#!/usr/local/bin/perl

# This is a program which replaces one set of text with another in all
# the files with ".htm" or ".html" extensions in the directory from which 
# the program is run and all its sub-directories.

# It's simple enough to use - just copy this file to the top of the directory tree
# you want text replacing, make it executable with the "chmod" command and
# then run it.  The program asks you what text you want replaced and what 
# text you wish to replace it with.  Each time it replaces text it sends a 
# message to standard output with the path and name of the file in which it
# has replaced text.  It also lists the directories it has searched.

# NOTE: You may need to include the full path with the call to run 
# the program. e.g. /users/p4/httpd/htdocs/ciaran/replace

# The program match is case insensative i.e. upper case will 
# match to lower case and visa versa, but, for the moment, it will NOT 
# ignore spaces, carriage returns etc. i.e. two spaces in a line instead of
# one will not return a match and therefore not replace the text.

# Tested on a Sparc workstation, for a UNIX platform with Perl 5.0? installed.

#************************** The Main Program ******************************

print "What text do you wish to replace?\n";
# Input the text to be replaced.
$match = <STDIN>;
# Remove the carriage return
chop($match);

# Transform the text into uppercase in a temporary variable.
$tempmatch = $match;
$tempmatch =~ tr/a-z/A-Z/;

print "Replacement text?\n";
# Input the text to take its place.
$replacement = <STDIN>;
# Remove the carriage return
chop($replacement);


# Start at the directory in which the program is being run.
$dirname = ".";
#Call the Replace function.
&ReplaceText($dirname);
print "The Replace program is now complete", "\n";


#************************** The Subroutine *******************************

sub ReplaceText
{

# Make some of the variables local to the function - for recursion purposes.
	local(@allfiles, $i, $totalfiles, @subdir, @htmfiles);
	local($nextline, $found);
	local($subdirnumber, $htmfilenumber, $j);	

# Open the directory, read all the files in it and close it again.
	opendir(DIR, $_[0]) || die "Can't open directory", $_[0], "\n";
#pick up all the files in the directory.
	@allfiles = grep(/[a-zA-Z0-9_-]*/, readdir(DIR)); 
	closedir(DIR);
	
# Store the number of files that are in the directory.
	$totalfiles = @allfiles;
	
# Pick out the subdirectories.
	$i = 0;
	$j = 0;

	while($i < $totalfiles)
	{
		if (($allfiles[$i] ne ".") && ($allfiles[$i] ne ".."))
		{
			$allfiles[$i] = $_[0] . "/" . $allfiles[$i];
#			print "allfiles = ", $allfiles[$i], "\n" ;
			if(opendir(SUBDIR, $allfiles[$i]))
			{
				$subdir[$j] = $allfiles[$i];
				closedir(SUBDIR);
				++$j;
			}
		}
		++$i;
	}

	$subdirnumber = @subdir;


# Open the directory again, select the *.htm & *.html files and 
# replace the text.
	opendir(DIR, $_[0]) || die "Can't open directory", $_[0], "\n";

	print "Searching directory: ", $_[0], "\n";


# For files ending in *.htm or *.html
	@htmfiles = grep(/\.html?$/, readdir(DIR));
# Store the number of htm files there are.
	$htmfilenumber = @htmfiles;
	$i = 0;

	while($i < $htmfilenumber)
	{
# Add the files directory path.
		$htmfiles[$i] = $_[0] . "/" . $htmfiles[$i];
# Make the file readable and writable.
		chmod(0666,$htmfiles[$i]);
		$tempfile = "tempfile.htm";
# Open the file.
		open(SRCHFILE, $htmfiles[$i]);
		open(TEMPFILE, ">$tempfile");

# Input the contents of the file into a buffer.
		$buff = "";
		while(<SRCHFILE>)
		{
			$buff = $buff . $_;
		}
# Transform the text into uppercase in a temporary buffer.
		$tempbuff = $buff;
		$tempbuff =~ tr/a-z/A-Z/;
		
		$startindex = 0;
		$offset = 0;

# Find the first instance of matching text.
		$found = index($tempbuff, $tempmatch);
		while($found != -1)
		{
			$startindex = $found + length($match);
# Replace the old text with the new text.
			substr($buff, $found + $offset, length($match)) = $replacement;
			$offset = $offset + (length($replacement) - length($match));
# Tell the user of the replacement.
			print "Replacing text in: ", $htmfiles[$i], "\n";
# Find the next instance of the matching text.
			$found = index($tempbuff, $tempmatch, $startindex);
		}
		print TEMPFILE $buff;

		close(TEMPFILE);
		close(SRCHFILE);
		rename ($tempfile, $htmfiles[$i]);
		++$i;
	}

	$i = 0;
# Store the number of subdirectories.
	$subdirnumber = @subdir;

	closedir(DIR);
# For each subdirectory, call this function giving recursion to the end of 
# the directory tree.

	while($i < $subdirnumber)
	{
		&ReplaceText($subdir[$i]);
		++$i;
	}
}