Accented characters issue when using java.nio.file.Path

[Manokha]

Describe the issue Accented characters issue when using Path.of(...)

Steps to reproduce the issue

1. Using this test sample (GraalVMAccentedTest.java):

import java.nio.file.Path;

public final class GraalVMAccentedTest {
    public static void main(String[] args) {
        var file = Path.of("àéïôù.txt").toFile();
        System.out.println("Could instantiate file " + file.getName());
    }
}

2. Using eclipse-temurin java 17 (locally):

javac GraalVMAccentedTest.java
java GraalVMAccentedTest
# Outputs: "Could instantiate file àéïôù.txt"

3. Run a GraalVM image:

docker run --name graalvm-accented-test -it "ghcr.io/graalvm/graalvm-ce:ol9-java17" /bin/bash

4. Copy the file there (from another terminal):

docker cp GraalVMAccentedTest.java graalvm-accented-test:/app/GraalVMAccentedTest.java

5. Compile and run (from the 3rd step terminal):

# Fails without -encoding utf8
javac -encoding utf8 GraalVMAccentedTest.java
java GraalVMAccentedTest

6. Outputs:

Exception in thread "main" java.nio.file.InvalidPathException: Malformed input or input contains unmappable characters: ?????.txt
	at java.base/sun.nio.fs.UnixPath.encode(UnixPath.java:121)
	at java.base/sun.nio.fs.UnixPath.<init>(UnixPath.java:68)
	at java.base/sun.nio.fs.UnixFileSystem.getPath(UnixFileSystem.java:279)
	at java.base/java.nio.file.Path.of(Path.java:147)
	at GraalVMAccentedTest.main(GraalVMAccentedTest.java:5)

Describe GraalVM and your environment:

• GraalVM version: ol9-java17-22.3.3-b1
• JDK major version: 17
• OS: OL9
• Architecture: AMD64

More details I tried with different bases (ol7, ol8), turns out ol7 works. And doesn't need the "-encoding utf-8" when compiling:

• ghcr.io/graalvm/graalvm-ce:ol9-java17 and container-registry.oracle.com/graalvm/jdk:17-ol7 are failing
• ghcr.io/graalvm/graalvm-ce:ol8-java17 and container-registry.oracle.com/graalvm/jdk:17-ol8 are failing
• ghcr.io/graalvm/graalvm-ce:ol7-java17 and container-registry.oracle.com/graalvm/jdk:17-ol7 are OK!

docker run --name graalvm-accented-test -it "container-registry.oracle.com/graalvm/jdk:17-ol7" /bin/bash
# ...
bash-4.2# javac GraalVMAccentedTest.java 
bash-4.2# java GraalVMAccentedTest
Could instantiate file àéïôù.txt

[fernando-valdez]

Thanks for reporting this. I will try to reproduce it on my side.

[Manokha]

@fernando-valdez Thanks :) Just updated the description (more details section) after trying different images, turns out it works with ol7 based images.

[Karm]

@fernando-valdez Speaking with CentOS/RHEL on mind, regardless of what you describe as locales for the native image, the host system actually have to have langpacks installed. I hit the issue back in the day, wrote this doc:

    /**
     * This test in Native won't work on a barebone system,
     * just with C.UTF-8 default fallback locale.
     *
     * For example, this package satisfies the dependency on a RHEL 9 type of OS:
     * glibc-all-langpacks
     *
     */

So I'd suggest checking glibc langpacks on those ol base images....

[Manokha]

Thanks @Karm I can confirm that running:

microdnf --nobest install glibc-all-langpacks

fixes it. No more "-encoding utf-8" needed for compilation, nor exception when running.

[fernando-valdez]

Thanks @Manokha for confirming. And thanks @Karm for your recommendation!