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 |
use Carp; |
16 |
|
17 |
use DBI; |
18 |
|
19 |
# bench time |
20 |
my $bench_time = time(); |
21 |
|
22 |
my $debug = 1; |
23 |
|
24 |
sub bench { |
25 |
my $self = shift; |
26 |
my $msg = shift; |
27 |
|
28 |
print STDERR "last operation took ",time()-$bench_time," seconds...\n"; |
29 |
$bench_time=time(); |
30 |
print STDERR "$msg\n"; |
31 |
} |
32 |
|
33 |
sub new { |
34 |
my $class = shift; |
35 |
my $self = {}; |
36 |
bless($self, $class); |
37 |
|
38 |
my $dbd = shift || die "need dbi_dbd= in [global] section of configuration file"; |
39 |
my $dsn = shift || die "need dbi_dsn= in [global] section of configuration file"; |
40 |
my $user = shift || die "need dbi_user= in [global] section of configuration file"; |
41 |
my $passwd = shift || die "need dbi_passwd= in [global] section of configuration file"; |
42 |
|
43 |
$self->{dbd} = $dbd; |
44 |
|
45 |
$self->{dbh} = DBI->connect("DBI:$dbd:$dsn",$user,$passwd) || die $DBI::errstr; |
46 |
$Count++; |
47 |
|
48 |
$self->bench("connected to $dbd as $user"); |
49 |
|
50 |
# force SQLite to support binary 0 in data (which shouldn't |
51 |
# happend, but it did to me) |
52 |
eval { |
53 |
no warnings 'all'; |
54 |
$self->{dbh}->{sqlite_handle_binary_nulls} = 1; |
55 |
}; |
56 |
|
57 |
return $self; |
58 |
} |
59 |
|
60 |
sub delete_and_create { |
61 |
my $self = shift; |
62 |
|
63 |
my $table = shift || croak "need table name!"; |
64 |
my $sql = shift || croak "need sql to create table!"; |
65 |
|
66 |
print STDERR "## delete_and_create($table)\n" if ($debug); |
67 |
|
68 |
my $sql_delete = "delete from $table"; |
69 |
my $sth = $self->{dbh}->prepare($sql_delete); |
70 |
|
71 |
if ($sth && $sth->execute()) { |
72 |
print STDERR "## deleted rows from table $table\n" if ($debug); |
73 |
} else { |
74 |
# can't delete from table, assume it doesn't exists! |
75 |
$self->{dbh}->rollback; |
76 |
$self->{dbh}->do($sql) || confess "SQL: $sql ".$self->{dbh}->errstr(); |
77 |
print STDERR "## creating table $table\n" if ($debug); |
78 |
$self->{dbh}->begin_work; |
79 |
} |
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 |
my $filter = shift; |
89 |
|
90 |
if (! $index_data) { |
91 |
print STDERR "\$index->insert() -- no value to insert\n"; |
92 |
return; |
93 |
} |
94 |
|
95 |
$index_data =~ s#&(\w)(acute|cedil|circ|grave|ring|slash|tilde|uml);#$1#gi; |
96 |
|
97 |
# strip spaces |
98 |
$index_data =~ s#^\s+##; |
99 |
$index_data =~ s#\s+$##; |
100 |
$index_data =~ s#\s\s+# #g; |
101 |
|
102 |
my $uc = uc($index_data); |
103 |
|
104 |
if (! $self->{c}->{$uc}->{$field}) { |
105 |
#print stderr "in index: $index_data\n"; |
106 |
$self->{c}->{$uc}->{$field}->{item} = $index_data; |
107 |
$self->{c}->{$uc}->{$field}->{display} = $display; |
108 |
} |
109 |
|
110 |
$self->{c}->{$uc}->{$field}->{count}++; |
111 |
$self->{c}->{$uc}->{$field}->{filter}->{$filter}++ if ($filter); |
112 |
} |
113 |
|
114 |
sub count { |
115 |
my $self = shift; |
116 |
|
117 |
my $field = shift; |
118 |
my $where = shift; |
119 |
|
120 |
my $filter = shift; |
121 |
|
122 |
my $tables_sql = 'data'; |
123 |
my $where_sql = ''; |
124 |
my @sql_args = ( $field, $where ); |
125 |
|
126 |
if ($filter) { |
127 |
$tables_sql .= ",filters"; |
128 |
$where_sql .= " |
129 |
and data.ord = filters.ord |
130 |
and filter = ? |
131 |
"; |
132 |
push @sql_args, $filter; |
133 |
} |
134 |
|
135 |
my $sql = qq{ |
136 |
select count(*) |
137 |
from $tables_sql |
138 |
where name = ? and upper(item) like upper(?)||'%' |
139 |
$where_sql |
140 |
}; |
141 |
|
142 |
my $sth = $self->{dbh}->prepare($sql) || confess $self->{dbh}->errstr(); |
143 |
$sth->execute(@sql_args) || confess "sql: $sql; ".$self->{dbh}->errstr(); |
144 |
|
145 |
my ($total) = $sth->fetchrow_array(); |
146 |
|
147 |
# no results, count all |
148 |
if (! $total) { |
149 |
my $sql = qq{ |
150 |
select count(*) |
151 |
from $tables_sql |
152 |
where data.name = ? |
153 |
$where_sql |
154 |
}; |
155 |
|
156 |
@sql_args = ( $field ); |
157 |
push @sql_args, $filter if ($filter); |
158 |
|
159 |
my $sth = $self->{dbh}->prepare($sql) || confess $self->{dbh}->errstr(); |
160 |
$sth->execute(@sql_args) || confess "sql: $sql; ".$self->{dbh}->errstr(); |
161 |
$total = $sth->fetchrow_array(); |
162 |
|
163 |
} |
164 |
|
165 |
return $total || '0'; |
166 |
} |
167 |
|
168 |
|
169 |
sub fetch { |
170 |
my $self = shift; |
171 |
|
172 |
my $field = shift; |
173 |
my $where = shift; |
174 |
|
175 |
my $offset = shift || 0; |
176 |
my $rows = shift || 10; |
177 |
my $filter = shift; |
178 |
|
179 |
my $from_ord = 0; |
180 |
|
181 |
my $tables_sql = 'data'; |
182 |
my $where_sql = ''; |
183 |
|
184 |
my @sql_args = ( $field, $where ); |
185 |
|
186 |
if ($filter) { |
187 |
$tables_sql .= ",filters"; |
188 |
$where_sql .= " |
189 |
and data.ord = filters.ord |
190 |
and filter = ? |
191 |
"; |
192 |
push @sql_args, $filter; |
193 |
} |
194 |
|
195 |
if ($where) { |
196 |
my $sql2 = qq{ |
197 |
select data.ord as ord |
198 |
from $tables_sql |
199 |
where name = ? and upper(item) like upper(?)||'%' |
200 |
$where_sql |
201 |
}; |
202 |
my $sth = $self->{dbh}->prepare($sql2) || confess "sql2: $sql2; ".$self->{dbh}->errstr(); |
203 |
|
204 |
$sth->execute(@sql_args) || confess "sql2: $sql2; ".$self->{dbh}->errstr(); |
205 |
if (my $row = $sth->fetchrow_hashref) { |
206 |
$from_ord += $row->{ord} - 1; |
207 |
} else { |
208 |
# if no match is found when searching from beginning |
209 |
# of word in index, try substring match anywhere |
210 |
$sql2 = qq{ |
211 |
select data.ord as ord |
212 |
from $tables_sql |
213 |
where name = ? and upper(item) like '% '||upper(?)||'%' |
214 |
$where_sql |
215 |
}; |
216 |
|
217 |
$sth = $self->{dbh}->prepare($sql2) || confess "sql2: $sql2; ".$self->{dbh}->errstr(); |
218 |
$sth->execute(@sql_args) || confess "sql2: $sql2; ".$self->{dbh}->errstr(); |
219 |
|
220 |
if (my $row = $sth->fetchrow_hashref) { |
221 |
$from_ord += $row->{ord} - 1; |
222 |
} |
223 |
} |
224 |
} |
225 |
|
226 |
@sql_args = ( $field, $from_ord ); |
227 |
push @sql_args, $filter if ($filter); |
228 |
|
229 |
my $sql = qq{ |
230 |
select item,display,data.count as count |
231 |
from $tables_sql |
232 |
where name = ? |
233 |
and data.ord > ? |
234 |
$where_sql |
235 |
order by data.ord |
236 |
}; |
237 |
|
238 |
# fix SQLite problem which doesn't allow placeholders in limit and offset |
239 |
# http://thread.gmane.org/gmane.comp.db.sqlite.general/9707 |
240 |
$sql .= "limit $rows offset $offset"; |
241 |
|
242 |
my $sth = $self->{dbh}->prepare($sql) || confess "prepare: $sql; ".$self->{dbh}->errstr(); |
243 |
$sth->execute(@sql_args) || confess "execute: $sql; ".join("|",@sql_args)." ".$self->{dbh}->errstr(); |
244 |
my @arr; |
245 |
while (my $row = $sth->fetchrow_hashref) { |
246 |
$row->{item} = HTML::Entities::encode($row->{item},' <>&"'); |
247 |
$row->{display} = HTML::Entities::encode($row->{display},'<>&"'); |
248 |
$row->{item} =~ s#&(\w)(acute|cedil|circ|grave|ring|slash|tilde|uml);#$1#gi; |
249 |
$row->{display} =~ s#&(\w)(acute|cedil|circ|grave|ring|slash|tilde|uml);#&$1$2;#gi; |
250 |
push @arr,$row; |
251 |
} |
252 |
return @arr; |
253 |
} |
254 |
|
255 |
sub close { |
256 |
my $self = shift; |
257 |
|
258 |
return if (! $self->{dbh}); |
259 |
|
260 |
$self->{dbh}->begin_work || confess $self->{dbh}->errstr(); |
261 |
|
262 |
$self->delete_and_create('data', qq{ |
263 |
create table data ( |
264 |
name varchar(255), |
265 |
ord int, |
266 |
item text, |
267 |
display text, |
268 |
count int, |
269 |
primary key (name,ord) |
270 |
); |
271 |
}); |
272 |
|
273 |
$self->delete_and_create('filters', qq{ |
274 |
create table filters ( |
275 |
filter varchar(255), |
276 |
ord int, |
277 |
count int, |
278 |
primary key (filter,ord) |
279 |
); |
280 |
}); |
281 |
|
282 |
$self->bench("getting all entries"); |
283 |
my @items = keys %{$self->{c}}; |
284 |
$self->bench("got ".($#items+1)." items, now sorting"); |
285 |
@items = sort @items; |
286 |
|
287 |
my $sql = "insert into data (name,ord,item,display,count) values (?,?,?,?,?)"; |
288 |
my $sth_index = $self->{dbh}->prepare($sql) || confess "$sql: ".$self->{dbh}->errstr(); |
289 |
|
290 |
$sql = "insert into filters (filter, ord, count) values (?,?,?)"; |
291 |
my $sth_filter = $self->{dbh}->prepare($sql) || confess "$sql: ".$self->{dbh}->errstr(); |
292 |
|
293 |
my $ord = 0; |
294 |
foreach my $key (@items) { |
295 |
|
296 |
foreach my $field (keys %{$self->{c}->{$key}}) { |
297 |
# store items |
298 |
$sth_index->execute( |
299 |
$field, |
300 |
++$ord, |
301 |
$self->{c}->{$key}->{$field}->{item}, |
302 |
$self->{c}->{$key}->{$field}->{display}, |
303 |
$self->{c}->{$key}->{$field}->{count}, |
304 |
); |
305 |
|
306 |
# store filters |
307 |
next unless ($self->{c}->{$key}->{$field}->{filter}); |
308 |
|
309 |
foreach my $filter (keys %{$self->{c}->{$key}->{$field}->{filter}}) { |
310 |
$sth_filter->execute( $filter, $ord, $self->{c}->{$key}->{$field}->{filter}->{$filter} ); |
311 |
} |
312 |
} |
313 |
|
314 |
|
315 |
} |
316 |
|
317 |
$self->{dbh}->commit || confess $self->{dbh}->errstr(); |
318 |
|
319 |
$self->bench("vacuuming"); |
320 |
|
321 |
if ($self->{dbd} =~ m/(Pg|SQLite)/) { |
322 |
$self->{dbh}->do(qq{vacuum}) || carp "vacumming failed. It shouldn't if you are using PostgreSQL or SQLite: ".$self->{dbh}->errstr(); |
323 |
} |
324 |
|
325 |
$self->bench("disconnecting from database"); |
326 |
|
327 |
$self->{dbh}->disconnect; |
328 |
undef $self->{dbh}; |
329 |
} |
330 |
|
331 |
END { |
332 |
$Count--; |
333 |
print STDERR "index_DBI fatal error: \$index->close() not called... $Count references left!\n" if ($Count > 0); |
334 |
# FIX: debug output |
335 |
# print STDERR "usage\ttable\n"; |
336 |
# foreach (keys %Table) { |
337 |
# print STDERR $Table{$_},"\t$_\n"; |
338 |
# } |
339 |
} |
340 |
|
341 |
1; |