1 |
# |
2 |
# this file implements index functions using DBI |
3 |
# and huge amounts of memory for cache speedup |
4 |
# |
5 |
|
6 |
package index_DBI; |
7 |
use strict qw(vars); |
8 |
use vars qw($Count); |
9 |
use HTML::Entities; |
10 |
|
11 |
use DBI; |
12 |
|
13 |
my %Table; # index tables which where visited in this run |
14 |
my %sth_cache; # cache prepared statements |
15 |
|
16 |
# cache var |
17 |
my $c_table; |
18 |
my $c_count; |
19 |
|
20 |
# bench time |
21 |
my $t = time(); |
22 |
|
23 |
sub new { |
24 |
my $class = shift; |
25 |
my $self = {}; |
26 |
bless($self, $class); |
27 |
|
28 |
my $dbd = shift || die "need dbi_dbd= in [global] section of configuration file"; |
29 |
my $dsn = shift || die "need dbi_dsn= in [global] section of configuration file"; |
30 |
my $user = shift || die "need dbi_user= in [global] section of configuration file"; |
31 |
my $passwd = shift || die "need dbi_passwd= in [global] section of configuration file"; |
32 |
|
33 |
$self->{dbh} = DBI->connect("DBI:$dbd:$dsn",$user,$passwd) || die $DBI::errstr; |
34 |
# begin transaction |
35 |
$self->{dbh}->begin_work || die $self->{dbh}->errstr(); |
36 |
|
37 |
$Count++; |
38 |
|
39 |
return $self; |
40 |
} |
41 |
|
42 |
sub delete_and_create { |
43 |
my $self = shift; |
44 |
|
45 |
my $field = shift; |
46 |
|
47 |
#print "#### delete_and_create($field)\n"; |
48 |
|
49 |
$self->{dbh}->commit; |
50 |
|
51 |
my $sql = "select count(*) from $field"; |
52 |
my $sth = $self->{dbh}->prepare($sql) || die $self->{dbh}->errstr(); |
53 |
# FIX: this is not a good way to check if table exists! |
54 |
if ($sth->execute() && $sth->fetchrow_hashref) { |
55 |
my $sql = "drop table $field"; |
56 |
$self->{dbh}->begin_work; |
57 |
my $sth = $self->{dbh}->do($sql) || die "SQL: $sql ".$self->{dbh}->errstr(); |
58 |
$self->{dbh}->commit; |
59 |
} |
60 |
$sql = "create table $field ( |
61 |
item varchar(255), |
62 |
ident varchar(255), |
63 |
count int, |
64 |
ord int, |
65 |
primary key (item,ident) |
66 |
)"; |
67 |
|
68 |
$self->{dbh}->begin_work; |
69 |
$sth = $self->{dbh}->do($sql); # || warn "SQL: $sql ".$self->{dbh}->errstr(); |
70 |
$self->{dbh}->commit; |
71 |
|
72 |
$self->{dbh}->begin_work; |
73 |
} |
74 |
|
75 |
sub insert { |
76 |
my $self = shift; |
77 |
|
78 |
my $field = shift; |
79 |
my $index_data = shift || print STDERR "\$index->insert($field,NULL,...)"; |
80 |
my $ident = shift || ''; # e.g. library id |
81 |
|
82 |
if (! $index_data) { |
83 |
print STDERR "\$index->insert() -- no value to insert\n"; |
84 |
return; |
85 |
} |
86 |
|
87 |
if (! $Table{$field}) { |
88 |
$self->delete_and_create($field); |
89 |
|
90 |
my $sql = "select item from $field where upper(item)=upper(?)"; |
91 |
$sth_cache{$field."select"} = $self->{dbh}->prepare($sql) || die $self->{dbh}->errstr(); |
92 |
|
93 |
$sql = "insert into $field (item,ident,count) values (?,?,?)"; |
94 |
$sth_cache{$field."insert"} = $self->{dbh}->prepare($sql) || die $self->{dbh}->errstr(); |
95 |
|
96 |
$sql = "update $field set count = count + 1 where item = ? and ident = ?"; |
97 |
$sth_cache{$field."update"} = $self->{dbh}->prepare($sql) || die $self->{dbh}->errstr(); |
98 |
} |
99 |
$Table{$field}++; |
100 |
|
101 |
#$sth_cache{$field."select"}->execute($index_data) || die "cache: $field select; ".$self->{dbh}->errstr(); |
102 |
|
103 |
# XXX for some strange reason, it seems that some entries in my |
104 |
# database produce strings which start with null byte. I suspect |
105 |
# this to be bug in OpenIsis 0.9.0. |
106 |
# This should fix it.. |
107 |
$index_data =~ s/^[^\w]+//; |
108 |
$index_data = substr($index_data,0,255); |
109 |
|
110 |
my $uc = uc($index_data); |
111 |
if (! $c_table->{$field}->{$ident}->{$uc}) { |
112 |
$sth_cache{$field."insert"}->execute($index_data,$ident,0) || warn "cache: $field insert ($index_data,$ident); ".$self->{dbh}->errstr(); |
113 |
#print stderr "in index: $index_data\n"; |
114 |
$c_table->{$field}->{$ident}->{$uc} = $index_data; |
115 |
$c_count->{$field}->{$ident}->{$uc} = 1; |
116 |
} else { |
117 |
$c_count->{$field}->{$ident}->{$uc}++; |
118 |
} |
119 |
} |
120 |
|
121 |
sub check { |
122 |
my $self = shift; |
123 |
|
124 |
my $field = shift; |
125 |
|
126 |
my $sql = "select count(*) from $field"; |
127 |
|
128 |
my $sth = $self->{dbh}->prepare($sql) || die $self->{dbh}->errstr(); |
129 |
$sth->execute() || die "sql: $sql; ".$self->{dbh}->errstr(); |
130 |
|
131 |
my ($total) = $sth->fetchrow_array(); |
132 |
|
133 |
return $total; |
134 |
} |
135 |
|
136 |
|
137 |
sub fetch { |
138 |
my $self = shift; |
139 |
|
140 |
my $field = shift; |
141 |
my $what = shift || 'item'; # 'item,ident' |
142 |
my $where = shift; |
143 |
|
144 |
my $from_ord = shift || 0; |
145 |
my $rows = shift || 10; |
146 |
|
147 |
my @sql_args; |
148 |
|
149 |
my $sql = "select $what,ord from $field"; |
150 |
|
151 |
if ($where) { |
152 |
my $sql2 = " select ord from $field where upper($what) like upper(?)||'%'"; |
153 |
my $sth = $self->{dbh}->prepare($sql2) || die "sql2: $sql2; ".$self->{dbh}->errstr(); |
154 |
|
155 |
$sth->execute($where) || die "sql2: $sql2; ".$self->{dbh}->errstr(); |
156 |
if (my $row = $sth->fetchrow_hashref) { |
157 |
$from_ord += $row->{ord} - 1; |
158 |
} |
159 |
} |
160 |
$sql .= " order by ord limit $rows offset $from_ord"; |
161 |
|
162 |
my $sth = $self->{dbh}->prepare($sql) || die "prepare: $sql; ".$self->{dbh}->errstr(); |
163 |
$sth->execute() || die "execute: $sql; ".$self->{dbh}->errstr(); |
164 |
my @arr; |
165 |
while (my $row = $sth->fetchrow_hashref) { |
166 |
$row->{item} = HTML::Entities::encode($row->{item},'<>&"'); |
167 |
push @arr,$row; |
168 |
} |
169 |
return @arr; |
170 |
} |
171 |
|
172 |
sub close { |
173 |
my $self = shift; |
174 |
|
175 |
|
176 |
# re-create ord column (sorted order) in table |
177 |
sub create_ord { |
178 |
|
179 |
my $table = shift; |
180 |
|
181 |
$self->{dbh}->begin_work || die $self->{dbh}->errstr(); |
182 |
|
183 |
my $sql = "select oid from $table order by upper(item)"; |
184 |
my $sth = $self->{dbh}->prepare($sql) || die "sql: $sql; ".$self->{dbh}->errstr(); |
185 |
$sql = "update $table set ord=? where oid=?"; |
186 |
my $sth_update = $self->{dbh}->prepare($sql) || die "sql: $sql; ".$self->{dbh}->errstr(); |
187 |
$sth->execute() || die "sql: $sql; ".$self->{dbh}->errstr(); |
188 |
my $ord = 1; |
189 |
while (my $row = $sth->fetchrow_hashref) { |
190 |
$sth_update->execute($ord++,$row->{oid}); |
191 |
} |
192 |
|
193 |
$self->{dbh}->commit || die $self->{dbh}->errstr(); |
194 |
} |
195 |
#--- end of sub |
196 |
|
197 |
if ($self->{dbh}) { |
198 |
|
199 |
# commit |
200 |
$self->{dbh}->commit || die $self->{dbh}->errstr(); |
201 |
|
202 |
foreach my $table (keys %Table) { |
203 |
# FIX |
204 |
print STDERR "last operation took ",time()-$t," seconds...\n"; |
205 |
$t=time(); |
206 |
print STDERR "creating ord for $table...\n"; |
207 |
create_ord($table); |
208 |
undef $sth_cache{$table."select"}; |
209 |
undef $sth_cache{$table."insert"}; |
210 |
undef $sth_cache{$table."update"}; |
211 |
# XXX |
212 |
# $sth_cache{$field."update"}->execute($index_data,$ident) || die "cache: $field update; ".$self->{dbh}->errstr(); |
213 |
} |
214 |
|
215 |
$self->{dbh}->disconnect; |
216 |
undef $self->{dbh}; |
217 |
} |
218 |
} |
219 |
|
220 |
END { |
221 |
$Count--; |
222 |
print STDERR "index_DBI fatal error: \$index->close() not called... $Count references left!\n" if ($Count > 0); |
223 |
# FIX: debug output |
224 |
# print STDERR "usage\ttable\n"; |
225 |
# foreach (keys %Table) { |
226 |
# print STDERR $Table{$_},"\t$_\n"; |
227 |
# } |
228 |
} |
229 |
|
230 |
1; |