edn-java

edn-java is a library to parse (read) and print (write) edn.

This is very early days. There are still rough edges in the design which may see me moving things around some before I'm happy with the result.

Installation

This is a Maven project with the following coordinates:

<dependency>
    <groupId>us.bpsm</groupId>
    <artifactId>edn-java</artifactId>
    <version>0.4.2</version>
</dependency>

It is available through the OSS Sonatype Releases repository:

https://oss.sonatype.org/content/repositories/releases

Parsing

You'll need to create a Parser and supply it with some input. Factory methods to create Parseable input are provided which accept either a java.lang.CharSequence or a java.lang.Readable. You can then call nextValue() on the Parser to get values form the input. When the input is exhausted, nextValue() will return Parser.END_OF_INPUT.

package us.bpsm.edn.examples;

import static org.junit.Assert.assertEquals;
import static us.bpsm.edn.Keyword.newKeyword;
import static us.bpsm.edn.parser.Parsers.defaultConfiguration;
import java.io.IOException;
import java.util.Map;
import org.junit.Test;
import us.bpsm.edn.parser.Parseable;
import us.bpsm.edn.parser.Parser;
import us.bpsm.edn.parser.Parsers;

public class ParseASingleMapTest {
    @Test
    public void simpleUsageExample() throws IOException {
        Parseable pbr = Parsers.newParseable("{:x 1, :y 2}");
        Parser p = Parsers.newParser(defaultConfiguration());
        Map<?, ?> m = (Map<?, ?>) p.nextValue(pbr);
        assertEquals(m.get(newKeyword("x")), 1L);
        assertEquals(m.get(newKeyword("y")), 2L);
        assertEquals(Parser.END_OF_INPUT, p.nextValue(pbr));
    }
}

Mapping from EDN to Java

Most edn values map to regular Java types, except in such cases where Java doesn't provide something suitable. Implementations of the types peculiar to edn are provided by the package us.bpsm.edn.

Symbol and Keyword have an optional prefix and a mandatory name. Both implement the interface Named.

Integers map to, Long or BigInteger depending on the magnitude of the number. Appending N to an integer literal maps to BigInteger irrespective of the magnitude.

Floating point numbers with the suffix M are mapeped to BigDecimal. All others are mapped to Double.

Characters are mapped to Character, booleans to Boolean and strings to String. No great shock there, I trust.

Lists "(...)" and vectors "[...]" are both mapped to implementations of java.util.List. A vector maps to a List implementation that also implements the marker interface java.util.RandomAccess.

Maps map to java.util.HashMap and sets to java.util.HashSet.

The parser is provided a configuration when created:

Parsers.newParser(Parsers.defaultConfiguration())

The parser can be customized to use different collection classes by first building the appropriate Parser.Config:

package us.bpsm.edn.examples;

import static org.junit.Assert.assertEquals;
import static us.bpsm.edn.parser.Parsers.newParseable;
import java.io.IOException;
import java.util.ArrayList;
import java.util.Arrays;
import java.util.SortedSet;
import java.util.TreeSet;
import org.junit.Test;
import us.bpsm.edn.parser.CollectionBuilder;
import us.bpsm.edn.parser.Parseable;
import us.bpsm.edn.parser.Parser;
import us.bpsm.edn.parser.Parsers;

public class SimpleParserConfigTest {
    @Test
    public void test() throws IOException {
        Parser.Config cfg =
            Parsers.newParserConfigBuilder().setSetFactory(
                    new CollectionBuilder.Factory() {
                public CollectionBuilder builder() {
                    return new CollectionBuilder() {
                        SortedSet<Object> s = new TreeSet<Object>();
                        public void add(Object o) { s.add(o); }
                        public Object build() { return s; }
                    };
                }
            }).build();
        Parseable pbr = newParseable("#{1 0 2 9 3 8 4 7 5 6}");
        Parser p = Parsers.newParser(cfg);
        SortedSet<?> s = (SortedSet<?>) p.nextValue(pbr);
        // The elements of s are sorted since our SetFactory
        // builds a SortedSet, not a (Hash)Set.
        assertEquals(Arrays.asList(0L, 1L, 2L, 3L, 4L, 5L, 6L, 7L, 8L, 9L),
            new ArrayList<Object>(s));
    }
}

Tagged Values

By default, handlers are provided automatically for #inst and #uuid, which return a java.util.Date and a java.util.UUID respectively. Tagged values with an unrecognized tag are mapped to us.bpsm.edn.TaggedValue.

Customizing the parsing of instants

The package us.bpsm.edn.parser makes three handlers for #inst available:

• InstantToDate is the default and converts each #inst to a java.util.Date.
• InstantToCalendar converts each #inst to a java.util.Calendar, which preserves the original GTM offset.
• InstantToTimestamp converts each #inst to a java.sql.Timstamp, which presrves nanoseconds.

Extend AbstractInstantHandler to provide your own implementation of #inst.

Adding support for your own tags

Use custom handlers may by building an appropriate Parser.Config:

package us.bpsm.edn.examples;

import static org.junit.Assert.assertEquals;
import java.io.IOException;
import java.net.URI;
import java.net.URISyntaxException;
import org.junit.Test;
import us.bpsm.edn.Tag;
import us.bpsm.edn.parser.Parseable;
import us.bpsm.edn.parser.Parser;
import us.bpsm.edn.parser.Parsers;
import us.bpsm.edn.parser.TagHandler;

public class CustomTagHandler {
    @Test
    public void test() throws IOException, URISyntaxException {
        Parser.Config cfg =
            Parsers.newParserConfigBuilder()
            .putTagHandler(Tag.newTag("us.bpsm", "uri"),
                new TagHandler() {
                public Object transform(Tag tag, Object value) {
                    return URI.create((String) value);
                }
            }).build();
        Parser p = Parsers.newParser(cfg);
        Parseable pbr = Parsers.newParseable(
            "#us.bpsm/uri \"http://example.com\"");
        assertEquals(new URI("http://example.com"), p.nextValue(pbr));
    }
}

Using pseudo-tags to influence the parsing of numbers

By default, integers not marked as arbitrary precision by the suffix "N" will parse as java.lang.Long. This can be influenced by installing handlers for the tag named by the constant Parser.Config.LONG_TAG.

package us.bpsm.edn.examples;

import static org.junit.Assert.assertEquals;
import java.io.IOException;
import java.math.BigInteger;
import org.junit.Test;
import us.bpsm.edn.Tag;
import us.bpsm.edn.parser.Parseable;
import us.bpsm.edn.parser.Parser;
import us.bpsm.edn.parser.Parsers;
import us.bpsm.edn.parser.TagHandler;

public class CustomLongHandler {
    @Test
    public void test() throws IOException {
        Parser.Config cfg =
            Parsers.newParserConfigBuilder()
                .putTagHandler(Parser.Config.LONG_TAG, new TagHandler() {
                    public Object transform(Tag tag, Object value) {
                        long n = (Long) value;
                        if (Integer.MIN_VALUE <= n && n <= Integer.MAX_VALUE) {
                            return Integer.valueOf((int) n);
                        } else {
                            return BigInteger.valueOf(n);
                        }
                    }
                }).build();
        Parser p = Parsers.newParser(cfg);
        Parseable pbr = Parsers.newParseable("1024, 2147483648");
        assertEquals(1024, p.nextValue(pbr));
        assertEquals(BigInteger.valueOf(2147483648L), p.nextValue(pbr));
    }
}

Parser also provides BIG_DECIMAL_TAG, DOUBLE_TAG and BIG_INTEGER_TAG to cover customizing all varieties of numbers.

Printing

The package us.bpsm.edn.printer provides an extensible printer for converting java data structures to valid edn text. The default configuration can print values of the following types, as well as Java's null, which prints as nil:

• us.bpsm.edn.Keyword
• us.bpsm.edn.Symbol
• us.bpsm.edn.TaggedValue
• java.lang.Boolean
• java.lang.Byte
• java.lang.CharSequence, which includes java.lang.String.
• java.lang.Character
• java.lang.Double
• java.lang.Float
• java.lang.Integer
• java.lang.Long
• java.lang.Short
• java.math.BigInteger
• java.meth.BigDecimal
• java.sql.Timestamp, as #inst.
• java.util.Date, as #inst.
• java.util.GregorianCalendar, as #inst.
• java.util.List, as [...] or (...).
• java.util.Map
• java.util.Set
• java.util.UUID, as #uuid.

The Printer writes characters to the underlying Writer. To serialize this text to a file or across a network you'll need to arrange to convert the characters to bytes. Use UTF-8, as edn specifies.

Supporting additional types

To support additional types, you'll need to provide a Protocol<Printer.Fn<?>> to the Printer which binds your custom Printer.Fn implementations to the class (or interface) it is responsible for.

As an example, we'll add printing support for URIs:

package us.bpsm.edn.examples;

import static org.junit.Assert.assertEquals;
import java.io.IOException;
import java.io.StringWriter;
import java.net.URI;
import org.junit.Test;
import us.bpsm.edn.Tag;
import us.bpsm.edn.printer.Printer;
import us.bpsm.edn.printer.Printer.Fn;
import us.bpsm.edn.printer.Printers;
import us.bpsm.edn.protocols.Protocol;

public class CustomTagPrinter {
    private static final Tag BPSM_URI = Tag.newTag("us.bpsm", "uri");
    @Test
    public void test() throws IOException {
        Protocol<Fn<?>> fns = Printers.defaultProtocolBuilder()
                .put(URI.class, new Printer.Fn<URI>() {
                    @Override
                    public void eval(URI self, Printer writer) {
                        writer.printValue(BPSM_URI).printValue(self.toString());
                    }})
                    .build();
        StringWriter w = new StringWriter();
        Printer p = Printers.newPrinter(fns, w);
        p.printValue(URI.create("http://example.com"));
        p.close();
        assertEquals("#us.bpsm/uri\"http://example.com\"", w.toString());
    }
}

Limitations

• Edn values must be acyclic. Any attempt to print a data structure containing cycles will surely end in a stack overflow.
• The current Printing support stikes me a as a bit of a hack.
• Currently, the printer doesn't pretty-print. This is fine when using edn for communication, but less than helpful for debugging or storage in version control.
• Edn-Java does not provide much by way of "convenience" methods. As a library it's still to young to really know what would be convenient.