Old McDonald had a File, IO IO IO

One of the most basic and common functions we programmers have to implement involves reading and writing data files. So get used to doing so. The good news is that Java is well equipped with many library functions for doing this. The bad news is exactly the same. There are many library functions, so many that you can get lost in them. In fact, just when we thought we had a handle on the 'java.io' package, along came 'java.nio' (New I/O) to cloud the picture even more. Even an experienced programmer can get a little lost in the multiple classes that are in place for dealing with persistent data, and I certainly don't want to overwhelm my audience, so I'm going to stick to some basics.

 First off, I'm going to pretend that 'new I/O' never happened. I'm also going to offer you only a subset of all the functionality that's out there even in the java.io library. When it comes down to it, when you've got a file to read or write, you want to be able to just do so, without a lot of muss or fuss.

I am going to have to assume that you're reasonably familiar with directories (or folders, if you must) and files as seen by your operating system. I'm going to assume that you understand what a text file is, and that you feel reasonably comfortable with viewing them or editing them with notepad, or UltraEdit, or whatever your choice of text editor might happen to be.

First, a note on file types. There are of course a huge number of file formats out there, to support applications from AutoCAD to PhotoShop to Visual Studio, but all files can be pretty much divided into to major categories: Text and Other. Text files are broken down into lines, and many programs deal with them on a line by line basis. Other files tend to consist of plain old blocks of information, whether those be pixels for an image, or a song or movie, or quite literally anything else you can save.

Java supports both major file types of course, and has objects specifically for dealing with each. When dealing with text files it's very convenient to read or write things one line at a time, so we have that ability in our libraries.

We've used an object we call 'PersonData' before, and we'll make use of it again now.  Suppose we had a file in which we had a bunch of last names and first names stored, for now we'll assume that last names and first names each appear on a line, so the file would look something like this:

PEOPLE.TXT

    Schmoe
    Joe
    Doe
    John
    Smith
    Bill

So we've got the information here to support three PersonData objects, all contained within a file called PEOPLE.TXT.

If we wanted to read the names from this file and create PersonData objects, it's conceptually pretty simple:

Open the file for reading
As long as there's something to read
    Read a last name
    Read a first name
    Create a PersonData object with those values

That makes sense, right?

Enter the Scanner.  The Scanner object is pretty new, and frankly I'm still not completely used to it.  But it's so darned handy I have no excuse not to use it.  The Scanner has several different constructors, but we're going to make use of the one that uses the File object right now.

First, let's define the file that we intend to read.  This sets us up for the next step:

File personFile = new File("PEOPLE.TXT");

Note that this isn't something that will fail, even if PEOPLE.TXT does not actually exist.  Maybe we intend to create it next, or we want to test for its existence.  The File object is handy that way.

Scanner personScanner = new Scanner(personFile);

This can fail.  If personFile refers to a non-existent file bad things can and will happen starting here.  Proper code will be wrapped up in try/catch blocks as I discussed when we talked about exception handling.

At this point we have covered the first line of our pseudo-code algorithm:  "Open the file for reading".  The next part was "As long as there's something to read", which a Scanner can handle quite easily:

while (personScanner.hasNextLine()) {

This loop will now run until we're out of lines.  That's pretty convenient.  Now let's extract the last and first names:

    String lastName = personScanner.nextLine();
    String firstName = personScanner.nextLine();

Yes, there's a weakness or three here.  We're making the assumption that our file is correctly formatted and complete, with two lines for every person.  Good coding practice would include checking 'hasNextLine' in between those two lines and dealing with it if that isn't the case.

We're actually almost done here.  Once we've got the last name and the first name, we can go ahead and create a PersonData object:

    PersonData personFromFile = new PersonData( firstName, lastName );

At this point, technically we can end the loop.  We've done the specified job, although realistically this is useless.  There's not much point in reading a data file only to toss away the data.  You'd want to do something with these 'personFromFile' objects, like adding them to a List or Map, or storing them in some other sort of data structure your program will use.

But for now, we'll just pretend we're doing that and close down the file:

}
Scanner.close();

And that's about all there is to reading text files since Java 7 came out.  It was a bit more difficult before, when we had to create a couple of other objects in place of Scanner, and if the data included numbers or multiple items on one line it was even more involved.

Let's say that the file was formatted like this instead:

PEOPLE.TXT

    Schmoe Joe
    Doe John
    Smith Bill

That seems reasonable, right?  It is, and it's pretty easy to deal with.  All we'd need to do is change the following two lines from this:

    String lastName = personScanner.nextLine();
    String firstName = personScanner.nextLine();

To this:

    String lastName = personScanner.next();
    String firstName = personScanner.nextLine();

And we will have accomplished our goal.  We could also have done this:

    String lastName = personScanner.next();
    String firstName = personScanner.next();

    String anythingLeftover = personScanner.nextLine();

What is important is to realize that 'nextLine' is critical to advance us from one line to the next, so we have to use it before we go through the loop again.

Before I go, I also want to mention that Scanner is capable of reading not just Strings, but also any of the primitive data types as well.  It supports nextInt(), nextDouble(), etc.  It's also capable of reading more complicated input via the use of Regular Expressions, which are a huge topic in their own right.  Regular Expressions essentially provide an entire language for matching complicated text, and an entire series of articles could no doubt be written just on that topic alone.

In our next episode, we'll talk about writing text files, which is just the reverse of what we've done here.

Don't Treat me Like an Object

I originally posted this on Reddit, and got some positive feedback on it, so I thought I would put it up here as an early lesson in object orientation.  That is, after all, our goal here and a dozen posts in I haven't really mentioned it at all yet.

Moving from procedural code to object oriented code is a major leap and takes a real adjustment in thinking.

Very often, multiple pieces of data logically belong together. For instance, a person's given name and family name should generally not be separated. There's little point in having separate references to "John" and "Doe", after all.

This concept existed before object oriented programming did, in the form of 'structures'. In c we'd do this

struct PersonData {
    char givenName[20];
    char surname[20];
}

Which helped some. Now we could declare a variable of type PersonData and pass it around, knowing that the fields would be kept together. Believe it or not that was a big step forward.  Of course a bit of energy was spent on doing things like determining if '20' was enough storage for a name.  

Objects are basically structs with functions added, and with the ability to hide internal details from users of said objects.

So while it would be possible to recreate the above struct as an object like this:

public class PersonData {
    public String givenName;
    public String surName;
}

That doesn't really take advantage of the power of the language yet. Sure, we can now go:

PersonData johnDoe = new PersonData(); //Use default constructor that comes free
johnDoe.givenName = "John";
johnDoe.surName  = "Doe";

But we can do better. For instance, a person ALWAYS has those names, right?  So why allow a PersonData object to even be created without them? Enter the 'Constructor'. It's just a method designed for one purpose: To initialize a new object. We could insert the following into our PersonData class:

public PersonData(String given, String sur) {
    givenName = given;
    surName = sur;
}

And now we could write the earlier code like this:

PersonData johnDoe = new PersonData("John", "Doe");

This PersonData is quite limited as it stands right now, since it's got no methods on it. But you know, that's OK, there's a place for objects that just store data, too. It is, however, a starting point to think about how objects work.

Of course, a constructor can do much more than set values for a few fields, too, performing any sorts of setup and initialization that might be required.

Please note, I am not saying that exposing those variables as public is a good idea, and I would not write them quite that way for a production application, but I'm trying to keep this example simple to understand.

Since whenever you write Java code you're forced to put it into classes anyway, I suggest that you always treat your classes as though they're going to be objects.  Make your main() methods as minimal as possible.  In short,don't do this:

public class InputData {
    public static void main(String args[]) {
        Scanner s = new Scanner(System.in);
        while(s.hasNextLine()) {
        ...

Instead, write it like this from the start:

public class InputData {
    public static void main(String args[]) {
        InputData inputData = new InputData();
        inputData.execute();
    }

    public void inputData() {
        Scanner s = new Scanner(System.in);
        while (s.hasNextLine()) {
        ...

Note that by adding just a few lines I've turned this into slightly more object oriented code.  I've avoided the static cascade problem.  I've also made it so that if I want to I can use this object from some other piece of code without a great deal of trouble.  That code I have in main could be duplicated anywhere.