renamed "old" index to swish, and introduced index which is -- index;

implemented using PostgreSQL for now.

git-svn-id: file:///home/dpavlin/private/svn/webpac/trunk@9 13eb9ef6-21d5-0310-b721-a9d68796d827

diff --git a/WebPac.pm b/WebPac.pm
index a80f991..843519c 100644 (file)
--- a/WebPac.pm
+++ b/WebPac.pm
@@ -8,6 +8,7 @@ use HTML::FillInForm;
 use SWISH;
 use Unicode::String qw(utf8 utf16);
 require Unicode::Map8;
+use DBI;
 
 # configuration options
 # FIX: they really should go in configuration file!
@@ -25,7 +26,8 @@ sub setup {
        $self->run_modes(
                'search' => 'show_search_form',
                'results' => 'show_results_list',
-#              'user' => 'show_user_detail'
+#              'user' => 'show_user_detail',
+               'index' => 'show_index',
        );
        $self->start_mode('search');
        $self->mode_param('rm');
@@ -56,13 +58,13 @@ sub show_results_list {
        my @swish_results;      # results from swish
 
        # load template for this page
-       my $tmpl = $self->load_tmpl('results.html');
 
        my @s_arr;      # all queries are located here
 
        for(my $i = 1; $i <=10; $i++) {
 
-               last if (! $q->param("f$i"));
+               return show_index($self, $i) if ($q->param("f".$i."_index"));
+               next if (! $q->param("f$i"));
                next if (! $q->param("v$i"));
 
                # re-write query from +/- to and/and not
@@ -84,9 +86,11 @@ sub show_results_list {
                        }
                }
 
-               push @s_arr,$q->param("f$i")."=($s)";
+               push @s_arr,$q->param("f$i")."_swish=($s)";
        }
 
+       my $tmpl = $self->load_tmpl('results.html');
+
        my $l2_map = Unicode::Map8->new($CHARSET) || die;
        my $us = Unicode::String->new();
 
@@ -161,4 +165,18 @@ sub show_results_list {
        return $html;
 }
  
+sub show_index {
+       my $self = shift;
+       my $i = shift;          # field number
+
+       my $q = $self->query();
+
+       my $html;
+
+       $html .= "show index of ".$q->param("f$i")." for ".$q->param("v$i");
+
+
+       return $html;
+}
+
 1;

diff --git a/all2xml.pl b/all2xml.pl
index 94e1a47..5bbc307 100755 (executable)
--- a/all2xml.pl
+++ b/all2xml.pl
@@ -7,8 +7,15 @@ use Data::Dumper;
 use XML::Simple;
 use Text::Unaccent 1.02;       # 1.01 won't compile on my platform,
 require Unicode::Map8;
+use DBI;
 
 my $config=XMLin(undef, forcearray => [ 'isis' ], forcecontent => 1);
+my $dbh = DBI->connect("DBI:Pg:dbname=webpac","","") || die $DBI::errstr; # FIX
+# FIX; select relname from pg_class where relname like 'index_%' ;
+$dbh->begin_work || die $dbh->errstr();
+
+$dbh->do("delete from index_author") || die $dbh->errstr();
+$dbh->do("delete from index_title") || die $dbh->errstr();
 
 my %opts;
 
@@ -57,19 +64,23 @@ sub isis2xml {
        foreach my $field (keys %{$config->{indexer}}) {
 
                my $display_data = "";
+               my $swish_data = "";
                my $index_data = "";
 
                foreach my $x (@{$config->{indexer}->{$field}->{isis}}) {
 
                        my $display_tmp = "";
+                       my $swish_tmp = "";
                        my $index_tmp = "";
 
                        my $format = $x->{content};
-                       my $i = 1;      # index only
+                       my $s = 1;      # swish only
                        my $d = 1;      # display only
-                       $i = 0 if (lc($x->{type}) eq "display");
-                       $d = 0 if (lc($x->{type}) eq "index");
-#print "## i: $i d: $d ## $format ##"; 
+                       my $i = 0;      # index only
+                       $s = 0 if (lc($x->{type}) eq "display");
+                       $d = 0 if (lc($x->{type}) eq "swish");
+                       ($s,$d,$i) = (0,0,1) if (lc($x->{type}) eq "index");
+#print STDERR "## s: $s d: $d i: $i ## $format ##\n";  
                        # parse format
                        my $prefix = "";
                        if ($format =~ s/^([^\d]+)//) {
@@ -81,8 +92,9 @@ sub isis2xml {
                                        if ($isis_tmp) {
 #                                              $display_tmp .= $prefix . "/$1/$2/".$isis_tmp if ($d);
                                                $display_tmp .= $prefix . $isis_tmp if ($d);
-                                               $index_tmp .= $isis_tmp." " if ($i);
-#print " $isis_tmp <--\n";
+                                               $swish_tmp .= $isis_tmp." " if ($s);
+                                               $index_tmp .= $prefix . $isis_tmp if ($i);
+#print STDERR " $isis_tmp <--\n";
                                        }
                                        $prefix = "";
                                } elsif ($format =~ s/^([^\d]+)//) {
@@ -94,19 +106,35 @@ sub isis2xml {
                        }
                        # add suffix
                        $display_tmp .= $prefix if ($display_tmp);
+                       $index_tmp .= $prefix if ($index_tmp);
 
 #                      $display_data .= $display_tmp if ($display_tmp ne "");
-#                      $index_data .= $index_tmp if ($index_tmp ne "");
+#                      $swish_data .= $swish_tmp if ($swish_tmp ne "");
                        $display_data .= $display_tmp;
+                       $swish_data .= $swish_tmp;
                        $index_data .= $index_tmp;
 
                }
-#print "--display:$display_data\n--index:$index_data\n";
+#print "--display:$display_data\n--swish:$swish_data\n";
                #$xml->{$field."_display"} = $isis_map->tou($display_data)->utf8 if ($display_data);
-               #$xml->{$field."_index"} = unac_string($config->{isis_codepage},$index_data) if ($index_data);
+               #$xml->{$field."_swish"} = unac_string($config->{isis_codepage},$swish_data) if ($swish_data);
                $xml->{$field."_display" } = [ $isis_map->tou($display_data)->utf8 ] if ($display_data);
-               $xml->{$field."_index"} = [ unac_string($config->{isis_codepage},$index_data)." jabuka" ] if ($index_data);
-       
+               $xml->{$field."_swish"} = [ unac_string($config->{isis_codepage},$swish_data) ] if ($swish_data);
+
+               # index
+               if ($index_data && $index_data ne "") {
+                       my $sql = "select $field from index_$field where upper($field)=upper(?)";
+                       my $sth = $dbh->prepare($sql) || die $dbh->errstr();
+                       $sth->execute($index_data) || die "SQL: $sql; ".$dbh->errstr();
+#print STDERR "--->$index_data<---\n";
+                       if (! $sth->fetchrow_hashref) {
+                               my $sql = "insert into index_$field values (?)";
+                               my $sth = $dbh->prepare($sql) || die $dbh->errstr();
+#print STDERR "$sql: $index_data<!----\n";
+                               $sth->execute($index_data) || die "SQL: $sql; ".$dbh->errstr();
+                       }
+               }
+
        }
        if ($xml) {
                return XMLout($xml, rootname => 'xml', keeproot => 0, noattr => 0 );
@@ -189,6 +217,7 @@ foreach my $isis_db (@isis_dbs) {
        print STDERR "\n";
 }
 
+$dbh->commit || die $dbh->errstr();
 
 1;
 __END__

diff --git a/isis2xml.xml b/isis2xml.xml
index 4751319..c0cc837 100644 (file)
--- a/isis2xml.xml
+++ b/isis2xml.xml
@@ -9,7 +9,7 @@
 >
 
 <!--
-       <isis type="index|display">_pre_000x_sep_000x_sep_000x_post_</isis>
+       <isis type="swish|display|index">_pre_000x_sep_000x_sep_000x_post_</isis>
 -->
     <indexer>
        <headline>
@@ -17,7 +17,7 @@
        </headline>
 
        <author>
-               <isis>700a 700b</isis>
+               <isis>700a, 700b</isis>
                <isis>701a 701b</isis>
                <isis>701c 701d</isis>
                <isis>710a : 710b , 710c</isis>
@@ -25,26 +25,26 @@
                <isis>711a : 711b , 711c</isis>
                <isis>(711d ; 711f ; 711e)</isis>
                <isis>503a, 503b</isis>
-               <isis type="index">702</isis>
-               <isis type="index">200f 200g</isis>
-               <isis type="display">205</isis>
+               <isis type="swish">702</isis>
+               <isis type="swish">200f 200g</isis>
+               <isis type="index">700a, 700b</isis>
        </author>
        <title>
-               <isis>200a ; $200k = $200d : 200e / 200f ; 200g. 200c / 200x ; 200y</isis>
-               <isis type="index">532</isis>
-               <isis type="index">424</isis>
-               <isis type="index">230a 230e</isis>
-               <isis type="index">231a 231e</isis>
-               <isis type="index">232a 232e</isis>
-               <isis type="index">233a 233e</isis>
+               <isis>200a ; 200k = 200d : 200e / 200f ; 200g. 200c / 200x ; 200y</isis>
+               <isis type="swish">532</isis>
+               <isis type="swish">424</isis>
+               <isis type="swish">230a 230e</isis>
+               <isis type="swish">231a 231e</isis>
+               <isis type="swish">232a 232e</isis>
+               <isis type="swish">233a 233e</isis>
                <isis type="display">&lt;br&gt;230v : 230a. - 250. - 260&lt;br&gt;ISBN 290</isis>
                <isis type="display">&lt;br&gt;231v : 231a. - 251. - 261&lt;br&gt;ISBN 291</isis>
                <isis type="display">&lt;br&gt;232v : 232a. - 252. - 262&lt;br&gt;ISBN 292</isis>
                <isis type="display">&lt;br&gt;233v : 233a. - 253. - 263&lt;br&gt;ISBN 293</isis>
-               <isis type="index">270</isis>
-               <isis type="index">271</isis>
-               <isis type="index">272</isis>
-               <isis type="index">273</isis>
+               <isis type="swish">270</isis>
+               <isis type="swish">271</isis>
+               <isis type="swish">272</isis>
+               <isis type="swish">273</isis>
        </title>
                
     </indexer>

diff --git a/swish_isis.conf b/swish_isis.conf
index 11c3cb5..e4de493 100644 (file)
--- a/swish_isis.conf
+++ b/swish_isis.conf
@@ -25,6 +25,7 @@ MetaNames xml
 #MetaNameAlias swishtitle title
 PropertyNameAlias swishtitle headline_display
 #PropertyNameAlias swishdefault title_display
+MetaNameAlias xml xml_swish
 
 #MetaNamesRank -4 publisher
 #MetaNamesRank 0 title.index

diff --git a/template_html/search.html b/template_html/search.html
index e8fd65e..436340e 100644 (file)
--- a/template_html/search.html
+++ b/template_html/search.html
@@ -47,7 +47,8 @@
         <td class="inputfield" width="280" > \r
          <input type="hidden" name="f2" value="author">\r
           <input size="30" type="text" name="v2" value="">\r
-          index</td>\r
+         <input type=submit value="index" name="f2_index" class="navigacija">\r
+          </td>\r
         <td class="example" >npr. </td>\r
       </tr>\r
       <tr> \r
@@ -55,7 +56,8 @@
         <td class="inputfield" width="280"> \r
          <input type="hidden" name="f3" value="title">\r
           <input size="30" type="text" name="v3">\r
-          index </td>\r
+         <input type=submit value="index" name="f3_index" class="navigacija">\r
+          </td>\r
         <td class="example">npr.</td>\r
       </tr>\r
       <tr> \r
@@ -71,7 +73,8 @@
         <td class="inputfield" width="280"> \r
          <input type="hidden" name="f4" value="publisher">\r
           <input  size="30" type="text" name="v4">\r
-          index </td>\r
+         <input type=submit value="index" name="f4_index" class="navigacija">\r
+          </td>\r
         <td class="example">npr.</td>\r
       </tr>\r
       <tr> \r
@@ -79,7 +82,8 @@
         <td  class="inputfield" width="280"> \r
          <input type="hidden" name="f5" value="isn">\r
           <input  size="15" type="text" name="v5">\r
-          index </td>\r
+         <input type=submit value="index" name="f5_index" class="navigacija">\r
+          </td>\r
         <td class="example">npr.</td>\r
       </tr>\r
       <tr> \r
@@ -89,7 +93,7 @@
             <option value="610">Tema (podruèje)</option>\r
             <option value="225">Nakl. cjelina</option>\r
             <option value="300+">Napomene </option>\r
-            <option value="330">Sadrµaj </option>\r
+            <option value="330">Sadr¾aj </option>\r
             <option value="464">Analitika</option>\r
             <option value="675">UDK </option>\r
             <option value="686">CC </option>\r
@@ -102,7 +106,8 @@
         </td>\r
         <td  valign="bottom" width="280"> \r
           <input  size="30" name="v9">\r
-          index </td>\r
+         <input type=submit value="index" name="f9_index" class="navigacija">\r
+          </td>\r
         <td>&nbsp;</td>\r
       </tr>\r
       <tr> \r

diff --git a/webpac.sql b/webpac.sql
new file mode 100644 (file)
index 0000000..0cad954
--- /dev/null
+++ b/webpac.sql
@@ -0,0 +1,89 @@
+--
+-- Selected TOC Entries:
+--
+\connect - dpavlin
+
+DROP INDEX "ind_index_author2";
+DROP INDEX "ind_index_author";
+DROP INDEX "ind_index_title2";
+DROP INDEX "ind_index_title";
+DROP TABLE "index_author";
+DROP PROCEDURAL LANGUAGE 'plpgsql';
+\connect - postgres
+
+DROP FUNCTION "plpgsql_call_handler" ();
+\connect - dpavlin
+
+DROP TABLE "index_title";
+--
+-- TOC Entry ID 2 (OID 831643)
+--
+-- Name: index_title Type: TABLE Owner: dpavlin
+--
+
+CREATE TABLE "index_title" (
+       "title" character varying(255)
+);
+
+\connect - postgres
+
+--
+-- TOC Entry ID 8 (OID 834826)
+--
+-- Name: "plpgsql_call_handler" () Type: FUNCTION Owner: postgres
+--
+
+CREATE FUNCTION "plpgsql_call_handler" () RETURNS opaque AS '/usr/lib/postgresql/lib/plpgsql.so', 'plpgsql_call_handler' LANGUAGE 'C';
+
+--
+-- TOC Entry ID 9 (OID 834827)
+--
+-- Name: plpgsql Type: PROCEDURAL LANGUAGE Owner: 
+--
+
+CREATE TRUSTED PROCEDURAL LANGUAGE 'plpgsql' HANDLER "plpgsql_call_handler" LANCOMPILER 'PL/pgSQL';
+
+\connect - dpavlin
+
+--
+-- TOC Entry ID 3 (OID 834828)
+--
+-- Name: index_author Type: TABLE Owner: dpavlin
+--
+
+CREATE TABLE "index_author" (
+       "author" character varying(255)
+);
+
+--
+-- TOC Entry ID 4 (OID 831645)
+--
+-- Name: "ind_index_title" Type: INDEX Owner: dpavlin
+--
+
+CREATE INDEX ind_index_title ON index_title USING btree (title);
+
+--
+-- TOC Entry ID 5 (OID 831646)
+--
+-- Name: "ind_index_title2" Type: INDEX Owner: dpavlin
+--
+
+CREATE INDEX ind_index_title2 ON index_title USING btree (upper(title));
+
+--
+-- TOC Entry ID 6 (OID 834830)
+--
+-- Name: "ind_index_author" Type: INDEX Owner: dpavlin
+--
+
+CREATE UNIQUE INDEX ind_index_author ON index_author USING btree (author);
+
+--
+-- TOC Entry ID 7 (OID 834831)
+--
+-- Name: "ind_index_author2" Type: INDEX Owner: dpavlin
+--
+
+CREATE UNIQUE INDEX ind_index_author2 ON index_author USING btree (upper(author));
+