GHSA-Q43M-VHCP-MHVM

Vulnerability from github – Published: 2026-10-06 00:02 – Updated: 2026-10-06 00:02
VLAI
Summary
Docling: `enable_local_fetch` is not enforced in HTML browser-rendering mode
Details

Summary

When the HTML backend renders pages in a headless browser (HTMLBackendOptions(render_page=True)), the enable_local_fetch option is not enforced. A crafted HTML file can embed an arbitrary local file (for example with <iframe src="file:///...">), and that file's contents appear in the page image attached to the returned DoclingDocument.

Details

In render mode, Playwright requests are filtered by HTMLDocumentBackend._get_browser_request_block_reason. In affected versions, this check allowed file: URLs unconditionally, before reading any option. As a result:

  • enable_local_fetch=False did not block local file access, and
  • even with enable_local_fetch=True, file access was not limited to the source document's directory, unlike the non-render path (ImageResourceLoader), which rejects absolute paths and path traversal.

Versions 2.82.0–2.90.x did no request filtering in render mode at all.

The browser runs with JavaScript disabled (from 2.91.0), so disclosure is passive: only what Chromium renders visibly inside the page viewport ends up in the page image.

Only Path inputs are affected. They are loaded through a file:// URL. Stream inputs are loaded with page.set_content() into an opaque origin, from which Chromium does not load file:// subresources.

Impact

An attacker who can submit HTML for conversion can read any text file the conversion process can read (for example .env files, credential files, or other users' documents on a shared host) by having it rendered into the page image.

Only applications that meet all of these conditions are affected:

  • they set HTMLBackendOptions(render_page=True) in Python,
  • they have the optional playwright dependency installed, and
  • they pass untrusted HTML as a filesystem Path.

The following are not affected: the default configuration (render_page=False), the docling CLI, docling-serve, and the EPUB, Markdown, XBRL and email backends.

Patches

Fixed in 2.118.1 (#3948). In render mode, file: requests are now blocked unless enable_local_fetch=True, and allowed requests are limited to the source document's directory.

Workarounds

If you can't upgrade, don't use render_page=True on untrusted HTML, or pass the input as a stream instead of a Path.

Credits

Reported by @priyankn.

Severity
5.9 (Medium)
CVSS:3.1/AV:N/AC:H/PR:N/UI:N/S:U/C:H/I:N/A:N
Show details on source website
To clipboard 

{
  "affected": [
    {
      "package": {
        "ecosystem": "PyPI",
        "name": "docling"
      },
      "ranges": [
        {
          "events": [
            {
              "introduced": "2.82.0"
            },
            {
              "fixed": "2.118.1"
            }
          ],
          "type": "ECOSYSTEM"
        }
      ]
    },
    {
      "package": {
        "ecosystem": "PyPI",
        "name": "docling-slim"
      },
      "ranges": [
        {
          "events": [
            {
              "introduced": "2.92.0"
            },
            {
              "fixed": "2.118.1"
            }
          ],
          "type": "ECOSYSTEM"
        }
      ]
    }
  ],
  "aliases": [
    "CVE-2026-105750"
  ],
  "database_specific": {
    "cwe_ids": [
      "CWE-552",
      "CWE-863"
    ],
    "github_reviewed": true,
    "github_reviewed_at": "2026-10-06T00:02:27Z",
    "nvd_published_at": null,
    "severity": "MODERATE"
  },
  "details": "### Summary\n\nWhen the HTML backend renders pages in a headless browser (`HTMLBackendOptions(render_page=True)`), the `enable_local_fetch` option is not enforced. A crafted HTML file can embed an arbitrary local file (for example with `\u003ciframe src=\"file:///...\"\u003e`), and that file\u0027s contents appear in the page image attached to the returned `DoclingDocument`.\n\n### Details\n\nIn render mode, Playwright requests are filtered by `HTMLDocumentBackend._get_browser_request_block_reason`. In affected versions, this check allowed `file:` URLs unconditionally, before reading any option. As a result:\n\n- `enable_local_fetch=False` did not block local file access, and\n- even with `enable_local_fetch=True`, file access was not limited to the source document\u0027s directory, unlike the non-render path (`ImageResourceLoader`), which rejects absolute paths and path traversal.\n\nVersions 2.82.0\u20132.90.x did no request filtering in render mode at all.\n\nThe browser runs with JavaScript disabled (from 2.91.0), so disclosure is passive: only what Chromium renders visibly inside the page viewport ends up in the page image.\n\nOnly `Path` inputs are affected. They are loaded through a `file://` URL. Stream inputs are loaded with `page.set_content()` into an opaque origin, from which Chromium does not load `file://` subresources.\n\n### Impact\n\nAn attacker who can submit HTML for conversion can read any text file the conversion process can read (for example `.env` files, credential files, or other users\u0027 documents on a shared host) by having it rendered into the page image.\n\nOnly applications that meet **all** of these conditions are affected:\n\n- they set `HTMLBackendOptions(render_page=True)` in Python,\n- they have the optional `playwright` dependency installed, and\n- they pass untrusted HTML as a filesystem `Path`.\n\nThe following are **not** affected: the default configuration (`render_page=False`), the `docling` CLI, `docling-serve`, and the EPUB, Markdown, XBRL and email backends.\n\n### Patches\n\nFixed in **2.118.1** (#3948). In render mode, `file:` requests are now blocked unless `enable_local_fetch=True`, and allowed requests are limited to the source document\u0027s directory.\n\n### Workarounds\n\nIf you can\u0027t upgrade, don\u0027t use `render_page=True` on untrusted HTML, or pass the input as a stream instead of a `Path`.\n\n### Credits\n\nReported by @priyankn.",
  "id": "GHSA-q43m-vhcp-mhvm",
  "modified": "2026-10-06T00:02:27Z",
  "published": "2026-10-06T00:02:27Z",
  "references": [
    {
      "type": "WEB",
      "url": "https://github.com/docling-project/docling/security/advisories/GHSA-q43m-vhcp-mhvm"
    },
    {
      "type": "WEB",
      "url": "https://github.com/docling-project/docling/pull/3948"
    },
    {
      "type": "WEB",
      "url": "https://github.com/docling-project/docling/commit/1612b8875b0937447ce3122536fb5360a7102a0a"
    },
    {
      "type": "PACKAGE",
      "url": "https://github.com/docling-project/docling"
    },
    {
      "type": "WEB",
      "url": "https://github.com/docling-project/docling/releases/tag/v2.118.1"
    }
  ],
  "schema_version": "1.4.0",
  "severity": [
    {
      "score": "CVSS:3.1/AV:N/AC:H/PR:N/UI:N/S:U/C:H/I:N/A:N",
      "type": "CVSS_V3"
    }
  ],
  "summary": "Docling: `enable_local_fetch` is not enforced in HTML browser-rendering mode"
}



Log in or create an account to share your comment.




Tags
Taxonomy of the tags.





Forecast uses a logistic model when the trend is rising, or an exponential decay model when the trend is falling. Fitted via linearized least squares.

Sightings

Author Source Type Date Other

Nomenclature

  • Seen: The vulnerability was mentioned, discussed, or observed by the user.
  • Confirmed: The vulnerability has been validated from an analyst's perspective.
  • Published Proof of Concept: A public proof of concept is available for this vulnerability.
  • Exploited: The vulnerability was observed as exploited by the user who reported the sighting.
  • Patched: The vulnerability was observed as successfully patched by the user who reported the sighting.
  • Not exploited: The vulnerability was not observed as exploited by the user who reported the sighting.
  • Not confirmed: The user expressed doubt about the validity of the vulnerability.
  • Not patched: The vulnerability was not observed as successfully patched by the user who reported the sighting.



Related by attack behaviour

Vulnerabilities whose description is nearest to this one in the vector space of the CIRCL/vulnerability-attack-technique-biencoder model. This is a similarity search over the bi-encoder space (plain cosine), not a classification, and it has no measured accuracy.