Changeset 11070

Timestamp:

2006-01-19T16:48:58+13:00 (18 years ago)

Author:

mdewsnip

Message:

A much tidier Kea.pm that now also works on Windows.

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 @@
 
 # returns a string containing comma-separated keyphrases
-sub extract_KeyPhrases {
+sub extract_KeyPhrases
+{
+    my $kea_version = shift(@_);
+    my $doc = shift(@_);  # Document's text  
+    my $args = shift(@_);  # Options
 
-    # Parsing arguments of the function
-    my $kea_version = shift(@_);
-    my $doc = shift(@_); # documents text  
-    my $args = shift(@_); # any options
-    my @optionlist = split(/\s+/, $args) if (defined($args)); #list of options
-
-    # Specifying directory names
-    my $keahome = &get_Kea_directory($kea_version);
-    my $defaultmodel = &util::filename_cat($keahome, "CSTR-20");
+    # Set default models
+    my $kea_home = &get_Kea_directory($kea_version);
+    my $default_model_path = &util::filename_cat($kea_home, "CSTR-20");
     if ($kea_version eq "4.0") {
-    # Use a different model for Kea 4.0
-    $defaultmodel = &util::filename_cat($keahome, "FAO-20docs");
+    # Use a different default model for Kea 4.0
+    $default_model_path = &util::filename_cat($kea_home, "FAO-20docs");
     }
 
-    # Initializing variables:
-    my $command = ""; 
-    my @keylist; 
-    my @options = ();
-    $modelspec = 0;
-
-    # Settings for the java executable: 
-
-    # CLASSPATH:
-    $java_classpath = ".:$keahome";
-
-    # See if java executable is on path
-    my $java_exec="";
-    if (system("which java >/dev/null 2>/dev/null")==0) {
-    $java_exec=`which java`;
-    chomp $java_exec;
-    } else {
-    $java_exec="$java_home/bin/java";
-    }
-    
-    # The actual java command is based on these other variables:
-    $java_command = "$java_exec -classpath \"$java_classpath\"";
-    
-    # end of java settings
-
-    # Parsing options for keyphrase extraction:
-    if (@optionlist) {
-    foreach $element (@optionlist){ #for each option
-        if (length($element) == 1) {
-        push(@options, "-$element");
-        } else {
-        $option = substr($element, 0, 1);
-        $value = substr($element,1);
-        if (($option eq "m") && (-e "$keahome/$value")) {
-            $modelspec = 1;
-            push(@options, "-$option $keahome/$value");
-        } elsif ($option eq "m") {          
-            $modelspec = 1;
-            print STDERR "Couldn't find model $value. Using the default model instead\n";
-            push(@options, "-$option $defaultmodel");
-        } else {
-            push(@options, "-$option $value");
+    # Parse the Kea options
+    my $options_string;
+    my @args_list = split(/\s+/, $args) if (defined($args));
+    if (@args_list) {
+    my $model_specified = 0;
+    foreach my $arg (@args_list) {
+        if (length($arg) == 1) {
+        $options_string .= " -$arg";
+        }
+        else {
+        my $option = substr($arg, 0, 1);
+        my $value = substr($arg, 1);
+        if ($option eq "m") {
+            my $model_path = &util::filename_cat($kea_home, $value);
+            if (-e $model_path) {
+            $options_string .= " -m $model_path";
+            }
+            else {
+            print STDERR "Warning: Couldn't find model $model_path; using the default model instead.\n";
+            $options_string .= " -m $default_model_path";
+            }
+            $model_specified = 1;
         }
-        
+        else {
+            $options_string .= " -$option $value";
+        }
         }
     }
-    # if none of the option specifies the model, set the default one:
-    if ($modelspec != 1) {
-        push(@options, "-m $defaultmodel"); 
+
+    # If none of the option specifies the model, use the default one
+    if ($model_specified != 1) {
+        $options_string .= " -m $default_model_path";
     }
-    $options = join(" ",@options);
-    # print STDERR "OPTIONS: $options\n";
-    } else {
-    # If no options were specified: Set default value for the model
-    $options = "-m $defaultmodel";  
+    }
+    else {
+    # If no options were specified, use the default model
+    $options_string = "-m $default_model_path";
     }
 
@@ -100 +82 @@
     $doc =~ tr/\n/\n/s;
 
-    # Write text to a temporary file doc.txt 
-    my $gsdlhome = $ENV{'GSDLHOME'}; 
-    open(OUT, ">$gsdlhome/tmp/doc.txt") or die "In Kea.pm doc.txt could not be created\n";  
-    print OUT $doc;
-    close(OUT);
+    # Write text to a temporary file doc.txt
+    my $tmp_directory_path = &util::filename_cat($ENV{'GSDLHOME'}, "tmp");
+    my $doc_txt_file_path = &util::filename_cat($tmp_directory_path, "doc.txt");
+    open(DOC_TXT, ">$doc_txt_file_path") or die "Error: Could not write $doc_txt_file_path in Kea.pm.\n";  
+    print DOC_TXT $doc;
+    close(DOC_TXT);
 
+    # Run Kea with the specified options
+    system("java -classpath \"$kea_home\" KEAKeyphraseExtractor -l $tmp_directory_path $options_string");
 
-    # EXECUTE KEA with specific options:
-    $command = "$java_command KEAKeyphraseExtractor -l $gsdlhome/tmp $options";
-    system ("$command");
+    # Read the resulting doc.key file which contains the keyphrases
+    my $doc_key_file_path = &util::filename_cat($tmp_directory_path, "doc.key");
+    if (!open(IN, "<$doc_key_file_path")) {
+    # The doc.key file does not exist (either an option was wrongly specified, or no keyphrases were found)
+    return "";
+    }
 
-    # Read the resulting doc.key, which contains keyphrases:
-
-    open(IN, "<$gsdlhome/tmp/doc.key") or return "";
-                                          #this means doc.key does not exist
-                                          #either because an option was wrongly specified
-                                      #or no keyphrases were found
-    while(<IN>){
+    my @keyphrase_list = ();
+    while (<IN>) {
     chomp;
-    push(@keylist,$_);
+    push(@keyphrase_list, $_);
     }
     close(IN);
 
-    $keylist = join(", ", @keylist);
+    # Delete doc.key so that in future it will not be opened and read (otherwise KEA sees it as more keyphrases!)
+    unlink($doc_key_file_path);
 
-    # Delete doc.key so that in future it will not be opened and read.
-    # Otherwise KEA sees it as more keyphrases!
-
-    unlink("$gsdlhome/tmp/doc.key");
-    
-    return $keylist;
+    my $keyphrases = join(", ", @keyphrase_list);
+    return $keyphrases;
 }