Summary
pypdf: Possible large memory usage when retrieving Roman page labels
Advisory details
Impact
An attacker who uses this vulnerability can craft a PDF which leads to large memory consumption. This requires accessing the page labels of a document with large Roman numerals.
Patches
This has been fixed in pypdf==6.17.0.
Workarounds
If you cannot upgrade yet, consider applying the changes from PR #4047.
References
- https://github.com/advisories/GHSA-qv6h-rv94-w285
- https://github.com/py-pdf/pypdf/security/advisories/GHSA-qv6h-rv94-w285
- https://nvd.nist.gov/vuln/detail/CVE-2026-102993
- https://github.com/py-pdf/pypdf/pull/4047
- https://github.com/py-pdf/pypdf/commit/89db7c4fe9315ecc964bfdf05a4e8c4b94175163
- https://github.com/py-pdf/pypdf/releases/tag/6.17.0
Related vulnerabilities
All Supply chain →- MEDIUMCVE-2026-100350
kin-openapi deepObject query decoding repeatedly compiles an invariant regular expression
- MEDIUMCVE-2026-59951
MCP Python SDK: Unbounded Session Accumulation in StreamableHTTP Transport Enables Resource Exhaustion
- HIGHCVE-2026-72923
Microsoft.OpenApi.YamlReader vulnerable to denial of service via YAML alias expansion
- HIGHCVE-2026-91979
Vikunja: Denial of service via decompression bomb in the data import
- MEDIUMCVE-2026-107294
Pydantic AI: Unbounded memory use when downloading remote content via web_fetch or FileUrl
- HIGHCVE-2026-89425
jackson-core: UTF8DataInputJsonParser._reportInvalidToken() missing maxErrorTokenLength limit -> unbounded StringBuilder growth (DoS)