{"id":"GHSA-r7w7-vpq8-c5vv","summary":"mail-parser is vulnerable to polynomial ReDoS in email header parsing","details":"### Summary\n\nTwo regular expressions used while parsing email headers backtrack\npolynomially on attacker-controlled input. Parsing a single small, well-formed\nemail can consume minutes to hours of CPU, denying service to any process that\nparses untrusted mail.\n\nBoth are reachable from an ordinary `parse_from_string()`, `parse_from_bytes()`\nor `parse_from_file()` call — no special API, option, or malformed structure is\nrequired by the victim.\n\n### Details\n\n**1. Address-header fallback regex (`src/mailparser/utils.py`)**\n\n`get_addresses()` falls back to `_ADDR_FALLBACK_RE` whenever\n`email.utils.getaddresses(strict=True)` yields no usable address. The\nalternative `([^\u003c,]*?)\\s*\u003c([^\u003e]+)\u003e` overlaps a lazy quantifier with an adjacent\n`\\s*` — a space matches both. Driven by `finditer`, which retries at every\noffset and rescans the remainder, matching degrades to **quadratic** on `\u003c`\npadding and **cubic** on whitespace padding.\n\nThe fallback engages on the feature's own happy path, e.g.\n`From: alice@example.com \u003cbob@example.com\u003e`.\n\nMeasured (4.6.0, CPython 3.14, macOS arm64; reproduced identically on Linux and\non CPython 3.9, 3.11, 3.12, 3.13):\n\n| Padding                | Parse time         |\n| ---------------------- | ------------------ |\n| valid address + 750 sp | 0.38 s             |\n| + 1500 sp              | 2.97 s             |\n| + 3000 sp              | 23.4 s (×7.9/2×)   |\n| + 6 KB sp              | 185 s              |\n| ~100 KB sp             | hours              |\n\nFully RFC 5322-compliant folded headers (all lines ≤ 71 octets) trigger the same\nblowup, so the input passes normal MTA and library checks. ~100 KB of headers is\nwithin Postfix's default `header_size_limit = 102400`.\n\nAll six address headers parsed by the library (`From`, `To`, `Cc`, `Bcc`,\n`Reply-To`, `Delivered-To`) are affected and their cost adds up linearly.\n\n**2. Received-header clause splitter (`src/mailparser/const.py`)**\n\n`parse_received()` splits on `_CLAUSE_SPLITTER`,\n`(?:^|\\s+)(from|by|via|with|id|for|envelope-from)\\s+`. Under `re.split` this\nbacktracks **quadratically** over any whitespace run that is not followed by a\nclause keyword. The pre-existing normalization collapsed only `[\\t\\n]`, so\nliteral spaces — including folded-header padding — reached the splitter intact.\n\nMeasured: 24 000 spaces in one `Received:` header → 22 s; ~100 KB → minutes.\n\n### Impact\n\nAvailability. A handful of small emails keeps every parsing worker of a mail\ngateway, ticketing system, quarantine UI, or archiving pipeline permanently\nbusy. No confidentiality or integrity impact.\n\nAnyone who parses email they did not author is affected — which is the library's\npurpose.\n\n### Patches\n\nFixed in **4.6.1**.\n\n- The address fallback is now a linear scan: the header is split into\n  comma-separated items (quote-aware, in Python), then each item is matched\n  against a bounded angle-address `\u003c([^\u003c\u003e\\s]+)\u003e` with the display name derived\n  in Python, falling back to a bare-email match. Each pattern has a single\n  non-nested quantifier over a negated character class → O(n).\n  2000 padding spaces: `3.5 s → \u003c0.01 s`. 256 KB padding: `hours → 0.087 s`.\n  Output is preserved on every previously supported shape.\n- `parse_received()` collapses whitespace runs to a single space before\n  splitting, so the splitter's `\\s+` only ever matches one character → O(n).\n  The date is extracted from the raw header beforehand, so date formatting is\n  unaffected. 24 000 spaces: `22 s → 0.003 s`; 200 000 spaces: `0.004 s`.\n\nRegression tests assert linear-time behaviour for both paths.\n\n### Workarounds\n\nNone within the library. Upgrade to `4.6.1`.\n\nUsers who cannot upgrade immediately should cap total header size before\nparsing and enforce a wall-clock timeout around each parse.","aliases":["CVE-2026-129295"],"modified":"2026-11-16T22:15:05.620597997Z","published":"2026-11-16T22:11:08Z","database_specific":{"github_reviewed":true,"github_reviewed_at":"2026-11-16T22:11:08Z","nvd_published_at":"2026-11-16T18:17:30Z","cwe_ids":["CWE-1333"],"severity":"HIGH"},"references":[{"type":"WEB","url":"https://github.com/SpamScope/mail-parser/security/advisories/GHSA-r7w7-vpq8-c5vv"},{"type":"ADVISORY","url":"https://nvd.nist.gov/vuln/detail/CVE-2026-129295"},{"type":"WEB","url":"https://github.com/SpamScope/mail-parser/pull/166"},{"type":"WEB","url":"https://github.com/SpamScope/mail-parser/commit/037293158cb291caa96635a54449243d32cd4d96"},{"type":"WEB","url":"https://github.com/SpamScope/mail-parser/commit/5aab122f0d763991604d2c2bcc807c6bf2489464"},{"type":"WEB","url":"https://github.com/SpamScope/mail-parser/commit/85e53f820d78ae7c955dafbbcd0e20989a735c2c"},{"type":"PACKAGE","url":"https://github.com/SpamScope/mail-parser"},{"type":"WEB","url":"https://github.com/SpamScope/mail-parser/releases/tag/4.6.1"}],"affected":[{"package":{"name":"mail-parser","ecosystem":"PyPI","purl":"pkg:pypi/mail-parser"},"ranges":[{"type":"ECOSYSTEM","events":[{"introduced":"4.2.0"},{"fixed":"4.6.1"}]}],"versions":["4.2.0","4.2.1","4.3.0","4.4.0","4.5.0","4.6.0"],"database_specific":{"source":"https://github.com/github/advisory-database/blob/main/advisories/github-reviewed/2026/11/GHSA-r7w7-vpq8-c5vv/GHSA-r7w7-vpq8-c5vv.json"}}],"schema_version":"1.9.2","severity":[{"type":"CVSS_V3","score":"CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:N/I:N/A:H"}]}