Experimental support for dBase .dbf files. Usege like this in all2xml.conf:

[webpac] / index_DBI_cache.pm

diff --git a/index_DBI_cache.pm b/index_DBI_cache.pm
index c544fbf..b3741fa 100644 (file)
--- a/index_DBI_cache.pm
+++ b/index_DBI_cache.pm
@@ -10,6 +10,8 @@ package index_DBI;
 use strict qw(vars);
 use vars qw($Count);
 use HTML::Entities;
+use URI::Escape;
+use locale;
 
 use DBI;
 
@@ -42,11 +44,20 @@ sub new {
        my $user = shift || die "need dbi_user= in [global] section of configuration file";
        my $passwd = shift || die "need dbi_passwd= in [global] section of configuration file";
 
+       $self->{dbd} = $dbd;
+
        $self->{dbh} = DBI->connect("DBI:$dbd:$dsn",$user,$passwd) || die $DBI::errstr;
        $Count++;
 
        $self->bench("connected to $dbd as $user");
 
+       # force SQLite to support binary 0 in data (which shouldn't
+       # happend, but it did to me)
+       eval {
+               no warnings 'all';
+               $self->{dbh}->{sqlite_handle_binary_nulls} = 1;
+       };
+
        return $self;
 }
 
@@ -58,14 +69,15 @@ sub delete_and_create {
 #print "#### delete_and_create($field)\n";
 
        my $sql = "select count(*) from $field";
-       my $sth = $self->{dbh}->prepare($sql) || die $self->{dbh}->errstr();
+       my $sth = $self->{dbh}->prepare($sql);
 # FIX: this is not a good way to check if table exists!
-       if ($sth->execute() && $sth->fetchrow_hashref) {
+       if ($sth && $sth->execute() && $sth->fetchrow_hashref) {
                my $sql = "drop table $field";
-               my $sth = $self->{dbh}->do($sql) || die "SQL: $sql ".$self->{dbh}->errstr();
+               my $sth = $self->{dbh}->do($sql) || warn "SQL: $sql - ".$sth->errstr();
        }
        $sql = "create table $field (
                        item varchar(255),
+                       display text,
                        count int,
                        ord int,
                        primary key (item)
@@ -79,7 +91,7 @@ sub insert {
 
        my $field = shift;
        my $index_data = shift || print STDERR "\$index->insert($field,NULL,...)";
-       my $ident = shift || '';        # e.g. library id
+       my $display = shift || $index_data;
 
        if (! $index_data) {
                print STDERR "\$index->insert() -- no value to insert\n";
@@ -90,17 +102,13 @@ sub insert {
 
        #$sth_cache{$field."select"}->execute($index_data) || die "cache: $field select; ".$self->{dbh}->errstr();
 
-       # XXX for some strange reason, it seems that some entries in my
-       # database produce strings which start with null byte. I suspect
-       # this to be bug in OpenIsis 0.9.0.
-       # This should fix it..
-       $index_data =~ s/^[^\w]+//;
-       $index_data = substr($index_data,0,255);
+       $index_data =~ s#&(\w)(acute|cedil|circ|grave|ring|slash|tilde|uml);#$1#gi;
 
        my $uc = uc($index_data);
        if (! $c_table->{$field}->{$uc}) {
 #print stderr "in index: $index_data\n";
                $c_table->{$field}->{$uc} = $index_data;
+               $c_table->{$field}->{$uc}->{display} = $display;
                $c_count->{$field}->{$uc} = 1;
        } else {
                $c_count->{$field}->{$uc}++;
@@ -145,7 +153,7 @@ sub fetch {
 
        my @sql_args;
 
-       my $sql = "select item,ord from $field";
+       my $sql = "select item,display,ord from $field";
 
        if ($where) {
                my $sql2 = "select ord from $field where upper(item) like upper(?)||'%'";
@@ -157,7 +165,7 @@ sub fetch {
                } else {
                        # if no match is found when searching from beginning
                        # of word in index, try substring match anywhere
-                       $sql2 = "select ord from $field where upper(item) like '%'||upper(?)||'%'";
+                       $sql2 = "select ord from $field where upper(item) like '% '||upper(?)||'%'";
                        $sth = $self->{dbh}->prepare($sql2) || die "sql2: $sql2; ".$self->{dbh}->errstr();
                        $sth->execute($where) || die "sql2: $sql2; ".$self->{dbh}->errstr();
                        if (my $row = $sth->fetchrow_hashref) {
@@ -171,7 +179,10 @@ sub fetch {
        $sth->execute() || die "execute: $sql; ".$self->{dbh}->errstr();
        my @arr;
        while (my $row = $sth->fetchrow_hashref) {
-               $row->{item} = HTML::Entities::encode($row->{item},'<>&"');
+               $row->{item} = HTML::Entities::encode($row->{item},' <>&"');
+               $row->{display} = HTML::Entities::encode($row->{display},'<>&"');
+               $row->{item} =~ s#&amp;(\w)(acute|cedil|circ|grave|ring|slash|tilde|uml);#$1#gi;
+               $row->{display} =~ s#&amp;(\w)(acute|cedil|circ|grave|ring|slash|tilde|uml);#&$1$2;#gi;
                push @arr,$row;
        }
        return @arr;
@@ -188,23 +199,29 @@ sub close {
 
                $self->{dbh}->begin_work || die $self->{dbh}->errstr();
 
-               $self->bench("Sorting ".$Table{$table}." items in $table");
+               $self->bench("Sorting ".$Table{$table}." (with duplicates) items in $table");
                my @keys = sort keys %{$c_table->{$table}};
 
-               $self->bench("Dumping data into $table");
-               my $sql = "insert into $table (ord,item,count) values (?,?,?)";
+               $self->bench("Dumping ".($#keys+1)." items into $table");
+               my $sql = "insert into $table (ord,item,display,count) values (?,?,?,?)";
                my $sth = $self->{dbh}->prepare($sql) || die "sql: $sql; ".$self->{dbh}->errstr();
 
                my $ord = 0;
                foreach my $key (@keys) {
                        $sth->execute(++$ord,
                                $c_table->{$table}->{$key},
+                               $c_table->{$table}->{$key}->{display},
                                $c_count->{$table}->{$key}
                        );
                }
 
                $self->{dbh}->commit || die $self->{dbh}->errstr();
        }
+
+       if ($self->{dbd} =~ m/(Pg|SQLite)/) {
+               $self->{dbh}->do(qq{vacuum}) || warn "vacumming failed. It shouldn't if you are using PostgreSQL or SQLite: ".$self->{dbh}->errstr();
+       }
+
        $self->bench("disconnecting from database");
 
        $self->{dbh}->disconnect;