plyj

plyj is a Java 7 parser written in Python. It has the awesome PLY as its sole dependency.

Status

plyj is officially in maintenance mode. For the foreseeable future no new developments will be made by me. As explained below plyj is basically a manual transcription of JDT's grammar. This has served me well in the past but with the Java 8 features JDT's developers did some things I'm unable to reproduce with PLY. I'm not smart enough to do this on my own. Writing parsers is still black magic to me. I am not thrilled by this development.

Synopsis

import plyj.parser as plyj

parser = plyj.Parser()

# parse a compilation unit from a file
tree = parser.parse_file(file('/foo/bar/Baz.java'))

# parse a compilation unit from a string
tree = parser.parse_string('class Foo { }')

# parse expression from string
tree = parser.parse_expression('1 / 2 * (float) 3')

# slightly bigger example: parse from an installed JDK with sources
import zipfile
srczip = zipfile.ZipFile('/usr/lib/jvm/java-6-openjdk/src.zip', mode='r')
info = srczip.getinfo('java/lang/Object.java')
srcfile = srczip.open(info)
tree = parser.parse_file(srcfile)

Acknowledgement

plyj is more or less a 1:1 translation of the grammar used in the Java Development Tools for Eclipse.

Completeness

The grammar is complete. There may still be errors left though. It successfully parsed every source file of the Oracle JDK. A lot of bugs were found that way but for all I know there may be many more. Time will tell.

Contributions

Contributions are always welcome. Depending on the type of work it may take a little while until I get around to accepting them.

• commit test that demonstrates a bug (optional)
• commit the fix
• open pull request

The test is required but does not have to be provided by you. If you do provide it, committing it first shows appropriate messages in the pull request and makes it easier to accept via Web.

Performance

A word of caution: Since plyj is pure Python, it is quite slow. Based on my laptop (which has an i7-3517U @ 1.90 GHz) I can present the following numbers (running inside a virtual machine):

• 619 rules
• 1149 states
• ~3.28 seconds to compile the grammar
• java/util/Collections.java takes ~0.44 seconds to parse (it's quite big though)

The timings are obviously highly dependent on the used hardware. My old laptop (Core 2 Duo @ 1 GHz) took 17 and 1.8 seconds respectively.

History

0.2 (in development)

• added ExpressionStatement

0.1 (2014-12-25) - The Christmas Release

• initial release
• supports complete Java 7 syntax (minus bugs)