2 |
# this file implements index functions using DBI |
# this file implements index functions using DBI |
3 |
# and huge amounts of memory for cache speedup |
# 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; |
package WebPac::Index; |
10 |
use strict qw(vars); |
use strict qw(vars); |
11 |
use vars qw($Count); |
use vars qw($Count); |
12 |
use HTML::Entities; |
use HTML::Entities; |
13 |
|
use URI::Escape; |
14 |
|
use locale; |
15 |
|
use Carp; |
16 |
|
|
17 |
use DBI; |
use DBI; |
18 |
|
|
24 |
my $c_count; |
my $c_count; |
25 |
|
|
26 |
# bench time |
# bench time |
27 |
my $t = time(); |
my $bench_time = time(); |
28 |
|
|
29 |
|
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 |
sub new { |
sub new { |
39 |
my $class = shift; |
my $class = shift; |
40 |
my $self = {}; |
my $self = {@_}; |
41 |
bless($self, $class); |
bless($self, $class); |
42 |
|
|
43 |
my $dbd = shift || die "need dbi_dbd= in [global] section of configuration file"; |
croak "need dbd" if (! $self->{dbd}); |
44 |
my $dsn = shift || die "need dbi_dsn= in [global] section of configuration file"; |
croak "need dsn" if (! $self->{dsn}); |
45 |
my $user = shift || die "need dbi_user= in [global] section of configuration file"; |
croak "need user" if (! $self->{user}); |
46 |
my $passwd = shift || die "need dbi_passwd= in [global] section of configuration file"; |
croak "need passwd" if (! $self->{passwd}); |
|
|
|
|
$self->{dbh} = DBI->connect("DBI:$dbd:$dsn",$user,$passwd) || die $DBI::errstr; |
|
|
# begin transaction |
|
|
$self->{dbh}->begin_work || die $self->{dbh}->errstr(); |
|
47 |
|
|
48 |
|
$self->{dbh} = DBI->connect("DBI:$self->{dbd}:$self->{dsn}",$self->{user},$self->{passwd}) || die $DBI::errstr; |
49 |
$Count++; |
$Count++; |
50 |
|
|
51 |
|
$self->bench("connected to $self->{dbd} as $self->{user}"); |
52 |
|
|
53 |
|
# 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 |
return $self; |
return $self; |
61 |
} |
} |
62 |
|
|
67 |
|
|
68 |
#print "#### delete_and_create($field)\n"; |
#print "#### delete_and_create($field)\n"; |
69 |
|
|
|
$self->{dbh}->commit; |
|
|
|
|
70 |
my $sql = "select count(*) from $field"; |
my $sql = "select count(*) from $field"; |
71 |
my $sth = $self->{dbh}->prepare($sql) || die $self->{dbh}->errstr(); |
my $sth = $self->{dbh}->prepare($sql) || die $self->{dbh}->errstr(); |
72 |
# FIX: this is not a good way to check if table exists! |
# FIX: this is not a good way to check if table exists! |
73 |
if ($sth->execute() && $sth->fetchrow_hashref) { |
if ($sth->execute() && $sth->fetchrow_hashref) { |
74 |
my $sql = "drop table $field"; |
my $sql = "drop table $field"; |
75 |
$self->{dbh}->begin_work; |
my $sth = $self->{dbh}->do($sql) || warn "SQL: $sql - ".$sth->errstr(); |
|
my $sth = $self->{dbh}->do($sql) || die "SQL: $sql ".$self->{dbh}->errstr(); |
|
|
$self->{dbh}->commit; |
|
76 |
} |
} |
77 |
$sql = "create table $field ( |
$sql = "create table $field ( |
78 |
item varchar(255), |
item varchar(255), |
79 |
ident varchar(255), |
display text, |
80 |
count int, |
count int, |
81 |
ord int, |
ord int, |
82 |
primary key (item,ident) |
primary key (item) |
83 |
)"; |
)"; |
84 |
|
|
85 |
$self->{dbh}->begin_work; |
$sth = $self->{dbh}->do($sql) || warn "SQL: $sql ".$self->{dbh}->errstr(); |
|
$sth = $self->{dbh}->do($sql); # || warn "SQL: $sql ".$self->{dbh}->errstr(); |
|
|
$self->{dbh}->commit; |
|
|
|
|
|
$self->{dbh}->begin_work; |
|
86 |
} |
} |
87 |
|
|
88 |
sub insert { |
sub insert { |
90 |
|
|
91 |
my $field = shift; |
my $field = shift; |
92 |
my $index_data = shift || print STDERR "\$index->insert($field,NULL,...)"; |
my $index_data = shift || print STDERR "\$index->insert($field,NULL,...)"; |
93 |
my $ident = shift || ''; # e.g. library id |
my $display = shift || $index_data; |
94 |
|
|
95 |
if (! $index_data) { |
if (! $index_data) { |
96 |
print STDERR "\$index->insert() -- no value to insert\n"; |
print STDERR "\$index->insert() -- no value to insert\n"; |
97 |
return; |
return; |
98 |
} |
} |
99 |
|
|
|
if (! $Table{$field}) { |
|
|
$self->delete_and_create($field); |
|
|
|
|
|
my $sql = "select item from $field where upper(item)=upper(?)"; |
|
|
$sth_cache{$field."select"} = $self->{dbh}->prepare($sql) || die $self->{dbh}->errstr(); |
|
|
|
|
|
$sql = "insert into $field (item,ident,count) values (?,?,?)"; |
|
|
$sth_cache{$field."insert"} = $self->{dbh}->prepare($sql) || die $self->{dbh}->errstr(); |
|
|
|
|
|
$sql = "update $field set count = count + 1 where item = ? and ident = ?"; |
|
|
$sth_cache{$field."update"} = $self->{dbh}->prepare($sql) || die $self->{dbh}->errstr(); |
|
|
} |
|
100 |
$Table{$field}++; |
$Table{$field}++; |
101 |
|
|
102 |
#$sth_cache{$field."select"}->execute($index_data) || die "cache: $field select; ".$self->{dbh}->errstr(); |
#$sth_cache{$field."select"}->execute($index_data) || die "cache: $field select; ".$self->{dbh}->errstr(); |
109 |
$index_data = substr($index_data,0,255); |
$index_data = substr($index_data,0,255); |
110 |
|
|
111 |
my $uc = uc($index_data); |
my $uc = uc($index_data); |
112 |
if (! $c_table->{$field}->{$ident}->{$uc}) { |
if (! $c_table->{$field}->{$uc}) { |
|
$sth_cache{$field."insert"}->execute($index_data,$ident,0) || warn "cache: $field insert ($index_data,$ident); ".$self->{dbh}->errstr(); |
|
113 |
#print stderr "in index: $index_data\n"; |
#print stderr "in index: $index_data\n"; |
114 |
$c_table->{$field}->{$ident}->{$uc} = $index_data; |
$c_table->{$field}->{$uc} = $index_data; |
115 |
$c_count->{$field}->{$ident}->{$uc} = 1; |
$c_table->{$field}->{$uc}->{display} = $display; |
116 |
|
$c_count->{$field}->{$uc} = 1; |
117 |
} else { |
} else { |
118 |
$c_count->{$field}->{$ident}->{$uc}++; |
$c_count->{$field}->{$uc}++; |
119 |
} |
} |
120 |
} |
} |
121 |
|
|
122 |
sub check { |
sub count { |
123 |
my $self = shift; |
my $self = shift; |
124 |
|
|
125 |
my $field = shift; |
my $field = shift; |
126 |
|
my $where = shift; |
127 |
|
|
128 |
my $sql = "select count(*) from $field"; |
my $sql = "select count(*) from $field where upper(item) like upper(?)||'%'"; |
129 |
|
|
130 |
my $sth = $self->{dbh}->prepare($sql) || die $self->{dbh}->errstr(); |
my $sth = $self->{dbh}->prepare($sql) || die $self->{dbh}->errstr(); |
131 |
$sth->execute() || die "sql: $sql; ".$self->{dbh}->errstr(); |
$sth->execute($where) || die "sql: $sql; ".$self->{dbh}->errstr(); |
132 |
|
|
133 |
my ($total) = $sth->fetchrow_array(); |
my ($total) = $sth->fetchrow_array(); |
134 |
|
|
135 |
return $total; |
# 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 |
} |
} |
147 |
|
|
148 |
|
|
150 |
my $self = shift; |
my $self = shift; |
151 |
|
|
152 |
my $field = shift; |
my $field = shift; |
|
my $what = shift || 'item'; # 'item,ident' |
|
153 |
my $where = shift; |
my $where = shift; |
154 |
|
|
155 |
my $from_ord = shift || 0; |
my $from_ord = shift || 0; |
157 |
|
|
158 |
my @sql_args; |
my @sql_args; |
159 |
|
|
160 |
my $sql = "select $what,ord from $field"; |
my $sql = "select item,display,ord from $field"; |
161 |
|
|
162 |
if ($where) { |
if ($where) { |
163 |
my $sql2 = " select ord from $field where upper($what) like upper(?)||'%'"; |
my $sql2 = "select ord from $field where upper(item) like upper(?)||'%'"; |
164 |
my $sth = $self->{dbh}->prepare($sql2) || die "sql2: $sql2; ".$self->{dbh}->errstr(); |
my $sth = $self->{dbh}->prepare($sql2) || die "sql2: $sql2; ".$self->{dbh}->errstr(); |
165 |
|
|
166 |
$sth->execute($where) || die "sql2: $sql2; ".$self->{dbh}->errstr(); |
$sth->execute($where) || die "sql2: $sql2; ".$self->{dbh}->errstr(); |
167 |
if (my $row = $sth->fetchrow_hashref) { |
if (my $row = $sth->fetchrow_hashref) { |
168 |
$from_ord += $row->{ord} - 1; |
$from_ord += $row->{ord} - 1; |
169 |
|
} else { |
170 |
|
# if no match is found when searching from beginning |
171 |
|
# of word in index, try substring match anywhere |
172 |
|
$sql2 = "select ord from $field where upper(item) like '% '||upper(?)||'%'"; |
173 |
|
$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 |
} |
} |
179 |
} |
} |
180 |
$sql .= " order by ord limit $rows offset $from_ord"; |
$sql .= " order by ord limit $rows offset $from_ord"; |
183 |
$sth->execute() || die "execute: $sql; ".$self->{dbh}->errstr(); |
$sth->execute() || die "execute: $sql; ".$self->{dbh}->errstr(); |
184 |
my @arr; |
my @arr; |
185 |
while (my $row = $sth->fetchrow_hashref) { |
while (my $row = $sth->fetchrow_hashref) { |
186 |
$row->{item} = HTML::Entities::encode($row->{item},'<>&"'); |
$row->{item} = HTML::Entities::encode($row->{item},' <>&"'); |
187 |
|
$row->{display} = HTML::Entities::encode($row->{display},'<>&"'); |
188 |
push @arr,$row; |
push @arr,$row; |
189 |
} |
} |
190 |
return @arr; |
return @arr; |
193 |
sub close { |
sub close { |
194 |
my $self = shift; |
my $self = shift; |
195 |
|
|
196 |
|
return if (! $self->{dbh}); |
197 |
|
|
198 |
# re-create ord column (sorted order) in table |
foreach my $table (keys %Table) { |
199 |
sub create_ord { |
$self->bench("Crating table $table"); |
200 |
|
$self->delete_and_create($table); |
|
my $table = shift; |
|
201 |
|
|
202 |
$self->{dbh}->begin_work || die $self->{dbh}->errstr(); |
$self->{dbh}->begin_work || die $self->{dbh}->errstr(); |
203 |
|
|
204 |
my $sql = "select oid from $table order by upper(item)"; |
$self->bench("Sorting ".$Table{$table}." (with duplicates) items in $table"); |
205 |
|
my @keys = sort keys %{$c_table->{$table}}; |
206 |
|
|
207 |
|
$self->bench("Dumping ".($#keys+1)." items into $table"); |
208 |
|
my $sql = "insert into $table (ord,item,display,count) values (?,?,?,?)"; |
209 |
my $sth = $self->{dbh}->prepare($sql) || die "sql: $sql; ".$self->{dbh}->errstr(); |
my $sth = $self->{dbh}->prepare($sql) || die "sql: $sql; ".$self->{dbh}->errstr(); |
210 |
$sql = "update $table set ord=? where oid=?"; |
|
211 |
my $sth_update = $self->{dbh}->prepare($sql) || die "sql: $sql; ".$self->{dbh}->errstr(); |
my $ord = 0; |
212 |
$sth->execute() || die "sql: $sql; ".$self->{dbh}->errstr(); |
foreach my $key (@keys) { |
213 |
my $ord = 1; |
$sth->execute(++$ord, |
214 |
while (my $row = $sth->fetchrow_hashref) { |
$c_table->{$table}->{$key}, |
215 |
$sth_update->execute($ord++,$row->{oid}); |
$c_table->{$table}->{$key}->{display}, |
216 |
|
$c_count->{$table}->{$key} |
217 |
|
); |
218 |
} |
} |
219 |
|
|
220 |
$self->{dbh}->commit || die $self->{dbh}->errstr(); |
$self->{dbh}->commit || die $self->{dbh}->errstr(); |
221 |
} |
} |
|
#--- end of sub |
|
222 |
|
|
223 |
if ($self->{dbh}) { |
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 |
# commit |
} |
|
$self->{dbh}->commit || die $self->{dbh}->errstr(); |
|
226 |
|
|
227 |
foreach my $table (keys %Table) { |
$self->bench("disconnecting from database"); |
|
# FIX |
|
|
print STDERR "last operation took ",time()-$t," seconds...\n"; |
|
|
$t=time(); |
|
|
print STDERR "creating ord for $table...\n"; |
|
|
create_ord($table); |
|
|
undef $sth_cache{$table."select"}; |
|
|
undef $sth_cache{$table."insert"}; |
|
|
undef $sth_cache{$table."update"}; |
|
|
# XXX |
|
|
# $sth_cache{$field."update"}->execute($index_data,$ident) || die "cache: $field update; ".$self->{dbh}->errstr(); |
|
|
} |
|
228 |
|
|
229 |
$self->{dbh}->disconnect; |
$self->{dbh}->disconnect; |
230 |
undef $self->{dbh}; |
undef $self->{dbh}; |
|
} |
|
231 |
} |
} |
232 |
|
|
233 |
END { |
END { |