2 |
# this file implements index functions using DBI |
# this file implements index functions using DBI |
3 |
# and huge amounts of memory for cache speedup |
# and huge amounts of memory for cache speedup |
4 |
# |
# |
5 |
|
# this version doesn't support ident (which sould be location in |
6 |
|
# library). But, that functionality is not used anyway... |
7 |
|
# |
8 |
|
|
9 |
package index_DBI; |
package index_DBI; |
10 |
use strict qw(vars); |
use strict qw(vars); |
21 |
my $c_count; |
my $c_count; |
22 |
|
|
23 |
# bench time |
# bench time |
24 |
my $t = time(); |
my $bench_time = time(); |
25 |
|
|
26 |
|
sub bench { |
27 |
|
my $self = shift; |
28 |
|
my $msg = shift; |
29 |
|
|
30 |
|
print STDERR "last operation took ",time()-$bench_time," seconds...\n"; |
31 |
|
$bench_time=time(); |
32 |
|
print STDERR "$msg\n"; |
33 |
|
} |
34 |
|
|
35 |
sub new { |
sub new { |
36 |
my $class = shift; |
my $class = shift; |
43 |
my $passwd = shift || die "need dbi_passwd= in [global] section of configuration file"; |
my $passwd = shift || die "need dbi_passwd= in [global] section of configuration file"; |
44 |
|
|
45 |
$self->{dbh} = DBI->connect("DBI:$dbd:$dsn",$user,$passwd) || die $DBI::errstr; |
$self->{dbh} = DBI->connect("DBI:$dbd:$dsn",$user,$passwd) || die $DBI::errstr; |
|
# begin transaction |
|
|
$self->{dbh}->begin_work || die $self->{dbh}->errstr(); |
|
|
|
|
46 |
$Count++; |
$Count++; |
47 |
|
|
48 |
|
$self->bench("connected to $dbd as $user"); |
49 |
|
|
50 |
return $self; |
return $self; |
51 |
} |
} |
52 |
|
|
57 |
|
|
58 |
#print "#### delete_and_create($field)\n"; |
#print "#### delete_and_create($field)\n"; |
59 |
|
|
|
$self->{dbh}->commit; |
|
|
|
|
60 |
my $sql = "select count(*) from $field"; |
my $sql = "select count(*) from $field"; |
61 |
my $sth = $self->{dbh}->prepare($sql) || die $self->{dbh}->errstr(); |
my $sth = $self->{dbh}->prepare($sql) || die $self->{dbh}->errstr(); |
62 |
# FIX: this is not a good way to check if table exists! |
# FIX: this is not a good way to check if table exists! |
63 |
if ($sth->execute() && $sth->fetchrow_hashref) { |
if ($sth->execute() && $sth->fetchrow_hashref) { |
64 |
my $sql = "drop table $field"; |
my $sql = "drop table $field"; |
|
$self->{dbh}->begin_work; |
|
65 |
my $sth = $self->{dbh}->do($sql) || die "SQL: $sql ".$self->{dbh}->errstr(); |
my $sth = $self->{dbh}->do($sql) || die "SQL: $sql ".$self->{dbh}->errstr(); |
|
$self->{dbh}->commit; |
|
66 |
} |
} |
67 |
$sql = "create table $field ( |
$sql = "create table $field ( |
68 |
item varchar(255), |
item varchar(255), |
|
ident varchar(255), |
|
69 |
count int, |
count int, |
70 |
ord int, |
ord int, |
71 |
primary key (item,ident) |
primary key (item) |
72 |
)"; |
)"; |
73 |
|
|
74 |
$self->{dbh}->begin_work; |
$sth = $self->{dbh}->do($sql) || warn "SQL: $sql ".$self->{dbh}->errstr(); |
|
$sth = $self->{dbh}->do($sql); # || warn "SQL: $sql ".$self->{dbh}->errstr(); |
|
|
$self->{dbh}->commit; |
|
|
|
|
|
$self->{dbh}->begin_work; |
|
75 |
} |
} |
76 |
|
|
77 |
sub insert { |
sub insert { |
86 |
return; |
return; |
87 |
} |
} |
88 |
|
|
|
if (! $Table{$field}) { |
|
|
$self->delete_and_create($field); |
|
|
|
|
|
my $sql = "select item from $field where upper(item)=upper(?)"; |
|
|
$sth_cache{$field."select"} = $self->{dbh}->prepare($sql) || die $self->{dbh}->errstr(); |
|
|
|
|
|
$sql = "insert into $field (item,ident,count) values (?,?,?)"; |
|
|
$sth_cache{$field."insert"} = $self->{dbh}->prepare($sql) || die $self->{dbh}->errstr(); |
|
|
|
|
|
$sql = "update $field set count = count + 1 where item = ? and ident = ?"; |
|
|
$sth_cache{$field."update"} = $self->{dbh}->prepare($sql) || die $self->{dbh}->errstr(); |
|
|
} |
|
89 |
$Table{$field}++; |
$Table{$field}++; |
90 |
|
|
91 |
#$sth_cache{$field."select"}->execute($index_data) || die "cache: $field select; ".$self->{dbh}->errstr(); |
#$sth_cache{$field."select"}->execute($index_data) || die "cache: $field select; ".$self->{dbh}->errstr(); |
92 |
|
|
93 |
|
# XXX for some strange reason, it seems that some entries in my |
94 |
|
# database produce strings which start with null byte. I suspect |
95 |
|
# this to be bug in OpenIsis 0.9.0. |
96 |
|
# This should fix it.. |
97 |
|
$index_data =~ s/^[^\w]+//; |
98 |
$index_data = substr($index_data,0,255); |
$index_data = substr($index_data,0,255); |
99 |
|
|
100 |
my $uc = uc($index_data); |
my $uc = uc($index_data); |
101 |
if (! $c_table->{$field}->{$ident}->{$uc}) { |
if (! $c_table->{$field}->{$uc}) { |
|
$sth_cache{$field."insert"}->execute($index_data,$ident,0) || warn "cache: $field insert ($index_data,$ident); ".$self->{dbh}->errstr(); |
|
102 |
#print stderr "in index: $index_data\n"; |
#print stderr "in index: $index_data\n"; |
103 |
$c_table->{$field}->{$ident}->{$uc} = $index_data; |
$c_table->{$field}->{$uc} = $index_data; |
104 |
$c_count->{$field}->{$ident}->{$uc} = 1; |
$c_count->{$field}->{$uc} = 1; |
105 |
} else { |
} else { |
106 |
$c_count->{$field}->{$ident}->{$uc}++; |
$c_count->{$field}->{$uc}++; |
107 |
} |
} |
108 |
} |
} |
109 |
|
|
138 |
my $sql = "select $what,ord from $field"; |
my $sql = "select $what,ord from $field"; |
139 |
|
|
140 |
if ($where) { |
if ($where) { |
141 |
my $sql2 = " select ord from $field where upper($what) like upper(?)||'%'"; |
my $sql2 = "select ord from $field where upper($what) like upper(?)||'%'"; |
142 |
my $sth = $self->{dbh}->prepare($sql2) || die "sql2: $sql2; ".$self->{dbh}->errstr(); |
my $sth = $self->{dbh}->prepare($sql2) || die "sql2: $sql2; ".$self->{dbh}->errstr(); |
143 |
|
|
144 |
$sth->execute($where) || die "sql2: $sql2; ".$self->{dbh}->errstr(); |
$sth->execute($where) || die "sql2: $sql2; ".$self->{dbh}->errstr(); |
145 |
if (my $row = $sth->fetchrow_hashref) { |
if (my $row = $sth->fetchrow_hashref) { |
146 |
$from_ord += $row->{ord} - 1; |
$from_ord += $row->{ord} - 1; |
147 |
|
} else { |
148 |
|
# if no match is found when searching from beginning |
149 |
|
# of word in index, try substring match anywhere |
150 |
|
$sql2 = "select ord from $field where upper($what) like '%'||upper(?)||'%'"; |
151 |
|
$sth = $self->{dbh}->prepare($sql2) || die "sql2: $sql2; ".$self->{dbh}->errstr(); |
152 |
|
$sth->execute($where) || die "sql2: $sql2; ".$self->{dbh}->errstr(); |
153 |
|
if (my $row = $sth->fetchrow_hashref) { |
154 |
|
$from_ord += $row->{ord} - 1; |
155 |
|
} |
156 |
} |
} |
157 |
} |
} |
158 |
$sql .= " order by ord limit $rows offset $from_ord"; |
$sql .= " order by ord limit $rows offset $from_ord"; |
170 |
sub close { |
sub close { |
171 |
my $self = shift; |
my $self = shift; |
172 |
|
|
173 |
|
return if (! $self->{dbh}); |
174 |
|
|
175 |
# re-create ord column (sorted order) in table |
foreach my $table (keys %Table) { |
176 |
sub create_ord { |
$self->bench("Crating table $table"); |
177 |
|
$self->delete_and_create($table); |
|
my $table = shift; |
|
178 |
|
|
179 |
$self->{dbh}->begin_work || die $self->{dbh}->errstr(); |
$self->{dbh}->begin_work || die $self->{dbh}->errstr(); |
180 |
|
|
181 |
my $sql = "select oid from $table order by upper(item)"; |
$self->bench("Sorting ".$Table{$table}." items in $table"); |
182 |
|
my @keys = sort keys %{$c_table->{$table}}; |
183 |
|
|
184 |
|
$self->bench("Dumping data into $table"); |
185 |
|
my $sql = "insert into $table (ord,item,count) values (?,?,?)"; |
186 |
my $sth = $self->{dbh}->prepare($sql) || die "sql: $sql; ".$self->{dbh}->errstr(); |
my $sth = $self->{dbh}->prepare($sql) || die "sql: $sql; ".$self->{dbh}->errstr(); |
187 |
$sql = "update $table set ord=? where oid=?"; |
|
188 |
my $sth_update = $self->{dbh}->prepare($sql) || die "sql: $sql; ".$self->{dbh}->errstr(); |
my $ord = 0; |
189 |
$sth->execute() || die "sql: $sql; ".$self->{dbh}->errstr(); |
foreach my $key (@keys) { |
190 |
my $ord = 1; |
$sth->execute(++$ord, |
191 |
while (my $row = $sth->fetchrow_hashref) { |
$c_table->{$table}->{$key}, |
192 |
$sth_update->execute($ord++,$row->{oid}); |
$c_count->{$table}->{$key} |
193 |
|
); |
194 |
} |
} |
195 |
|
|
196 |
$self->{dbh}->commit || die $self->{dbh}->errstr(); |
$self->{dbh}->commit || die $self->{dbh}->errstr(); |
197 |
} |
} |
198 |
#--- end of sub |
$self->bench("disconnecting from database"); |
199 |
|
|
200 |
if ($self->{dbh}) { |
$self->{dbh}->disconnect; |
201 |
|
undef $self->{dbh}; |
|
# commit |
|
|
$self->{dbh}->commit || die $self->{dbh}->errstr(); |
|
|
|
|
|
foreach my $table (keys %Table) { |
|
|
# FIX |
|
|
print STDERR "last operation took ",time()-$t," seconds...\n"; |
|
|
$t=time(); |
|
|
print STDERR "creating ord for $table...\n"; |
|
|
create_ord($table); |
|
|
undef $sth_cache{$table."select"}; |
|
|
undef $sth_cache{$table."insert"}; |
|
|
undef $sth_cache{$table."update"}; |
|
|
# XXX |
|
|
# $sth_cache{$field."update"}->execute($index_data,$ident) || die "cache: $field update; ".$self->{dbh}->errstr(); |
|
|
} |
|
|
|
|
|
$self->{dbh}->disconnect; |
|
|
undef $self->{dbh}; |
|
|
} |
|
202 |
} |
} |
203 |
|
|
204 |
END { |
END { |