Changeset 6792

Timestamp:

2004-02-12T15:56:37+13:00 (20 years ago)

Author:

jrm21

Message:

tidied up code quite a bit, remove tmp files after use.

File:

• trunk/gsdl/perllib/Kea.pm (modified) (3 diffs)

trunk/gsdl/perllib/Kea.pm

@@ -1 +1 @@
 package Kea;
+
+use strict;
 
 # This function is called by BasPlug.pm when a flag in a collection
@@ -21 +23 @@
     my @optionlist = split(/ +/, $args) if (defined($args)); #list of options
     my $suffix = 'kea'; #default file will be called .kea
-    my $command = ""; 
+    my @kea_options;
     my @keylist; 
     my @stemlist;
 
-    print STDERR "optionlist: @optionlist\n";
-       
-    foreach $element (@optionlist){ #for each option
+
+    foreach my $element (@optionlist){ #for each option
     my ($option, $file) = split(/,/, $element); #split option letter and file (if file exist)
 
     $option  = "-".$option; #place dash in front of option
-    $file = "" if(!defined($file)); #no file options specified
-    $suffix = $file if($option eq '-E'); #if option is extension (suffix) option
-    $command .= " $option $file "; #add to list of commands
+    push @kea_options, $option;
+    if (defined($file)) {
+        push @kea_options, $file;
+    }
+
+    if ($option eq '-E')  # option is extension (suffix) option
+        { $suffix = $file }
     }
 
-    print STDERR "Using output suffix: $suffix\n"; 
+    # print STDERR "Using output suffix: $suffix\n"; 
 
     # remove all HTML tags
-    $doc =~ s/<P[^>]*>/\n/sgi;
-    $doc =~ s/<H[^>]*>/\n/sgi;
-    $doc =~ s/<[^>]*>//sgi;
-    $doc =~ tr/\n/\n/s;
+    $doc =~ s/<[ph][^>]*>/\n/sgi; # replace headings/paragraphs with newline
+    $doc =~ s/<[^>]*>/ /sgi; # replace all others with a space
 
-    #write text to a file eg doc.txt 
-    open(OUT, ">$gsdlhome/tmp/doc.txt") or die "In Kea.pm doc.txt could not be created\n";  
+    # &gt; lt amp
+    $doc =~ s/\&(?:gt|lt|amp)\;/ /gi;
+
+    my $tmpfile="$gsdlhome/tmp/doc.txt";
+    open(OUT, ">$tmpfile") or die "Kea.pm could not create doc.txt: $!\n";  
     print OUT $doc;
     close(OUT);
 
-    #call Kea with specifed options
-    `$gsdlhome/perllib/Kea-1.1.4/Kea $command $gsdlhome/tmp/doc.txt`; 
+    # call Kea with specifed options
+    system("$gsdlhome/perllib/Kea-1.1.4/Kea", @kea_options,
+       $tmpfile); 
 
-    #read doc.kea with keywords
-    open(IN, "<$gsdlhome/tmp/doc.$suffix") or return @emptykeylist;
-                                          #this means doc.kea does not exist
-                                          #either because an option was wrongly specified
-                                      #or no keyphrases were found
+    unlink($tmpfile); # don't need this file anymore
+
+    # read doc.kea with keywords
+    my $inputfile="$gsdlhome/tmp/doc.$suffix";
+
+    # If this file doesn't exist, then either an option was wrongly specified
+    # or no keyphrases were found
+    open(IN, "<$inputfile") or return ();
+
     while(<IN>){
     chomp;
-    @key = split(/\t/); #split into array separated by a tab
+    my @key = split(/\t/); #split into array separated by a tab
     push(@keylist, $key[0]); #add to list of keywords
     push(@stemlist, $key[1]); #add to list of stems  
@@ -66 +77 @@
 
     #put data into appropriate format 
-    $keylist = join(", ", @keylist);
-    $stemlist = join(", ", @stemlist);  
+    my $keylistref = join(", ", @keylist);
+    my $stemlistref = join(", ", @stemlist);    
  
-    #delete doc.extension so that in future it will not be opened and read
-    `rm $gsdlhome/tmp/doc.$suffix`;
+    # delete doc.extension so that in future it will not be opened and read
+    unlink($inputfile);
 
-    #return keywords + stems to basplug
-    my @keystemlist = ($keylist, $stemlist);
-    return @keystemlist;
-
+    # return keywords + stems to basplug
+    return ($keylistref, $stemlistref);
 }