Squashed initial commit

2024-09-10 13:47:29 -04:00
commit 8ebb6ad265
6221 changed files with 2512206 additions and 0 deletions
--- a/qhtcp-workflow/apps/java/weka-clustering/src/ExecMain.java
+++ b/qhtcp-workflow/apps/java/weka-clustering/src/ExecMain.java
@@ -0,0 +1,102 @@
+/*
+ * This is the main class to execute the computation and get the GOID.
+ */
+
+import java.io.File;
+import java.io.FileOutputStream;
+import java.io.OutputStreamWriter;
+import java.util.Vector;
+import weka.core.Instances;
+
+/**
+ *
+ * @author DTian
+ */
+public class ExecMain {
+
+	public static void main(String[] args) {
+        
+// SGD2AttrTable sgd = new SGD2AttrTable();
+// Get clusters
+
+    try {
+			// Check the input arguments
+			checkParameters(args.length);
+      GetClusters myGetClusters = new GetClusters();
+	
+			// To get the input file type
+			Instances oriData = myGetClusters.input(args[0]);
+
+			// Get the output files name
+			String outputWholeCluster = args[0] + "-WholeTree.txt";
+			String outputFinalTable = args[0] + "-finalTable.csv";
+			String outputSummary = args[0] + "-summary.csv";
+      boolean fromFunction = true?false:args[4].trim().compareToIgnoreCase("true") == 0;
+
+			// Create the root cluster name
+			int round = 1; //for tree level
+			int position = 0; // for node position in same level
+			String rootName = (round-1)+"-"+position+"-"+0;
+
+			// System.out.println("root cluster is:" + rootName);
+			myGetClusters.printRootName(outputWholeCluster,rootName);
+
+			// Create a vector for fianl table
+			Vector vecFinalTable = new Vector();
+			Vector vecSummary = new Vector();
+
+			//get the variable name
+			vecFinalTable.addElement(myGetClusters.printTableHead(oriData));
+
+			// Create the root node
+			TreeNode root = new TreeNode(rootName,0.0,oriData,null);
+			OutputStreamWriter xmlWriter = new OutputStreamWriter(
+					new FileOutputStream(new File("tree.xml"), true));
+			xmlWriter.write(" <tree>\n <declarations> \n <attributeDecl name=\"name\" type=\"String\"/>\n </declarations>\n");
+			xmlWriter.write(" <branch>\n  <attribute name=\"name\" value=\""+"root" +"\"/>\n");
+			xmlWriter.flush();
+
+			// Recursive clustering the data
+			myGetClusters.clustering(
+				root,
+				round,
+				""+position,
+				vecFinalTable,
+				vecSummary,
+				outputWholeCluster,
+        xmlWriter,
+				args[1],
+				args[2],
+				args[3], 
+				fromFunction
+			);
+
+			xmlWriter.write(" </branch>\n");
+			xmlWriter.write("</tree>\n");
+			xmlWriter.close();
+
+			// Output final result
+			myGetClusters.printVector(vecFinalTable,outputFinalTable);
+			myGetClusters.printVector(vecSummary,outputSummary);
+		} catch (Exception e) {
+			e.printStackTrace();
+			System.exit(1);
+		}
+  }
+
+	/**
+	* check the number of the arguments:
+	* java GetCluster arg1 arg2 ...
+	*
+	* @param length  the length of the arguments
+	* in this program, length should be 1
+	*/
+	private static void checkParameters(int length) {
+		if(length != 5) {
+			// there are 5 parameters, 1,file for clustering; 2, lookup table file;
+			// 3, backgroud file; 4, repeat counts(an integer) 5, GoIDfromFunction (boolean)
+			System.out.println("Usage: java  ExecMain clusterFileName GoMatrixFilename backGroundFilename repeatTime true|false");
+			System.exit(1);
+		}
+	}
+}