#!/usr/bin/perl
# searchtext
# Google-style searching through text files on your *nix computer
# http://nodivisions.com/software/searchtext/
# By Anthony DiSante, http://nodivisions.com/contact/
#
# Changelog:
#
# 20041007:
#
# Added "-type f" to the find command, put its -iname argument in
# single-quotes, and added *.shtml as one of the "alltext" extensions.
#
# 20040614:
#
# Came across a webpage using "stm" for its extension, so I added
# that to the default text file list.
#
# 20040521:
#
# First public release.

use strict;
my (%PREF, $msg, $datapath_qs);
my $opts = $ARGV[0];
my $path = $ARGV[1];
my $ext = $ARGV[2];
my $terms = $ARGV[3];
$terms =~ s/'/"/g;
#die "terms: $terms\n";
my $me = `basename $0`;
chomp $me;

die qq`
$me: does a Google-style search of files on your Unix/Linux computer.

Usage: $me <options> <searchpath> <extension(s)|alltext> <terms>

- options is either h or t (for html or plaintext output).

- searchpath will be searched recursively.

- extension is a comma-separated list of one or more extensions (no spaces);
  only files with these extensions will be searched for your search terms.
  or you can say "alltext" here to search most common text file extensions
  (txt,log,htm,html,shtml,php,php3,asp,cfm,conf,stm).

- terms is any number of words or phrases; put the whole thing in double-
  quotes and put phrases in single-quotes.

Examples:
$me -t /home/me/files/ txt,log "foobar"
$me -t /home/me/files/ alltext "foobar 'kick the can' ketchup"
` unless ($opts && $path && $ext && $terms);


$path = `pwd` if $path eq '.';
chomp $path;


search_posts($path,$ext,$terms);


sub search_posts($$$)
{
	my($path,$ext,$string) = @_;

	my $original_string = $string;
	my ($front, $mid, $back);

	# Get the string ready for searching.  We're going to do this Google-stylee...
	# to get a match, multiple terms separated by spaces must ALL be found in the
	# post, unless (some of) the terms are in quotes, in which case, that exact
	# sequence of terms must be found in the post.

	# First condense any multiple spaces:
	$string =~ s/\s+/ /g;

	# If the string contains any quoted phrases, temporarily replace spaces
	# within those phrases with a token that we can check for later:
	while($string =~ /(.*)"(.+?)"(.*)/)
	{
		$front	= $1;
		$mid	= $2;
		$back	= $3;

		$mid =~ s/\s/\+\+TOKEN\+\+/g;

		$string = $front . $mid . $back;
	}

	# So now a search like this:
	#
	# 	blue green "black and white"
	#
	# ...becomes this:
	#
	#	blue green black++TOKEN++and++TOKEN++white

	# Now split the string on its spaces into an array:
	my @search_terms = split(/\s/, $string);

	# Now we're all set.  To find a post that matches, we just need to make sure
	# that it contains every term (some of which may be phrases) in @search_terms.

	my @files = get_filenames($path,$ext);
	#print join "\n", @files;
	#die;

	my @results;
	my $num_found = 0;
	my ($highlight_start, $highlight_end) = $opts =~ /t/ ? ('','') : ('<b>','</b>');

	for(@files)
	{
		chomp;
		my $file = $_;
		my $summary = '';

		if( !(-e "$file") )
		{
			next;
		}

		open(IN,"<$file") or die "$0: couldn't open dfg4 $file: $!\n";
		flock IN, 2;
		seek IN, 0, 0;
		my $whole_post = '';
		while(<IN>)
		{
			$whole_post .= $_;
		}
		close IN or die "$0: couldn't close $file: $!\n";

		compress_whitespace($whole_post);
		strip_html($whole_post) if $opts =~ /t/;


		# Assume all terms are present, and if we find one that isn't, then we'll unset it.
		my $it_matches = 1;
		for(@search_terms)
		{
			# First re-replace our tokens with spaces:
			s/\+\+TOKEN\+\+/ /g;

			# Now do the actual checking.  We surround the search term with "\b"s which are
			# word boundaries, because otherwise a search for "no" would match "no" but also
			# match "know" and "not" which is probably not what was intended.
			if($whole_post !~ /\b$_\b/i)
			{
				$it_matches = 0;
				last;
			}
		}


		if($it_matches)
		{
			$num_found++;

			for(@search_terms)
			{
				my ($pre, $term, $post) = ($whole_post =~ /(\b.{0,200}\b)($_)(\b.{0,200}\b)/si);
				$summary = "$summary ... $pre$highlight_start$term$highlight_end$post";
			}

			$summary .= '... ';

			if($opts =~ /t/)
			{
				push @results, "$num_found. File:\nfile://$file\nSummary:\n$summary\n\n";
			}
			else
			{
				push @results, "<p style=\"width: 500px;\">$num_found. <a href=\"$file\">$file</a><br>$summary</p>\n\n\n\n";
			}
		}
	}


	print	  "<p style=\"border: 1px solid; padding: 3px 3px 3px 3px;\">Note: this search "
		. "works just like Google.&nbsp; To find posts containing "
		. "the word <b>black</b> and the word <b>white</b>, just type <b>black white</b> into the box.&nbsp; To "
		. "find posts containing the <i>phrase </i> <b>black and white</b>, type <b>\"black and white\"</b> in "
		. "quotes.&nbsp; Without the quotes, you'll find posts containing all the words <b>black</b>, <b>and</b>, "
		. "and <b>white</b>, even if they're not together in a phrase.&nbsp; Note also that you can search for "
		. "multiple phrases and/or single terms at the same time.\n</p>\n\n" unless $opts =~ /t/;




	print "Search Results:\n\n";
	if($num_found < 1)	{ print "No files";		}
	if($num_found == 1)	{ print "1 file";		}
	if($num_found > 1)	{ print "$num_found files";	}
	print " found containing: $original_string.\n\n";


	print @results;


	print	  "<br><br><form style=\"border: 1px solid; padding: 3px 3px 3px 3px;\" method=POST \n"
		. "action=\"$PREF{'here_script'}?${datapath_qs}search\">Advanced search:<br><br><input type=text "
		. "name=string class=default value=\"$original_string\">\n<br>"
		. "<input type=checkbox name=names>search through poster names"
		. "<br><input type=checkbox name=emails>search through poster email addresses"
		. "<br><input type=checkbox name=dates>search through post dates"
		. "<br><input type=checkbox name=subjects>search through post subjects"
		. "<br><input type=checkbox name=bodies>search through post bodies"
		. "<br><br>(You must check at least one.&nbsp; The normal search checks through all of these.)"
		. "</form>\n\n" unless $opts =~ /t/;

	print "\n$num_found files matched.\n" if $opts =~ /t/;

}


sub get_filenames($$)
{
	my($path,$ext) = @_;
	my (@exts,@files);

	if($ext eq 'alltext')
	{
		@exts = ('txt','log','htm','html','shtml','php','php3','asp','cfm','conf','stm');
	}
	else
	{
		@exts = split(/,\s*/, $ext);
	}

	for(@exts)
	{
		#print "debug: find $path -type f -iname '*.$_'\n";
		my @f = `find $path -type f -iname '*.$_'`;
		push @files, @f;
	}

	return @files;
}


sub strip_html
{
	for(@_) { s/<.+?>//g; }
}


sub compress_whitespace
{
	# turns any \s (which means newlines, etc) into a single space.
	for(@_) { s/\s+/ /g; }
}