From text rows to numbers

Part 14 starts with data visualization: showing numbers as a picture. A table of 50 numbers is hard to read, while a line that climbs and then falls tells its story at a glance. JavaFX, the library of part 13, has classes for charts, and this module uses them: line charts, bar charts and charts that change when the user moves a slider.

Charts in Java Arena

As with the windows of part 13 (module 48 explained it), Java Arena's JavaFX is a practice version written for this site, and the page draws each chart from what your program told JavaFX to show: the kind of chart, its title, its axes and every data point. The drawing is approximate, as the windows were: real JavaFX picks its own colors, fonts and tick marks. Under each chart, "The window as text" lists every data point, and that's what the checks compare. The check that runs every model solution on the real JavaFX (module 48) covers charts too: it confirms that real JavaFX gets the same kind of chart, title, legend, axes and points, and the same slider values. It doesn't check how the chart is drawn. (Module 52 draws on a canvas and works with images, which are read and written pixel by pixel exactly; it also explains why sound can't play on this page.)

A chart needs numbers, and numbers usually come from a file. So this first step reads files of data, and step 2 draws the first chart. The numbers in this module's files are made up for the exercises: they look like measurements, but nobody measured them.

Rows and columns

A weather station on a hill measures the rain each month. Its file, rain.csv:

month,mm
1,52
2,41
3,-
4,38
5,45
6,60

This is a CSV file (comma-separated values): one row per line, and the values of a row separated by commas, so that they form columns. The first line is a header: it names the columns, and it isn't data. The - in March means that there's no measurement for that month (the rain gauge was broken). Files of data often mark a missing value like that, so a program must expect it.

The program reads the lines (module 48 used Files.readAllLines too), skips the header, and splits each row into its values with split:

import java.nio.file.Files;
import java.nio.file.Paths;
import java.util.List;

public class Main {
    public static void main(String[] args) throws Exception {
        List<String> lines = Files.readAllLines(Paths.get("rain.csv"));
        int total = 0;
        int months = 0;
        for (int i = 1; i < lines.size(); i++) {
            String[] parts = lines.get(i).split(",");
            if (parts[1].equals("-")) {
                System.out.println("Month " + parts[0] + ": no measurement");
            } else {
                total += Integer.valueOf(parts[1]);
                months++;
            }
        }
        System.out.println("Months measured: " + months);
        System.out.println("Rain in them: " + total + " mm");
    }
}

Month 3: no measurement
Months measured: 5
Rain in them: 236 mm

• The loop starts at index 1, so the header line is left out. Integer.valueOf("mm") would throw a NumberFormatException.
• "4,38".split(",") gives the array ["4", "38"]: text, still. Integer.valueOf(parts[1]) turns the text into a number. A value with a decimal point, such as 12.5, needs Double.valueOf.
• The check for - comes before Integer.valueOf, since Integer.valueOf("-") throws too. A missing value is skipped: it isn't counted as 0, which would be a measurement of no rain at all, and so a lie.

Two more things that files of data do:

• Tabs. A tab-separated file (often named .tsv) has a tab between its values instead of a comma: split("\t"). Tabs are handy when the values themselves may contain commas, such as Smith, Ann.
• Empty values at the end. split drops empty values at the end of a row: "5,".split(",") gives only ["5"], an array of length 1, and parts[1] throws an ArrayIndexOutOfBoundsException. That's one reason why files mark a missing value with a - rather than with nothing.

Keeping the numbers

A sum is fine for a report, but a chart needs every value, and which month it belongs to. A map from the month to the rain keeps both, and a month without a measurement simply isn't in it. Put the reading in a class of its own, and the rest of the program asks it for the map. The same file again:

month,mm
1,52
2,41
3,-
4,38
5,45
6,60

import java.util.Map;

public class Main {
    public static void main(String[] args) throws Exception {
        Map<Integer, Integer> rain = RainFile.read("rain.csv");
        System.out.println(rain.size() + " months");
        System.out.println("April: " + rain.get(4) + " mm");
        System.out.println("March measured: " + rain.containsKey(3));
    }
}

import java.io.IOException;
import java.nio.file.Files;
import java.nio.file.Paths;
import java.util.HashMap;
import java.util.List;
import java.util.Map;

public class RainFile {

    public static Map<Integer, Integer> read(String fileName) throws IOException {
        Map<Integer, Integer> rain = new HashMap<>();
        List<String> lines = Files.readAllLines(Paths.get(fileName));
        for (int i = 1; i < lines.size(); i++) {
            String[] parts = lines.get(i).split(",");
            if (!parts[1].equals("-")) {
                rain.put(Integer.valueOf(parts[0]), Integer.valueOf(parts[1]));
            }
        }
        return rain;
    }
}

5 months
April: 38 mm
March measured: false

read declares throws IOException (module 42), since Files.readAllLines can throw it, and leaves it to the caller. When the values are only a sequence, such as the cups of coffee on each day of a week in order, a list is enough: the position in the list says where each value belongs.