1 |
# |
2 |
# this file implements index functions using DBI |
3 |
# 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; |
10 |
use strict qw(vars); |
11 |
use vars qw($Count); |
12 |
use HTML::Entities; |
13 |
use URI::Escape; |
14 |
use locale; |
15 |
|
16 |
use DBI; |
17 |
|
18 |
my %Table; # index tables which where visited in this run |
19 |
my %sth_cache; # cache prepared statements |
20 |
|
21 |
# cache var |
22 |
my $c_table; |
23 |
my $c_count; |
24 |
|
25 |
# bench time |
26 |
my $bench_time = time(); |
27 |
|
28 |
sub bench { |
29 |
my $self = shift; |
30 |
my $msg = shift; |
31 |
|
32 |
print STDERR "last operation took ",time()-$bench_time," seconds...\n"; |
33 |
$bench_time=time(); |
34 |
print STDERR "$msg\n"; |
35 |
} |
36 |
|
37 |
sub new { |
38 |
my $class = shift; |
39 |
my $self = {}; |
40 |
bless($self, $class); |
41 |
|
42 |
my $dbd = shift || die "need dbi_dbd= in [global] section of configuration file"; |
43 |
my $dsn = shift || die "need dbi_dsn= in [global] section of configuration file"; |
44 |
my $user = shift || die "need dbi_user= in [global] section of configuration file"; |
45 |
my $passwd = shift || die "need dbi_passwd= in [global] section of configuration file"; |
46 |
|
47 |
$self->{dbd} = $dbd; |
48 |
|
49 |
$self->{dbh} = DBI->connect("DBI:$dbd:$dsn",$user,$passwd) || die $DBI::errstr; |
50 |
$Count++; |
51 |
|
52 |
$self->bench("connected to $dbd as $user"); |
53 |
|
54 |
return $self; |
55 |
} |
56 |
|
57 |
sub delete_and_create { |
58 |
my $self = shift; |
59 |
|
60 |
my $field = shift; |
61 |
|
62 |
#print "#### delete_and_create($field)\n"; |
63 |
|
64 |
my $sql = "select count(*) from $field"; |
65 |
my $sth = $self->{dbh}->prepare($sql) || die $self->{dbh}->errstr(); |
66 |
# FIX: this is not a good way to check if table exists! |
67 |
if ($sth->execute() && $sth->fetchrow_hashref) { |
68 |
my $sql = "drop table $field"; |
69 |
my $sth = $self->{dbh}->do($sql) || die "SQL: $sql ".$self->{dbh}->errstr(); |
70 |
} |
71 |
$sql = "create table $field ( |
72 |
item varchar(255), |
73 |
display text, |
74 |
count int, |
75 |
ord int, |
76 |
primary key (item) |
77 |
)"; |
78 |
|
79 |
$sth = $self->{dbh}->do($sql) || warn "SQL: $sql ".$self->{dbh}->errstr(); |
80 |
} |
81 |
|
82 |
sub insert { |
83 |
my $self = shift; |
84 |
|
85 |
my $field = shift; |
86 |
my $index_data = shift || print STDERR "\$index->insert($field,NULL,...)"; |
87 |
my $display = shift || $index_data; |
88 |
|
89 |
if (! $index_data) { |
90 |
print STDERR "\$index->insert() -- no value to insert\n"; |
91 |
return; |
92 |
} |
93 |
|
94 |
$Table{$field}++; |
95 |
|
96 |
#$sth_cache{$field."select"}->execute($index_data) || die "cache: $field select; ".$self->{dbh}->errstr(); |
97 |
|
98 |
# XXX for some strange reason, it seems that some entries in my |
99 |
# database produce strings which start with null byte. I suspect |
100 |
# this to be bug in OpenIsis 0.9.0. |
101 |
# This should fix it.. |
102 |
$index_data =~ s/^[^\w]+//; |
103 |
$index_data = substr($index_data,0,255); |
104 |
|
105 |
my $uc = uc($index_data); |
106 |
if (! $c_table->{$field}->{$uc}) { |
107 |
#print stderr "in index: $index_data\n"; |
108 |
$c_table->{$field}->{$uc} = $index_data; |
109 |
$c_table->{$field}->{$uc}->{display} = $display; |
110 |
$c_count->{$field}->{$uc} = 1; |
111 |
} else { |
112 |
$c_count->{$field}->{$uc}++; |
113 |
} |
114 |
} |
115 |
|
116 |
sub count { |
117 |
my $self = shift; |
118 |
|
119 |
my $field = shift; |
120 |
my $where = shift; |
121 |
|
122 |
my $sql = "select count(*) from $field where upper(item) like upper(?)||'%'"; |
123 |
|
124 |
my $sth = $self->{dbh}->prepare($sql) || die $self->{dbh}->errstr(); |
125 |
$sth->execute($where) || die "sql: $sql; ".$self->{dbh}->errstr(); |
126 |
|
127 |
my ($total) = $sth->fetchrow_array(); |
128 |
|
129 |
# no results, count all |
130 |
if (! $total) { |
131 |
my $sql = "select count(*) from $field"; |
132 |
|
133 |
my $sth = $self->{dbh}->prepare($sql) || die $self->{dbh}->errstr(); |
134 |
$sth->execute() || die "sql: $sql; ".$self->{dbh}->errstr(); |
135 |
$total = $sth->fetchrow_array(); |
136 |
|
137 |
} |
138 |
|
139 |
return $total || 1; |
140 |
} |
141 |
|
142 |
|
143 |
sub fetch { |
144 |
my $self = shift; |
145 |
|
146 |
my $field = shift; |
147 |
my $where = shift; |
148 |
|
149 |
my $from_ord = shift || 0; |
150 |
my $rows = shift || 10; |
151 |
|
152 |
my @sql_args; |
153 |
|
154 |
my $sql = "select item,display,ord from $field"; |
155 |
|
156 |
if ($where) { |
157 |
my $sql2 = "select ord from $field where upper(item) like upper(?)||'%'"; |
158 |
my $sth = $self->{dbh}->prepare($sql2) || die "sql2: $sql2; ".$self->{dbh}->errstr(); |
159 |
|
160 |
$sth->execute($where) || die "sql2: $sql2; ".$self->{dbh}->errstr(); |
161 |
if (my $row = $sth->fetchrow_hashref) { |
162 |
$from_ord += $row->{ord} - 1; |
163 |
} else { |
164 |
# if no match is found when searching from beginning |
165 |
# of word in index, try substring match anywhere |
166 |
$sql2 = "select ord from $field where upper(item) like '% '||upper(?)||'%'"; |
167 |
$sth = $self->{dbh}->prepare($sql2) || die "sql2: $sql2; ".$self->{dbh}->errstr(); |
168 |
$sth->execute($where) || die "sql2: $sql2; ".$self->{dbh}->errstr(); |
169 |
if (my $row = $sth->fetchrow_hashref) { |
170 |
$from_ord += $row->{ord} - 1; |
171 |
} |
172 |
} |
173 |
} |
174 |
$sql .= " order by ord limit $rows offset $from_ord"; |
175 |
|
176 |
my $sth = $self->{dbh}->prepare($sql) || die "prepare: $sql; ".$self->{dbh}->errstr(); |
177 |
$sth->execute() || die "execute: $sql; ".$self->{dbh}->errstr(); |
178 |
my @arr; |
179 |
while (my $row = $sth->fetchrow_hashref) { |
180 |
$row->{item} = HTML::Entities::encode($row->{item},' <>&"'); |
181 |
$row->{display} = HTML::Entities::encode($row->{display},'<>&"'); |
182 |
push @arr,$row; |
183 |
} |
184 |
return @arr; |
185 |
} |
186 |
|
187 |
sub close { |
188 |
my $self = shift; |
189 |
|
190 |
return if (! $self->{dbh}); |
191 |
|
192 |
foreach my $table (keys %Table) { |
193 |
$self->bench("Crating table $table"); |
194 |
$self->delete_and_create($table); |
195 |
|
196 |
$self->{dbh}->begin_work || die $self->{dbh}->errstr(); |
197 |
|
198 |
$self->bench("Sorting ".$Table{$table}." (with duplicates) items in $table"); |
199 |
my @keys = sort keys %{$c_table->{$table}}; |
200 |
|
201 |
$self->bench("Dumping ".($#keys+1)." items into $table"); |
202 |
my $sql = "insert into $table (ord,item,display,count) values (?,?,?,?)"; |
203 |
my $sth = $self->{dbh}->prepare($sql) || die "sql: $sql; ".$self->{dbh}->errstr(); |
204 |
|
205 |
my $ord = 0; |
206 |
foreach my $key (@keys) { |
207 |
$sth->execute(++$ord, |
208 |
$c_table->{$table}->{$key}, |
209 |
$c_table->{$table}->{$key}->{display}, |
210 |
$c_count->{$table}->{$key} |
211 |
); |
212 |
} |
213 |
|
214 |
$self->{dbh}->commit || die $self->{dbh}->errstr(); |
215 |
} |
216 |
|
217 |
if ($self->{dbd} =~ m/(Pg|SQLite)/) { |
218 |
$self->{dbh}->do(qq{vacuum}) || warn "vacumming failed. It shouldn't if you are using PostgreSQL or SQLite: ".$self->{dbh}->errstr(); |
219 |
} |
220 |
|
221 |
$self->bench("disconnecting from database"); |
222 |
|
223 |
$self->{dbh}->disconnect; |
224 |
undef $self->{dbh}; |
225 |
} |
226 |
|
227 |
END { |
228 |
$Count--; |
229 |
print STDERR "index_DBI fatal error: \$index->close() not called... $Count references left!\n" if ($Count > 0); |
230 |
# FIX: debug output |
231 |
# print STDERR "usage\ttable\n"; |
232 |
# foreach (keys %Table) { |
233 |
# print STDERR $Table{$_},"\t$_\n"; |
234 |
# } |
235 |
} |
236 |
|
237 |
1; |