igor@5: #!/usr/bin/perl
igor@5: 
igor@29: use utf8;
igor@29: use Encode;
igor@29: binmode STDIN,":utf8";
igor@29: binmode STDOUT,":utf8";
igor@5: 
igor@6: $regexp=$ARGV[0];
igor@29: $regexp=decode("utf8",$regexp);
igor@29: 
igor@6: $page=$ARGV[1];
igor@62: shift @ARGV;
igor@31: $number_of_printed = 0;
igor@31: $number_of_printed_max = 10;
igor@5: #if (open(PAGE, "lynx -dump '$page'|")) {
igor@62: for $page (@ARGV) {
igor@62:     if (open(PAGE, "$page")) {
igor@62:         binmode PAGE,":utf8";
igor@62:         local $/;
igor@62:         $text=<PAGE>;
igor@62:         $text =~ s@http://[a-zA-Z&_.:/0-9%?=,\#+()\[\]~-]*@@g;
igor@62:         $text =~ s@\n@@g;
igor@62:         $text =~ s@(Mr|Mrs|viz)\.@\1POINT@g;
igor@62:         $text =~ s@e\.g\.@ePOINTgPOINT@g;
igor@62:         $text =~ s@i\.e\.@iPOINTePOINT@g;
igor@62:         @sentences=split /[.!?]/, $text;
igor@62:         for (@sentences) {
igor@62:             s@iPOINTePOINT@i\.e\.@g;
igor@62:             s@ePOINTgPOINT@e\.g\.@g;
igor@62:             s@(Mr|Mrs|viz)POINT@\1.@g;
igor@62:             s/^\s*//;
igor@62:             s/\s*$//;
igor@62:             s/\[[0-9]+\]//g;
igor@62:             s/\s+/ /g;
igor@62:             if (/\b$regexp\b/ and $number_of_printed < $number_of_printed_max ) {
igor@62:                 $mark_start = "\e[35;1m";
igor@62:                 $mark_stop = "\e[0m";
igor@62:                 s/\b($regexp)\b/$mark_start$1$mark_stop/;
igor@62:                 print "$_.\n\n";
igor@62:                 $number_of_printed++;
igor@62:             }
igor@62:             if ($number_of_printed >= $number_of_printed_max) {
igor@62:                 exit(0);
igor@62:             };
igor@31:         }
igor@5:     }
igor@5: }