palmr / classfile-parser Goto Github PK

nom, the main library this project is based on, has release version 4.0 which we should upgrade to.

They have information on upgrading from previous versions to help with breaking changes.

Relevant Reference: https://docs.oracle.com/javase/specs/jvms/se8/html/jvms-4.html#jvms-4.7.4
The VerificationTypeInfo in src/attribute_info/types.rs only parses the tags of the code, and ignores the data that would come after it (only two variants have it).

Currently almost everything is public, which isn't great.

The only way to parse your own byte array requires calling the parsing code yourself and dealing with the result from nom, which leaks parsing implementation details outside of the crate.

There should probably be some more top level functions in lib for parsing from paths, jars, byte arrays etc.
Structs for the public data types should be together rather than scattered about.

Parsers should be together and non-public.

A parser is required for LocalVariableTypeTable attributes

https://docs.oracle.com/javase/specs/jvms/se8/html/jvms-4.html#jvms-4.7.14

I would like to use this library, but it has no license. Could you provide a license for the code?

A parser is required for SourceFile attributes

https://docs.oracle.com/javase/specs/jvms/se8/html/jvms-4.html#jvms-4.7.10

A parser is required for LocalVariableTable attributes

https://docs.oracle.com/javase/specs/jvms/se8/html/jvms-4.html#jvms-4.7.13

At the moment, the code parser is not tested against all opcodes. Ideally, we can reuse an existing corpus of opcodes to test our implementation.

Could class files be written, as well as parsed, via serde and serde-bytes ?
I could work on this...

I was using this as a dependency and wanted to confirm if the project was still supported.

Hey is this project dead?

I want to use it in my project to write a full virtual machine (with cranelift backend) in rust.
But I need all features in README.md.

Now that the parsing of valid Java strings works as expected (thanks!), I discovered an other issue: Java supports unpaired surrogates (U+D800 to U+DFFF). They aren't officially allowed, which is why Rust does not support them. However, some Java libraries use them to encode binary data (e.g. libgcj). At the moment, class_parser is unable to handle such classes because the decoding error is propagated back to it.

Possible solutions:

• keep as is
• change utf8_string to Result<String, Vec<u8>> and let the user decide (e.g. apply String::from_utf8_lossy)
• apply String::from_utf8_lossy ourselves
• ...

A parser is required for LineNumberTable attributes

https://docs.oracle.com/javase/specs/jvms/se8/html/jvms-4.html#jvms-4.7.12

A parser is required for BootstrapMethods attributes

https://docs.oracle.com/javase/specs/jvms/se8/html/jvms-4.html#jvms-4.7.23