Line # Revision Author
1 60 dpavlin #
2 # this file implements index functions using DBI
3 # and huge amounts of memory for cache speedup
4 #
5 94 dpavlin # this version doesn't support ident (which sould be location in
6 # library). But, that functionality is not used anyway...
7 #
8 60 dpavlin
9 package index_DBI;
10 use strict qw(vars);
11 use vars qw($Count);
12 use HTML::Entities;
13 188 dpavlin use URI::Escape;
14 192 dpavlin use locale;
15 60 dpavlin
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 88 dpavlin # bench time
26 94 dpavlin my $bench_time = time();
27 88 dpavlin
28 94 dpavlin 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 60 dpavlin 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
47 206 dpavlin $self->{dbd} = $dbd;
48
49 60 dpavlin $self->{dbh} = DBI->connect("DBI:$dbd:$dsn",$user,$passwd) || die $DBI::errstr;
50 $Count++;
51
52 94 dpavlin $self->bench("connected to $dbd as $user");
53
54 226 dpavlin # 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 60 dpavlin 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";
72 400 dpavlin my $sth = $self->{dbh}->prepare($sql);
73 60 dpavlin # FIX: this is not a good way to check if table exists!
74 400 dpavlin if ($sth && $sth->execute() && $sth->fetchrow_hashref) {
75 60 dpavlin my $sql = "drop table $field";
76 225 dpavlin my $sth = $self->{dbh}->do($sql) || warn "SQL: $sql - ".$sth->errstr();
77 60 dpavlin }
78 $sql = "create table $field (
79 item varchar(255),
80 188 dpavlin display text,
81 60 dpavlin count int,
82 ord int,
83 94 dpavlin primary key (item)
84 60 dpavlin )";
85
86 94 dpavlin $sth = $self->{dbh}->do($sql) || warn "SQL: $sql ".$self->{dbh}->errstr();
87 60 dpavlin }
88
89 sub insert {
90 my $self = shift;
91
92 my $field = shift;
93 my $index_data = shift || print STDERR "\$index->insert($field,NULL,...)";
94 188 dpavlin my $display = shift || $index_data;
95 60 dpavlin
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 93 dpavlin
105 60 dpavlin my $uc = uc($index_data);
106 94 dpavlin if (! $c_table->{$field}->{$uc}) {
107 60 dpavlin #print stderr "in index: $index_data\n";
108 94 dpavlin $c_table->{$field}->{$uc} = $index_data;
109 188 dpavlin $c_table->{$field}->{$uc}->{display} = $display;
110 94 dpavlin $c_count->{$field}->{$uc} = 1;
111 60 dpavlin } else {
112 94 dpavlin $c_count->{$field}->{$uc}++;
113 60 dpavlin }
114 }
115
116 140 dpavlin sub count {
117 60 dpavlin my $self = shift;
118
119 my $field = shift;
120 140 dpavlin my $where = shift;
121 60 dpavlin
122 140 dpavlin my $sql = "select count(*) from $field where upper(item) like upper(?)||'%'";
123 60 dpavlin
124 my $sth = $self->{dbh}->prepare($sql) || die $self->{dbh}->errstr();
125 140 dpavlin $sth->execute($where) || die "sql: $sql; ".$self->{dbh}->errstr();
126 60 dpavlin
127 my ($total) = $sth->fetchrow_array();
128
129 142 dpavlin # no results, count all
130 if (! $total) {
131 my $sql = "select count(*) from $field";
132
133 my $sth = $self->{dbh}->prepare($sql) || die $self->{dbh}->errstr();
134 $sth->execute() || die "sql: $sql; ".$self->{dbh}->errstr();
135 $total = $sth->fetchrow_array();
136
137 }
138
139 return $total || 1;
140 60 dpavlin }
141
142
143 sub fetch {
144 my $self = shift;
145
146 my $field = shift;
147 my $where = shift;
148
149 my $from_ord = shift || 0;
150 my $rows = shift || 10;
151
152 my @sql_args;
153
154 188 dpavlin my $sql = "select item,display,ord from $field";
155 60 dpavlin
156 if ($where) {
157 140 dpavlin my $sql2 = "select ord from $field where upper(item) like upper(?)||'%'";
158 60 dpavlin my $sth = $self->{dbh}->prepare($sql2) || die "sql2: $sql2; ".$self->{dbh}->errstr();
159
160 $sth->execute($where) || die "sql2: $sql2; ".$self->{dbh}->errstr();
161 if (my $row = $sth->fetchrow_hashref) {
162 $from_ord += $row->{ord} - 1;
163 127 dpavlin } else {
164 # if no match is found when searching from beginning
165 # of word in index, try substring match anywhere
166 201 dpavlin $sql2 = "select ord from $field where upper(item) like '% '||upper(?)||'%'";
167 127 dpavlin $sth = $self->{dbh}->prepare($sql2) || die "sql2: $sql2; ".$self->{dbh}->errstr();
168 $sth->execute($where) || die "sql2: $sql2; ".$self->{dbh}->errstr();
169 if (my $row = $sth->fetchrow_hashref) {
170 $from_ord += $row->{ord} - 1;
171 }
172 60 dpavlin }
173 }
174 $sql .= " order by ord limit $rows offset $from_ord";
175
176 my $sth = $self->{dbh}->prepare($sql) || die "prepare: $sql; ".$self->{dbh}->errstr();
177 $sth->execute() || die "execute: $sql; ".$self->{dbh}->errstr();
178 my @arr;
179 while (my $row = $sth->fetchrow_hashref) {
180 188 dpavlin $row->{item} = HTML::Entities::encode($row->{item},' <>&"');
181 $row->{display} = HTML::Entities::encode($row->{display},'<>&"');
182 60 dpavlin push @arr,$row;
183 }
184 return @arr;
185 }
186
187 sub close {
188 my $self = shift;
189
190 94 dpavlin return if (! $self->{dbh});
191 60 dpavlin
192 94 dpavlin foreach my $table (keys %Table) {
193 $self->bench("Crating table $table");
194 $self->delete_and_create($table);
195 60 dpavlin
196 $self->{dbh}->begin_work || die $self->{dbh}->errstr();
197
198 219 dpavlin $self->bench("Sorting ".$Table{$table}." (with duplicates) items in $table");
199 94 dpavlin my @keys = sort keys %{$c_table->{$table}};
200
201 219 dpavlin $self->bench("Dumping ".($#keys+1)." items into $table");
202 188 dpavlin my $sql = "insert into $table (ord,item,display,count) values (?,?,?,?)";
203 60 dpavlin my $sth = $self->{dbh}->prepare($sql) || die "sql: $sql; ".$self->{dbh}->errstr();
204 94 dpavlin
205 my $ord = 0;
206 foreach my $key (@keys) {
207 95 dpavlin $sth->execute(++$ord,
208 94 dpavlin $c_table->{$table}->{$key},
209 188 dpavlin $c_table->{$table}->{$key}->{display},
210 94 dpavlin $c_count->{$table}->{$key}
211 );
212 60 dpavlin }
213
214 $self->{dbh}->commit || die $self->{dbh}->errstr();
215 }
216 206 dpavlin
217 if ($self->{dbd} =~ m/(Pg|SQLite)/) {
218 $self->{dbh}->do(qq{vacuum}) || warn "vacumming failed. It shouldn't if you are using PostgreSQL or SQLite: ".$self->{dbh}->errstr();
219 }
220
221 94 dpavlin $self->bench("disconnecting from database");
222 60 dpavlin
223 94 dpavlin $self->{dbh}->disconnect;
224 undef $self->{dbh};
225 60 dpavlin }
226
227 END {
228 $Count--;
229 print STDERR "index_DBI fatal error: \$index->close() not called... $Count references left!\n" if ($Count > 0);
230 # FIX: debug output
231 # print STDERR "usage\ttable\n";
232 # foreach (keys %Table) {
233 # print STDERR $Table{$_},"\t$_\n";
234 # }
235 }
236
237 1;