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 |
# force SQLite to support binary 0 in data (which shouldn't |
55 |
# happend, but it did to me) |
56 |
eval { |
57 |
no warnings 'all'; |
58 |
$self->{dbh}->{sqlite_handle_binary_nulls} = 1; |
59 |
}; |
60 |
|
61 |
return $self; |
62 |
} |
63 |
|
64 |
sub delete_and_create { |
65 |
my $self = shift; |
66 |
|
67 |
my $field = shift; |
68 |
|
69 |
#print "#### delete_and_create($field)\n"; |
70 |
|
71 |
my $sql = "select count(*) from $field"; |
72 |
my $sth = $self->{dbh}->prepare($sql); |
73 |
# FIX: this is not a good way to check if table exists! |
74 |
if ($sth && $sth->execute() && $sth->fetchrow_hashref) { |
75 |
my $sql = "drop table $field"; |
76 |
my $sth = $self->{dbh}->do($sql) || warn "SQL: $sql - ".$sth->errstr(); |
77 |
} |
78 |
$sql = "create table $field ( |
79 |
item varchar(255), |
80 |
display text, |
81 |
count int, |
82 |
ord int, |
83 |
primary key (item) |
84 |
)"; |
85 |
|
86 |
$sth = $self->{dbh}->do($sql) || warn "SQL: $sql ".$self->{dbh}->errstr(); |
87 |
} |
88 |
|
89 |
sub insert { |
90 |
my $self = shift; |
91 |
|
92 |
my $field = shift; |
93 |
my $index_data = shift || print STDERR "\$index->insert($field,NULL,...)"; |
94 |
my $display = shift || $index_data; |
95 |
|
96 |
if (! $index_data) { |
97 |
print STDERR "\$index->insert() -- no value to insert\n"; |
98 |
return; |
99 |
} |
100 |
|
101 |
$Table{$field}++; |
102 |
|
103 |
#$sth_cache{$field."select"}->execute($index_data) || die "cache: $field select; ".$self->{dbh}->errstr(); |
104 |
|
105 |
$index_data =~ s#&(\w)(acute|cedil|circ|grave|ring|slash|tilde|uml);#$1#gi; |
106 |
|
107 |
my $uc = uc($index_data); |
108 |
if (! $c_table->{$field}->{$uc}) { |
109 |
#print stderr "in index: $index_data\n"; |
110 |
$c_table->{$field}->{$uc} = $index_data; |
111 |
$c_table->{$field}->{$uc}->{display} = $display; |
112 |
$c_count->{$field}->{$uc} = 1; |
113 |
} else { |
114 |
$c_count->{$field}->{$uc}++; |
115 |
} |
116 |
} |
117 |
|
118 |
sub count { |
119 |
my $self = shift; |
120 |
|
121 |
my $field = shift; |
122 |
my $where = shift; |
123 |
|
124 |
my $sql = "select count(*) from $field where upper(item) like upper(?)||'%'"; |
125 |
|
126 |
my $sth = $self->{dbh}->prepare($sql) || die $self->{dbh}->errstr(); |
127 |
$sth->execute($where) || die "sql: $sql; ".$self->{dbh}->errstr(); |
128 |
|
129 |
my ($total) = $sth->fetchrow_array(); |
130 |
|
131 |
# no results, count all |
132 |
if (! $total) { |
133 |
my $sql = "select count(*) from $field"; |
134 |
|
135 |
my $sth = $self->{dbh}->prepare($sql) || die $self->{dbh}->errstr(); |
136 |
$sth->execute() || die "sql: $sql; ".$self->{dbh}->errstr(); |
137 |
$total = $sth->fetchrow_array(); |
138 |
|
139 |
} |
140 |
|
141 |
return $total || 1; |
142 |
} |
143 |
|
144 |
|
145 |
sub fetch { |
146 |
my $self = shift; |
147 |
|
148 |
my $field = shift; |
149 |
my $where = shift; |
150 |
|
151 |
my $from_ord = shift || 0; |
152 |
my $rows = shift || 10; |
153 |
|
154 |
my @sql_args; |
155 |
|
156 |
my $sql = "select item,display,ord from $field"; |
157 |
|
158 |
if ($where) { |
159 |
my $sql2 = "select ord from $field where upper(item) like upper(?)||'%'"; |
160 |
my $sth = $self->{dbh}->prepare($sql2) || die "sql2: $sql2; ".$self->{dbh}->errstr(); |
161 |
|
162 |
$sth->execute($where) || die "sql2: $sql2; ".$self->{dbh}->errstr(); |
163 |
if (my $row = $sth->fetchrow_hashref) { |
164 |
$from_ord += $row->{ord} - 1; |
165 |
} else { |
166 |
# if no match is found when searching from beginning |
167 |
# of word in index, try substring match anywhere |
168 |
$sql2 = "select ord from $field where upper(item) like '% '||upper(?)||'%'"; |
169 |
$sth = $self->{dbh}->prepare($sql2) || die "sql2: $sql2; ".$self->{dbh}->errstr(); |
170 |
$sth->execute($where) || die "sql2: $sql2; ".$self->{dbh}->errstr(); |
171 |
if (my $row = $sth->fetchrow_hashref) { |
172 |
$from_ord += $row->{ord} - 1; |
173 |
} |
174 |
} |
175 |
} |
176 |
$sql .= " order by ord limit $rows offset $from_ord"; |
177 |
|
178 |
my $sth = $self->{dbh}->prepare($sql) || die "prepare: $sql; ".$self->{dbh}->errstr(); |
179 |
$sth->execute() || die "execute: $sql; ".$self->{dbh}->errstr(); |
180 |
my @arr; |
181 |
while (my $row = $sth->fetchrow_hashref) { |
182 |
$row->{item} = HTML::Entities::encode($row->{item},' <>&"'); |
183 |
$row->{display} = HTML::Entities::encode($row->{display},'<>&"'); |
184 |
$row->{item} =~ s#&(\w)(acute|cedil|circ|grave|ring|slash|tilde|uml);#$1#gi; |
185 |
$row->{display} =~ s#&(\w)(acute|cedil|circ|grave|ring|slash|tilde|uml);#&$1$2;#gi; |
186 |
push @arr,$row; |
187 |
} |
188 |
return @arr; |
189 |
} |
190 |
|
191 |
sub close { |
192 |
my $self = shift; |
193 |
|
194 |
return if (! $self->{dbh}); |
195 |
|
196 |
foreach my $table (keys %Table) { |
197 |
$self->bench("Crating table $table"); |
198 |
$self->delete_and_create($table); |
199 |
|
200 |
$self->{dbh}->begin_work || die $self->{dbh}->errstr(); |
201 |
|
202 |
$self->bench("Sorting ".$Table{$table}." (with duplicates) items in $table"); |
203 |
my @keys = sort keys %{$c_table->{$table}}; |
204 |
|
205 |
$self->bench("Dumping ".($#keys+1)." items into $table"); |
206 |
my $sql = "insert into $table (ord,item,display,count) values (?,?,?,?)"; |
207 |
my $sth = $self->{dbh}->prepare($sql) || die "sql: $sql; ".$self->{dbh}->errstr(); |
208 |
|
209 |
my $ord = 0; |
210 |
foreach my $key (@keys) { |
211 |
$sth->execute(++$ord, |
212 |
$c_table->{$table}->{$key}, |
213 |
$c_table->{$table}->{$key}->{display}, |
214 |
$c_count->{$table}->{$key} |
215 |
); |
216 |
} |
217 |
|
218 |
$self->{dbh}->commit || die $self->{dbh}->errstr(); |
219 |
} |
220 |
|
221 |
if ($self->{dbd} =~ m/(Pg|SQLite)/) { |
222 |
$self->{dbh}->do(qq{vacuum}) || warn "vacumming failed. It shouldn't if you are using PostgreSQL or SQLite: ".$self->{dbh}->errstr(); |
223 |
} |
224 |
|
225 |
$self->bench("disconnecting from database"); |
226 |
|
227 |
$self->{dbh}->disconnect; |
228 |
undef $self->{dbh}; |
229 |
} |
230 |
|
231 |
END { |
232 |
$Count--; |
233 |
print STDERR "index_DBI fatal error: \$index->close() not called... $Count references left!\n" if ($Count > 0); |
234 |
# FIX: debug output |
235 |
# print STDERR "usage\ttable\n"; |
236 |
# foreach (keys %Table) { |
237 |
# print STDERR $Table{$_},"\t$_\n"; |
238 |
# } |
239 |
} |
240 |
|
241 |
1; |