1 |
dpavlin |
60 |
# |
2 |
|
|
# this file implements index functions using DBI |
3 |
|
|
# and huge amounts of memory for cache speedup |
4 |
|
|
# |
5 |
dpavlin |
94 |
# this version doesn't support ident (which sould be location in |
6 |
|
|
# library). But, that functionality is not used anyway... |
7 |
|
|
# |
8 |
dpavlin |
60 |
|
9 |
|
|
package index_DBI; |
10 |
|
|
use strict qw(vars); |
11 |
|
|
use vars qw($Count); |
12 |
|
|
use HTML::Entities; |
13 |
dpavlin |
188 |
use URI::Escape; |
14 |
dpavlin |
192 |
use locale; |
15 |
dpavlin |
60 |
|
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 |
dpavlin |
88 |
# bench time |
26 |
dpavlin |
94 |
my $bench_time = time(); |
27 |
dpavlin |
88 |
|
28 |
dpavlin |
94 |
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 |
dpavlin |
60 |
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 |
dpavlin |
206 |
$self->{dbd} = $dbd; |
48 |
|
|
|
49 |
dpavlin |
60 |
$self->{dbh} = DBI->connect("DBI:$dbd:$dsn",$user,$passwd) || die $DBI::errstr; |
50 |
|
|
$Count++; |
51 |
|
|
|
52 |
dpavlin |
94 |
$self->bench("connected to $dbd as $user"); |
53 |
|
|
|
54 |
dpavlin |
226 |
# 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 |
dpavlin |
60 |
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) || die $self->{dbh}->errstr(); |
73 |
|
|
# FIX: this is not a good way to check if table exists! |
74 |
|
|
if ($sth->execute() && $sth->fetchrow_hashref) { |
75 |
|
|
my $sql = "drop table $field"; |
76 |
dpavlin |
225 |
my $sth = $self->{dbh}->do($sql) || warn "SQL: $sql - ".$sth->errstr(); |
77 |
dpavlin |
60 |
} |
78 |
|
|
$sql = "create table $field ( |
79 |
|
|
item varchar(255), |
80 |
dpavlin |
188 |
display text, |
81 |
dpavlin |
60 |
count int, |
82 |
|
|
ord int, |
83 |
dpavlin |
94 |
primary key (item) |
84 |
dpavlin |
60 |
)"; |
85 |
|
|
|
86 |
dpavlin |
94 |
$sth = $self->{dbh}->do($sql) || warn "SQL: $sql ".$self->{dbh}->errstr(); |
87 |
dpavlin |
60 |
} |
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 |
dpavlin |
188 |
my $display = shift || $index_data; |
95 |
dpavlin |
60 |
|
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 |
dpavlin |
93 |
|
105 |
|
|
# XXX for some strange reason, it seems that some entries in my |
106 |
|
|
# database produce strings which start with null byte. I suspect |
107 |
|
|
# this to be bug in OpenIsis 0.9.0. |
108 |
|
|
# This should fix it.. |
109 |
|
|
$index_data =~ s/^[^\w]+//; |
110 |
dpavlin |
60 |
$index_data = substr($index_data,0,255); |
111 |
dpavlin |
93 |
|
112 |
dpavlin |
60 |
my $uc = uc($index_data); |
113 |
dpavlin |
94 |
if (! $c_table->{$field}->{$uc}) { |
114 |
dpavlin |
60 |
#print stderr "in index: $index_data\n"; |
115 |
dpavlin |
94 |
$c_table->{$field}->{$uc} = $index_data; |
116 |
dpavlin |
188 |
$c_table->{$field}->{$uc}->{display} = $display; |
117 |
dpavlin |
94 |
$c_count->{$field}->{$uc} = 1; |
118 |
dpavlin |
60 |
} else { |
119 |
dpavlin |
94 |
$c_count->{$field}->{$uc}++; |
120 |
dpavlin |
60 |
} |
121 |
|
|
} |
122 |
|
|
|
123 |
dpavlin |
140 |
sub count { |
124 |
dpavlin |
60 |
my $self = shift; |
125 |
|
|
|
126 |
|
|
my $field = shift; |
127 |
dpavlin |
140 |
my $where = shift; |
128 |
dpavlin |
60 |
|
129 |
dpavlin |
140 |
my $sql = "select count(*) from $field where upper(item) like upper(?)||'%'"; |
130 |
dpavlin |
60 |
|
131 |
|
|
my $sth = $self->{dbh}->prepare($sql) || die $self->{dbh}->errstr(); |
132 |
dpavlin |
140 |
$sth->execute($where) || die "sql: $sql; ".$self->{dbh}->errstr(); |
133 |
dpavlin |
60 |
|
134 |
|
|
my ($total) = $sth->fetchrow_array(); |
135 |
|
|
|
136 |
dpavlin |
142 |
# no results, count all |
137 |
|
|
if (! $total) { |
138 |
|
|
my $sql = "select count(*) from $field"; |
139 |
|
|
|
140 |
|
|
my $sth = $self->{dbh}->prepare($sql) || die $self->{dbh}->errstr(); |
141 |
|
|
$sth->execute() || die "sql: $sql; ".$self->{dbh}->errstr(); |
142 |
|
|
$total = $sth->fetchrow_array(); |
143 |
|
|
|
144 |
|
|
} |
145 |
|
|
|
146 |
|
|
return $total || 1; |
147 |
dpavlin |
60 |
} |
148 |
|
|
|
149 |
|
|
|
150 |
|
|
sub fetch { |
151 |
|
|
my $self = shift; |
152 |
|
|
|
153 |
|
|
my $field = shift; |
154 |
|
|
my $where = shift; |
155 |
|
|
|
156 |
|
|
my $from_ord = shift || 0; |
157 |
|
|
my $rows = shift || 10; |
158 |
|
|
|
159 |
|
|
my @sql_args; |
160 |
|
|
|
161 |
dpavlin |
188 |
my $sql = "select item,display,ord from $field"; |
162 |
dpavlin |
60 |
|
163 |
|
|
if ($where) { |
164 |
dpavlin |
140 |
my $sql2 = "select ord from $field where upper(item) like upper(?)||'%'"; |
165 |
dpavlin |
60 |
my $sth = $self->{dbh}->prepare($sql2) || die "sql2: $sql2; ".$self->{dbh}->errstr(); |
166 |
|
|
|
167 |
|
|
$sth->execute($where) || die "sql2: $sql2; ".$self->{dbh}->errstr(); |
168 |
|
|
if (my $row = $sth->fetchrow_hashref) { |
169 |
|
|
$from_ord += $row->{ord} - 1; |
170 |
dpavlin |
127 |
} else { |
171 |
|
|
# if no match is found when searching from beginning |
172 |
|
|
# of word in index, try substring match anywhere |
173 |
dpavlin |
201 |
$sql2 = "select ord from $field where upper(item) like '% '||upper(?)||'%'"; |
174 |
dpavlin |
127 |
$sth = $self->{dbh}->prepare($sql2) || die "sql2: $sql2; ".$self->{dbh}->errstr(); |
175 |
|
|
$sth->execute($where) || die "sql2: $sql2; ".$self->{dbh}->errstr(); |
176 |
|
|
if (my $row = $sth->fetchrow_hashref) { |
177 |
|
|
$from_ord += $row->{ord} - 1; |
178 |
|
|
} |
179 |
dpavlin |
60 |
} |
180 |
|
|
} |
181 |
|
|
$sql .= " order by ord limit $rows offset $from_ord"; |
182 |
|
|
|
183 |
|
|
my $sth = $self->{dbh}->prepare($sql) || die "prepare: $sql; ".$self->{dbh}->errstr(); |
184 |
|
|
$sth->execute() || die "execute: $sql; ".$self->{dbh}->errstr(); |
185 |
|
|
my @arr; |
186 |
|
|
while (my $row = $sth->fetchrow_hashref) { |
187 |
dpavlin |
188 |
$row->{item} = HTML::Entities::encode($row->{item},' <>&"'); |
188 |
|
|
$row->{display} = HTML::Entities::encode($row->{display},'<>&"'); |
189 |
dpavlin |
60 |
push @arr,$row; |
190 |
|
|
} |
191 |
|
|
return @arr; |
192 |
|
|
} |
193 |
|
|
|
194 |
|
|
sub close { |
195 |
|
|
my $self = shift; |
196 |
|
|
|
197 |
dpavlin |
94 |
return if (! $self->{dbh}); |
198 |
dpavlin |
60 |
|
199 |
dpavlin |
94 |
foreach my $table (keys %Table) { |
200 |
|
|
$self->bench("Crating table $table"); |
201 |
|
|
$self->delete_and_create($table); |
202 |
dpavlin |
60 |
|
203 |
|
|
$self->{dbh}->begin_work || die $self->{dbh}->errstr(); |
204 |
|
|
|
205 |
dpavlin |
219 |
$self->bench("Sorting ".$Table{$table}." (with duplicates) items in $table"); |
206 |
dpavlin |
94 |
my @keys = sort keys %{$c_table->{$table}}; |
207 |
|
|
|
208 |
dpavlin |
219 |
$self->bench("Dumping ".($#keys+1)." items into $table"); |
209 |
dpavlin |
188 |
my $sql = "insert into $table (ord,item,display,count) values (?,?,?,?)"; |
210 |
dpavlin |
60 |
my $sth = $self->{dbh}->prepare($sql) || die "sql: $sql; ".$self->{dbh}->errstr(); |
211 |
dpavlin |
94 |
|
212 |
|
|
my $ord = 0; |
213 |
|
|
foreach my $key (@keys) { |
214 |
dpavlin |
95 |
$sth->execute(++$ord, |
215 |
dpavlin |
94 |
$c_table->{$table}->{$key}, |
216 |
dpavlin |
188 |
$c_table->{$table}->{$key}->{display}, |
217 |
dpavlin |
94 |
$c_count->{$table}->{$key} |
218 |
|
|
); |
219 |
dpavlin |
60 |
} |
220 |
|
|
|
221 |
|
|
$self->{dbh}->commit || die $self->{dbh}->errstr(); |
222 |
|
|
} |
223 |
dpavlin |
206 |
|
224 |
|
|
if ($self->{dbd} =~ m/(Pg|SQLite)/) { |
225 |
|
|
$self->{dbh}->do(qq{vacuum}) || warn "vacumming failed. It shouldn't if you are using PostgreSQL or SQLite: ".$self->{dbh}->errstr(); |
226 |
|
|
} |
227 |
|
|
|
228 |
dpavlin |
94 |
$self->bench("disconnecting from database"); |
229 |
dpavlin |
60 |
|
230 |
dpavlin |
94 |
$self->{dbh}->disconnect; |
231 |
|
|
undef $self->{dbh}; |
232 |
dpavlin |
60 |
} |
233 |
|
|
|
234 |
|
|
END { |
235 |
|
|
$Count--; |
236 |
|
|
print STDERR "index_DBI fatal error: \$index->close() not called... $Count references left!\n" if ($Count > 0); |
237 |
|
|
# FIX: debug output |
238 |
|
|
# print STDERR "usage\ttable\n"; |
239 |
|
|
# foreach (keys %Table) { |
240 |
|
|
# print STDERR $Table{$_},"\t$_\n"; |
241 |
|
|
# } |
242 |
|
|
} |
243 |
|
|
|
244 |
|
|
1; |