Wiz rejoint Google Cloud : Faire de la magie ensemble

CVE-2026-132277: 
Python Analyse et atténuation de la vulnérabilité

Aperçu

CVE-2026-132277 is a quadratic memory growth (denial of service) vulnerability in Inscriptis, a Python library for converting HTML to text. Affecting versions 2.0.0 through 2.7.5, the flaw allows an unauthenticated remote attacker to exhaust server memory by supplying a small HTML document containing multiple sibling <table> elements. The vulnerability was discovered by researcher gistrec and disclosed via a GitHub Security Advisory on October 23, 2026, with a fix released in version 2.7.6 (GitHub Advisory). It carries a CVSS v3.1 base score of 5.3 (Medium) (GitHub Advisory).

Détails techniques

The root cause is an unbalanced prefix stack in the HTML-to-text rendering pipeline, classified as CWE-400 (Uncontrolled Resource Consumption) and CWE-407 (Inefficient Algorithmic Complexity). Specifically, table_end_handler() in model/tag/table_tag.py renders each sibling table via HtmlElement.write_verbatim_text(), which calls Canvas.open_block() — pushing the parent's padding_inline prefix onto the stack. However, Canvas.close_block() never pops this prefix; only Canvas.close_tag() calls prefix.remove_last_prefix(). As a result, each successive sibling table inherits all previously accumulated indentation prefixes, causing get_text() output size and memory consumption to grow quadratically with the number of tables. Benchmarks show that a ~240 KB HTML document with 16,000 sibling tables produces approximately 1 billion characters of output and consumes ~1 GB of RAM (GitHub Advisory, Fix Commit). The bug has been present since version 2.0.0 (commit 84ec720) and requires no authentication or user interaction to trigger over the network.

Impact

Successful exploitation causes denial of service through memory exhaustion in any application that passes untrusted HTML to Inscriptis's get_text() function. Affected use cases include web crawlers, RAG (Retrieval-Augmented Generation) and LLM ingestion pipelines, and the inscriptis-api service. There is no confidentiality or integrity impact; the vulnerability is limited to availability, but a single small crafted request can render the host application unresponsive or crash it entirely (GitHub Advisory).

Exploitabilité

A proof-of-concept is publicly documented in the GitHub Security Advisory, demonstrating that a Python one-liner generating a small HTML payload with many sibling <table> elements is sufficient to trigger the vulnerability (GitHub Advisory). No authentication or user interaction is required, and the attack vector is network-accessible. The EPSS score is 0.0 as of the disclosure date, and there is no evidence of in-the-wild exploitation or CISA KEV catalog inclusion at this time (Feedly).

Étapes d’exploitation

  1. Identify target: Locate an application that accepts external HTML input and processes it with Inscriptis versions 2.0.0–2.7.5 (e.g., a web crawler API, an LLM ingestion endpoint, or inscriptis-api).
  2. Craft malicious payload: Generate a small HTML document containing a large number of sibling <table> elements inside a block-level parent, e.g.:
html = "<div>" + "<table><tr><td>a</td></tr></table>" * N + "</div>"

where N is tuned to exhaust the target's available memory (e.g., N=16,000 produces ~1 GB output from a ~240 KB input). 3. Submit payload: Send the crafted HTML to the target application via its normal input channel (HTTP POST, file upload, API call, etc.). 4. Trigger memory exhaustion: The application calls get_text(html), which accumulates indentation prefixes quadratically, exhausting available memory and causing the process to crash or become unresponsive (GitHub Advisory).

Indicateurs de compromis

  • Network: Unusual HTTP requests submitting HTML payloads containing a large number of repeated <table> tags to HTML-processing endpoints.
  • Process: Rapid, sustained memory growth in Python processes running Inscriptis; OOM (Out of Memory) killer events targeting the application process.
  • Logs: Application logs showing MemoryError exceptions or process crashes originating from inscriptis.get_text() calls; abnormally long processing times for small HTML inputs.
  • System: Sudden spikes in system memory usage correlated with incoming HTML conversion requests; swap exhaustion or system-level OOM events (GitHub Advisory).

Atténuation et solutions de contournement

Upgrade Inscriptis to version 2.7.6 or later, which fixes the issue by calling prefix.remove_last_prefix() in table_end_handler() after the block prefix has been used, preventing prefix accumulation across sibling tables (GitHub Release, Fix Commit). As a short-term workaround where upgrading is not immediately possible, applications should validate or sanitize untrusted HTML input before passing it to Inscriptis — for example, by limiting the number of <table> elements or overall document size. Rate limiting and memory limits on conversion processes can also reduce the blast radius of exploitation.

Réactions de la communauté

The vulnerability was reported and fixed by community contributor gistrec, who also provided the regression test suite included in the 2.7.6 release. The Inscriptis maintainer (AlbertWeichselbraun) acknowledged the issue and credited gistrec in the release notes, noting that users of versions 2.0.0–2.7.5 are encouraged to upgrade (GitHub Release). No broader media coverage or significant social media discussion has been observed beyond the GitHub advisory.

Ressources supplémentaires


Source: Ce rapport a été généré à l’aide de l’IA

Connexe Python Vulnérabilités :

CVE ID

Gravité

Partition

Technologies

Nom du composant

Exploit CISA KEV

A corrigé

Date de publication

CVE-2026-132275HIGH7.5
  • Python logoPython
  • svglib
NonOuiNov 20, 2026
CVE-2026-132276MEDIUM5.9
  • Python logoPython
  • wcmatch
NonOuiNov 20, 2026
CVE-2026-132277MEDIUM5.3
  • Python logoPython
  • inscriptis
NonOuiNov 20, 2026
CVE-2026-77332LOW2.3
  • Python logoPython
  • memtomem-stm
NonOuiNov 20, 2026
CVE-2026-77331LOW2
  • Python logoPython
  • memtomem-stm
NonOuiNov 20, 2026

Évaluation gratuite des vulnérabilités

Évaluez votre posture de sécurité cloud

Évaluez vos pratiques de sécurité cloud dans 9 domaines de sécurité afin d’évaluer votre niveau de risque et d’identifier les lacunes dans vos défenses.

Évaluation de la demande

Ressources Wiz supplémentaires

Obtenez une démo personnalisée

Prêt(e) à voir Wiz en action ?

"La meilleure expérience utilisateur que j’ai jamais vue, offre une visibilité totale sur les workloads cloud."
David EstlickRSSI
"Wiz fournit une interface unique pour voir ce qui se passe dans nos environnements cloud."
Adam FletcherChef du service de sécurité
"Nous savons que si Wiz identifie quelque chose comme critique, c’est qu’il l’est réellement."
Greg PoniatowskiResponsable de la gestion des menaces et des vulnérabilités