0 if ($line =~ m@</pre>@ig);
$line = $_;
$SCRIPT = 1 if ($line =~ m@(<script|<%)@ig);
$COMMENT = 1 if ($line =~ m@(<!--|<comment)@ig);
$PRE = 1 if ($line =~ m@<pre@ig);
if (!$SCRIPT && !$COMMENT && !$PRE) {
$line =~ s/\t//ig;
$line =~ s/<\ /</ig;
$line =~ s/\ >/>/ig;
if ($line =~ />$/) {
$temp .= $line;
} else {
$temp .= $line." ";
}
} else {
$temp .= "\n".$line."\n";
}
}
push @temp, $temp;
splitlines();
$SCRIPT = 0;
$COMMENT = 0;
$PRE = 0;
foreach(@lines) {
$SCRIPT = 0 if ($line =~ m@(</script|%>)@ig);
$COMMENT = 0 if ($line =~ m@(-->|</comment)@ig);
$PRE = 0 if ($line =~ m@</pre>@ig);
$line = $_;
$SCRIPT = 1 if ($line =~ m@(<script|<%)@ig);
$COMMENT = 1 if ($line =~ m@(<!--|<comment)@ig);
$PRE = 1 if ($line =~ m@<pre@ig);
if (!$SCRIPT && !$COMMENT && !$PRE) {
$line =~ s/\ {2,}/\ /ig;
$line =~ s@($tags)@\n$1@ig;
if ($convert) {
$line =~ s@©@©@ig;
$line =~ s@®@®@ig;
}
if ($upper || $lower || $convert) {
for($i=0; $i<length($line); $i++) {
$char = substr($line,$i,1);
if ($char eq '<') { $in = 1; }
if ($char eq '>') { $in = 0; }
if ($char eq '"') {
if ($quote) { $quote = 0; }
else { $quote = 1; }
}
if ($in && !$quote) {
substr($line,$i,1) = uc($char) if $upper;
substr($line,$i,1) = lc($char) if $lower;
}
if (!$quote) {
if (ord($char) == 169) { substr($line,$i,1) = "©"; }
if (ord($char) == 174) { substr($line,$i,1) = "®"; }
}
}
}
}
push @temp, $line;
}
splitlines();
$indent = 0;
$SCRIPT = 0;
$COMMENT = 0;
$PRE = 0;
foreach (@lines) {
$SCRIPT = 0 if ($line =~ m@(</script|%>)@ig);
$COMMENT = 0 if ($line =~ m@(-->|</comment)@ig);
$PRE = 0 if ($line =~ m@</pre>@ig);
$line = $_;
$SCRIPT = 1 if ($line =~ m@(<script|<%)@ig);
$COMMENT = 1 if ($line =~ m@(<!--|<comment)@ig);
$PRE = 1 if ($line =~ m@<pre@ig);
$spaces = "";
if (!$SCRIPT && !$COMMENT && !$PRE) {
$line =~ s@(\ $)@@ig;
$indent -= $line =~ s@($tagunindent)@$1@ig;
$spaces = "";
for ($j=0; $j<$indent; $j++) {
for ($k=0; $k<$amount; $k++) {
if ($usetabs) {
$spaces .= "\t";
} else {
$spaces .= " ";
}
}
}
}
push @temp, $spaces.$line;
if (!$SCRIPT && !$COMMENT && !$PRE) {
$indent += $line =~ s/($tagindent)/$1/ig;
}
}
splitlines();
open (o, ">$fileout");
foreach (@lines) { print o "$_\n"; }
close(o);
}
open(x, ">Results.txt")||die"can not open result file\n";
chdir('.') ? print "+\n" : print "-\n";
opendir(DIR,'.');
@files = readdir(DIR);
foreach(@files) {
if($_ !~ m/\.(out.*?)/gi){
next;
}
#final fucking stretch - thank fuck for that!!!!!
#load the shit into memory and finish it's sorry ass off(sorry, I'm a little tired now)
#yer... you own me a beer for this one... 'heh' 'heh'
open(p,"$_")||die"can not open $file\n";
my @loadit=<p>;
close(p);
foreach (@loadit){
$_ =~ s/\s+//;
$_ =~ s/\<L.*//;
if($_ eq ''){
next;
}
if($_ =~ m/class=listingName/gi){
if(!($_ =~ /\<A/gi)){
($name) = $_ =~ /\>(.*)/;
$name =~ s/&/&/gi;
print x "\n$name\n";
} else {
($name) = $_ =~ /\">(.*)\</mgi;
$name =~ s/&/&/gi;
print x "\n$name\n";
}
}
if($_ =~ m/class=gold/gi){
my ($address) = $_ =~ /\>(.*)/;
if(!($address =~ m/^(ph)/gi)){
print x "$address\n\n";
}
}elsif($_ =~ m/class=free/gi){
($address) = $_ =~ /\>(.*)/gi;
print x "$address\n\n";
}
}
}
close(x);
cleaner();
print "\n\nExtraction Complete\n\n";
exit;
sub splitlines {
@lines = ();
foreach(@temp) {
$line = $_;
if ($line eq "\n") {
# This preserves blank lines in script and comments.
push @lines, " ";
} else {
push @lines, split(/\n/, $line);
}
}
@temp = ();
}
sub cleaner(){
if(!(-e 'processed')){
system("md processed");
}
if(!(-e 'html_files')){
system("md html_files");
}
chdir('.') ? print "+\n" : print "-\n";
opendir(DIR,'.');
my @files = readdir(DIR);
foreach(@files) {
if($_ =~ m/\.(out.*?)/gi){
system("copy \"$_\" processed");
system ("del \"$_\"");
} elsif ($_ =~ m/\.(htm.*?)/gi){
system("copy \"$_\" html_files");
system ("del \"$_\"");
}
}
}
sub usage {
print <<EOT;
$credit
Usage: $progname [-options] filespec [filespec...]
filespec is a filename or filename pattern (e.g. *.htm)
-n Number of tabs/spaces to indent (default: 0)
-t Uses tabs instead of spaces for indenting
-r Replace original with new file
-l Convert tags to lower-case
-u Convert tags to upper-case
-c Convert (C) -> © and (R) -> ®
-h Help
-todo To do list
EOT
}
(This post was edited by chetwyn on Apr 26, 2006, 6:12 AM)
|
MAIN
SEARCH
WHO'S 
LOG 
