Last Updated: April 26, 2023

This is part 4 of the Bleeding edge Java series. Start at the introduction if you haven’t already.

Parsing is the process of interpreting free form JSON text and converting it into a stream of JSON tokens. The JSON Spec defines the manner in which
parsing should be done: what constitutes whitespace, escape sequences in strings, etc. In short, take a stream of characters and interpret those characters as a stream of JSON tokens.

We can define a Java interface for this:

public interface JsonParser
    Stream<JsonToken> parse(IntStream stream);  // Note: Java Strings can be streamed via the chars() method

Instead of a normal Java Stream we need to use an IntStream as we’re dealing with primitive values: characters. Java does not have a CharStream so IntStream will suffice.

There are a number of wrinkles we’ll have to deal with when interpreting characters:

  • We sometimes need to look ahead before we know what token type is implied (for literals, strings, numbers)
  • Our library introduces an object name token. This means that when we parse a string we need to wait for the next token. If it’s a : we have to convert the string into an object name.
  • Number tokens can possibly end with a parseable character: , [ {. When this happens we need to reprocess that character.

Here is the general flow for parsing JSON text:

The code

Java streams process an element at a time and don’t allow arbitrary forward/backward access. This limitation would make writing a parser very difficult. Fortunately, a stream can be converted into an Iterator which allows the parser to continue accessing characters in the stream as needed. The stream method iterator() is used to access this iterator whose type has the unfortunately complex name PrimitiveIterator.OfInt.

Another issue with Java streams is that it’s not simple to generate a stream with an unknown length. The management of stream ending occurs in the JDK and there aren’t exposed methods to control this. However, you can easily convert an Iterator to a stream. Our parser can create an Iterator internally which can continue to return tokens until parsing is complete and then its hasNext( method can return false.

Putting this together we need these variables:

  • PrimitiveIterator.OfInt iterator – the iterator of the character stream being processed
  • LinkedList<JsonToken> tokens – the list of tokens that should be returned from our iterator. We need a list because we will sometimes have more than one token to return.
  • int pushedBack – this is the possibly pushed back character from number parsing or -1

Our parser implementation looks like this:

Stream<JsonToken> parse(IntStream stream)
    Iterator<JsonToken> tokenIterator = new Iterator<>() {
    return StreamSupport.stream(Spliterators.spliteratorUnknownSize(tokenIterator, 0), false);  // standard method of converting an Iterator to a stream

The first step in our internal iterator is to prepare our state variables:

private final StringBuilder builder = new StringBuilder();          // contains currently being parsed string, number, etc.
private final PrimitiveIterator.OfInt iterator = stream.iterator(); // stream of characters being processed as an Iterator
private final LinkedList<JsonToken> tokens = new LinkedList<>();    // list of tokens that need to be returned
private int pushedBack = -1;                                        // possibly pushed back character from number parsing or -1

The hasNext() implementation for our iterator is straightforward:

public boolean hasNext()
    return !tokens.isEmpty();

The majority of the implementation in our iterator will be in the method advanceToNextToken() which advances to the next token in the stream updating the state variables in the process. Before we show this let’s show the implementation for the next() method of our iterator:

public JsonToken next()
    if (tokens.isEmpty()) {
        throw new RuntimeException();
    JsonToken result = tokens.removeFirst();    // remove the first token and save it
    advanceToNextToken();                       // advance
    return result;                              // return the token

Now we implement advanceToNextToken(). First let’s advance past any whitespace unless there’s a pushed-back character:

if (pushedBack < 0) {
    pushedBack = advanceWhitespace(iterator);   // note: advanceWhitespace() can generate a new pushed-back character

Next we loop over characters from the character stream until we have a token.

String possibleObjectName = null;   // set when a string is processed
JsonToken nextToken = null;
while ((nextToken == null) && ((pushedBack >= 0) || iterator.hasNext())) {
    char c = (char) (((pushedBack >= 0) ? pushedBack : iterator.nextInt()) & 0xffff);
    pushedBack = -1;

    ... process character ...
if (possibleObjectName != null) {
    // wasn't an object name - we now have to emit this string token first
    tokens.add(new StringToken(possibleObjectName));
if (nextToken != null) {

We only need to process each character – again using enhanced switch:

nextToken = switch (c) {
    case '{' -> new BeginObjectToken();
    case '}' -> new EndObjectToken();
    case '[' -> new BeginArrayToken();
    case ']' -> new EndArrayToken();
    case ',' -> new ValueSeparatorToken();
    case '"' -> {
        if (possibleObjectName != null) {
            throw new RuntimeException();   // can't have more than one string 
        possibleObjectName = StringUtils.parseString(builder, iterator);    // it may be an object name so we have to keep parsing
        yield null;
    case 't' -> parseLiteral(iterator, "rue", new BooleanToken(true));    // see the Java source for details on parseLiteral()
    case 'f' -> parseLiteral(iterator, "alse", new BooleanToken(false));
    case 'n' -> parseLiteral(iterator, "ull", new NullToken());
    case '0', '1', '2', '3', '4', '5', '6', '7', '8', '9', '-', '+', '.', 'e', 'E' -> new NumberToken(StringUtils.parseNumber(builder, c, iterator, i -> pushedBack = i)); // see Java source for details on StringUtils.parseNumber
    case ':' -> {
        if (possibleObjectName == null) {
            throw new RuntimeException();   // illegal state - ':' without a preceeding string
        ObjectNameToken objectNameToken = new ObjectNameToken(possibleObjectName);
        possibleObjectName = null;
        yield objectNameToken;
    default -> {
        if (StringUtils.isWhitespace(c)) {
            yield null; // ignore whitespace between tokens
        throw new RuntimeException();   // unexpected character

See JsonParser.java for the full implementation.

Test it out for yourself!

In the previous articles we developed a serializer and a printer. Now we can reverse the process and take JSON text to produce a stream of JSON tokens. Let’s put this together in jshell. The example use these files:

From a terminal with Java 19 installed, run the following (note you’ll need the wget utility):

wget -nc https://raw.githubusercontent.com/starburstdata/developer-blog-assets/main/bleeding-edge-java/code/TypeToken.java
wget -nc https://raw.githubusercontent.com/starburstdata/developer-blog-assets/main/bleeding-edge-java/code/JsonToken.java
wget -nc https://raw.githubusercontent.com/starburstdata/developer-blog-assets/main/bleeding-edge-java/code/JsonSerializer.java
wget -nc https://raw.githubusercontent.com/starburstdata/developer-blog-assets/main/bleeding-edge-java/code/StringUtils.java
wget -nc https://raw.githubusercontent.com/starburstdata/developer-blog-assets/main/bleeding-edge-java/code/JsonPrinter.java
wget -nc https://raw.githubusercontent.com/starburstdata/developer-blog-assets/main/bleeding-edge-java/code/JsonParser.java
jshell --enable-preview TypeToken.java JsonToken.java JsonSerializer.java StringUtils.java JsonPrinter.java JsonParser.java

Inside jshell let’s serialize a Java record into JSON text and then parse that text back into a stream of JSON tokens:

var serializer = JsonSerializer.instance();
var printer = JsonPrinter.instance();
var parser = JsonParser.instance();

record Person(String name, int age) {}
var person = new Person("someone", 28);

serializer.serialize(person)        // serialize to stream of JsonToken
    .map(printer::print)            // map each JsonToken to a String (as a CharSequence)
    .map(CharSequence::chars)       // map each CharSequence to an IntStream
    .flatMap(parser::parse)         // pass each IntStream to the parser and flatten the resulting stream of tokens
    .forEach(System.out::println);  // print each token to standard out


We can now serialize a Java record into JSON text and then parse that text back into a stream of JSON tokens. Next up is deserialization.

We’re hiring

Want to be able to use the latest features of Java? We’re hiring!

Jordan Zimmerman is a Senior Software Engineer working on Starburst Galaxy.

Start Free with
Starburst Galaxy

Up to $500 in usage credits included

  • Query your data lake fast with Starburst's best-in-class MPP SQL query engine
  • Get up and running in less than 5 minutes
  • Easily deploy clusters in AWS, Azure and Google Cloud
For more deployment options:
Download Starburst Enterprise

Please fill in all required fields and ensure you are using a valid email address.