/[webpac]/trunk/index_DBI_cache.pm
This is repository of my old source code which isn't updated any more. Go to git.rot13.org for current projects!
ViewVC logotype

Annotation of /trunk/index_DBI_cache.pm

Parent Directory Parent Directory | Revision Log Revision Log


Revision 219 - (hide annotations)
Thu Feb 5 10:56:55 2004 UTC (20 years, 1 month ago) by dpavlin
File size: 6244 byte(s)
Display total and unique count when dumping index data to tables

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 60 return $self;
55     }
56    
57     sub delete_and_create {
58     my $self = shift;
59    
60     my $field = shift;
61    
62     #print "#### delete_and_create($field)\n";
63    
64     my $sql = "select count(*) from $field";
65     my $sth = $self->{dbh}->prepare($sql) || die $self->{dbh}->errstr();
66     # FIX: this is not a good way to check if table exists!
67     if ($sth->execute() && $sth->fetchrow_hashref) {
68     my $sql = "drop table $field";
69     my $sth = $self->{dbh}->do($sql) || die "SQL: $sql ".$self->{dbh}->errstr();
70     }
71     $sql = "create table $field (
72     item varchar(255),
73 dpavlin 188 display text,
74 dpavlin 60 count int,
75     ord int,
76 dpavlin 94 primary key (item)
77 dpavlin 60 )";
78    
79 dpavlin 94 $sth = $self->{dbh}->do($sql) || warn "SQL: $sql ".$self->{dbh}->errstr();
80 dpavlin 60 }
81    
82     sub insert {
83     my $self = shift;
84    
85     my $field = shift;
86     my $index_data = shift || print STDERR "\$index->insert($field,NULL,...)";
87 dpavlin 188 my $display = shift || $index_data;
88 dpavlin 60
89     if (! $index_data) {
90     print STDERR "\$index->insert() -- no value to insert\n";
91     return;
92     }
93    
94     $Table{$field}++;
95    
96     #$sth_cache{$field."select"}->execute($index_data) || die "cache: $field select; ".$self->{dbh}->errstr();
97 dpavlin 93
98     # XXX for some strange reason, it seems that some entries in my
99     # database produce strings which start with null byte. I suspect
100     # this to be bug in OpenIsis 0.9.0.
101     # This should fix it..
102     $index_data =~ s/^[^\w]+//;
103 dpavlin 60 $index_data = substr($index_data,0,255);
104 dpavlin 93
105 dpavlin 60 my $uc = uc($index_data);
106 dpavlin 94 if (! $c_table->{$field}->{$uc}) {
107 dpavlin 60 #print stderr "in index: $index_data\n";
108 dpavlin 94 $c_table->{$field}->{$uc} = $index_data;
109 dpavlin 188 $c_table->{$field}->{$uc}->{display} = $display;
110 dpavlin 94 $c_count->{$field}->{$uc} = 1;
111 dpavlin 60 } else {
112 dpavlin 94 $c_count->{$field}->{$uc}++;
113 dpavlin 60 }
114     }
115    
116 dpavlin 140 sub count {
117 dpavlin 60 my $self = shift;
118    
119     my $field = shift;
120 dpavlin 140 my $where = shift;
121 dpavlin 60
122 dpavlin 140 my $sql = "select count(*) from $field where upper(item) like upper(?)||'%'";
123 dpavlin 60
124     my $sth = $self->{dbh}->prepare($sql) || die $self->{dbh}->errstr();
125 dpavlin 140 $sth->execute($where) || die "sql: $sql; ".$self->{dbh}->errstr();
126 dpavlin 60
127     my ($total) = $sth->fetchrow_array();
128    
129 dpavlin 142 # 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 dpavlin 60 }
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 dpavlin 188 my $sql = "select item,display,ord from $field";
155 dpavlin 60
156     if ($where) {
157 dpavlin 140 my $sql2 = "select ord from $field where upper(item) like upper(?)||'%'";
158 dpavlin 60 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 dpavlin 127 } else {
164     # if no match is found when searching from beginning
165     # of word in index, try substring match anywhere
166 dpavlin 201 $sql2 = "select ord from $field where upper(item) like '% '||upper(?)||'%'";
167 dpavlin 127 $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 dpavlin 60 }
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 dpavlin 188 $row->{item} = HTML::Entities::encode($row->{item},' <>&"');
181     $row->{display} = HTML::Entities::encode($row->{display},'<>&"');
182 dpavlin 60 push @arr,$row;
183     }
184     return @arr;
185     }
186    
187     sub close {
188     my $self = shift;
189    
190 dpavlin 94 return if (! $self->{dbh});
191 dpavlin 60
192 dpavlin 94 foreach my $table (keys %Table) {
193     $self->bench("Crating table $table");
194     $self->delete_and_create($table);
195 dpavlin 60
196     $self->{dbh}->begin_work || die $self->{dbh}->errstr();
197    
198 dpavlin 219 $self->bench("Sorting ".$Table{$table}." (with duplicates) items in $table");
199 dpavlin 94 my @keys = sort keys %{$c_table->{$table}};
200    
201 dpavlin 219 $self->bench("Dumping ".($#keys+1)." items into $table");
202 dpavlin 188 my $sql = "insert into $table (ord,item,display,count) values (?,?,?,?)";
203 dpavlin 60 my $sth = $self->{dbh}->prepare($sql) || die "sql: $sql; ".$self->{dbh}->errstr();
204 dpavlin 94
205     my $ord = 0;
206     foreach my $key (@keys) {
207 dpavlin 95 $sth->execute(++$ord,
208 dpavlin 94 $c_table->{$table}->{$key},
209 dpavlin 188 $c_table->{$table}->{$key}->{display},
210 dpavlin 94 $c_count->{$table}->{$key}
211     );
212 dpavlin 60 }
213    
214     $self->{dbh}->commit || die $self->{dbh}->errstr();
215     }
216 dpavlin 206
217     if ($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 dpavlin 94 $self->bench("disconnecting from database");
222 dpavlin 60
223 dpavlin 94 $self->{dbh}->disconnect;
224     undef $self->{dbh};
225 dpavlin 60 }
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;

Properties

Name Value
cvs2svn:cvs-rev 1.15

  ViewVC Help
Powered by ViewVC 1.1.26