Changeset 4192

Timestamp:

2003-04-18T14:23:57+12:00 (21 years ago)

Author:

sjboddie

Message:

Added maxnumeric build time configuration option. This is useful for collections containing
numeric values longer than 4 digits which need to be searchable.

Location:

trunk/gsdl/perllib

Files:

• colcfg.pm (modified) (6 diffs)
• mgbuilder.pm (modified) (7 diffs)

trunk/gsdl/perllib/colcfg.pm

@@ -43 +43 @@
 # {'textcompress'}->string
 # {'buildtype'}->string
+# {'maxnumeric'}->string
 
 # {'maintainer'}->array of strings
@@ -72 +73 @@
                     q/textcompress|buildtype|no_text|keepold|gzip/ .
                     q/verbosity|allclassifications|OIDtype|maxdocs|/ .
-                    q/groupsize|sortmeta|debug|mode|create_images)$/,
+                    q/groupsize|sortmeta|debug|mode|create_images|/ .
+                    q/maxnumeric)$/,
                     q/(maintainer|languages|indexsubcollections|/ .
                        q/indexes|dontbuild|dontgdbm|mirror|phind|levels)$/,
@@ -86 +88 @@
                  q/^(creator|public|defaultindex|importdir|/ .
                  q/archivedir|cachedir|builddir|removeold|/ .
-                 q/textcompress|buildtype|collectdir|no_text|allclassifications)$/,
+                 q/textcompress|buildtype|collectdir|no_text|/ .
+                 q/allclassifications|maxnumeric)$/,
                  q/^(maintainer|languages|indexsubcollections|/ .
                  q/indexes|dontbuild|dontgdbm|levels)$/,
@@ -104 +107 @@
 # {'numwords'}->string
 # {'numbytes'}->string
+# {'maxnumeric'}->string
 # {'indexfields'}->array of strings
 # {'indexfieldmap'}->array of strings in the form "field->FI"
@@ -112 +116 @@
 
     return &cfgread::read_cfg_file ($filename, 
-                    q/^(builddate|buildtype|numdocs|numwords|numbytes)$/, 
+                    q/^(builddate|buildtype|numdocs|numwords|numbytes|maxnumeric)$/, 
                     q/^(metadata|languages|indexfields|indexfieldmap|indexmap)$/);
 }
@@ -120 +124 @@
 
     &cfgread::write_cfg_file($filename, $data,
-           q/^(builddate|buildtype|numdocs|numwords|numbytes)$/, 
+           q/^(builddate|buildtype|numdocs|numwords|numbytes|maxnumeric)$/, 
            q/^(metadata|languages|indexfieldsindexfieldmap|indexmap)$/);
 }

trunk/gsdl/perllib/mgbuilder.pm

@@ -202 +202 @@
     my $outhandle = $self->{'outhandle'};
 
+    my $maxnumeric = 4;
+    if (defined($self->{'collect_cfg'}->{'maxnumeric'}) &&
+    $self->{'collect_cfg'}->{'maxnumeric'} =~ /^\d+$/) {
+    $maxnumeric = $self->{'collect_cfg'}->{'maxnumeric'};
+    }
+
     &util::mk_all_dir (&util::filename_cat($self->{'build_dir'}, "text"));
     my $basefilename = "text/$self->{'collection'}";
@@ -224 +230 @@
     } else {
     if (!-e "$mg_passes_exe" || 
-        !open (PIPEOUT, "| mg_passes$exe -f \"$fulltextprefix\" -b $maxdocsize -T1 $osextra")) {
+        !open (PIPEOUT, "| mg_passes$exe -f \"$fulltextprefix\" -b $maxdocsize -T1 -M $maxnumeric $osextra")) {
         die "mgbuilder::compress_text - couldn't run $mg_passes_exe\n";
     }
@@ -264 +270 @@
     # -b $maxdocsize sets the maximum document size to be 12 meg
     if (!-e "$mg_passes_exe" || 
-        !open ($handle, "| mg_passes$exe -f \"$fulltextprefix\" -b $maxdocsize -T2 $osextra")) {
+        !open ($handle, "| mg_passes$exe -f \"$fulltextprefix\" -b $maxdocsize -T2 -M $maxnumeric $osextra")) {
         die "mgbuilder::compress_text - couldn't run $mg_passes_exe\n";
     }
@@ -475 +481 @@
     &util::filename_cat ($exedir, "mg_stem_idx$exe");
 
+    my $maxnumeric = 4;
+    if (defined($self->{'collect_cfg'}->{'maxnumeric'}) &&
+    $self->{'collect_cfg'}->{'maxnumeric'} =~ /^\d+$/) {
+    $maxnumeric = $self->{'collect_cfg'}->{'maxnumeric'};
+    }
+
     my $osextra = "";
     if ($ENV{'GSDLOS'} =~ /^windows$/i) {
@@ -538 +550 @@
     if (!-e "$mg_passes_exe" || 
         !open (PIPEOUT, "| mg_passes$exe -f \"$fullindexprefix\" -b $maxdocsize " .
-           "-$index_level -m 32 -s 0 -G -t 10 -N1 $osextra")) {
+           "-$index_level -m 32 -s 0 -G -t 10 -N1 -M $maxnumeric $osextra")) {
         die "mgbuilder::build_index - couldn't run $mg_passes_exe\n";
     }
@@ -567 +579 @@
     if (!-e "$mg_passes_exe" || 
         !open ($handle, "| mg_passes$exe -f \"$fullindexprefix\" -b $maxdocsize " .
-           "-$index_level -c 3 -G -t 10 -N2 $osextra")) {
+           "-$index_level -c 3 -G -t 10 -N2 -M $maxnumeric $osextra")) {
         die "mgbuilder::build_index - couldn't run $mg_passes_exe\n";
     }
@@ -822 +834 @@
     $build_cfg->{'notbuilt'} = $self->{'notbuilt'} if scalar @{$self->{'notbuilt'}};
 
+    $build_cfg->{'maxnumeric'} = 4;
+    if (defined($self->{'collect_cfg'}->{'maxnumeric'}) &&
+    $self->{'collect_cfg'}->{'maxnumeric'} =~ /^\d+$/) {
+    $build_cfg->{'maxnumeric'} = $self->{'collect_cfg'}->{'maxnumeric'};
+    }
+
     # write out the build information
     &cfgread::write_cfg_file("$self->{'build_dir'}/build.cfg", $build_cfg,
-                 '^(builddate|numdocs|numbytes|numwords|numsections)$', 
+                 '^(builddate|numdocs|numbytes|numwords|numsections|maxnumeric)$', 
                              '^(indexmap|subcollectionmap|languagemap|notbuilt)$');