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 |
dpavlin |
348 |
package WebPac::Index; |
10 |
dpavlin |
60 |
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 |
348 |
use Carp; |
16 |
dpavlin |
60 |
|
17 |
|
|
use DBI; |
18 |
|
|
|
19 |
|
|
my %Table; # index tables which where visited in this run |
20 |
|
|
my %sth_cache; # cache prepared statements |
21 |
|
|
|
22 |
|
|
# cache var |
23 |
|
|
my $c_table; |
24 |
|
|
my $c_count; |
25 |
|
|
|
26 |
dpavlin |
88 |
# bench time |
27 |
dpavlin |
94 |
my $bench_time = time(); |
28 |
dpavlin |
88 |
|
29 |
dpavlin |
94 |
sub bench { |
30 |
|
|
my $self = shift; |
31 |
|
|
my $msg = shift; |
32 |
|
|
|
33 |
|
|
print STDERR "last operation took ",time()-$bench_time," seconds...\n"; |
34 |
|
|
$bench_time=time(); |
35 |
|
|
print STDERR "$msg\n"; |
36 |
|
|
} |
37 |
|
|
|
38 |
dpavlin |
60 |
sub new { |
39 |
|
|
my $class = shift; |
40 |
dpavlin |
348 |
my $self = {@_}; |
41 |
dpavlin |
60 |
bless($self, $class); |
42 |
|
|
|
43 |
dpavlin |
348 |
croak "need dbd" if (! $self->{dbd}); |
44 |
|
|
croak "need dsn" if (! $self->{dsn}); |
45 |
|
|
croak "need user" if (! $self->{user}); |
46 |
|
|
croak "need passwd" if (! $self->{passwd}); |
47 |
dpavlin |
60 |
|
48 |
dpavlin |
348 |
$self->{dbh} = DBI->connect("DBI:$self->{dbd}:$self->{dsn}",$self->{user},$self->{passwd}) || die $DBI::errstr; |
49 |
dpavlin |
60 |
$Count++; |
50 |
|
|
|
51 |
dpavlin |
348 |
$self->bench("connected to $self->{dbd} as $self->{user}"); |
52 |
dpavlin |
94 |
|
53 |
dpavlin |
226 |
# force SQLite to support binary 0 in data (which shouldn't |
54 |
|
|
# happend, but it did to me) |
55 |
|
|
eval { |
56 |
|
|
no warnings 'all'; |
57 |
|
|
$self->{dbh}->{sqlite_handle_binary_nulls} = 1; |
58 |
|
|
}; |
59 |
|
|
|
60 |
dpavlin |
60 |
return $self; |
61 |
|
|
} |
62 |
|
|
|
63 |
|
|
sub delete_and_create { |
64 |
|
|
my $self = shift; |
65 |
|
|
|
66 |
|
|
my $field = shift; |
67 |
|
|
|
68 |
|
|
#print "#### delete_and_create($field)\n"; |
69 |
|
|
|
70 |
|
|
my $sql = "select count(*) from $field"; |
71 |
|
|
my $sth = $self->{dbh}->prepare($sql) || die $self->{dbh}->errstr(); |
72 |
|
|
# FIX: this is not a good way to check if table exists! |
73 |
|
|
if ($sth->execute() && $sth->fetchrow_hashref) { |
74 |
|
|
my $sql = "drop table $field"; |
75 |
dpavlin |
225 |
my $sth = $self->{dbh}->do($sql) || warn "SQL: $sql - ".$sth->errstr(); |
76 |
dpavlin |
60 |
} |
77 |
|
|
$sql = "create table $field ( |
78 |
|
|
item varchar(255), |
79 |
dpavlin |
188 |
display text, |
80 |
dpavlin |
60 |
count int, |
81 |
|
|
ord int, |
82 |
dpavlin |
94 |
primary key (item) |
83 |
dpavlin |
60 |
)"; |
84 |
|
|
|
85 |
dpavlin |
94 |
$sth = $self->{dbh}->do($sql) || warn "SQL: $sql ".$self->{dbh}->errstr(); |
86 |
dpavlin |
60 |
} |
87 |
|
|
|
88 |
|
|
sub insert { |
89 |
|
|
my $self = shift; |
90 |
|
|
|
91 |
|
|
my $field = shift; |
92 |
|
|
my $index_data = shift || print STDERR "\$index->insert($field,NULL,...)"; |
93 |
dpavlin |
188 |
my $display = shift || $index_data; |
94 |
dpavlin |
60 |
|
95 |
|
|
if (! $index_data) { |
96 |
|
|
print STDERR "\$index->insert() -- no value to insert\n"; |
97 |
|
|
return; |
98 |
|
|
} |
99 |
|
|
|
100 |
|
|
$Table{$field}++; |
101 |
|
|
|
102 |
|
|
#$sth_cache{$field."select"}->execute($index_data) || die "cache: $field select; ".$self->{dbh}->errstr(); |
103 |
dpavlin |
93 |
|
104 |
|
|
# XXX for some strange reason, it seems that some entries in my |
105 |
|
|
# database produce strings which start with null byte. I suspect |
106 |
|
|
# this to be bug in OpenIsis 0.9.0. |
107 |
|
|
# This should fix it.. |
108 |
|
|
$index_data =~ s/^[^\w]+//; |
109 |
dpavlin |
60 |
$index_data = substr($index_data,0,255); |
110 |
dpavlin |
93 |
|
111 |
dpavlin |
60 |
my $uc = uc($index_data); |
112 |
dpavlin |
94 |
if (! $c_table->{$field}->{$uc}) { |
113 |
dpavlin |
60 |
#print stderr "in index: $index_data\n"; |
114 |
dpavlin |
94 |
$c_table->{$field}->{$uc} = $index_data; |
115 |
dpavlin |
188 |
$c_table->{$field}->{$uc}->{display} = $display; |
116 |
dpavlin |
94 |
$c_count->{$field}->{$uc} = 1; |
117 |
dpavlin |
60 |
} else { |
118 |
dpavlin |
94 |
$c_count->{$field}->{$uc}++; |
119 |
dpavlin |
60 |
} |
120 |
|
|
} |
121 |
|
|
|
122 |
dpavlin |
140 |
sub count { |
123 |
dpavlin |
60 |
my $self = shift; |
124 |
|
|
|
125 |
|
|
my $field = shift; |
126 |
dpavlin |
140 |
my $where = shift; |
127 |
dpavlin |
60 |
|
128 |
dpavlin |
140 |
my $sql = "select count(*) from $field where upper(item) like upper(?)||'%'"; |
129 |
dpavlin |
60 |
|
130 |
|
|
my $sth = $self->{dbh}->prepare($sql) || die $self->{dbh}->errstr(); |
131 |
dpavlin |
140 |
$sth->execute($where) || die "sql: $sql; ".$self->{dbh}->errstr(); |
132 |
dpavlin |
60 |
|
133 |
|
|
my ($total) = $sth->fetchrow_array(); |
134 |
|
|
|
135 |
dpavlin |
142 |
# no results, count all |
136 |
|
|
if (! $total) { |
137 |
|
|
my $sql = "select count(*) from $field"; |
138 |
|
|
|
139 |
|
|
my $sth = $self->{dbh}->prepare($sql) || die $self->{dbh}->errstr(); |
140 |
|
|
$sth->execute() || die "sql: $sql; ".$self->{dbh}->errstr(); |
141 |
|
|
$total = $sth->fetchrow_array(); |
142 |
|
|
|
143 |
|
|
} |
144 |
|
|
|
145 |
|
|
return $total || 1; |
146 |
dpavlin |
60 |
} |
147 |
|
|
|
148 |
|
|
|
149 |
|
|
sub fetch { |
150 |
|
|
my $self = shift; |
151 |
|
|
|
152 |
|
|
my $field = shift; |
153 |
|
|
my $where = shift; |
154 |
|
|
|
155 |
|
|
my $from_ord = shift || 0; |
156 |
|
|
my $rows = shift || 10; |
157 |
|
|
|
158 |
|
|
my @sql_args; |
159 |
|
|
|
160 |
dpavlin |
188 |
my $sql = "select item,display,ord from $field"; |
161 |
dpavlin |
60 |
|
162 |
|
|
if ($where) { |
163 |
dpavlin |
140 |
my $sql2 = "select ord from $field where upper(item) like upper(?)||'%'"; |
164 |
dpavlin |
60 |
my $sth = $self->{dbh}->prepare($sql2) || die "sql2: $sql2; ".$self->{dbh}->errstr(); |
165 |
|
|
|
166 |
|
|
$sth->execute($where) || die "sql2: $sql2; ".$self->{dbh}->errstr(); |
167 |
|
|
if (my $row = $sth->fetchrow_hashref) { |
168 |
|
|
$from_ord += $row->{ord} - 1; |
169 |
dpavlin |
127 |
} else { |
170 |
|
|
# if no match is found when searching from beginning |
171 |
|
|
# of word in index, try substring match anywhere |
172 |
dpavlin |
201 |
$sql2 = "select ord from $field where upper(item) like '% '||upper(?)||'%'"; |
173 |
dpavlin |
127 |
$sth = $self->{dbh}->prepare($sql2) || die "sql2: $sql2; ".$self->{dbh}->errstr(); |
174 |
|
|
$sth->execute($where) || die "sql2: $sql2; ".$self->{dbh}->errstr(); |
175 |
|
|
if (my $row = $sth->fetchrow_hashref) { |
176 |
|
|
$from_ord += $row->{ord} - 1; |
177 |
|
|
} |
178 |
dpavlin |
60 |
} |
179 |
|
|
} |
180 |
|
|
$sql .= " order by ord limit $rows offset $from_ord"; |
181 |
|
|
|
182 |
|
|
my $sth = $self->{dbh}->prepare($sql) || die "prepare: $sql; ".$self->{dbh}->errstr(); |
183 |
|
|
$sth->execute() || die "execute: $sql; ".$self->{dbh}->errstr(); |
184 |
|
|
my @arr; |
185 |
|
|
while (my $row = $sth->fetchrow_hashref) { |
186 |
dpavlin |
188 |
$row->{item} = HTML::Entities::encode($row->{item},' <>&"'); |
187 |
|
|
$row->{display} = HTML::Entities::encode($row->{display},'<>&"'); |
188 |
dpavlin |
60 |
push @arr,$row; |
189 |
|
|
} |
190 |
|
|
return @arr; |
191 |
|
|
} |
192 |
|
|
|
193 |
|
|
sub close { |
194 |
|
|
my $self = shift; |
195 |
|
|
|
196 |
dpavlin |
94 |
return if (! $self->{dbh}); |
197 |
dpavlin |
60 |
|
198 |
dpavlin |
94 |
foreach my $table (keys %Table) { |
199 |
|
|
$self->bench("Crating table $table"); |
200 |
|
|
$self->delete_and_create($table); |
201 |
dpavlin |
60 |
|
202 |
|
|
$self->{dbh}->begin_work || die $self->{dbh}->errstr(); |
203 |
|
|
|
204 |
dpavlin |
219 |
$self->bench("Sorting ".$Table{$table}." (with duplicates) items in $table"); |
205 |
dpavlin |
94 |
my @keys = sort keys %{$c_table->{$table}}; |
206 |
|
|
|
207 |
dpavlin |
219 |
$self->bench("Dumping ".($#keys+1)." items into $table"); |
208 |
dpavlin |
188 |
my $sql = "insert into $table (ord,item,display,count) values (?,?,?,?)"; |
209 |
dpavlin |
60 |
my $sth = $self->{dbh}->prepare($sql) || die "sql: $sql; ".$self->{dbh}->errstr(); |
210 |
dpavlin |
94 |
|
211 |
|
|
my $ord = 0; |
212 |
|
|
foreach my $key (@keys) { |
213 |
dpavlin |
95 |
$sth->execute(++$ord, |
214 |
dpavlin |
94 |
$c_table->{$table}->{$key}, |
215 |
dpavlin |
188 |
$c_table->{$table}->{$key}->{display}, |
216 |
dpavlin |
94 |
$c_count->{$table}->{$key} |
217 |
|
|
); |
218 |
dpavlin |
60 |
} |
219 |
|
|
|
220 |
|
|
$self->{dbh}->commit || die $self->{dbh}->errstr(); |
221 |
|
|
} |
222 |
dpavlin |
206 |
|
223 |
|
|
if ($self->{dbd} =~ m/(Pg|SQLite)/) { |
224 |
|
|
$self->{dbh}->do(qq{vacuum}) || warn "vacumming failed. It shouldn't if you are using PostgreSQL or SQLite: ".$self->{dbh}->errstr(); |
225 |
|
|
} |
226 |
|
|
|
227 |
dpavlin |
94 |
$self->bench("disconnecting from database"); |
228 |
dpavlin |
60 |
|
229 |
dpavlin |
94 |
$self->{dbh}->disconnect; |
230 |
|
|
undef $self->{dbh}; |
231 |
dpavlin |
60 |
} |
232 |
|
|
|
233 |
|
|
END { |
234 |
|
|
$Count--; |
235 |
|
|
print STDERR "index_DBI fatal error: \$index->close() not called... $Count references left!\n" if ($Count > 0); |
236 |
|
|
# FIX: debug output |
237 |
|
|
# print STDERR "usage\ttable\n"; |
238 |
|
|
# foreach (keys %Table) { |
239 |
|
|
# print STDERR $Table{$_},"\t$_\n"; |
240 |
|
|
# } |
241 |
|
|
} |
242 |
|
|
|
243 |
|
|
1; |