version 1.136, 2004/09/05 14:57:05
|
version 1.138, 2004/11/21 13:37:29
|
Line 11 use locale;
|
Line 11 use locale;
|
use vars qw($opt_z); |
use vars qw($opt_z); |
use Getopt::Std; |
use Getopt::Std; |
#my ($dbuser,$dbname,$dbpass,$dbhost); |
#my ($dbuser,$dbname,$dbpass,$dbhost); |
require "dbdefs.pl"; |
eval {require "dbdefs.pl";} ; |
my $url=url||''; |
my $url=url||''; |
|
my @used_stop=(); |
$dbuser||="piataev"; |
$dbuser||="piataev"; |
$dbname||="chgk"; |
$dbname||="chgk"; |
$dbpass||=""; |
$dbpass||=""; |
Line 500 sub russearch {
|
Line 501 sub russearch {
|
my %relevance; |
my %relevance; |
my @blob; |
my @blob; |
my %count; |
my %count; |
|
my %stop_word; |
POSIX::setlocale( &POSIX::LC_ALL, $thislocale ); |
POSIX::setlocale( &POSIX::LC_ALL, $thislocale ); |
$sstr=~tr/йцукенгшщзхъфывапролджэячсмитьбю/ЙЦУКЕНГШЩЗХЪФЫВАПРОЛДЖЭЯЧСМИТЬБЮ/; |
$sstr=~tr/йцукенгшщзхъфывапролджэячсмитьбю/ЙЦУКЕНГШЩЗХЪФЫВАПРОЛДЖЭЯЧСМИТЬБЮ/; |
# @qw=@w =split (' ', uc $sstr); |
# @qw=@w =split (' ', uc $sstr); |
my $ts=uc $sstr; |
my $ts=uc $sstr; |
@qw=@w= $ts=~m/(?:(?:${RLrl})+)|(?:[A-Za-z0-9]+)/gom; |
@qw=@w= $ts=~m/(?:(?:${RLrl})+)|(?:[A-Za-z0-9]+)/gom; |
|
$query="select nf.word from nf where number>=50000"; |
|
$sth=$dbh->prepare($query); |
|
$sth->execute(); |
|
%stop_word=(); |
|
while (@arr = $sth->fetchrow) |
|
{ |
|
$stop_word{$arr[0]}=1; |
|
} |
|
$sth->finish; |
|
|
|
|
#----------- |
#----------- |
foreach $i (0..$#w) # заполняем массив @nf начальных форм |
foreach $i (0..$#w) # заполняем массив @nf начальных форм |
# $nf[$i] -- ссылка на массив возможных |
# $nf[$i] -- ссылка на массив возможных |
# начальных форм словоформы $i |
# начальных форм словоформы $i |
{ |
{ |
|
(push @used_stop, uc $w[$i]),next if $stop_word{uc $w[$i]}; |
$qw= $dbh->quote (uc $w[$i]); |
$qw= $dbh->quote (uc $w[$i]); |
|
|
$query=" select distinct w2 from nests |
$query=" select distinct w2 from nests |
where w1=$qw"; |
where w1=$qw"; |
$sth=$dbh -> prepare($query); |
$sth=$dbh -> prepare($query); |
Line 557 $sstr=~tr/йцукенгшщзхъфывапролджэячсмить
|
Line 570 $sstr=~tr/йцукенгшщзхъфывапролджэячсмить
|
$_= " word2question.word=$_" foreach @arr; |
$_= " word2question.word=$_" foreach @arr; |
$_= " nf.id=".$_. ' ' foreach @arr1; |
$_= " nf.id=".$_. ' ' foreach @arr1; |
# @arr=(0) unless @arr; |
# @arr=(0) unless @arr; |
$query="select questions from word2question where (". (join ' OR ', @arr).") AND length(questions)<80000"; |
$query="select questions from word2question where (". (join ' OR ', @arr).") "; |
|
|
$sth=$dbh -> prepare($query); |
$sth=$dbh -> prepare($query); |
$sth->execute; |
$sth->execute; |
Line 945 sub PrintSearch {
|
Line 958 sub PrintSearch {
|
|
|
|
|
$Output.= p. "Время поиска: " . (time-$t) ." сек.".p; |
$Output.= p. "Время поиска: " . (time-$t) ." сек.".p; |
|
$_="\"$_\"" foreach @used_stop; |
|
$Output.= p. (join ', ',@used_stop) ." ignored".p if @used_stop; |
my ($output, $i, $suffix, $hits) = ('', 0, '', $#Questions + 1); |
my ($output, $i, $suffix, $hits) = ('', 0, '', $#Questions + 1); |
|
|
my $shablon; |
my $shablon; |