Added input data viewing, started xml doc construction

svn path=/; revision=103

diff --git a/preecej/semantic_wiki/PS_TransformForImport.pl b/preecej/semantic_wiki/PS_TransformForImport.pl
index d58cc7da4ad1881270e62454119adb8648e530e3..9667f2b7332fdc93752fceb3a55462c6a9dbdd7a 100644 (file)
--- a/preecej/semantic_wiki/PS_TransformForImport.pl
+++ b/preecej/semantic_wiki/PS_TransformForImport.pl
@@ -36,7 +36,7 @@ data.
 
     [Provenance]
     Source Date Time Stamp=Apr 2 2008
-    Source Data=Oryzabase
+    Source Database=Oryzabase
     Source Version=rel. 10
     Source URI=http://www.shigen.nig.ac.jp/rice/oryzabase/
     Source File=http://www.shigen.nig.ac.jp/rice/oryzabase/genes/...
@@ -97,11 +97,15 @@ my %provenance;
 my $template_name;
 my @field_names;
 my @field_data;
+my $xml;
+my $output_data;
 
 # ---------------------------------------------------------------------------
 # functions
 # ---------------------------------------------------------------------------
 
+
+# ---------------------------------------------------------------------------
 sub init
 {
     # read and set options
@@ -120,29 +124,30 @@ sub init
     
     $file_del = ($file_type eq "csv") ? ',' : '\t';
     
-    if ($verbose) {
-        system "clear";
-        print "\n"
-            . "------------------------------------------------------------\n"
-            . "------- Plant Semantics Import Transformation Script -------\n"
-            . "------------------------------------------------------------\n"
-            . "\n\n"
-            . "Input File: $input_file\n"
-            . "Designated input file type: $file_type\n"
-            . "Output File: $output_file\n"
-            . "Running in verbose mode? " . ($verbose ? "Yes" : "No") . "\n"
-            . "Running in debug mode? " . ($verbose ? "Yes" : "No") . "\n"
-            . "\n"
-            . "------------------------------------------------------------\n"
-            . "------------------------------------------------------------\n"
-            . "------------------------------------------------------------\n"
-            . "\n";
-    }
+    system "clear";
+    print "\n"
+        . "------------------------------------------------------------\n"
+        . "------- Plant Semantics Import Transformation Script -------\n"
+        . "------------------------------------------------------------\n"
+        . "\n"
+        . "Input File: $input_file\n"
+        . "Designated input file type: $file_type\n"
+        . "Output File: $output_file\n"
+        . "Running in verbose mode? " . ($verbose ? "Yes" : "No") . "\n"
+        . "Running in debug mode? " . ($verbose ? "Yes" : "No") . "\n"
+        . "\n"
+        . "------------------------------------------------------------\n"
+        . "------------------------------------------------------------\n"
+        . "------------------------------------------------------------\n"
+        . "\n";
 }
 
+
+# read, parse, and store provenance, template, and annotations
+# ---------------------------------------------------------------------------
 sub import_data
 {
-    if ($verbose) { print "Opening file and reading header info...\n\n"; }
+    print "Opening file and reading header info...\n\n";
 
     # open file
     open(INPUT_FILE,$input_file) or die("Could not open input file.");
@@ -157,7 +162,7 @@ sub import_data
        my $data_val = (split('=',$line))[1];
        switch ($count) {
            case 2 { $provenance{'SourceDateTimeStamp'} = $data_val; }
-           case 3 { $provenance{'SourceDate'} = $data_val; }
+           case 3 { $provenance{'SourceDatabase'} = $data_val; }
            case 4 { $provenance{'SourceVersion'} = $data_val; }
            case 5 { $provenance{'SourceURI'} = $data_val; }
            case 6 { $provenance{'SourceFile'} = $data_val; }
@@ -168,7 +173,7 @@ sub import_data
        if ($count == 10) { last; }
     }
 
-    if ($verbose) { print "Reading data...\n\n"; }
+    print "Reading data...\n\n";
     
     # loop through data rows and add all data fields to an array of hashes
     while (<INPUT_FILE>)
@@ -185,9 +190,9 @@ sub import_data
             if ($debug) { print "$i: $field_names[$i]: $tmp_data_ary[$i]\n"; }           
             $tmp_hash{$field_names[$i]} = $tmp_data_ary[$i];
         }
-        push @field_data, %tmp_hash;
-        %tmp_hash = (); # clear your hash as a precaution
+        push @field_data, \%tmp_hash;
     }
+    close INPUT_FILE;
 }
 
 
@@ -200,28 +205,61 @@ sub show_input
         print "$key: $provenance{$key}\n";
     }
     print "\n";
-    print "Template: $template_name\n";
-    print "Fields: " . join(', ',@field_names) . "\n";
+    print "[Template]\n$template_name\n\n";
+    print "[Fields]\n" . join(', ',@field_names) . "\n\n";
     
     print "[Data]\n";
     foreach my $row (@field_data) {
         foreach my $key (keys %$row) {
-            print "$key: (%$row){$key}\n";
+            print "$key => " . $row->{$key} . "\n";
         }
+        print "\n";
     }
+    print "\n";
 }
 
 
+# loop through the hash and build annotation data and provenance xml doc
 # ---------------------------------------------------------------------------
 sub write_xml
 {
-    # loop through the hash and build annotation data and provenance xml docs
-    # simultaneously
+    print "Transforming data...\n\n";
+    my $current_node; # placeholder for node cursor
+    
+    $xml = new XML::Smart($output_data, 'XML::Smart::Parser');
+
+    # set root element, provenance page and elements
+    $xml->{Pages}{Page} = {Title => 'PROV:3'};
+
+    $current_node = $xml->{Pages}{Page};
+    $current_node->{Template} = {Name => 'Provenance'};
+
+    $current_node = $current_node->{Template};    
+    $current_node->{Field} = {Name => 'Source Data Time Stamp'};
+    $current_node = $current_node->{Field};
+    $current_node->content(0,$provenance{'SourceDateTimeStamp'});
+    
+    # iterate through the data
     
-    # write xml docs to a single ImportXML file
+        # build the annotation page
+
+        # add a template to the annotation provenance page
+        
+    # write out xml doc to a single ImportXML file
+    $xml->save($output_file);
 }
 
 
+# print the transformed data (as xml)
+# ---------------------------------------------------------------------------
+sub show_output
+{
+    print "[XML]\n";
+    $output_data = $xml->data;
+    print $output_data;
+    print "\n";
+}
+
 # ---------------------------------------------------------------------------
 # main
 # ---------------------------------------------------------------------------
@@ -230,6 +268,7 @@ init;
 import_data;
 if ($verbose) { show_input; }
 write_xml;
+if ($verbose) { show_output; }
 
 exit;