GHSA-24MQ-4VC6-2PRF

Vulnerability from github – Published: 2026-11-16 22:12 – Updated: 2026-11-16 22:12
VLAI
Summary
mail-parser: Uncaught exceptions and silent header loss in mail-parser on attacker-controlled input
Details

Impact

Two effects, sharing the root cause of Advisory 1 — header names chosen by the sender were resolved through Python attribute lookup and rewritten before lookup.

Uncaught exceptions. A crafted message raises an exception that is outside the MailParser* hierarchy, so a caller catching only mail-parser exceptions does not catch it and the worker dies.

Header named after a method (CWE-248, reported by @iam-niranjan). Python resolves real class attributes before __getattr__, so a header named after a method put a bound method object into the parsed mail dict. A 12-byte email is sufficient. Nine colliding names were confirmed: Parse, Message, Write_Attachments, From_File, From_File_Obj, From_File_Msg, From_String, From_Bytes, Get_Server_Ipaddress. Also reproducible from the CLI (mail-parser -f evil.eml -j).

import mailparser
mailparser.parse_from_string("Parse: x\r\n\r\n").mail_json
# 4.6.1 -> TypeError: Object of type method is not JSON serializable
# 4.6.2 -> {"parse": "x", "has_defects": false}

Header carrying 8-bit bytes (CWE-248). For a header containing raw 8-bit bytes, Python's compat32 policy returns an email.header.Header rather than a str, which the _raw accessor passed straight to json.dumps. This fires on entirely benign non-ASCII mail via parse_from_bytes(), the normal entry point for mail off the wire.

import mailparser
mailparser.parse_from_bytes(b"Subject: caf\xe9\r\n\r\nbody\r\n").subject_raw
# 4.6.1 -> TypeError: Object of type Header is not JSON serializable
# 4.6.2 -> '["caf�"]'

Silent loss of message content (CWE-436). A sender can make a header's value vanish from every output surface, or be replaced by a different header's value, with has_defects remaining false. This is an evasion primitive: the tool's output diverges from the bytes actually in the message, and a pipeline scanning parsed output for indicators sees nothing and no failure signal.

Header names were normalised before lookup — underscores folded to dashes, and _json / _raw suffixes interpreted — so a header genuinely named Subject_json was reported with the value of Subject and its own value discarded, and any header whose name contains an underscore resolved to nothing.

import mailparser

raw = ("From: Real Sender <real@good.example>\r\n"
       "Subject: innocuous\r\n"
       "Subject_json: MALWARE-C2 http://evil.example/payload.exe\r\n"
       "X_Spam_Flag: YES\r\n\r\nbody\r\n")
mail = mailparser.parse_from_string(raw)
print(mail.mail["subject_json"])       # 4.6.1 -> '"innocuous"'   4.6.2 -> the real value
print(mail.headers["X_Spam_Flag"])     # 4.6.1 -> ''              4.6.2 -> 'YES'
print("MALWARE-C2" in mail.mail_json)  # 4.6.1 -> False           4.6.2 -> True
print(mail.has_defects)                # both  -> False

A related case: email.message.Message.__len__ returns the header count, so a message with no headers is falsy. parse() returned before initialising state, and the unset attributes were then absorbed by __getattr__ and answered as absent headers. Every property returned '' — including mail and attachments, which are documented to return a dict and a list — with no error and no defect flag.

import mailparser
mail = mailparser.parse_from_string("Click http://evil.example/pay.exe now\r\n")
print(repr(mail.body))         # 4.6.1 -> ''    4.6.2 -> the text
print(repr(mail.attachments))  # 4.6.1 -> ''    4.6.2 -> []
print(mail.has_defects)        # 4.6.1 -> ''    4.6.2 -> True

Patches

Fixed in 4.6.2. Upgrade with pip install --upgrade mail-parser.

Header names taken from a message are now looked up literally — no attribute access, no case or separator folding, no suffix interpretation — so no header name can collide with a method or property, and no value can be substituted or dropped. 8-bit values are coerced to str before serialisation. Parser state is initialised before the message check, which now tests is None, and attribute names beginning with _ raise AttributeError instead of being answered as headers.

Behaviour changes. Header names now appear in mail and headers exactly as sent, so a header named X_Spam_Flag keeps that name and its own value. X_raw always returns a JSON array — an absent header yields [] where 4.6.1 returned null.

Workarounds

None for the data loss. For the uncaught exceptions, callers who cannot upgrade should wrap mail_json, mail_partial_json, headers_json and any X_raw access in a bare except Exception rather than catching only MailParserError subclasses.

Show details on source website

{
  "affected": [
    {
      "database_specific": {
        "last_known_affected_version_range": "\u003c= 4.6.1"
      },
      "package": {
        "ecosystem": "PyPI",
        "name": "mail-parser"
      },
      "ranges": [
        {
          "events": [
            {
              "introduced": "0"
            },
            {
              "fixed": "4.6.2"
            }
          ],
          "type": "ECOSYSTEM"
        }
      ]
    }
  ],
  "aliases": [
    "CVE-2026-129299"
  ],
  "database_specific": {
    "cwe_ids": [
      "CWE-248",
      "CWE-436"
    ],
    "github_reviewed": true,
    "github_reviewed_at": "2026-11-16T22:12:51Z",
    "nvd_published_at": "2026-11-16T18:17:30Z",
    "severity": "MODERATE"
  },
  "details": "### Impact\n\nTwo effects, sharing the root cause of Advisory 1 \u2014 header names chosen by the sender were\nresolved through Python attribute lookup and rewritten before lookup.\n\n**Uncaught exceptions.** A crafted message raises an exception that is *outside* the\n`MailParser*` hierarchy, so a caller catching only mail-parser exceptions does not catch it\nand the worker dies.\n\n*Header named after a method* (CWE-248, reported by @iam-niranjan). Python resolves real\nclass attributes before `__getattr__`, so a header named after a method put a bound method\nobject into the parsed mail dict. A 12-byte email is sufficient. Nine colliding names were\nconfirmed: `Parse`, `Message`, `Write_Attachments`, `From_File`, `From_File_Obj`,\n`From_File_Msg`, `From_String`, `From_Bytes`, `Get_Server_Ipaddress`. Also reproducible from\nthe CLI (`mail-parser -f evil.eml -j`).\n\n```python\nimport mailparser\nmailparser.parse_from_string(\"Parse: x\\r\\n\\r\\n\").mail_json\n# 4.6.1 -\u003e TypeError: Object of type method is not JSON serializable\n# 4.6.2 -\u003e {\"parse\": \"x\", \"has_defects\": false}\n```\n\n*Header carrying 8-bit bytes* (CWE-248). For a header containing raw 8-bit bytes, Python\u0027s\n`compat32` policy returns an `email.header.Header` rather than a `str`, which the `_raw`\naccessor passed straight to `json.dumps`. This fires on entirely benign non-ASCII mail via\n`parse_from_bytes()`, the normal entry point for mail off the wire.\n\n```python\nimport mailparser\nmailparser.parse_from_bytes(b\"Subject: caf\\xe9\\r\\n\\r\\nbody\\r\\n\").subject_raw\n# 4.6.1 -\u003e TypeError: Object of type Header is not JSON serializable\n# 4.6.2 -\u003e \u0027[\"caf\ufffd\"]\u0027\n```\n\n**Silent loss of message content** (CWE-436). A sender can make a header\u0027s value vanish from\nevery output surface, or be replaced by a different header\u0027s value, with `has_defects`\nremaining `false`. This is an evasion primitive: the tool\u0027s output diverges from the bytes\nactually in the message, and a pipeline scanning parsed output for indicators sees nothing\nand no failure signal.\n\nHeader names were normalised before lookup \u2014 underscores folded to dashes, and `_json` /\n`_raw` suffixes interpreted \u2014 so a header genuinely named `Subject_json` was reported with\nthe value of `Subject` and its own value discarded, and any header whose name contains an\nunderscore resolved to nothing.\n\n```python\nimport mailparser\n\nraw = (\"From: Real Sender \u003creal@good.example\u003e\\r\\n\"\n       \"Subject: innocuous\\r\\n\"\n       \"Subject_json: MALWARE-C2 http://evil.example/payload.exe\\r\\n\"\n       \"X_Spam_Flag: YES\\r\\n\\r\\nbody\\r\\n\")\nmail = mailparser.parse_from_string(raw)\nprint(mail.mail[\"subject_json\"])       # 4.6.1 -\u003e \u0027\"innocuous\"\u0027   4.6.2 -\u003e the real value\nprint(mail.headers[\"X_Spam_Flag\"])     # 4.6.1 -\u003e \u0027\u0027              4.6.2 -\u003e \u0027YES\u0027\nprint(\"MALWARE-C2\" in mail.mail_json)  # 4.6.1 -\u003e False           4.6.2 -\u003e True\nprint(mail.has_defects)                # both  -\u003e False\n```\n\nA related case: `email.message.Message.__len__` returns the header count, so a message with\n**no headers** is falsy. `parse()` returned before initialising state, and the unset\nattributes were then absorbed by `__getattr__` and answered as absent headers. Every\nproperty returned `\u0027\u0027` \u2014 including `mail` and `attachments`, which are documented to return\na `dict` and a `list` \u2014 with no error and no defect flag.\n\n```python\nimport mailparser\nmail = mailparser.parse_from_string(\"Click http://evil.example/pay.exe now\\r\\n\")\nprint(repr(mail.body))         # 4.6.1 -\u003e \u0027\u0027    4.6.2 -\u003e the text\nprint(repr(mail.attachments))  # 4.6.1 -\u003e \u0027\u0027    4.6.2 -\u003e []\nprint(mail.has_defects)        # 4.6.1 -\u003e \u0027\u0027    4.6.2 -\u003e True\n```\n\n### Patches\n\nFixed in **4.6.2**. Upgrade with `pip install --upgrade mail-parser`.\n\nHeader names taken from a message are now looked up literally \u2014 no attribute access, no\ncase or separator folding, no suffix interpretation \u2014 so no header name can collide with a\nmethod or property, and no value can be substituted or dropped. 8-bit values are coerced to\n`str` before serialisation. Parser state is initialised before the message check, which now\ntests `is None`, and attribute names beginning with `_` raise `AttributeError` instead of\nbeing answered as headers.\n\n**Behaviour changes.** Header names now appear in `mail` and `headers` exactly as sent, so a\nheader named `X_Spam_Flag` keeps that name and its own value. `X_raw` always returns a JSON\narray \u2014 an absent header yields `[]` where 4.6.1 returned `null`.\n\n### Workarounds\n\nNone for the data loss. For the uncaught exceptions, callers who cannot upgrade should wrap\n`mail_json`, `mail_partial_json`, `headers_json` and any `X_raw` access in a bare\n`except Exception` rather than catching only `MailParserError` subclasses.",
  "id": "GHSA-24mq-4vc6-2prf",
  "modified": "2026-11-16T22:12:52Z",
  "published": "2026-11-16T22:12:51Z",
  "references": [
    {
      "type": "WEB",
      "url": "https://github.com/SpamScope/mail-parser/security/advisories/GHSA-24mq-4vc6-2prf"
    },
    {
      "type": "ADVISORY",
      "url": "https://nvd.nist.gov/vuln/detail/CVE-2026-129299"
    },
    {
      "type": "WEB",
      "url": "https://github.com/SpamScope/mail-parser/pull/167"
    },
    {
      "type": "WEB",
      "url": "https://github.com/SpamScope/mail-parser/commit/ce3cf3559db1ff806bda598b378de9f8c2a5be96"
    },
    {
      "type": "PACKAGE",
      "url": "https://github.com/SpamScope/mail-parser"
    },
    {
      "type": "WEB",
      "url": "https://github.com/SpamScope/mail-parser/releases/tag/4.6.2"
    }
  ],
  "schema_version": "1.4.0",
  "severity": [
    {
      "score": "CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:N/I:L/A:L",
      "type": "CVSS_V3"
    }
  ],
  "summary": "mail-parser: Uncaught exceptions and silent header loss in mail-parser on attacker-controlled input"
}



Log in or create an account to share your comment.




Tags
Taxonomy of the tags.


Loading…

Loading…

Loading…

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.

Loading…

Loading…

Loading…

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.


Loading…