#!/usr/bin/perl
# ====================================================================
# @(#)  grab the contents of an html doc within a given element class
# ====================================================================
# $Id:$
#
#    Written by:        David Boldt                     Sept 4, 1998
#  Revisions:
# --------------------------------------------------------------------

$DocumentRoot = $ENV{'DOCUMENT_ROOT'};
$Env_Path_Translated = $ENV{'PATH_TRANSLATED'};

print &classgrep(@ARGV);

# pull out the text surrounded by an element of a given class 
# from one or more html files, with option for random selection
# of the matches
#
# USAGE:  classgrep [-r[n]] class (file|virtual)="file" ...
#     n     - number of random matches to keep (1 by default)
#     class - classname to search
#     file  - fullpathname or relative pathname of file,
#             (must be in quotes)
sub classgrep {
    local($/);         # input record separator, read whole file at once
    local($args,@args);# the list of arguments
    local($class);     # class to search for
    local($ref);       # file reference
    local($type);      # reference type, 'file' or 'virtual'
    local($file);      # pathname
    local(*DOC);       # file handle
    local($piece);     # bit of doc we have found as we are parsing
    local($tag);       # element name
    local($starting);  # boolean: new element of the right class
    local($count);     # start tag counter
    local(@results);   # array of matching pieces
    local($random);    # boolean: random selection of matching pieces
    local($selection); # text we will return
    local($index);     # array index for random selection

    # -- deal with args passed any old way

    $args = "@_";
    $args =~ s/^\s+//;
    $args =~ s/\s+$//;
    @args = split(/\s+/, $args);

    # -- command line option for random selections

    if ($args[0] =~ /^-r/) {
	$random = 1;
	if ($args[0] =~ /(\d+)$/) {
	    $random = $1;
	    shift(@args);
	}
	else {
	    shift (@args);
	    $random = shift(@args)  if ($args[0] =~ /^\d+$/);
	}
    }
    $results[0] = '';

    $class = shift(@args);
    foreach $ref (@args) {  # for each file listed
	# -- is the file a full pathname or relative (virtual)
	$ref =~ /(virtual|file)=\"([^\"]+)\"/;
	($type, $file) = ($1, $2);
	$type =~ tr/A-Z/a-z/;
	$file = &FullPath($file)  if ($type eq 'virtual');
	open(DOC, "<$file");
	$text = <DOC>; 
	close(DOC);

	foreach $piece ( split(/(<\/?[^>]+>)/, $text) ) {
	    if ($piece =~ /^<(\w+).*[cC][lL][aA][sS][sS]="$class"/o) {
		$tag = $1;
		$starting = 1;
		$count = 1;
	    }
	    elsif ($tag) {
		++$count  if ($piece =~ /^<$tag[\W]/i); # nested elements 
		if ($piece =~ /^<\/$tag>/i) {  # closing an element
		    --$count;
		    if ($count == 0) {
			$tag = ''  ;
			$results[$#results] .= "$piece\n";
			push(@results,'');
		    }
		}
	    }
	    $results[$#results] .= $piece  if ($tag);
	}
    }
    pop(@results)  unless ($results[$#results]);

    # -- random selection of results, or all of them

    if ($random) {
	srand(time||$$);
	for ($i=1;$i<=$random;++$i) {
	    $index = int(rand($#results+1));  # random array index
	    $selection .= $results[$index];   # basket of goodies to keep
	    splice(@results,$index,1);  # remove selected item from list
	}
    }
    else {  # no random arg
	$selection = join('', @results);  # return everything
    }

    return $selection;
}

sub FullPath {
    #Finds full path of a file.
    local ($file) = $_[0];
    local ($basedir);
    return $file  if ($file =~ /^$DocumentRoot/);
    return "$DocumentRoot$file"  if ($file =~ /^\//);
    ($basedir = $Env_Path_Translated) =~ s/[^\/]+$//;
    while ($file =~ s/^\.\.\///) {$basedir =~ s/[^\/]+\/$//}
    $file =~ s/\.\///;
    $file = $basedir . $file;
    $file =~ s#//#/#g;
    return $file;
}