This article, also published on https://dev.java, highlights how common I/O tasks can be done easily with the Java API. I wrote it because web searches and AI chatbots often lead to unnecessarily convoluted code from the ancient past.
This article focuses on tasks that application programmers are likely to encounter, particularly in web applications, such as:
The Java API supports many other tasks, which are explained in detail in the Java I/O API tutorial.
Modern, at the time of this writing, means features that are out of preview in Java 21. In particular:
java.nio.file.Files
class, which first appeared in Java 7, added useful methods in Java 8, 11, and 12java.io.InputStream
gained useful methods in Java 9, 11, and 12java.io.File
and java.io.BufferedReader
classes are now thoroughly obsolete.Tou can read a text file into a string like this:
String content = Files.readString(path);
Here, path
is an instance of java.nio.Path
, obtained like this:
var path = Path.of("/usr/share/dict/words");
If you want the file as a sequence of lines, call
List<String> lines = Files.readAllLines(path);
If the file is large, process the lines lazily as a Stream<String>
:
try (Stream<String> lines = Files.lines(path)) { . . . }
Also use Files.lines
if you can naturally process lines with stream operations (such as map
, filter
).
Note that the stream returned by Files.lines
needs to be closed. To ensure that this happens, use a try
-with-resources statement, as in the preceding code snippet.
There is no longer a good reason to use the readLine
method of java.io.BufferedReader
.
To split your input into something else than lines, use a java.util.Scanner
. For example, here is how you can read words, separated by non-letters:
Stream<String> tokens = new Scanner(path).useDelimiter("\\PL+").tokens();
Be careful when reading numbers from text files. Number formatting is locale-dependent. For example, the input 100.000
means 100.0 in the US locale and 100000 in the German locale. If the numbers may be formatted with decimal separators, use a java.text.NumberFormatter
or a Scanner
with the appropriate locale. Or, if you know that the numbers follow the same format as Java literals, use Integer.parseInt
/Double.parseDouble
.
It is possible to parse a mixture of strings and numbers with the Scanner
methods nextInt
/nextDouble
/next
/nextLine
, but it can be confusing or tedious. Just consider a file consisting of records:
Zapper 2000 Microwave 49.95 Blackwell Toaster 29.95
It is clearly simpler to read the input as one string per line, and then parse it.
As anecdotal evidence, when solving the 2023 Advent of Code problems, I always found it easier to parse the input lines than to use a Scanner
.
You can write a string to a text file with a single call:
String content = . . .; Files.writeString(path, content);
If you have a list of lines rather than a single string, use:
List<String> lines = . . .; Files.write(path, lines);
For more general output, use a PrintWriter
if you want to use the printf
method:
var writer = new PrintWriter(path.toFile()); writer.printf(locale, "Hello, %s, next year you'll be %d years old!%n", name, age + 1);
Note that printf
is locale-specific. When writing numbers, be sure to write them in the appropriate format. Instead of using printf
, consider java.text.NumberFormat
or Integer.toString
/Double.toString
.
Weirdly enough, as of Java 21, there is no PrintWriter
constructor with a Path
parameter.
If you don't use printf
, you can use the BufferedWriter
class and write strings with the write
method:
var writer = Files.newBufferedWriter(path); writer.write(line); // Does not write a line separator writer.newLine();
Remember to close the writer
when you are done.
Perhaps the most common reason to use a stream is to read something from a web site.
If you need to set request headers or read response headers, use the HttpClient
:
HttpClient client = HttpClient.newBuilder().build(); HttpRequest request = HttpRequest.newBuilder() .uri(URI.create("https://horstmann.com/index.html")) .GET() .build(); HttpResponse<String> response = client.send(request, HttpResponse.BodyHandlers.ofString()); String result = response.body();
That is overkill if all you want is the data. Instead, use:
InputStream in = new URI("https://horstmann.com/index.html").toURL().openStream();
Then read the data into a byte array and optionally turn them into a string:
byte[] bytes = in.readAllBytes(); String result = new String(bytes);
Or transfer the data to an output stream:
OutputStream out = Files.newOutputStream(path); in.transferTo(out);
Note that no loop is required if you simply want to read all bytes of an input stream.
But do you really need an input stream? Many APIs give you the option to read from a file or URL.
Your favorite JSON library is likely to have methods for reading from a file or URL. For example, with Jackson jr:
URL url = new URI("https://dog.ceo/api/breeds/image/random").toURL(); Map<String, Object> result = JSON.std.mapFrom(url);
Here is how to read the dog image from the preceding call:
url = new URI(result.get("message").toString()).toURL(); BufferedImage img = javax.imageio.ImageIO.read(url)
This is better than passing an input stream to the read
method, because the library can use additional information from the URL to determine the image type.
The java.nio.file.Files
class provides a comprehensive set of file operations, such as creating, copying, moving, and deleting fies and directories. The File System Basics tutorial provides a thorough description. In this section, I highlight a few common tasks.
For most situations you can use one of two methods. The Files.list
method visits all entries (files, subdirectories, symbolic links) of a directory.
try (Stream<Path> entries = Files.list(pathToDirectory)) { . . . }
Use a try
-with-resources statement to ensure that the stream object, which keeps track of the iteration, will be closed.
If you also want to visit the entries of descendant directories, instead use the method
Stream<Path> entries = Files.walk(pathToDirectory);
Then simply use stream methods to home in on the entries that you are interested in, and to collect the results:
try (Stream<Path> entries = Files.walk(pathToDirectory)) { List<Path> htmlFiles = entries.filter(p -> p.toString().endsWith("html")).toList(); . . . }
Here are the other methods for traversing directory entries:
Files.walk
lets you limit the depth of the traversed tree.Files.walkFileTree
methods provide more control over the iteration process, by notifying a FileVisitor
when a directory is visited for the first and last time. This can be occasionally useful, in particularly for emptying and deleting a tree of directories. See the tutorial =https://dev.java/learn/java-io/file-system/walking-tree/ Walking the File Tree for details. Unless you need this control, use the simpler Files.walk
method.Files.find
method is just like Files.walk
, but you provide a filter that inspects each path and its BasicFileAttributes
. This is slightly more efficient than reading the attributes separately for each file.Files.newDirectoryStream
methods yields DirectoryStream
instances, which can be used in enhanced for
loops. There is no advantage over using Files.list
. File.list
and File.listFiles
methods return file names or File
objects. These are now obsolete.Ever since Java 1.1, the ZipInputStream
and ZipOutputStream
classes provide an API for processing zip files. But the API is a bit clunky. Java 8 introduced a much nicer zip file system:
FileSystem fs = FileSystems.newFileSystem(pathToZipFile);
You can then use the methods of the Files
class. Here we get a list of all files in the zip file:
try (Stream<Path> entries = Files.walk(fs.getPath("/"))) { List<Path> filesInZip = entries.filter(Files::isRegularFile).toList(); }
To read the file contents, just use Files.readString
or Files.readAllBytes
:
String contents = Files.readString(fs.getPath("/LICENSE"));
You can remove files with Files.delete
. To add or replace files, simply use Files.writeString
or Files.write
.
You must close the file system so that the changes are written to the zip file. Call
fs.close();
or use a try
-with-resources statement.
Fairly often, I need to collect user input, produce files, and run an external process. Then I use temporary files, which are gone after the next reboot, or a temporary directory that I erase after the process has completed.
The calls
Path filePath = Files.createTempFile("myapp", ".txt"); Path dirPath = Files.createTempDirectory("myapp");
create a temporary file or directory in a suitable location (/tmp
in Linux) with the given prefix and, for a file, suffix.
Web searches and AI chats can suggest needlessly complex code for common I/O operations. There are often better alternatives:
Files
methods for creating, copying, moving, and deleting files and directories.Files.list
or Files.walk
to traverse directory entries.File
class.
Comments powered by Talkyard.