从JNI（C ++）代码向Java返回字符串

Question

Java Code:

public class MainClass 
{       
    static
    {
        System.load("/home/chronic/workspace/ramRead/src/libRamRead.so");
    }
    private native String readRam(int len, long addr, int pid);
    private native int readSize();

    public static void main(String[] args) 
    {
        MainClass app = new MainClass();


        String x = app.readRam(4096, 0x55c5520b5000L, 4435);

        //System.out.println("HEY THERE");
        //  System.out.println("I BEGIN HERE");
        System.out.println(x.length());

        //int test = app.readSize();
        //System.out.println(test);
    }
}

This is my native function (C++)

JNIEXPORT jstring JNICALL Java_ramRead_MainClass_readRam
  (JNIEnv *env, jobject, jint len, jlong addr, jint pid)
{
        struct iovec local[1];
        struct iovec remote[1];
        char buf[len];
        jstring result;
        //jcharArray buf = (*env).NewCharArray(len);
        //jcharArray buf1 = (*env).NewCharArray(len);

        local[0].iov_base = buf;
        local[0].iov_len = len;

        remote[0].iov_base = (void *) addr;
        remote[0].iov_len = len;

        nread = process_vm_readv(pid, local, 1, remote, 1, 0);

        //printf("len %d, pid %d, addr %li, buf %c, size of buf %d ", len, pid, addr, buf, sizeof(buf));
        printf("\nTHIS MUCH: %d\n", nread);

        for(int i=0; i<4096; i++){
                printf("%c", buf[i]);
        }

        //(*env).SetCharArrayRegion(buf1, 0, len, buf);
 result = (*env).NewStringUTF(buf);
        //printf("RESULT: %s", (*env).GetStringUTFLength(result));

        return result;
}

My goal:

Have the native function in C++ read the contents of ram and then return it as a string or as a char array to java for further processing.

Problem:

The return value of the jstring result is not what I expected it to be in java, there are several problems, first of all it has a length of 7 instead of 4096 as was intended. Antoher problem is the output is nonsensical. I have another program written in C++ which basically reads the contents of the RAM and then just prints them to a console. Here is the code bellow it works like a charm.I know from which address to read and I know what result to expect, and it does work without any problems so long as I am using C++ but as soon as I use JNI with the same methods before mentioned problems occur.

    #include <jni.h>
#include "ramRead_MainClass.h"
#include <sys/uio.h>

static ssize_t nread;

JNIEXPORT jbyteArray JNICALL Java_ramRead_MainClass_readRam
  (JNIEnv *env, jobject, jint len, jlong addr, jint pid)
//JNIEXPORT jbyteArray JNICALL Java_ramRead_MainClass_readRam
  //(JNIEnv *env, jobject, jobject len, jobject addr, jint pid)
{
        struct iovec local[1];
        struct iovec remote[1];
        jbyte buf[len];
        jbyteArray buf1 = (*env).NewByteArray(len);

        local[0].iov_base = buf;
        local[0].iov_len = len;

        remote[0].iov_base = (void *) addr;
        remote[0].iov_len = len;

        nread = process_vm_readv(pid, local, 1, remote, 1, 0);

        (*env).SetByteArrayRegion(buf1, 0, len, buf);
        return buf1;
}

JNIEXPORT jint JNICALL Java_ramRead_MainClass_readSize
  (JNIEnv *, jobject)
{
        return nread;
}

What I have tried thus far:

Check list 1. Not a permission problem 2. Ram address is correct and I know what is stored there. How do I know? I am reading RAM of a calculator, and I am looking for a specific set of numbers that I type in. With C++ I find them, but with jni not really. 3. Read more about UTF_8 and UTF_16 (there is a good chance the problem is here) 4. Seen posts here tried to reproduce the solutions, but to no avail.

OS: Linux 64 bit Fedora

Question:

How to convert a C++ string to a jstring and then return it in java and print it out so that the output would match the output of the C++ program listed above has.

Anyway this is how I intended to go about solving it, but if anyone has any better solutions or just functional solutions, I will more then gladly accept them.

Then you all for your time and effort.

Answer 1

JNI stores strings in modified utf-8 encoding (see bottom of the page).

If you want to keep using String, then you have to convert your bytes into this modified utf-8 encoding (so byte values >127 and zero have to be converted into 2 bytes).

But as Jorn suggested in the comments, you'll want to use ByteBuffer or a simple byte array ( byte[] ) instead, as it is more optimal: with String , you waste memory, as it contains 16-bit elements.

Answer 2

I managed to get it to work but it took some tinkering. Here is the code bellow which can be combined with the original code I have posted Many thx to all, I will try to up vote the answers.

In the main inside of Java code simply create this to convert the byte array into human readable chars. Keynote, not all chars will be properly converted but as far as my tests go all special chars such as !"£$%^&*()_+=-{}[]@~:;'#<>?,./ will be printed out no problems plus letters and numbers AZ az 0-9 anything else if you have some letters outside of English alphabet or something like that, you're likely to have problems.

public static void main(String[] args){
        MainClass app = new MainClass();

            byte[] x = app.readRam(4096, 0x556cbdb2a000L, 22444);

            System.out.println(x.length);
            System.out.write(x, 0, x.length);

            for(int i=0; i<x.length; i++)
            {   
                if((int)x[i] >= 32 && (int)x[i] < 127)
                {
                    System.out.print((char)x[i]);
                }
            }

            char c = 120;
            System.out.println("\n" + c);

    }