Changeset 28121

Timestamp:

2013-08-22T18:41:08+12:00 (11 years ago)

Author:

ak19

Message:

Better error checking after system commands in mgpbuilder too. Could not use bitwise flags for this, since mg needs to generate all the indexes, stemming, casefolding and casefolding and stemming, even when one of them (such as casefolding) is not turned on. And if ever one of them doesn't work, the original comment seemed to indicate that then none of it will work.

Location:

main/trunk/greenstone2/perllib

Files:

• mgbuilder.pm (modified) (5 diffs)
• mgppbuilder.pm (modified) (2 diffs)

main/trunk/greenstone2/perllib/mgbuilder.pm

@@ -205 +205 @@
         die "mgbuilder::compress_text - couldn't run $mg_compression_dict_exe\n";
     }
-    system ("mg_compression_dict$exe -f \"$fulltextprefix\" -S -H -2 -k 5120 $osextra");
+    my $comp_dict_status = system ("mg_compression_dict$exe -f \"$fulltextprefix\" -S -H -2 -k 5120 $osextra");
+    if($comp_dict_status != 0) {
+        print $outhandle "\nmgbuilder::compress_text - Warning: there's no compressed text\n";
+        $self->{'notbuilt'}->{'compressedtext'} = 1;
+        print STDERR "<Warning name='NoCompressedText'/>\n</Stage>\n" if $self->{'gli'};
+        return;
+    }
 
     # -b $maxdocsize sets the maximum document size to be 12 meg
@@ -456 +462 @@
     my $hash_status = system ($hash_cmd);
     print $outhandle "\nstatus from running hash_cmd: $hash_status\n" if ($self->{'verbosity'} >= 4);
+    # check that perf hash was generated - if not, don't carry on
+    if ($hash_status !=0) {
+        print $outhandle "mgbuilder::build_index - Couldn't create index $index as there are too few words in the index.\n";
+        print STDERR "<Warning name='NoIndex'/>\n</Stage>\n" if $self->{'gli'};
+        $self->{'notbuilt'}->{$index}=1;
+        return;
+        
+    }
 
     if (!-e "$mg_passes_exe" || 
@@ -493 +507 @@
     my $weights_cmd = "mg_weights_build$exe -f \"$fullindexprefix\" -t \"$fulltextprefix\" $osextra";
     print $outhandle "\ncmd: $weights_cmd\n" if ($self->{'verbosity'} >= 4);
-    system ($weights_cmd);
+    my $weights_status = system ($weights_cmd);
+    # check that it worked - if not, don't carry on
+    if ($weights_status !=0) {
+        print $outhandle "mgbuilder::build_index - No Index: couldn't create weights file, error calling mg_weights_build.\n";
+        print STDERR "<Warning name='NoIndex'/>\n</Stage>\n" if $self->{'gli'};
+        $self->{'notbuilt'}->{$index}=1;
+        return;
+        
+    }
 
     # create 'on-disk' stemmed dictionary
@@ -502 +524 @@
         die "mgbuilder::build_index - couldn't run $mg_invf_dict_exe\n";
     }
-    system ("mg_invf_dict$exe -f \"$fullindexprefix\" $osextra");
-
+    my $invdict_status = system ("mg_invf_dict$exe -f \"$fullindexprefix\" $osextra");
+    # check that it worked - if not, don't carry on
+    if ($invdict_status !=0) {
+        print $outhandle "mgbuilder::build_index - No Index: couldn't create on-disk stemmed dictionary, error calling mg_invf_dict.\n";
+        print STDERR "<Warning name='NoIndex'/>\n</Stage>\n" if $self->{'gli'};
+        $self->{'notbuilt'}->{$index}=1;
+        return;
+        
+    }
 
     # creates stem index files for the various stemming methods
@@ -515 +544 @@
     # so we generate them whatever, but don't advertise the fact
     #if ($self->{'casefold'}) {
-    system ("mg_stem_idx$exe -b 4096 -s1 -f \"$fullindexprefix\" $osextra");
+#   system ("mg_stem_idx$exe -b 4096 -s1 -f \"$fullindexprefix\" $osextra");
     #}
     #if ($self->{'stem'}) {
-    system ("mg_stem_idx$exe -b 4096 -s2 -f \"$fullindexprefix\" $osextra");
+#   system ("mg_stem_idx$exe -b 4096 -s2 -f \"$fullindexprefix\" $osextra");
     #}
     #if ($self->{'casefold'} && $self->{'stem'}) {
-    system ("mg_stem_idx$exe -b 4096 -s3 -f \"$fullindexprefix\" $osextra");
+#   system ("mg_stem_idx$exe -b 4096 -s3 -f \"$fullindexprefix\" $osextra");
     #}
+
+    # same as above: generate all the stem idexes. But don't bother stemming if
+    # casefolding failed, and don't try generating indexes for both if stemming failed
+    my $stem_index_status = system ("mg_stem_idx$exe -b 4096 -s1 -f \"$fullindexprefix\" $osextra");
+    if($stem_index_status != 0) {
+        print $outhandle "\nCase folding failed: mg_stem_idx exit status $stem_index_status\n" if ($self->{'verbosity'} >= 4);
+    } else {
+        $stem_index_status = system ("mg_stem_idx$exe -b 4096 -s2 -f \"$fullindexprefix\" $osextra");
+
+        if($stem_index_status != 0) {
+        print $outhandle "\nStemming failed: mg_stem_idx exit status $stem_index_status\n" if ($self->{'verbosity'} >= 4);
+        } else {
+        $stem_index_status = system ("mg_stem_idx$exe -b 4096 -s3 -f \"$fullindexprefix\" $osextra");
+
+        if($stem_index_status != 0) {
+            print $outhandle "\nCasefolding and stemming failed: mg_stem_idx exit status $stem_index_status\n" if ($self->{'verbosity'} >= 4);
+        }
+        }
+    }
 
     # remove unwanted files

main/trunk/greenstone2/perllib/mgppbuilder.pm

@@ -573 +573 @@
     # check that it worked - if not, don't carry on
     if ($weights_status !=0) {
-        print $outhandle "mgppbuilder::build_index - No Index: couldn't create weights file, error calling mggp_weights_build.\n";
+        print $outhandle "mgppbuilder::build_index - No Index: couldn't create weights file, error calling mgpp_weights_build.\n";
         print STDERR "<Warning name='NoIndex'/>\n</Stage>\n" if $self->{'gli'};
         $self->{'notbuilt'}->{$index}=1;
@@ -589 +589 @@
     # check that it worked - if not, don't carry on
     if ($invdict_status !=0) {
-        print $outhandle "mgppbuilder::build_index - No Index: couldn't create on-disk stemmed dictionary, error calling mggp_invf_dict.\n";
+        print $outhandle "mgppbuilder::build_index - No Index: couldn't create on-disk stemmed dictionary, error calling mgpp_invf_dict.\n";
         print STDERR "<Warning name='NoIndex'/>\n</Stage>\n" if $self->{'gli'};
         $self->{'notbuilt'}->{$index}=1;