fix: fail on truncated bodies instead of NUL-padding them

[predic8]

ByteUtil.readByteArray() returned its pre-allocated buffer when the stream ended before the requested number of bytes had been read, so the zero padding was indistinguishable from real content.

On the Content-Length path (Body.readLocal) this produced no error at all: the body reported isRead() == true with its content padded to the announced length, and the truncated message was forwarded. The chunked reader handed the padded chunk to the observers and to the consumer, then failed one chunk later with the misleading "Empty chunk-size field"; ChunkedBody.writeStreamed() had already written it to the peer. Frame.read() padded HTTP/2 frame payloads the same way, although Frame.readByte() already threw EOFException for the frame header.

readByteArray() now throws EOFException on a short read. As it extends IOException, every caller's existing handling turns it into a ReadingBodyException without a signature change. Reading until the end of the stream (length < 0) is unaffected.

Note this changes a path that previously appeared to work: a truncated Content-Length body is now a hard body-read failure rather than silently padded.

Closes #3193

Summary by CodeRabbit

• Bug Fixes
  • Incomplete fixed-length HTTP request and response bodies are now rejected instead of being treated as complete.
  • Truncated chunked transfers now fail safely without delivering partial data to consumers.
  • Complete bodies continue to be processed successfully, including data received across multiple short reads.
• Tests
  • Added coverage for truncated, complete, and fragmented body-reading scenarios.

[coderabbitai]

No actionable comments were generated in the recent review. 🎉

ℹ️ Recent review info

⚙️ Run configuration

Configuration used: Path: .coderabbit.yaml

Review profile: CHILL

Plan: Team

Run ID: 678a289d-9769-461f-97f2-1c6b1ace6a1d

📥 Commits

Reviewing files that changed from the base of the PR and between cfe7c76 and 7f1ae0d.

📒 Files selected for processing (4)

• core/src/main/java/com/predic8/membrane/core/util/ByteUtil.java
• core/src/test/java/com/predic8/membrane/core/http/BodyTest.java
• core/src/test/java/com/predic8/membrane/core/http/ChunkedBodyTest.java
• core/src/test/java/com/predic8/membrane/core/util/ByteUtilTest.java

Included review availability: Your plan provides up to 4 included reviews per hour; 2 remain after this review.

---

📝 Walkthrough

Walkthrough

readByteArray now enforces fixed-length reads and throws EOFException on premature EOF. Tests cover short reads, complete bodies, truncated fixed-length bodies, and truncated chunked bodies without partial delivery.

Changes

Body read integrity

Layer / File(s)	Summary
Exact-length read contract core/src/main/java/com/predic8/membrane/core/util/ByteUtil.java, core/src/test/java/com/predic8/membrane/core/util/ByteUtilTest.java	readByteArray now accumulates short reads and throws EOFException when fixed-length input ends early. Tests cover both behaviors.
HTTP body truncation handling core/src/test/java/com/predic8/membrane/core/http/BodyTest.java, core/src/test/java/com/predic8/membrane/core/http/ChunkedBodyTest.java	Tests verify that complete bodies remain successful and truncated bodies fail without marking them read or delivering partial data.

Estimated code review effort: 2 (Simple) | ~10 minutes

Merge Risk: ⚪ Minimal · up to 7f1ae

The change rejects truncated HTTP and HTTP/2 content instead of forwarding zero-padded data, with existing error handling preserved. No actionable merge-blocking risk remains beyond normal checks and review.

Poem

A rabbit reads each byte in line
No padded tails shall cross the sign
Short streams now raise EOF with care
Whole chunks travel everywhere
The burrow cheers for exact reads fine

🚥 Pre-merge checks | ✅ 4 | ❌ 1

❌ Failed checks (1 warning)

Check name	Status	Explanation	Resolution
Docstring Coverage	⚠️ Warning	Docstring coverage is 25.00% which is insufficient. The required threshold is 80.00%. Docstring coverage is scoped to functions touched by this diff. Analyzed 16 functions across 4 files.	Write docstrings for the functions missing them to satisfy the coverage threshold.

✅ Passed checks (4 passed)

Check name	Status	Explanation
Description Check	✅ Passed	Check skipped - CodeRabbit’s high-level summary is enabled.
Title check	✅ Passed	The title clearly and concisely describes the main fix: truncated bodies now fail instead of receiving NUL padding.
Linked Issues check	✅ Passed	The changes satisfy issue #3193. Fixed-length reads now fail with EOFException on premature EOF, existing IOException handling remains compatible, negative-length reads remain unchanged, and regressio…
Out of Scope Changes check	✅ Passed	All changes are directly related to issue #3193. The production change fixes truncated reads, and the added tests validate the affected fixed-length, chunked, and utility behavior. No unrelated code c…

Full details: Linked Issues check

Explanation

The changes satisfy issue #3193. Fixed-length reads now fail with EOFException on premature EOF, existing IOException handling remains compatible, negative-length reads remain unchanged, and regression tests cover Content-Length and chunked-body behavior. The shared ByteUtil fix also applies to HTTP/2 payload reads.

Full details: Out of Scope Changes check

Explanation

All changes are directly related to issue #3193. The production change fixes truncated reads, and the added tests validate the affected fixed-length, chunked, and utility behavior. No unrelated code changes are present.

• Fix all pre-merge checks with AI

✨ Finishing Touches 💡 1

📝 Generate docstrings 💡

• Create stacked PR
• Commit on current branch

🧪 Generate unit tests (beta)

• Create PR with unit tests
• Commit unit tests in branch fix/truncated-body-zero-padding

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

_{Comment @coderabbitai help to get the list of available commands.}