Changeset 2355

Timestamp:

2001-05-03T13:59:37+12:00 (23 years ago)

Author:

sjboddie

Message:

All options to import.pl and buildcol.pl may now be specified from
within a collect.cfg file

Location:

trunk/gsdl

Files:

• bin/script/buildcol.pl (modified) (3 diffs)
• bin/script/import.pl (modified) (4 diffs)
• perllib/colcfg.pm (modified) (1 diff)

trunk/gsdl/bin/script/buildcol.pl

@@ -85 +85 @@
     $create_images, $collectdir, $out, $buildtype, $textindex,
     $no_strip_html, $no_text);
+
+    # note that no defaults are passed for most options as they're set
+    # later (after we check the collect.cfg file)
     if (!parsargv::parse(\@ARGV, 
-             'verbosity/\d+/2', \$verbosity,
+             'verbosity/\d+/', \$verbosity,
              'archivedir/.*/', \$archivedir,
              'cachedir/.*/', \$cachedir,
              'builddir/.*/', \$builddir,
-             'maxdocs/^\-?\d+/-1', \$maxdocs,
+             'maxdocs/^\-?\d+/', \$maxdocs,
              'debug', \$debug,
-             'mode/^(all|compress_text|build_index|infodb)$/all', \$mode,
+             'mode/^(all|compress_text|build_index|infodb)$/', \$mode,
              'index/.*/', \$indexname,
              'no_text', \$no_text,
@@ -122 +125 @@
 
     # read the configuration file
-
     $configfilename = &util::filename_cat ($ENV{'GSDLCOLLECTDIR'}, "etc", "collect.cfg");
     if (-e $configfilename) {
     $collectcfg = &colcfg::read_collect_cfg ($configfilename);
 
+    if ($verbosity !~ /\d+/) {
+        if (defined $collectcfg->{'verbosity'} && $collectcfg->{'verbosity'} =~ /\d+/) {
+        $verbosity = $collectcfg->{'verbosity'};
+        } else {
+        $verbosity = 2; # the default
+        }
+    }
     if (defined $collectcfg->{'buildtype'} && $buildtype eq "") {
         $buildtype = $collectcfg->{'buildtype'};
@@ -142 +151 @@
         $builddir = $collectcfg->{'builddir'};
     }
-    if (defined $collectcfg->{'collectdir'} && $collectdir eq "") {
-        $collectdir = $collectcfg->{'collectdir'};
+    if ($maxdocs !~ /\-?\d+/) {
+        if (defined $collectcfg->{'maxdocs'} && $collectcfg->{'maxdocs'} =~ /\-?\d+/) {
+        $maxdocs = $collectcfg->{'maxdocs'};
+        } else {
+        $maxdocs = -1; # the default
+        }
+    }
+    if (defined $collectcfg->{'debug'} && $collectcfg->{'debug'} =~ /^true$/i) {
+        $debug = 1;
+    }
+    if ($mode !~ /^(all|compress_text|build_index|infodb)$/) {
+        if (defined $collectcfg->{'mode'} && $collectcfg->{'mode'} =~ /^(all|compress_text|build_index|infodb)$/) {
+        $mode = $collectcfg->{'mode'};
+        } else {
+        $mode = "all"; # the default
+        }
+    }
+    if (defined $collectcfg->{'index'} && $indexname eq "") {
+        $indexname = $collectcfg->{'index'};
     }
     if (defined $collectcfg->{'no_text'} && $no_text == 0) {
-        if ($collectcfg->{'no_text'} =~ /^true$/) {
+        if ($collectcfg->{'no_text'} =~ /^true$/i) {
         $no_text = 1;
         }
     }
     if (defined $collectcfg->{'allclassifications'} && $allclassifications == 0) {
-        if ($collectcfg->{'allclassifications'} =~ /^true$/) {
+        if ($collectcfg->{'allclassifications'} =~ /^true$/i) {
         $allclassifications = 1;
         }
+    }
+    if (defined $collectcfg->{'keepold'} && $collectcfg->{'keepold'} =~ /^true$/i) {
+        $keepold = 1;
+    }
+    if (defined $collectcfg->{'create_images'} && $collectcfg->{'create_images'} =~ /^true$/i) {
+        $create_images = 1;
     }
     if ($buildtype eq "mgpp" && defined $collectcfg->{'textcompress'}) {

trunk/gsdl/bin/script/import.pl

@@ -92 +92 @@
     $pluginfo, $sortmeta, $archive_info_filename, 
     $archive_info, $processor, $out, $collectdir);
-    if (!parsargv::parse(\@ARGV, 
-             'verbosity/\d+/2', \$verbosity,
+
+    # note that no defaults are passed for most options as they're set
+    # later (after we check the collect.cfg file)
+    if (!parsargv::parse(\@ARGV,
+             'verbosity/\d+/', \$verbosity,
              'importdir/.*/', \$importdir,
-             'archivedir/.*/', \$archivedir,
-             'keepold', \$keepold,
-             'removeold', \$removeold,
-             'gzip', \$gzip,
-             'groupsize/\d+/1', \$groupsize,
-             'OIDtype/^(hash|incremental)$/hash', \$OIDtype,
-             'sortmeta/.*/', \$sortmeta,
+             'archivedir/.*/', \$archivedir, 
+             'keepold', \$keepold, 
+             'removeold', \$removeold, 
+             'gzip', \$gzip, 
+             'groupsize/\d+/', \$groupsize, 
+             'OIDtype/^(hash|incremental)$/', \$OIDtype,
+             'sortmeta/.*/', \$sortmeta, 
              'debug', \$debug,
-             'maxdocs/^\-?\d+/-1', \$maxdocs,
+             'maxdocs/^\-?\d+/', \$maxdocs,
              'collectdir/.*/', \$collectdir,
              'out/.*/STDERR', \$out)) {
@@ -142 +145 @@
 
 
-    # get the list of plugins for this collection
+    # get the list of plugins for this collection and set any options that
+    # were specified in the collect.cfg (all import.pl options except
+    # -collectdir and -out may be specified in the collect.cfg (these
+    # options must be known before we read the collect.cfg))
     my $plugins = [];
     $configfilename = &util::filename_cat ($ENV{'GSDLCOLLECTDIR'}, "etc", "collect.cfg");
@@ -150 +156 @@
         $plugins = $collectcfg->{'plugin'};
     }
+
+    if ($verbosity !~ /\d+/) {
+        if (defined $collectcfg->{'verbosity'} && $collectcfg->{'verbosity'} =~ /\d+/) {
+        $verbosity = $collectcfg->{'verbosity'};
+        } else {
+        $verbosity = 2; # the default
+        }
+    }
     if (defined $collectcfg->{'importdir'} && $importdir eq "") {
         $importdir = $collectcfg->{'importdir'};
@@ -164 +178 @@
         }
     }
+    if (defined $collectcfg->{'keepold'}) {
+        if ($collectcfg->{'keepold'} =~ /^false$/i && !$keepold) {
+        $removeold = 1;
+        }
+    }
+    if (defined $collectcfg->{'gzip'} && !$gzip) {
+        if ($collectcfg->{'gzip'} =~ /^true$/i) {
+        $gzip = 1;
+        }
+    }
+    if ($maxdocs !~ /\-?\d+/) {
+        if (defined $collectcfg->{'maxdocs'} && $collectcfg->{'maxdocs'} =~ /\-?\d+/) {
+        $maxdocs = $collectcfg->{'maxdocs'};
+        } else {
+        $maxdocs = -1; # the default
+        }
+    }
+    if ($groupsize !~ /\d+/) {
+        if (defined $collectcfg->{'groupsize'} && $collectcfg->{'groupsize'} =~ /\d+/ && $groupsize !~ /\d+/) {
+        $groupsize = $collectcfg->{'groupsize'};
+        } else {
+        $groupsize = 1; # the default
+        }
+    }
+    if ($OIDtype !~ /^(hash|incremental)$/) {
+        if (defined $collectcfg->{'OIDtype'} && $collectcfg->{'OIDtype'} =~ /^(hash|incremental)$/) {
+        $OIDtype = $collectcfg->{'OIDtype'};
+        } else {
+        $OIDtype = "hash"; # the default
+        }
+    }
+    if (defined $collectcfg->{'sortmeta'} && $sortmeta eq "") {
+        $sortmeta = $collectcfg->{'sortmeta'};
+    }
+    if (defined $collectcfg->{'debug'} && $collectcfg->{'debug'} =~ /^true$/i) {
+        $debug = 1;
+    }
+
     } else {
     die "Couldn't find the configuration file $configfilename\n";

trunk/gsdl/perllib/colcfg.pm

@@ -67 +67 @@
                     q/^(creator|public|beta|defaultindex|importdir|/ .
                     q/archivedir|cachedir|builddir|removeold|/ .
-                    q/textcompress|buildtype|collectdir|no_text|allclassifications)$/,
+                    q/textcompress|buildtype|no_text|keepold|gzip/ .
+                    q/verbosity|allclassifications|OIDtype|maxdocs|/ .
+                    q/groupsize|sortmeta|debug|mode|create_images)$/,
                     q/(maintainer|languages|indexsubcollections|/ .
                        q/indexes|dontbuild|dontgdbm|mirror|phind|levels)$/,