#! /usr/local/bin/perl -c
package referer_keyword;
# -------------------------------------------------------------------
#   referer_keywords
# -------------------------------------------------------------------
#  与えられた URL から「検索用キーワードらしきもの」を抜き出します
#
#  version:
#    2000.11.11   aiba
# -------------------------------------------------------------------

#
sub get {
  local($za) = @_;

  if($za =~ /\?/) {
    if($za =~ /[\?\&](query|queryterm|general|suche|[Pp]arole|entry)\=[^\&]+/ ||
       $za =~ /[\?\&](word|[Kk]eywords?|searchfor|raw\_keywords)\=[^\&]+/ ||
       $za =~ /[\?\&](fi\_1)\=[^\&]+/ ||  # excite.com
       $za =~ /[\?\&](q|qr|qt|s|kw|key|ask|rawq)\=[^\&]+/ ||
       $za =~ /[\?\&](p|r|mt|MT|QRY|KW|AW0|st)\=[^\&]+/ ||
       $za =~ /[\?\&](phrase|QueryString|AK[0-9]|box|RAW\_KEYWORDS)\=[^\&]+/ ||
       $za =~ /[\?\&]([Ss]earch|request|qu|w|LsSearchReq)\=[^\&]+/ ||
       $za =~ /[\?\&](FI\_[0-9]|QUERY|genre|rads|saisie|Parola[0-9])|SearchWd\=[^\&]+/ ) {
      local($zb) = $&;
      $zb =~ s/^[^\=]*\=[ \t]*//;
      if($zb =~ /./) {
				# 以下はデータの整形部分
	$zb = &chr_normalize($zb);
	$zb = " $zb ";   $zb =~ y/A-Z/a-z/;
	$zb =~ s/[\"\t\r\+\|\/\(\),;:\[\]\<\>\=\+]/ /g;
	$zb =~ s/\*+/\*/g;
	$zb =~ s/ \-[^ ]*/ /g;
	$zb =~ s/ [\'\#\!\\\_]+/ /g;
	$zb =~ s/[\'\#\!\\\_\~]+ / /g;
	$zb =~ s/ (AND|and|OR|or|\&) / /g;

	# 先頭と末尾につけた空白を抜く
	$zb =~ s/^[ \t]*//; $zb =~ s/[ \t\r\n]*$//;
	return split(/ +/,$zb);
      }
    }
  }
  return undef;
}

# -----------------------------------------------
#  文字列の表記を統一する
#
sub chr_normalize {
  local($za) = @_;
  
  $za =~ s/%([0-9A-Fa-f][0-9A-Fa-f])/pack("c",hex($1))/eg;
  if($za =~ /[^ -~]/) {
    require 'jcode.pl';
    &jcode'convert(*za, 'sjis');
    &jcode'h2z_sjis(*za);
    &jcode'convert(*za, 'euc');
    &jcode'tr(*za, "Ａ-ｚ","A-z");
    &jcode'tr(*za, "０-９","0-9");
    &jcode'tr(*za, "　（）［］／￥！？｜"," ()[]/\\!?\|");
  }
  return $za;
}

1;