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@31: $number_of_printed = 0;
igor@31: $number_of_printed_max = 10;
igor@5: #if (open(PAGE, "lynx -dump '$page'|")) {
igor@5: if (open(PAGE, "$page")) {
igor@29:     binmode PAGE,":utf8";
igor@5:     local $/;
igor@5:     $text=<PAGE>;
igor@5:     $text =~ s@http://[a-zA-Z&_.:/0-9%?=,\#+()\[\]~-]*@@g;
igor@5:     $text =~ s@\n@@g;
igor@28:     $text =~ s@(Mr|Mrs|viz)\.@\1POINT@g;
igor@28:     $text =~ s@e\.g\.@ePOINTgPOINT@g;
igor@28:     $text =~ s@i\.e\.@iPOINTePOINT@g;
igor@19:     @sentences=split /[.!?]/, $text;
igor@5:     for (@sentences) {
igor@28:         s@iPOINTePOINT@i\.e\.@g;
igor@28:         s@ePOINTgPOINT@e\.g\.@g;
igor@28:         s@(Mr|Mrs|viz)POINT@\1.@g;
igor@5:         s/^\s*//;
igor@5:         s/\s*$//;
igor@5:         s/\[[0-9]+\]//g;
igor@5:         s/\s+/ /g;
igor@31:         if (/\b$regexp\b/ and $number_of_printed < $number_of_printed_max ) {
igor@31:             print "$_.\n\n";
igor@31:             $number_of_printed++;
igor@31:         }
igor@5:     }
igor@5: }