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); |
11 |
use vars qw($Count); |
use vars qw($Count); |
12 |
use HTML::Entities; |
use HTML::Entities; |
13 |
|
use URI::Escape; |
14 |
|
|
15 |
use DBI; |
use DBI; |
16 |
|
|
21 |
my $c_table; |
my $c_table; |
22 |
my $c_count; |
my $c_count; |
23 |
|
|
24 |
|
# bench time |
25 |
|
my $bench_time = time(); |
26 |
|
|
27 |
|
sub bench { |
28 |
|
my $self = shift; |
29 |
|
my $msg = shift; |
30 |
|
|
31 |
|
print STDERR "last operation took ",time()-$bench_time," seconds...\n"; |
32 |
|
$bench_time=time(); |
33 |
|
print STDERR "$msg\n"; |
34 |
|
} |
35 |
|
|
36 |
sub new { |
sub new { |
37 |
my $class = shift; |
my $class = shift; |
38 |
my $self = {}; |
my $self = {}; |
44 |
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"; |
45 |
|
|
46 |
$self->{dbh} = DBI->connect("DBI:$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(); |
|
|
|
|
47 |
$Count++; |
$Count++; |
48 |
|
|
49 |
|
$self->bench("connected to $dbd as $user"); |
50 |
|
|
51 |
return $self; |
return $self; |
52 |
} |
} |
53 |
|
|
58 |
|
|
59 |
#print "#### delete_and_create($field)\n"; |
#print "#### delete_and_create($field)\n"; |
60 |
|
|
|
$self->{dbh}->commit; |
|
|
|
|
61 |
my $sql = "select count(*) from $field"; |
my $sql = "select count(*) from $field"; |
62 |
my $sth = $self->{dbh}->prepare($sql) || die $self->{dbh}->errstr(); |
my $sth = $self->{dbh}->prepare($sql) || die $self->{dbh}->errstr(); |
63 |
# FIX: this is not a good way to check if table exists! |
# FIX: this is not a good way to check if table exists! |
64 |
if ($sth->execute() && $sth->fetchrow_hashref) { |
if ($sth->execute() && $sth->fetchrow_hashref) { |
65 |
my $sql = "drop table $field"; |
my $sql = "drop table $field"; |
|
$self->{dbh}->begin_work; |
|
66 |
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; |
|
67 |
} |
} |
68 |
$sql = "create table $field ( |
$sql = "create table $field ( |
69 |
item varchar(255), |
item varchar(255), |
70 |
ident varchar(255), |
display text, |
71 |
count int, |
count int, |
72 |
ord int, |
ord int, |
73 |
primary key (item,ident) |
primary key (item) |
74 |
)"; |
)"; |
75 |
|
|
76 |
$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; |
|
77 |
} |
} |
78 |
|
|
79 |
sub insert { |
sub insert { |
81 |
|
|
82 |
my $field = shift; |
my $field = shift; |
83 |
my $index_data = shift || print STDERR "\$index->insert($field,NULL,...)"; |
my $index_data = shift || print STDERR "\$index->insert($field,NULL,...)"; |
84 |
my $ident = shift || ''; # e.g. library id |
my $display = shift || $index_data; |
85 |
|
|
86 |
if (! $index_data) { |
if (! $index_data) { |
87 |
print STDERR "\$index->insert() -- no value to insert\n"; |
print STDERR "\$index->insert() -- no value to insert\n"; |
88 |
return; |
return; |
89 |
} |
} |
90 |
|
|
|
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(); |
|
|
} |
|
91 |
$Table{$field}++; |
$Table{$field}++; |
92 |
|
|
93 |
#$sth_cache{$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(); |
94 |
|
|
95 |
|
# XXX for some strange reason, it seems that some entries in my |
96 |
|
# database produce strings which start with null byte. I suspect |
97 |
|
# this to be bug in OpenIsis 0.9.0. |
98 |
|
# This should fix it.. |
99 |
|
$index_data =~ s/^[^\w]+//; |
100 |
$index_data = substr($index_data,0,255); |
$index_data = substr($index_data,0,255); |
101 |
|
|
102 |
my $uc = uc($index_data); |
my $uc = uc($index_data); |
103 |
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(); |
|
104 |
#print stderr "in index: $index_data\n"; |
#print stderr "in index: $index_data\n"; |
105 |
$c_table->{$field}->{$ident}->{$uc} = $index_data; |
$c_table->{$field}->{$uc} = $index_data; |
106 |
$c_count->{$field}->{$ident}->{$uc} = 1; |
$c_table->{$field}->{$uc}->{display} = $display; |
107 |
|
$c_count->{$field}->{$uc} = 1; |
108 |
} else { |
} else { |
109 |
$c_count->{$field}->{$ident}->{$uc}++; |
$c_count->{$field}->{$uc}++; |
110 |
} |
} |
111 |
} |
} |
112 |
|
|
113 |
sub check { |
sub count { |
114 |
my $self = shift; |
my $self = shift; |
115 |
|
|
116 |
my $field = shift; |
my $field = shift; |
117 |
|
my $where = shift; |
118 |
|
|
119 |
my $sql = "select count(*) from $field"; |
my $sql = "select count(*) from $field where upper(item) like upper(?)||'%'"; |
120 |
|
|
121 |
my $sth = $self->{dbh}->prepare($sql) || die $self->{dbh}->errstr(); |
my $sth = $self->{dbh}->prepare($sql) || die $self->{dbh}->errstr(); |
122 |
$sth->execute() || die "sql: $sql; ".$self->{dbh}->errstr(); |
$sth->execute($where) || die "sql: $sql; ".$self->{dbh}->errstr(); |
123 |
|
|
124 |
my ($total) = $sth->fetchrow_array(); |
my ($total) = $sth->fetchrow_array(); |
125 |
|
|
126 |
return $total; |
# no results, count all |
127 |
|
if (! $total) { |
128 |
|
my $sql = "select count(*) from $field"; |
129 |
|
|
130 |
|
my $sth = $self->{dbh}->prepare($sql) || die $self->{dbh}->errstr(); |
131 |
|
$sth->execute() || die "sql: $sql; ".$self->{dbh}->errstr(); |
132 |
|
$total = $sth->fetchrow_array(); |
133 |
|
|
134 |
|
} |
135 |
|
|
136 |
|
return $total || 1; |
137 |
} |
} |
138 |
|
|
139 |
|
|
141 |
my $self = shift; |
my $self = shift; |
142 |
|
|
143 |
my $field = shift; |
my $field = shift; |
|
my $what = shift || 'item'; # 'item,ident' |
|
144 |
my $where = shift; |
my $where = shift; |
145 |
|
|
146 |
my $from_ord = shift || 0; |
my $from_ord = shift || 0; |
148 |
|
|
149 |
my @sql_args; |
my @sql_args; |
150 |
|
|
151 |
my $sql = "select $what,ord from $field"; |
my $sql = "select item,display,ord from $field"; |
152 |
|
|
153 |
if ($where) { |
if ($where) { |
154 |
my $sql2 = " select ord from $field where upper($what) like upper(?)||'%'"; |
my $sql2 = "select ord from $field where upper(item) like upper(?)||'%'"; |
155 |
my $sth = $self->{dbh}->prepare($sql2) || die "sql2: $sql2; ".$self->{dbh}->errstr(); |
my $sth = $self->{dbh}->prepare($sql2) || die "sql2: $sql2; ".$self->{dbh}->errstr(); |
156 |
|
|
157 |
$sth->execute($where) || die "sql2: $sql2; ".$self->{dbh}->errstr(); |
$sth->execute($where) || die "sql2: $sql2; ".$self->{dbh}->errstr(); |
158 |
if (my $row = $sth->fetchrow_hashref) { |
if (my $row = $sth->fetchrow_hashref) { |
159 |
$from_ord += $row->{ord} - 1; |
$from_ord += $row->{ord} - 1; |
160 |
|
} else { |
161 |
|
# if no match is found when searching from beginning |
162 |
|
# of word in index, try substring match anywhere |
163 |
|
$sql2 = "select ord from $field where upper(item) like '%'||upper(?)||'%'"; |
164 |
|
$sth = $self->{dbh}->prepare($sql2) || die "sql2: $sql2; ".$self->{dbh}->errstr(); |
165 |
|
$sth->execute($where) || die "sql2: $sql2; ".$self->{dbh}->errstr(); |
166 |
|
if (my $row = $sth->fetchrow_hashref) { |
167 |
|
$from_ord += $row->{ord} - 1; |
168 |
|
} |
169 |
} |
} |
170 |
} |
} |
171 |
$sql .= " order by ord limit $rows offset $from_ord"; |
$sql .= " order by ord limit $rows offset $from_ord"; |
174 |
$sth->execute() || die "execute: $sql; ".$self->{dbh}->errstr(); |
$sth->execute() || die "execute: $sql; ".$self->{dbh}->errstr(); |
175 |
my @arr; |
my @arr; |
176 |
while (my $row = $sth->fetchrow_hashref) { |
while (my $row = $sth->fetchrow_hashref) { |
177 |
$row->{item} = HTML::Entities::encode($row->{item},'<>&"'); |
$row->{item} = HTML::Entities::encode($row->{item},' <>&"'); |
178 |
|
$row->{display} = HTML::Entities::encode($row->{display},'<>&"'); |
179 |
push @arr,$row; |
push @arr,$row; |
180 |
} |
} |
181 |
return @arr; |
return @arr; |
184 |
sub close { |
sub close { |
185 |
my $self = shift; |
my $self = shift; |
186 |
|
|
187 |
|
return if (! $self->{dbh}); |
188 |
|
|
189 |
# re-create ord column (sorted order) in table |
foreach my $table (keys %Table) { |
190 |
sub create_ord { |
$self->bench("Crating table $table"); |
191 |
|
$self->delete_and_create($table); |
|
my $table = shift; |
|
192 |
|
|
193 |
$self->{dbh}->begin_work || die $self->{dbh}->errstr(); |
$self->{dbh}->begin_work || die $self->{dbh}->errstr(); |
194 |
|
|
195 |
my $sql = "select oid from $table order by upper(item)"; |
$self->bench("Sorting ".$Table{$table}." items in $table"); |
196 |
|
my @keys = sort keys %{$c_table->{$table}}; |
197 |
|
|
198 |
|
$self->bench("Dumping data into $table"); |
199 |
|
my $sql = "insert into $table (ord,item,display,count) values (?,?,?,?)"; |
200 |
my $sth = $self->{dbh}->prepare($sql) || die "sql: $sql; ".$self->{dbh}->errstr(); |
my $sth = $self->{dbh}->prepare($sql) || die "sql: $sql; ".$self->{dbh}->errstr(); |
201 |
$sql = "update $table set ord=? where oid=?"; |
|
202 |
my $sth_update = $self->{dbh}->prepare($sql) || die "sql: $sql; ".$self->{dbh}->errstr(); |
my $ord = 0; |
203 |
$sth->execute() || die "sql: $sql; ".$self->{dbh}->errstr(); |
foreach my $key (@keys) { |
204 |
my $ord = 1; |
$sth->execute(++$ord, |
205 |
while (my $row = $sth->fetchrow_hashref) { |
$c_table->{$table}->{$key}, |
206 |
$sth_update->execute($ord++,$row->{oid}); |
$c_table->{$table}->{$key}->{display}, |
207 |
|
$c_count->{$table}->{$key} |
208 |
|
); |
209 |
} |
} |
210 |
|
|
211 |
$self->{dbh}->commit || die $self->{dbh}->errstr(); |
$self->{dbh}->commit || die $self->{dbh}->errstr(); |
212 |
} |
} |
213 |
#--- end of sub |
$self->bench("disconnecting from database"); |
|
|
|
|
if ($self->{dbh}) { |
|
|
|
|
|
# commit |
|
|
$self->{dbh}->commit || die $self->{dbh}->errstr(); |
|
|
|
|
|
foreach my $table (keys %Table) { |
|
|
# FIX |
|
|
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(); |
|
|
} |
|
214 |
|
|
215 |
$self->{dbh}->disconnect; |
$self->{dbh}->disconnect; |
216 |
undef $self->{dbh}; |
undef $self->{dbh}; |
|
} |
|
217 |
} |
} |
218 |
|
|
219 |
END { |
END { |