Zero Radar
← feed
HighVulnerabilityCVE-2026-119204updated 30 Oct 2026, 23:13 UTC

pypdf can use excessive memory when parsing a crafted PDF

A specially crafted PDF file can make the pypdf Python library use a huge amount of memory, which can slow down or crash the program reading it. Opening untrusted PDFs is the main risk.

Fixed in pypdf 6.21.4 · out Official, 2 sources
SeverityHighCVSS 7.1 of 10
Being exploited?UnclearNo reports either way yet
Fix available?Yespypdf 6.21.4 · since 30 Oct 2026
How sure is this?Official, 2 sourcesBacked by an official source

What happened

pypdf versions before 6.21.4 do not limit the total work when flattening a PDF page tree with many shared resources, so crafted resource structures cause disproportionate memory allocations.

Why it is dangerous: Services or scripts that parse untrusted PDFs with pypdf can run out of memory or become unresponsive, causing denial of service.

Are you affected?

You are affected if you run pypdf earlier than 6.21.4 and parse PDF files that may come from untrusted sources.

Package / productAffected versionsFixed in
pypdf · pip< 6.21.46.21.4

What to do

  1. Upgrade pypdf to 6.21.4 or later (pip install --upgrade pypdf)
  2. If you cannot upgrade yet, apply the changes from PR #4290 in the pypdf repository
  3. Limit memory for processes that parse untrusted PDFs until patched

Timeline

  1. 29 Oct 2026, 21:17 UTC
    Publicly disclosed
  2. 29 Oct 2026, 21:17 UTC
    Reported by NVD · NVD
  3. 29 Oct 2026, 21:21 UTC
    Picked up by Zero Radar
  4. 30 Oct 2026, 23:13 UTC
    GitHub Advisory reported on it · GitHub Advisory
  5. 30 Oct 2026, 23:13 UTC
    Fixed version published: pypdf 6.21.4 · GitHub Advisory

Sources 2 reports, oldest first

Official vendor, government CERT or advisory database

Technical details

ID
CVE-2026-119204 · GHSA-cggf-277g-3q54
CVSS
7.1 / 10
Disclosed
29 Oct 2026, 21:17 UTC
On Zero Radar
29 Oct 2026, 21:21 UTC (4 min after disclosure)
Last change
30 Oct 2026, 23:13 UTC

The plain-language parts and the attack diagram were written by an AI model (anthropic/claude-haiku-5.5) from the sources above. Facts like versions, exploitation and dates come from the sources themselves. How Zero Radar works

pypdf can use excessive memory when parsing a crafted PDF · Zero Radar