support for lookup fields. Implemented using GDBM or TDB (which I recommend

because it's fastest implementation)

git-svn-id: file:///home/dpavlin/private/svn/webpac/trunk@177 13eb9ef6-21d5-0310-b721-a9d68796d827

diff --git a/all2xml.pl b/all2xml.pl
index 98f5de1..158b866 100755 (executable)
--- a/all2xml.pl
+++ b/all2xml.pl
@@ -9,6 +9,9 @@ use Text::Unaccent 1.02;        # 1.01 won't compile on my platform,
 use Text::Iconv;
 use Config::IniFiles;
 use Encode;
+#use GDBM_File;
+use Fcntl;     # for O_RDWR
+use TDB_File;
 
 $|=1;
 
@@ -63,6 +66,12 @@ my %type2tag = (
 
 my $cache;     # for cacheing
 
+# lookup hash (tied to file)
+my %lhash;
+# this option will cache all lookup entries in memory.
+# if you are tight on memory, turn this off
+my $use_lhash_cache = 1;
+
 sub data2xml {
 
        use xmlify;
@@ -100,6 +109,9 @@ sub data2xml {
                $cache->{tags_by_order}->{$type} = \@sorted_tags;
        }
 
+       # lookup key
+       my $lookup_key;
+
        foreach my $field (@sorted_tags) {
 
                $field=x($field);
@@ -120,11 +132,15 @@ sub data2xml {
 
                        my $repeat_off = 0;             # repeatable offset
 
-                       my ($s,$se,$d,$i) = (1,0,1,0);  # swish, display default
+                       # swish, swish_exact, display, index, index_lookup
+                       # swish and display defaults
+                       my ($s,$se,$d,$i,$il) = (1,0,1,0,0);
                        $s = 0 if (lc($x->{type}) eq "display");
                        $d = 0 if (lc($x->{type}) eq "swish");
                        $se = 1 if (lc($x->{type}) eq "swish_exact");
                        ($s,$d,$i) = (0,0,1) if (lc($x->{type}) eq "index");
+                       $il = 1 if (lc($x->{type}) =~ /^lookup/);
+
 
                        # what will separate last line from this one?
                        if ($display_data && $x->{append} && $x->{append} eq "1") {
@@ -138,7 +154,6 @@ sub data2xml {
 
                        # placeholder for all repeatable entries for index
                        my @index_data;
-                       my $index_filter;
 
                        sub mkformat {
                                my $x = shift || die "mkformat needs tag reference";
@@ -173,7 +188,29 @@ sub data2xml {
                                        print STDERR "loop (more than 1000 repeatable fields) deteced in $row, $format\n";
                                        last;
                                }
-                               
+
+                               # is this field is lookup?
+                               if ($display && $x->{lookup}) {
+                                       if ($use_lhash_cache) {
+                                               if (!defined($cache->{lhash}->{$display})) {
+                                                       my $new_display = $lhash{$display};
+                                                       if ($new_display) {
+#print STDERR "lookup cache store '$display' = '$new_display'\n";
+                                                               $display = $new_display;
+                                                               $cache->{lhash}->{$display} = $new_display;
+                                                       } else {
+                                                               print STDERR "WARNING: lookup for '$display' didn't find anything.\n";
+                                                               $display = "";
+                                                               $cache->{lhash}->{$display} = "";
+                                                       }
+                                               } else {
+                                                       $display = $cache->{lhash}->{$display};
+                                               }
+                                       } else {
+                                               $display = $lhash{$display} || "";
+                                       }
+                               }
+
                                # filter="name" ; filter this field through
                                # filter/[name].pm
                                my $filter = $x->{filter};
@@ -219,24 +256,39 @@ sub data2xml {
                                                
                                # type="index" ; insert into index
                                if ($i && $display) {
-                                       push @index_data, $display;
-                                       $index_filter = $filter if ($filter);
+                                       if ($filter) {
+                                               no strict 'refs';
+                                               $display = &$filter($display);
+                                       }
+                                       if ($x->{append} && @index_data) {
+                                               $index_data[$#index_data].=$display;
+                                       } else {
+                                               push @index_data, $display;
+                                       }
                                }
-                       }
 
-                       # fill data in index
-                       if (@index_data) {
-                               if ($index_filter) {
-                                       no strict 'refs';
-                                       foreach my $d (@index_data) {
-                                               $index->insert($field, &$index_filter($d), $path);
-                                       }
-                               } else {
-                                       foreach my $d (@index_data) {
-                                               $index->insert($field, $d, $path);
+                               # store fields in lookup
+                               if ($il && $display) {
+                                       if (lc($x->{type}) eq "lookup_key") {
+                                               if ($lookup_key) {
+                                                       print STDERR "WARNING: try to redefine lookup_key (keys shouldn't be repeatable fields!)";
+                                               } else {
+                                                       $lookup_key = $display;
+                                               }
+                                       } elsif (lc($x->{type}) eq "lookup_val") {
+                                               if ($lookup_key) {
+                                                       $lhash{$lookup_key} = $display;
+                                               } else {
+                                                       print STDERR "WARNING: no lookup_key defined for  '$display'?";
+                                               }
                                        }
                                }
                        }
+
+                       # fill data in index
+                       foreach my $d (@index_data) {
+                               $index->insert($field, $d, $path);
+                       }
                }
 
                # now try to parse variables from configuration file
@@ -346,6 +398,22 @@ foreach my $database ($cfg->Sections) {
        my $type = lc($cfg -> val($database, 'type')) || die "$database doesn't have 'type' defined";
        my $add_xml = $cfg -> val($database, 'xml');    # optional
 
+       # create new lookup file
+       my $lookup_file = $cfg -> val($database, 'lookup_newfile'); # optional
+       if ($lookup_file) {
+               #tie %lhash, 'GDBM_File', $lookup_file, &GDBM_NEWDB, 0644;
+               tie %lhash, 'TDB_File', $lookup_file, TDB_CLEAR_IF_FIRST, O_RDWR, 0644;
+               print STDERR "creating lookup file '$lookup_file'\n";
+       }
+
+       # open existing lookup file
+       $lookup_file = $cfg -> val($database, 'lookup_open'); # optional
+       if ($lookup_file) {
+               #tie %lhash, 'GDBM_File', $lookup_file, &GDBM_READER, 0644;
+               tie %lhash, 'TDB_File', $lookup_file, TDB_DEFAULT, O_RDWR, 0644;
+               print STDERR "opening lookup file '$lookup_file'\n";
+       }
+
 print STDERR "reading ./import_xml/$type.xml\n";
 
        # extract just type basic
@@ -596,6 +664,8 @@ print STDERR "using: $type...\n";
                        fakeprogress($i);
 
                }
+               # close lookup
+               untie %lhash if (%lhash);
        }
 }