prerna.reactor.frame.py.HistogramReactor Maven / Gradle / Ivy

Go to download
Show more of this group Show more artifacts with this name
Show all versions of semoss Show documentation
SEMOSS
The newest version!
package prerna.reactor.frame.py;

import java.util.List;

import prerna.ds.py.PandasFrame;
import prerna.ds.py.PyTranslator;
import prerna.reactor.frame.AbstractFrameReactor;
import prerna.reactor.task.constant.ConstantTaskCreationHelper;
import prerna.sablecc2.om.GenRowStruct;
import prerna.sablecc2.om.PixelDataType;
import prerna.sablecc2.om.PixelOperationType;
import prerna.sablecc2.om.ReactorKeysEnum;
import prerna.sablecc2.om.nounmeta.NounMetadata;
import prerna.sablecc2.om.task.ITask;

public class HistogramReactor extends AbstractFrameReactor {

	
	/**
	 * This reactor gets a histogram
	 * The inputs to the reactor are: 
	 * 1) the column to base the histogram on
	 * 2) the number of breaks
	 * 3) the panel id - defaults to zero if nothing is entered
	 */
	
	public HistogramReactor() {
		this.keysToGet = new String[]{ReactorKeysEnum.COLUMN.getKey(), ReactorKeysEnum.BREAKS.getKey(), ReactorKeysEnum.PANEL.getKey()};
	}
	
	@Override
	public NounMetadata execute() {
		// get frame
		PandasFrame frame = (PandasFrame) getFrame();

		//get frame name
		String table = frame.getName();
		
		//get inputs
		String column = getColumn();
		//clean column name
		if (column.contains("__")) {
			column = column.split("__")[1];
		}
		//get number of breaks as an integer
		int numBreaks = getNumBreaks();
		
		//need to retrieve panel id to use in the task options
		String panelId = getPanelId();
		
		//build the py script to execute
		return getHistogram(frame, table, column, panelId, numBreaks);
	}
	
	//////////////////////////////////////////////////////////////////////
	//////////////////////////////////////////////////////////////////////
	///////////////////////// MAKE HISTOGRAM ////////////////////////////
	//////////////////////////////////////////////////////////////////////
	//////////////////////////////////////////////////////////////////////
	
	//method to make the histogram - now we can easily use the histogram code from other reactors
	protected NounMetadata getHistogram(PandasFrame frame, String table, String column, String panelId, int numBreaks) {
		PyTranslator pyT = this.insight.getPyTranslator();
		StringBuilder script = new StringBuilder();
		script.append("hist,bins = ");
		StringBuilder formatHist = new StringBuilder();
		StringBuilder formatBins = new StringBuilder();
		boolean format = false;

		String colSelector = table + "['" + column + "']";
		if(numBreaks > 1) {
//			np.histogram( FRAME940921[~np.isnan(FRAME940921['MovieBudget'])]['MovieBudget'], bins=5)
			script.append("np.histogram(").append(table).append("[~np.isnan(").append(colSelector).append(")]['").append(column).append("'], bins=").append(numBreaks).append(")");
		} else {
//			np.histogram( FRAME940921[~np.isnan(FRAME940921['MovieBudget'])]['MovieBudget'], bins='auto')
			script.append("np.histogram(").append(table).append("[~np.isnan(").append(colSelector).append(")]['").append(column).append("'], bins='auto')");
			formatHist.append("hist = list(map(int, hist))");
			formatBins.append("bins = list(bins)");
			format = true;
		}
		
		insight.getPyTranslator().runPyAndReturnOutput("import numpy as np", script.toString());

		if (format) {
			insight.getPyTranslator().runPyAndReturnOutput(formatHist.toString(), formatBins.toString());
		}

		List