Changeset 2785

Timestamp:

2001-10-09T17:21:29+13:00 (23 years ago)

Author:

sjboddie

Message:

The build process now creates a summary of how many files were included,
which were rejected, etc. A link to a page containing this summary is
provided from the final page of the collector (once the collection is built
successfully) and from the default "about this collection" text for
collections built by the collector.

Also did a little bit of tidying in a couple of places

Location:

trunk/gsdl

Files:

• bin/script/build (modified) (4 diffs)
• bin/script/buildcol.pl (modified) (6 diffs)
• bin/script/classinfo.pl (modified) (1 diff)
• bin/script/import.pl (modified) (7 diffs)
• bin/script/mirror.pl (modified) (1 diff)
• bin/script/nightly.pl (modified) (1 diff)
• bin/script/pluginfo.pl (modified) (1 diff)
• bin/script/update.pl (modified) (1 diff)
• etc/main.cfg (modified) (1 diff)
• macros/bsummary.dm (added)
• macros/collect.dm (modified) (1 diff)
• macros/english.dm (modified) (1 diff)
• macros/english2.dm (modified) (2 diffs)
• perllib/acronym.pm (modified) (1 diff)
• perllib/mgbuilder.pm (modified) (3 diffs)
• perllib/plugin.pm (modified) (3 diffs)
• perllib/plugins/BasPlug.pm (modified) (6 diffs)
• perllib/plugins/ConvertToPlug.pm (modified) (5 diffs)
• perllib/plugins/PDFPlug.pm (modified) (2 diffs)
• perllib/plugins/PSPlug.pm (modified) (4 diffs)
• perllib/plugins/RTFPlug.pm (modified) (2 diffs)
• perllib/plugins/WordPlug.pm (modified) (2 diffs)
• perllib/plugins/ZIPPlug.pm (modified) (1 diff)
• src/recpt/collectoraction.cpp (modified) (4 diffs)
• src/recpt/pageaction.cpp (modified) (7 diffs)
• src/recpt/pageaction.h (modified) (1 diff)

trunk/gsdl/bin/script/build

@@ -132 +132 @@
 &main();
 
-close OUT if $use_out;
+if ($use_out) {
+    close OUT;
+
+    # if we've created a build log we'll copy it to the collection's etc
+    # directory
+    my ($etcdir);
+    if ($dontinstall) {
+    $etcdir = &util::filename_cat($collectdir, "etc", "build.log");
+    } else {
+    $etcdir = &util::filename_cat($ENV{'GSDLHOME'}, "collect", $collection, "etc", "build.log");
+    }
+    
+    &util::cp($outfile, $etcdir);
+}
 
 sub print_usage {
@@ -168 +181 @@
     print STDOUT "   -out                    Filename or handle to print output status to.\n";
     print STDOUT "                           The default is STDERR\n";
+    print STDOUT "   -statsfile name         Filename or handle to print import statistics to.\n";
+    print STDOUT "                           The default is STDERR\n";
     print STDOUT "   -make_writable          If set build will make the collection and any\n";
     print STDOUT "                           temporary files it created globally writable after\n";
@@ -373 +388 @@
     $import_cmd .= " -removeold" unless $append;
     $import_cmd .= " -collectdir \"$collectdir\"" if $collectdir =~ /\w/;
+    $import_cmd .= " -statsfile \"$statsfile\"" if $statsfile =~ /\w/;
     $import_cmd .= " $maxdocs $collection";
+
+    print STDERR "\n**import_cmd: $import_cmd\n";
+
     system ($import_cmd);
     # if using output directory append the import output to it
@@ -558 +577 @@
              'email_events/.*/', \$email_events,
              'mail_server/.*/', \$mail_server,
+             'statsfile/.*/STDERR', \$statsfile,
              'event_header/.*/', \$event_header)) {
     

trunk/gsdl/bin/script/buildcol.pl

@@ -76 +76 @@
     print STDOUT "   -no_strip_html        Do not strip the html tags from the indexed text\n";
     print STDOUT "                         (only used for mgpp collections).\n\n";
+    print STDOUT "   -faillog name         Fail log filename. This log receives the filenames\n";
+    print STDOUT "                         of any files which fail to be processed (defaults.\n";
+    print STDOUT "                         to " . 
+    &util::filename_cat("<collectdir>", "colname", "etc", "fail.log") . ")\n";
     print STDOUT "  [Type \"perl -S buildcol.pl | more\" if this help text scrolled off your screen]";
     print STDOUT "\n" unless $ENV{'GSDLOS'} =~ /^windows$/i;
@@ -86 +90 @@
     $debug, $mode, $indexname, $keepold, $allclassifications,
     $create_images, $collectdir, $out, $buildtype, $textindex,
-    $no_strip_html, $no_text);
+    $no_strip_html, $no_text, $faillog);
 
     # note that no defaults are passed for most options as they're set
@@ -106 +110 @@
              'out/.*/STDERR', \$out,
              'no_strip_html', \$no_strip_html,
-             'buildtype/^(mg|mgpp)$/', \$buildtype)) {
+             'buildtype/^(mg|mgpp)$/', \$buildtype,
+             'faillog/.*/', \$faillog)) {
     &print_usage();
     die "\n";
@@ -125 +130 @@
     die "\n";
     }
+
+    if ($faillog eq "") {
+    $faillog = &util::filename_cat($ENV{'GSDLCOLLECTDIR'}, "etc", "fail.log");
+    }
+    # note that we're appending to the faillog here (import.pl clears it each time)
+    # this could potentially create a situation where the faillog keeps being added 
+    # to over multiple builds (if the import process is being skipped)
+    open (FAILLOG, ">>$faillog") || die "Couldn't open fail log $faillog\n";
+    $faillog = 'buildcol::FAILLOG';
+    $faillog->autoflush(1);
 
     # read the configuration file
@@ -268 +283 @@
      "\$realarchivedir, \$realbuilddir, \$verbosity, " .
      "\$maxdocs, \$debug, \$keepold, \$allclassifications, " .
-     "\$out, \$no_text)");
+     "\$out, \$no_text, \$faillog)");
     die "$@" if $@;
 
@@ -301 +316 @@
 
     close OUT if $close_out;
+    close FAILLOG;
 }
 

trunk/gsdl/bin/script/classinfo.pl

@@ -35 +35 @@
 }
 
-use strict;
 use classify;
 use util;

trunk/gsdl/bin/script/import.pl

@@ -81 +81 @@
     &util::filename_cat ($ENV{'GSDLHOME'}, "collect") . ")\n";
     print STDOUT "   -out name              Filename or handle to print output status to.\n";
-    print STDOUT "   -faillog name          Filename to log import failures to.\n";
-    print STDOUT "                          The default is <collectdir>/colname/etc/fail.log\n\n";
+    print STDOUT "                          The default is STDERR\n";
+    print STDOUT "   -statsfile name        Filename or handle to print import statistics to.\n";
+    print STDOUT "                          The default is STDERR\n";
+    print STDOUT "   -faillog name          Fail log filename. This log receives the filenames\n";
+    print STDOUT "                          of any files which fail to be processed (defaults.\n";
+    print STDOUT "                          to " . 
+    &util::filename_cat("<collectdir>", "colname", "etc", "fail.log") . ")\n";
     print STDOUT "  [Type \"perl -S import.pl | more\" if this help text scrolled off your screen]";
     print STDOUT "\n" unless $ENV{'GSDLOS'} =~ /^windows$/i;
@@ -93 +98 @@
     $removeold, $gzip, $groupsize, $OIDtype, $debug, 
     $maxdocs, $collection, $configfilename, $collectcfg, 
-    $pluginfo, $sortmeta, $archive_info_filename, 
-    $archive_info, $processor, $out, $faillogname, $collectdir);
+    $pluginfo, $sortmeta, $archive_info_filename, $statsfile,
+    $archive_info, $processor, $out, $faillog, $collectdir);
 
     # note that no defaults are passed for most options as they're set
@@ -112 +117 @@
              'collectdir/.*/', \$collectdir,
              'out/.*/STDERR', \$out,
-             'faillog/.*/', \$faillogname)) {
+             'statsfile/.*/STDERR', \$statsfile,
+             'faillog/.*/', \$faillog)) {
     &print_usage();
     die "\n";
@@ -133 +139 @@
     die "\n";
     }
-    
-    # check and/or set fail log file
-    if ($faillogname eq "") { 
-    $faillogname="$ENV{GSDLCOLLECTDIR}/etc/fail.log";
-    } else {
-    my $can_open=1;
-    open (TESTOPEN, ">$faillogname") || ($can_open=0);
-    close (TESTOPEN);
-    if ($can_open==0) {
-        warn "fail.log - cannot write to \"$faillogname\", using default\n \"$ENV{GSDLCOLLECTDIR}/etc/fail.log\" instead.\n";
-        $faillogname="$ENV{GSDLCOLLECTDIR}/etc/fail.log";
-    }
-    }
-    # test that default is writable...
-    if ($faillogname eq "$ENV{GSDLCOLLECTDIR}/etc/fail.log") {
-    my $can_open=1;
-    open (TESTOPEN, ">$faillogname") || ($can_open=0);
-    close (TESTOPEN);
-    if ($can_open==0) {
-        warn "warning - cannot write to \"$faillogname\".\n";
-        $faillogname="";
-    }
-    }
-
+
+    if ($faillog eq "") {
+    $faillog = &util::filename_cat($ENV{'GSDLCOLLECTDIR'}, "etc", "fail.log");
+    }
+    open (FAILLOG, ">$faillog") || die "Couldn't open fail log $faillog\n";
+    $faillog = 'import::FAILLOG';
+    $faillog->autoflush(1);
 
     # check sortmeta
@@ -256 +245 @@
 
     # load all the plugins
-    $pluginfo = &plugin::load_plugins ($plugins, $verbosity, $out, $faillogname);
+    $pluginfo = &plugin::load_plugins ($plugins, $verbosity, $out, $faillog);
     if (scalar(@$pluginfo) == 0) {
     print $out "No plugins were loaded.\n";
@@ -290 +279 @@
     
     &plugin::end($pluginfo, $processor);
-    
+
     # write out the archive information file
     if (!$debug) {
@@ -296 +285 @@
     $archive_info->save_info($archive_info_filename);
     }
+    
+    # write out import stats
+    my $close_stats = 0;
+    if ($statsfile !~ /^(STDERR|STDOUT)$/i) {
+    if (open (STATS, ">$statsfile")) {
+        $statsfile = 'import::STATS';
+        $close_stats = 1;
+    } else {
+        print $out "WARNING: couldn't open stats file $statsfile\n";
+        print $out "         will print stats to STDERR instead\n";
+        $statsfile = 'STDERR';
+    }
+    }
+
+    print $out "\n";
+    print $out "*********************************************\n";
+    print $out "Import Complete\n";
+    print $out "*********************************************\n";
+
+    &plugin::write_stats($pluginfo, $statsfile);
+    if ($close_stats) {
+    close STATS;
+    }
+
     close OUT if $close_out;
+    close FAILLOG;
 }
+

trunk/gsdl/bin/script/mirror.pl

@@ -48 +48 @@
 }
 
-use strict;
 use arcinfo;
 use colcfg;

trunk/gsdl/bin/script/nightly.pl

@@ -41 +41 @@
 }
 
-use strict;
 use arcinfo;
 use colcfg;

trunk/gsdl/bin/script/pluginfo.pl

@@ -36 +36 @@
 }
 
-use strict;
 use plugin;
 use util;

trunk/gsdl/bin/script/update.pl

@@ -46 +46 @@
 }
 
-use strict;
 use arcinfo;
 use colcfg;

trunk/gsdl/etc/main.cfg

@@ -74 +74 @@
                 spanish.dm dutch.dm port.dm russian.dm russian2.dm \
                 italian.dm italian2.dm home.dm collect.dm docs.dm \
-                hebrew.dm
-
+                hebrew.dm bsummary.dm
 
 # Define the interface languages and encodings supported by this receptionist

trunk/gsdl/macros/collect.dm

@@ -715 +715 @@
 <p>_textbildsuc_
 
+<p>_textviewbildsummary_
+
 <p><center>
 _collectorbar_

trunk/gsdl/macros/english.dm

@@ -62 +62 @@
 _collectorextra_ {
 <p>This collection contains _numdocs_ _If_("_numdocs_" eq "1",document,documents), a total of _numbytes_.
+<p><a href="_httppagex_(bsummary)">Click here</a> to view the build summary for this collection.
 }
 

trunk/gsdl/macros/english2.dm

@@ -470 +470 @@
 _textseconds_ {seconds}
 _textbildsuc_ {Collection built successfully.}
+
+_textviewbildsummary_ {
+You may <a href="_httppagex_(bsummary)&c=_cgiargbc1dirname_" target=_top>view the build
+summary</a> of this collection for further details.
+}
 
 _textfailmsg11_ {
@@ -917 +922 @@
 _textmus_ {main.cfg updated successfully}
 
+
+######################################################################
+# 'bsummary' pages
+package bsummary
+######################################################################
+
+
+#------------------------------------------------------------
+# text macros
+#------------------------------------------------------------
+
+_textbsummary_ {Build summary for _collectionname_ collection}
+_textflog_ {Fail log for _collectionname_ collection}
+
+

trunk/gsdl/perllib/acronym.pm

@@ -26 +26 @@
 ###########################################################################
 
-use strict;
 use diagnostics;
 

trunk/gsdl/perllib/mgbuilder.pm

@@ -62 +62 @@
     my ($class, $collection, $source_dir, $build_dir, $verbosity, 
     $maxdocs, $debug, $keepold, $allclassifications, 
-    $outhandle, $no_text) = @_;
+    $outhandle, $no_text, $failhandle) = @_;
 
     $outhandle = STDERR unless defined $outhandle;
     $no_text = 0 unless defined $no_text;
+    $failhandle = STDERR unless defined $failhandle;
 
     # create an mgbuilder object
@@ -78 +79 @@
               'outhandle'=>$outhandle,
               'no_text'=>$no_text,
+              'failhandle'=>$failhandle,
               'notbuilt'=>[]    # indexes not built
               }, $class;
@@ -134 +136 @@
     
     # load all the plugins
-    $self->{'pluginfo'} = &plugin::load_plugins ($plugins, $verbosity, $outhandle);
+    $self->{'pluginfo'} = &plugin::load_plugins ($plugins, $verbosity, $outhandle, $failhandle);
     if (scalar(@{$self->{'pluginfo'}}) == 0) {
     print $outhandle "No plugins were loaded.\n";

trunk/gsdl/perllib/plugin.pm

@@ -28 +28 @@
 require util;
 
+my $stats = {'num_processed' => 0,
+         'num_blocked' => 0,
+         'num_not_processed' => 0,
+         'num_archives' => 0
+         };
+
 sub load_plugins {
     my ($plugin_list) = shift @_;
-    ($verbosity, $outhandle, $faillogname) = @_; # globals
+    ($verbosity, $outhandle, $failhandle) = @_; # globals
     my @plugin_objects = ();
 
     $verbosity = 2 unless defined $verbosity;
     $outhandle = STDERR unless defined $outhandle;
+    $failhandle = STDERR unless defined $failhandle;
 
     foreach $pluginoptions (@$plugin_list) {
@@ -58 +65 @@
     
     # initialize plugin
-    $plugobj->init($verbosity, $outhandle, $faillogname);
+    $plugobj->init($verbosity, $outhandle, $failhandle);
 
     # add this object to the list
@@ -94 +101 @@
     return $rv if defined $rv;
     }
+
+    if ($verbosity >= 2) {
+    print $outhandle "WARNING - no plugin could process $file\n";
+    }
     
-    if ($verbosity >= 2) {
-    print $outhandle "WARNING - no plugin could process " . 
-        &util::filename_cat($base_dir,$file) . "\n";
+    $file =~ s/([^\\\/]+)$/$1/;
+    print $failhandle "$file: no plugin could process this file\n";
+    $stats->{'num_not_processed'} ++;
+
+    return 0;
+}
+
+# write out some general stats that the plugins have compiled - note that
+# the buildcol.pl process doesn't currently call this process so the stats
+# are only output after import.pl -
+sub write_stats {
+    my ($pluginfo, $statshandle) = @_;
+
+    foreach $plugobj (@$pluginfo) {
+    $plugobj->compile_stats($stats);
     }
-    return 0;
+
+    my $total = $stats->{'num_processed'} + $stats->{'num_blocked'} + 
+    $stats->{'num_not_processed'};
+
+    if ($total == 1) {
+    print $statshandle "* 1 document was considered for processing\n";
+    } else {
+    print $statshandle "* $total documents were considered for processing\n";
+    }
+    if ($stats->{'num_archives'}) {
+    print $statshandle "   (including the contents of " . $stats->{'num_archives'} .
+        " ZIP/TAR archive";
+    if ($stats->{'num_archives'} == 1) {print $statshandle ")\n";}
+    else {print $statshandle "s)\n";}
+    }
+    if ($stats->{'num_processed'} == 1) {
+    print $statshandle "* 1 was processed and included in the collection\n";
+    } else {
+    print $statshandle "* " . $stats->{'num_processed'} . " were processed and included in the collection\n";
+    }
+    if ($stats->{'num_not_processed'} == 1) {
+    print $statshandle "* 1 was rejected.";
+    } else {
+    print $statshandle "* " . $stats->{'num_not_processed'} . " were rejected.";
+    } 
+    print $statshandle " See the fail log for a list of rejected documents\n";
+    
 }
 

trunk/gsdl/perllib/plugins/BasPlug.pm

@@ -142 +142 @@
 
     $self->{'textcat'} = new textcat();
+
+    $self->{'num_processed'} = 0;
+    $self->{'num_not_processed'} = 0;
+    $self->{'num_blocked'} = 0;
+    $self->{'num_archives'} = 0;
     
     # general options available to all plugins
@@ -175 +180 @@
 sub init {
     my $self = shift (@_);
-    my ($verbosity, $outhandle, $faillogname) = @_;
+    my ($verbosity, $outhandle, $failhandle) = @_;
 
     # verbosity is passed through from the processor
     $self->{'verbosity'} = $verbosity;
 
-    # as is the outhandle ...
+    # as are the outhandle and failhandle
     $self->{'outhandle'} = $outhandle if defined $outhandle;
-    $self->{'faillogname'} = $faillogname;
+    $self->{'failhandle'} = $failhandle;
 
     # set process_exp and block_exp to defaults unless they were
@@ -258 +263 @@
 
     my $filename = &util::filename_cat($base_dir, $file);
-    return 0 if $self->{'block_exp'} ne "" && $filename =~ /$self->{'block_exp'}/;
+    if ($self->{'block_exp'} ne "" && $filename =~ /$self->{'block_exp'}/) {
+    $self->{'num_blocked'} ++;
+    return 0;
+    }
     if ($filename !~ /$self->{'process_exp'}/ || !-f $filename) {
     return undef;
@@ -298 +306 @@
     if (!length ($text)) {
     print $outhandle "$plugin_name: ERROR: $file contains no text\n" if $self->{'verbosity'};
+
+    my $failhandle = $self->{'failhandle'};
+    print $failhandle "$file: " . ref($self) . ": file contains no text\n";
+    $self->{'num_not_processed'} ++;
+
     return 0;
     }
@@ -316 +329 @@
     # process the document
     $processor->process($doc_obj);
+
+    $self->{'num_processed'} ++;
 
     return 1; # processed the file
@@ -671 +686 @@
 }
 
+sub compile_stats {
+    my $self = shift(@_);
+    my ($stats) = @_;
+
+    $stats->{'num_processed'} += $self->{'num_processed'};
+    $stats->{'num_not_processed'} += $self->{'num_not_processed'};
+
+}
+
+
 1;

trunk/gsdl/perllib/plugins/ConvertToPlug.pm

@@ -48 +48 @@
 #    @ISA = ('BasPlug'); #, 'HTMLPlug', 'TEXTPlug');
 }
-
-# use strict;  # this breaks 'print $outhandle ' error msgs.
 
 sub print_usage {
@@ -146 +144 @@
     my $outhandle = $self->{'outhandle'};
     my $convert_to = $self->{'convert_to'};
+    my $failhandle = $self->{'failhandle'};
 
     # softlink to collection tmp dir
@@ -184 +183 @@
     if ($output_type eq "fail") {
     print $outhandle "Could not convert $tailname$suffix to $convert_to format\n";
-    if ($self->{'faillogname'} ne "" && -s "$errlog") {
-        open(SAVELOG, ">>$self->{'faillogname'}");
+    print $failhandle "$tailname$suffix: " . ref($self) . " failed to convert to $convert_to\n";
+    $self->{'num_not_processed'} ++;
+    if (-s "$errlog") {
         open(ERRLOG, "$errlog");
-        print SAVELOG "$tailname$suffix (converting to $convert_to) failed:\n";
         while (<ERRLOG>) {
-        print SAVELOG "$_";
+        print $outhandle "$_";
         }
+        print $outhandle "\n";
         close ERRLOG;
-        print SAVELOG "\n";
-        close SAVELOG;
     }
     &util::rm("$errlog") if (-e "$errlog");
@@ -236 +234 @@
 
     my $filename = &util::filename_cat($base_dir, $file);
-    return 0 if $self->{'block_exp'} ne "" && $filename =~ /$self->{'block_exp'}/;
+    if ($self->{'block_exp'} ne "" && $filename =~ /$self->{'block_exp'}/) {
+    $self->{'num_blocked'} ++;
+    return 0;
+    }
     if ($filename !~ /$self->{'process_exp'}/ || !-f $filename) {
         return undef;
@@ -299 +300 @@
     $self->cleanup_tmp_area();
 
+    $self->{'num_processed'} ++;
 
     return 1;

trunk/gsdl/perllib/plugins/PDFPlug.pm

@@ -63 +63 @@
 
     my $outhandle = $self->{'outhandle'};
-    print $outhandle "PDFPlug: passing $_[3] onto $self->{'convert_to'} Plug\n" 
+    print $outhandle "PDFPlug: passing $_[3] on to $self->{'convert_to'}Plug\n" 
     if $self->{'verbosity'} > 1;
     
@@ -69 +69 @@
 }
 
-
 1;
-
-
-
-

trunk/gsdl/perllib/plugins/PSPlug.pm

@@ -33 +33 @@
 }
 
-use strict;
-
-
 sub new {
     my $class = shift (@_);
@@ -65 +62 @@
     print STDERR "using the PRESCRIPT package, which is available for download at\n   http://www.nzdl.org/html/software.html\n\n";
 }
-
 
 sub get_default_block_exp {
@@ -147 +143 @@
 }
 
-
 # do plugin specific processing of doc_obj for HTML type
 sub process {
@@ -153 +148 @@
     my ($trash, $trash2, $path, $file, $trash3, $doc) = @_;
 
-    print STDERR "PSPlug: passing $_[3] onto $self->{'convert_to'} Plug\n" 
+    my $outhandle = $self->{'outhandle'};
+    print $outhandle "PSPlug: passing $_[3] on to $self->{'convert_to'}Plug\n" 
     if $self->{'verbosity'} > 1;
     

trunk/gsdl/perllib/plugins/RTFPlug.pm

@@ -33 +33 @@
 }
 
-use strict;
-
 sub get_default_process_exp {
     my $self = shift (@_);
@@ -42 +40 @@
 sub process {
     my $self = shift (@_);
-    print STDERR "RTFPlug: passing $_[3] onto $self->{'convert_to'} Plug\n" 
+
+    my $outhandle = $self->{'outhandle'};
+    print $outhandle "RTFPlug: passing $_[3] on to $self->{'convert_to'}Plug\n" 
     if $self->{'verbosity'} > 1;
+
     return ConvertToPlug::process_type($self,"rtf",@_);
 }

trunk/gsdl/perllib/plugins/WordPlug.pm

@@ -45 +45 @@
 
     my $outhandle = $self->{'outhandle'};    
-    print $outhandle "WordPlug: passing $_[3] onto $self->{'convert_to'} Plug\n" 
+    print $outhandle "WordPlug: passing $_[3] on to $self->{'convert_to'}Plug\n" 
     if $self->{'verbosity'} > 1;
     
@@ -51 +51 @@
 }
 
-
 1;
-

trunk/gsdl/perllib/plugins/ZIPPlug.pm

@@ -119 +119 @@
     my $numdocs = &plugin::read ($pluginfo, "", $tmpdir, $metadata, $processor, $maxdocs);
     &util::rm_r ($tmpdir);
+
+    $self->{'num_archives'} ++;
+
     return $numdocs;
 

trunk/gsdl/src/recpt/collectoraction.cpp

@@ -1030 +1030 @@
   // _faillog_ - set to last 6 lines of .bld file if build failed
   // _gsdlhome_ - the gsdlhome path (dm_safe)
-
   // _sourcelist_ -- "input source" text boxes
   // _badsources_ -- will be set to "1" if we've come from the 
@@ -1154 +1153 @@
     collectorbar += get_button (collector_page, "grey", "view", false);
   }
+
+  collectorbar += "</tr></table>\n";
+  disp.setmacro ("collectorbar", "collector", collectorbar);
 
   if (collector_page == "bildfail") {
@@ -1175 +1177 @@
     disp.setmacro ("faillog", "collector", dm_safe(faillog));
   }
-
-  collectorbar += "</tr></table>\n";
-  disp.setmacro ("collectorbar", "collector", collectorbar);
 
   if (collector_page == "srce" || collector_page == "existing") 
@@ -1659 +1658 @@
   }
     
+  text_t collectdir = get_collectdir (args);
+
   // set up build options
   text_t options = "-make_writable -remove_import -out \"";
   options += filename_cat (tmpdir, collection + ".bld");
-  options += "\" -collectdir \"" + get_collectdir (args) + "\"";
+  options += "\" -collectdir \"" + collectdir + "\" -statsfile \"";
+  options += filename_cat(collectdir, collection, "etc", "import.log") + "\"";
 
   if (args["bc1esrce"] == 1) {

trunk/gsdl/src/recpt/pageaction.cpp

@@ -28 +28 @@
 #include "receptionist.h"
 #include "fileutil.h"
+#include "gsdltools.h"
 #include <time.h>
 
@@ -327 +328 @@
 }
 
+void pageaction::set_macro_to_file_contents (displayclass &disp, const text_t &macroname,
+                         const text_t &packagename, const text_t &filename) {
+
+  text_t filecontent;
+  char *filenamec = filename.getcstr();
+  ifstream file_in (filenamec);
+  delete filenamec;
+  if (file_in) {
+    char c;
+    file_in.get(c);
+    while (!file_in.eof ()) {
+      if (c == '\n') filecontent += "<br>";
+      filecontent.push_back(c);
+      file_in.get(c);
+    }
+    file_in.close();
+  }
+  disp.setmacro (macroname, packagename, dm_safe(filecontent));
+}
+
 void pageaction::define_internal_macros (displayclass &disp, cgiargsclass &args, 
                      recptprotolistclass *protos, ostream &logout) {
@@ -383 +404 @@
   // if page is "docs"
   // _documentation_       links to PDF documents if they're available
+
+
+  // if page is "bsummary"
+  // _importlog_           set to contents of collections import.log file
+  // _faillog_             set to contents of collections fail.log file
 
   
@@ -418 +444 @@
   //setting _queryformcontent_ so that the query interface is consistent.
   if(arg_p == "about"){
-    text_t collection = args["c"];
-     //check that the protocol is alive
-    recptproto* colproto = protos->getrecptproto (collection, logout);
-    if(colproto == NULL) {
-      logout << "ERROR: Null collection protocol trying to query"
-         << collection.getcstr() << "\n";
-      return;
-    }
-    
-    //check the collection is responding/in place
-    ColInfoResponse_t *colinfo = recpt->get_collectinfo_ptr(colproto, collection,
-                                logout);
-    if(colinfo == NULL){
-      logout << "ERROR: Null returned for get_collectinfo_ptr on "
-         << collection.getcstr() << "in queryaction::define_query_interface\n";
-      return;
-    }
-    
-    
-    text_tmap::iterator check = colinfo->format.find("QueryInterface");
-    if(check != colinfo->format.end()){
-      if((*check).second=="DateSearch"){
-    text_t current = "_basicqueryform_ _datesearch_";
+    if (cinfo != NULL) {
+      text_tmap::iterator check = cinfo->format.find("QueryInterface");
+      if(check != cinfo->format.end()){
+    if((*check).second=="DateSearch"){
+      text_t current = "_basicqueryform_ _datesearch_";
     
-    disp.setmacro("queryformcontent","query",current);
-      }
-    }
-  }
-
-
+      disp.setmacro("queryformcontent","query",current);
+    }
+      }
+    }
+  }
 
   if (arg_p == "home") {
@@ -454 +461 @@
     if (collector_disabled) disp.setmacro ("textgocollector", "home", "");
   }
-
-
-  
   
   else if (arg_p == "preferences") {
@@ -462 +466 @@
     // _collectionoption_
 
-    if (args["ccs"] == "1" && collectproto != NULL && (cinfo->ccsCols.size() > 1)) {
+    if ((args["ccs"] == "1") && (cinfo->ccsCols.size() > 1)) {
       text_t collectionoption = "_textcollectionoption_";
       text_tarray::const_iterator col_here = cinfo->ccsCols.begin();
@@ -675 +679 @@
     set_documentation_macro (disp);
 
+  } else if (arg_p == "bsummary" && !arg_c.empty()) {
+
+    set_macro_to_file_contents (disp, "importlog", "bsummary",
+                filename_cat(gsdlhome, "collect", arg_c, "etc", "import.log"));
+    set_macro_to_file_contents (disp, "faillog", "bsummary",
+                filename_cat(gsdlhome, "collect", arg_c, "etc", "fail.log"));
   }
 }

trunk/gsdl/src/recpt/pageaction.h

@@ -75 +75 @@
   void set_documentation_macro (displayclass &disp);
 
+  void set_macro_to_file_contents (displayclass &disp, const text_t &macroname,
+                   const text_t &packagename, const text_t &filename);
+
   bool status_disabled;
   bool collector_disabled;