Notice! This post is more than a year old. It may be outdated.
Unicode is a text encoding standard which supports a broad range of characters and symbols. Although the latest version of the standard is 9.0, JDK 8 supports Unicode 6.2 and JDK 9 is expected to be released with support for Unicode 8.0. Java allows you to insert any supported Unicode characters with Unicode escapes. These are essentially a sequence of hexadecimal digits representing a code point. In this post I’m going to cover how to use Unicode escapes in Java and how to avoid unexplainable compiler errors caused by Unicode escape misuse.
What are unicode escapes?
Let’s start from the beginning. Unicode escapes are used to represent Unicode symbols with only ASCII characters. This will come in handy when you need to insert a character that cannot be represented in the source file’s character set. According to section 3.3 of the Java Language Specification (JLS) a unicode escape consists of a backslash character (\) followed by one or more ‘u’ characters and four hexadecimal digits.
So for example
\u000A will be treated as a line feed.
The following is a piece of Java code containing a Unicode escape.
Take a moment to think about what will be printed out. If you want, copy and paste the code to a new file, compile and run it.
At first glance it looks like the program prints out
18. There’s 18 characters between the double quotes, so the length of the string should be 18. But if you run the program, the output is
13. As the comment suggests, the Unicode escape will be replaced with a single character.
Equipped with the knowledge that Unicode escapes are replaced with their respective Unicode characters, let’s look at the following example.
Can you guess what will be printed out now? The answer should be the same as before, right? I’m sure some of you might suspect that this is a trick question and as a matter of fact, it is. This example will not compile at all.
What!? So many errors! My IDE doesn’t show any squiggly red lines and I can’t seem to find any syntax errors myself. Error on line 3? But that’s a comment. What is going on?
What caused the error?
To get a better understanding of what is going on, we need to look at section 3.2 of the Java Language Specification - Lexical Translations. I cannot speak for all compilers that have ever existed but usually the first job of a compiler is to take the source code of a program, treat it as a sequence of characters and produce a sequence of tokens.
A token is something that has a meaning in the context of the language. For example in Java it can be a reserved word (
interface), an operator (
>>) or a literal (a notation for representing a fixed value). The process of generating tokens from a sequence of characters is called lexical analysis (or lexical translation as it is called in the Oracle docs) and the program that performs that is called a lexer or a tokenizer.
The Java Language Specification says that lexical translation is performed in the following 3 steps, where each step is applied to the result of the previous step:
- Translation of Unicode escapes.
- Divide stream of input characters into lines by recognizing line terminators (LF, CR or CR LF).
- Discard whitespace and comments and tokenize the result from the previous step.
As you can see, the very first step processes Unicode escapes. This is done before the compiler has had the chance to separate the source code into tokens. Broadly speaking, this is like applying a search and replace function on the source code, replacing all well formed Unicode escapes with their respective Unicode characters, and then letting the compiler work on the rest of the code.
Keep in mind, when Unicode escapes are being processed, the compiler does not differentiate comments from actual code. It can only see a sequence of characters. And this explains the erroneous code you saw in the introduction of this post. Let’s have a look at it again.
The Unicode escape representing the line feed character is replaced with a line feed and now part of the comment is on a new line. Unfortunately the new line does not start with a double-slash (
//) and the rest of the line is not valid Java code. Hence the confusing compiler error shown previously.
Quick Detour: native2ascii
You can play around with Unicode conversion yourself. Up to Java 8 the JRE is bundled with a tool called native2ascii, which converts a file with characters in any supported character encoding to one with ASCII and/or Unicode escapes, or visa versa.
What about Java 9 (and later)? Prior to it, Java property files use the ISO-8859-1 character set by default. Characters that cannot be represented in ISO-8859-1 are converted to Unicode escapes using the native2ascii tool. But JEP 226 changes that and property files can now be encoded in UTF-8, which means that the native2ascii tool is not needed anymore.
In this post, native2ascii is used to demonstrate what the Java source file would look like if Unicode escapes were replaced with actual Unicode characters. For Java 9 users, I recommend to use uni2ascii, which can achieve the same result.
Hiding code in comments
If Unicode escapes are processed before everything else, then can I cleverly hide code inside comments which will later be executed? The somewhat scary answer to this question is yes. Looking back at the previous example, we saw that a line feed was inserted and the rest of the comment was on the next line, resulting in invalid Java code. But we could have written the following
If the Unicode escape is replaced with a line feed, then it should be clear there’s actually two print statements executed.
Why does Java allow that?
This all seems weird, right? Why is Java designed like that? Is it a bug that was accidentally introduced and never fixed because it would break something else? To find an answer to that question we need to look at section 3.1 and section 3.3 of the Java Language Specification (JLS).
From section 3.1
The Java programming language represents text in sequences of 16-bit code units, using the UTF-16 encoding.
From section 3.3
The Java programming language specifies a standard way of transforming a program written in Unicode into ASCII that changes a program into a form that can be processed by ASCII-based tools. The transformation involves converting any Unicode escapes in the source text of the program to ASCII by adding an extra u - for example, \uxxxx becomes \uuxxxx - while simultaneously converting non-ASCII characters in the source text to Unicode escapes containing a single u each.
Unicode escapes were designed to ensure compatibility with a wide variety of character sets. Think of the following scenario. You receive a piece of code with an encoding your text editor does not understand (i.e. the code includes characters not available in the encoding you use). This can be solved by replacing all unknown characters with Unicode escapes. As ASCII is the lowest common denominator of character sets, it is always possible to represent Java code in any encoding by replacing characters that are not supported by the target encoding with Unicode escapes. Today Unicode is fairly common and this should not be an issue, but I guess back in the early days this was useful.
The transformed version is equal to the initial version and the compiler treats them as the same. As this process is reversible, the compiler can go back to the initial version by replacing Unicode escapes with respective Unicode characters.
From section 3.3
This transformed version is equally acceptable to a Java compiler and represents the exact same program. The exact Unicode source can later be restored from this ASCII form by converting each [Unicode] escape sequence where multiple u’s are present to a sequence of Unicode characters with one fewer u, while simultaneously converting each [Unicode] escape sequence with a single u to the corresponding single Unicode character.
Prefer escape sequences
Because Unicode escapes are processed before everything else in the compilation process, they can create a considerable amount of confusion. Therefore it is better to avoid them if possible. Instead prefer escape sequences, for example
\n for line feeds or
\” for double quotes, in string or character literals. There’s no need to use Unicode escapes for ASCII characters.
Unicode escapes have to be well formed
I mentioned previously that only well formed Unicode escapes are replaced with Unicode characters during the compilation process. You will get an error if there’s an ill-formed Unicode escape in your code. Have a look at the following example.
This seems like an innocent looking piece of code. The comment tries to be helpful and communicate something important to the reader. Unfortunately there’s a Unicode escape lurking in this code which is not well formed. As you know by now, Unicode escapes start with
\u and the compiler expects four hexadecimal digits to be followed. When this rule is not met, the compiler will throw an error.
Windows path names use backslashes to separate directory names. But if one of those backslashes is followed with the
u character, you can run into unexpected situations. The problem in this example is the sequence of characters
\users which is in fact an ill-formed Unicode escape.
Taking it to the extreme
We have looked at several examples where Unicode escapes can cause harm. Your eye should be trained enough to spot most of them by now. For the next example I’m going to show you a piece of code that I first saw when I read the book Java Puzzlers by Joshua Bloch and Neal Gafter.
What? Really? This looks like an entry to a code obfuscation contest. But if you think about it, this seems like it should compile, granted that all the Unicode escapes actually represent characters that make up a valid Java program. We learned that the very first thing the compiler does is look for Unicode escapes and replace them. It does not know anything about the program structure at that point.
You can try it yourself. Copy the text into a file called
Ugly.java. Then compile and run the program. By the way, there’s no point trying to run it from an IDE (at least IntelliJ IDEA is baffled and is only able to show squiggly red lines). Use command line tools instead.
Additionally you can use the
native2ascii tool to view what the code looks like if all Unicode escapes have been replaced.
I have only one thing to say. Just because you can doesn’t mean you should.
Professionally I have never had the need to insert a Unicode escape. Nowadays Unicode is fairly common and most text editors can display non-ASCII characters. If I find myself in a situation where I need to insert a character that’s not available on my keyboard, I can use methods provided by most operating systems to input them. If possible, avoid Unicode escapes because they create confusion. Prefer escape sequences instead.