[Bug] Bytes.getMD5(InputStream/File) returns incorrect MD5 for short final reads #16451

Pre-check

• I am sure that all the content I provide is in English.

Search before asking

• I had searched in the issues and found no similar issues.

Apache Dubbo Component

Java SDK (apache/dubbo)

Dubbo Version

Apache Dubbo 3.3 branch, JDK 8 or later, any operating system.

Steps to reproduce this issue

Bytes.getMD5(InputStream) uses InputStream.available() as an EOF check and updates the message digest with the entire buffer instead of the number of bytes actually read.

For example, the bundled md5.testfile.txt contains only:

hello world!

Calling Bytes.getMD5(File) returns the Base64 value:

iNZ+5qHafVNPLJxHwLKJ3w==

The correct MD5 of hello world! is:

/D/5joxqDTCH1RXARz+Gdw==

The same problem occurs for an InputStream whose final read is shorter than 8192 bytes.

What you expected to happen

Bytes.getMD5(InputStream) and Bytes.getMD5(File) should calculate the MD5 of exactly the bytes supplied by the input source.

Anything else

The implementation should read until read() returns -1 and update the digest with only the bytes returned by each read:

int read;
while ((read = is.read(buf)) != -1) {
    md.update(buf, 0, read);
}

A pull request is prepared with tests for 12, 8191, 8192, and 8193 byte inputs, plus an input stream whose available() method returns zero while data is still readable.

Do you have a (mini) reproduction demo?

• Yes, I have a minimal reproduction demo to help resolve this issue more effectively!

Are you willing to submit a pull request to fix on your own?

• Yes I am willing to submit a pull request on my own!

Code of Conduct

• I agree to follow this project's Code of Conduct