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 |
|
|
$self->{dbh} = DBI->connect("DBI:$dbd:$dsn",$user,$passwd) || die $DBI::errstr; |
48 |
|
|
$Count++; |
49 |
|
|
|
50 |
dpavlin |
94 |
$self->bench("connected to $dbd as $user"); |
51 |
|
|
|
52 |
dpavlin |
60 |
return $self; |
53 |
|
|
} |
54 |
|
|
|
55 |
|
|
sub delete_and_create { |
56 |
|
|
my $self = shift; |
57 |
|
|
|
58 |
|
|
my $field = shift; |
59 |
|
|
|
60 |
|
|
#print "#### delete_and_create($field)\n"; |
61 |
|
|
|
62 |
|
|
my $sql = "select count(*) from $field"; |
63 |
|
|
my $sth = $self->{dbh}->prepare($sql) || die $self->{dbh}->errstr(); |
64 |
|
|
# FIX: this is not a good way to check if table exists! |
65 |
|
|
if ($sth->execute() && $sth->fetchrow_hashref) { |
66 |
|
|
my $sql = "drop table $field"; |
67 |
|
|
my $sth = $self->{dbh}->do($sql) || die "SQL: $sql ".$self->{dbh}->errstr(); |
68 |
|
|
} |
69 |
|
|
$sql = "create table $field ( |
70 |
|
|
item varchar(255), |
71 |
dpavlin |
188 |
display text, |
72 |
dpavlin |
60 |
count int, |
73 |
|
|
ord int, |
74 |
dpavlin |
94 |
primary key (item) |
75 |
dpavlin |
60 |
)"; |
76 |
|
|
|
77 |
dpavlin |
94 |
$sth = $self->{dbh}->do($sql) || warn "SQL: $sql ".$self->{dbh}->errstr(); |
78 |
dpavlin |
60 |
} |
79 |
|
|
|
80 |
|
|
sub insert { |
81 |
|
|
my $self = shift; |
82 |
|
|
|
83 |
|
|
my $field = shift; |
84 |
|
|
my $index_data = shift || print STDERR "\$index->insert($field,NULL,...)"; |
85 |
dpavlin |
188 |
my $display = shift || $index_data; |
86 |
dpavlin |
60 |
|
87 |
|
|
if (! $index_data) { |
88 |
|
|
print STDERR "\$index->insert() -- no value to insert\n"; |
89 |
|
|
return; |
90 |
|
|
} |
91 |
|
|
|
92 |
|
|
$Table{$field}++; |
93 |
|
|
|
94 |
|
|
#$sth_cache{$field."select"}->execute($index_data) || die "cache: $field select; ".$self->{dbh}->errstr(); |
95 |
dpavlin |
93 |
|
96 |
|
|
# XXX for some strange reason, it seems that some entries in my |
97 |
|
|
# database produce strings which start with null byte. I suspect |
98 |
|
|
# this to be bug in OpenIsis 0.9.0. |
99 |
|
|
# This should fix it.. |
100 |
|
|
$index_data =~ s/^[^\w]+//; |
101 |
dpavlin |
60 |
$index_data = substr($index_data,0,255); |
102 |
dpavlin |
93 |
|
103 |
dpavlin |
60 |
my $uc = uc($index_data); |
104 |
dpavlin |
94 |
if (! $c_table->{$field}->{$uc}) { |
105 |
dpavlin |
60 |
#print stderr "in index: $index_data\n"; |
106 |
dpavlin |
94 |
$c_table->{$field}->{$uc} = $index_data; |
107 |
dpavlin |
188 |
$c_table->{$field}->{$uc}->{display} = $display; |
108 |
dpavlin |
94 |
$c_count->{$field}->{$uc} = 1; |
109 |
dpavlin |
60 |
} else { |
110 |
dpavlin |
94 |
$c_count->{$field}->{$uc}++; |
111 |
dpavlin |
60 |
} |
112 |
|
|
} |
113 |
|
|
|
114 |
dpavlin |
140 |
sub count { |
115 |
dpavlin |
60 |
my $self = shift; |
116 |
|
|
|
117 |
|
|
my $field = shift; |
118 |
dpavlin |
140 |
my $where = shift; |
119 |
dpavlin |
60 |
|
120 |
dpavlin |
140 |
my $sql = "select count(*) from $field where upper(item) like upper(?)||'%'"; |
121 |
dpavlin |
60 |
|
122 |
|
|
my $sth = $self->{dbh}->prepare($sql) || die $self->{dbh}->errstr(); |
123 |
dpavlin |
140 |
$sth->execute($where) || die "sql: $sql; ".$self->{dbh}->errstr(); |
124 |
dpavlin |
60 |
|
125 |
|
|
my ($total) = $sth->fetchrow_array(); |
126 |
|
|
|
127 |
dpavlin |
142 |
# no results, count all |
128 |
|
|
if (! $total) { |
129 |
|
|
my $sql = "select count(*) from $field"; |
130 |
|
|
|
131 |
|
|
my $sth = $self->{dbh}->prepare($sql) || die $self->{dbh}->errstr(); |
132 |
|
|
$sth->execute() || die "sql: $sql; ".$self->{dbh}->errstr(); |
133 |
|
|
$total = $sth->fetchrow_array(); |
134 |
|
|
|
135 |
|
|
} |
136 |
|
|
|
137 |
|
|
return $total || 1; |
138 |
dpavlin |
60 |
} |
139 |
|
|
|
140 |
|
|
|
141 |
|
|
sub fetch { |
142 |
|
|
my $self = shift; |
143 |
|
|
|
144 |
|
|
my $field = shift; |
145 |
|
|
my $where = shift; |
146 |
|
|
|
147 |
|
|
my $from_ord = shift || 0; |
148 |
|
|
my $rows = shift || 10; |
149 |
|
|
|
150 |
|
|
my @sql_args; |
151 |
|
|
|
152 |
dpavlin |
188 |
my $sql = "select item,display,ord from $field"; |
153 |
dpavlin |
60 |
|
154 |
|
|
if ($where) { |
155 |
dpavlin |
140 |
my $sql2 = "select ord from $field where upper(item) like upper(?)||'%'"; |
156 |
dpavlin |
60 |
my $sth = $self->{dbh}->prepare($sql2) || die "sql2: $sql2; ".$self->{dbh}->errstr(); |
157 |
|
|
|
158 |
|
|
$sth->execute($where) || die "sql2: $sql2; ".$self->{dbh}->errstr(); |
159 |
|
|
if (my $row = $sth->fetchrow_hashref) { |
160 |
|
|
$from_ord += $row->{ord} - 1; |
161 |
dpavlin |
127 |
} else { |
162 |
|
|
# if no match is found when searching from beginning |
163 |
|
|
# of word in index, try substring match anywhere |
164 |
dpavlin |
140 |
$sql2 = "select ord from $field where upper(item) like '%'||upper(?)||'%'"; |
165 |
dpavlin |
127 |
$sth = $self->{dbh}->prepare($sql2) || die "sql2: $sql2; ".$self->{dbh}->errstr(); |
166 |
|
|
$sth->execute($where) || die "sql2: $sql2; ".$self->{dbh}->errstr(); |
167 |
|
|
if (my $row = $sth->fetchrow_hashref) { |
168 |
|
|
$from_ord += $row->{ord} - 1; |
169 |
|
|
} |
170 |
dpavlin |
60 |
} |
171 |
|
|
} |
172 |
|
|
$sql .= " order by ord limit $rows offset $from_ord"; |
173 |
|
|
|
174 |
|
|
my $sth = $self->{dbh}->prepare($sql) || die "prepare: $sql; ".$self->{dbh}->errstr(); |
175 |
|
|
$sth->execute() || die "execute: $sql; ".$self->{dbh}->errstr(); |
176 |
|
|
my @arr; |
177 |
|
|
while (my $row = $sth->fetchrow_hashref) { |
178 |
dpavlin |
188 |
$row->{item} = HTML::Entities::encode($row->{item},' <>&"'); |
179 |
|
|
$row->{display} = HTML::Entities::encode($row->{display},'<>&"'); |
180 |
dpavlin |
60 |
push @arr,$row; |
181 |
|
|
} |
182 |
|
|
return @arr; |
183 |
|
|
} |
184 |
|
|
|
185 |
|
|
sub close { |
186 |
|
|
my $self = shift; |
187 |
|
|
|
188 |
dpavlin |
94 |
return if (! $self->{dbh}); |
189 |
dpavlin |
60 |
|
190 |
dpavlin |
94 |
foreach my $table (keys %Table) { |
191 |
|
|
$self->bench("Crating table $table"); |
192 |
|
|
$self->delete_and_create($table); |
193 |
dpavlin |
60 |
|
194 |
|
|
$self->{dbh}->begin_work || die $self->{dbh}->errstr(); |
195 |
|
|
|
196 |
dpavlin |
94 |
$self->bench("Sorting ".$Table{$table}." items in $table"); |
197 |
|
|
my @keys = sort keys %{$c_table->{$table}}; |
198 |
|
|
|
199 |
|
|
$self->bench("Dumping data into $table"); |
200 |
dpavlin |
188 |
my $sql = "insert into $table (ord,item,display,count) values (?,?,?,?)"; |
201 |
dpavlin |
60 |
my $sth = $self->{dbh}->prepare($sql) || die "sql: $sql; ".$self->{dbh}->errstr(); |
202 |
dpavlin |
94 |
|
203 |
|
|
my $ord = 0; |
204 |
|
|
foreach my $key (@keys) { |
205 |
dpavlin |
95 |
$sth->execute(++$ord, |
206 |
dpavlin |
94 |
$c_table->{$table}->{$key}, |
207 |
dpavlin |
188 |
$c_table->{$table}->{$key}->{display}, |
208 |
dpavlin |
94 |
$c_count->{$table}->{$key} |
209 |
|
|
); |
210 |
dpavlin |
60 |
} |
211 |
|
|
|
212 |
|
|
$self->{dbh}->commit || die $self->{dbh}->errstr(); |
213 |
|
|
} |
214 |
dpavlin |
94 |
$self->bench("disconnecting from database"); |
215 |
dpavlin |
60 |
|
216 |
dpavlin |
94 |
$self->{dbh}->disconnect; |
217 |
|
|
undef $self->{dbh}; |
218 |
dpavlin |
60 |
} |
219 |
|
|
|
220 |
|
|
END { |
221 |
|
|
$Count--; |
222 |
|
|
print STDERR "index_DBI fatal error: \$index->close() not called... $Count references left!\n" if ($Count > 0); |
223 |
|
|
# FIX: debug output |
224 |
|
|
# print STDERR "usage\ttable\n"; |
225 |
|
|
# foreach (keys %Table) { |
226 |
|
|
# print STDERR $Table{$_},"\t$_\n"; |
227 |
|
|
# } |
228 |
|
|
} |
229 |
|
|
|
230 |
|
|
1; |