OpenCC has an Out-of-bounds read when processing truncated UTF-8 inputGHSA-7fqq-q52p-2jjg
MEDIUMFix: BYVoid/OpenCC#1005GHSA-7fqq-q52p-2jjg is a medium-severity (CVSS 6.5) remote code execution vulnerability in opencc. A fix is available for opencc — see the affected versions and patch details below.
Real-World Exposure
How broadly this vulnerability is actually deployed: weekly install volume shows current usage, and reverse-dependency count shows how many other packages break if it stays unpatched.
openccnpmDescription
Summary
OpenCC versions before 1.2.0 contain two CWE-125: Out-of-bounds Read issues caused by length validation failures in UTF-8 processing. When handling malformed or truncated UTF-8 input, OpenCC trusted derived length values without enforcing the invariant that processed length must not exceed the remaining input buffer. This could result in out-of-bounds reads during segmentation or conversion.
Details
Two independent code paths in OpenCC failed to enforce the invariant:
matchedLength <= remainingLength
Both paths assumed derived length values were valid and within input bounds, but did not validate that assumption against the remaining buffer. This created the following failure chain:
invalid UTF-8 -> incorrect derived length -> incorrect pointer advance -> remaining-length desynchronization -> out-of-bounds read
In MaxMatchSegmentation::Segment, this could desynchronize remaining-length tracking and cause out-of-bounds reads during prefix matching.
In Conversion::Convert(const char*), similar logic could advance processing past the end of the input string and read beyond the null terminator into adjacent memory. In some cases, unintended heap bytes could be propagated into the conversion result.
PR #1005 fixes both issues by explicitly tracking input boundaries, recomputing remaining length on each iteration, and clamping processed lengths so the buffer-bound invariant is preserved.
Affected versions:
- All versions before 1.2.0
Patched version:
- 1.2.0
PoC
Build a vulnerable version with AddressSanitizer enabled and process input ending with a truncated UTF-8 sequence, such as a missing final byte of a 3-byte character. The original report and ASan reproduction are available in Issue #997.
Impact
This vulnerability may cause process crashes and limited, non-deterministic information disclosure when OpenCC processes malformed or attacker-controlled UTF-8 input. The issue does not indicate arbitrary write or code execution.
OpenCC is distributed through system and language-specific package managers, prebuilt binaries, container images, and downstream software, so affected versions may be present even when it is not listed as a direct dependency. Users should upgrade all installed or bundled copies of OpenCC to 1.2.0 or later.
Credit
OpenCC thanks @oneafter for reporting the issue.
Affected Packages
| Ecosystem | Package | Vulnerable range | Fix |
|---|---|---|---|
| 📦npm | opencc | all versions | 1.2.0npm install opencc@1.2.0 |
| 🐍PyPI | opencc | all versions | 1.2.0pip install --upgrade 'opencc==1.2.0' |
Detection & mitigation playbook
Open-source dependencyDetect
Scan your dependency tree (package-lock.json, pnpm-lock.yaml, requirements.txt, go.sum, etc.) for opencc, including transitive dependencies — a direct dependency you never call can still pull in a vulnerable version.
Fix
Update opencc to 1.2.0 or later, then make sure no transitive (indirect) dependency still pins the vulnerable range — O3 confirms GHSA-7fqq-q52p-2jjg is resolved across your whole dependency graph.
Workarounds
Stop feeding it untrusted input: reject or quarantine files and payloads from unverified sources until you can upgrade, restrict accepted formats to the ones you actually need, and run the parsing or decoding step in a least-privileged sandbox or short-lived worker so a crash or corrupted read cannot reach the rest of the process.
Frequently Asked Questions
Is GHSA-7fqq-q52p-2jjg in your dependencies?
Find it across npm, PyPI, including transitive dependencies.