NLTK TweetTokenizer URL Pattern Backtracks Catastrophically on Naked-Domain-Like Input
The URLS regular expression in nltk/tokenize/casual.py, compiled into TweetTokenizer.WORD_RE and applied by TweetTokenizer.tokenize, contains a naked-domain branch whose domain-label prefix [a-z0-9]+(?:[.\-][a-z0-9]+)* is unbounded. Input consisting of many alternating label separators can be partitioned in exponentially many ways, and because the branch also requires a trailing top-level domain that such input never supplies, the engine explores those partitions before failing at each offset. A few kilobytes of input therefore consumes seconds to minutes of single-threaded CPU, and the HANG_RE substitution performed before matching does not collapse the pattern. TweetTokenizer is intended for tokenizing untrusted social-media text, so any service that applies it, or the module-level casual_tokenize, to submitted text can be stalled per request without authentication. Version 3.10.1 bounds the label repetition.
- Products
- Not specified
- Affected versions
- Not specified
- Fixed versions
- Not specified
- CVEs
- CVE-2026-72818
- Confirmed fixed CVEs
- Not specified
- Updated in this feed
- 8 Nov 2026, 01:20 UTC · Revision 1
Source evidence
NLTK TweetTokenizer URL Pattern Backtracks Catastrophically on Naked-Domain-Like Input
core-osv · Original sourcePublished 20 Aug 2026, 21:57 UTC
The URLS regular expression in nltk/tokenize/casual.py, compiled into TweetTokenizer.WORD_RE and applied by TweetTokenizer.tokenize, contains a naked-domain branch whose domain-label prefix [a-z0-9]+(?:[.\-][a-z0-9]+)* is unbounded. Input consisting of many alternating label separators can be partitioned in exponentially many ways, and because the branch also requires a trailing top-level domain that such input never supplies, the engine explores those partitions before failing at each offset. A few kilobytes of input therefore consumes seconds to minutes of single-threaded CPU, and the HANG_RE substitution performed before matching does not collapse the pattern. TweetTokenizer is intended for tokenizing untrusted social-media text, so any service that applies it, or the module-level casual_tokenize, to submitted text can be stalled per request without authentication. Version 3.10.1 bounds the label repetition.
osv:CVE-2026-72818
Complete source record
{
"id": "c4bde865d287a6b31a1a070c05e7c4bc1e14cd0f64bb52744d2dd97625240c92",
"url": "https://osv.dev/vulnerability/CVE-2026-72818",
"method": "structured",
"payload": {
"kind": "advisory",
"title": "NLTK TweetTokenizer URL Pattern Backtracks Catastrophically on Naked-Domain-Like Input",
"summary": "The URLS regular expression in nltk/tokenize/casual.py, compiled into TweetTokenizer.WORD_RE and applied by TweetTokenizer.tokenize, contains a naked-domain branch whose domain-label prefix [a-z0-9]+(?:[.\\-][a-z0-9]+)* is unbounded. Input consisting of many alternating label separators can be partitioned in exponentially many ways, and because the branch also requires a trailing top-level domain that such input never supplies, the engine explores those partitions before failing at each offset. A few kilobytes of input therefore consumes seconds to minutes of single-threaded CPU, and the HANG_RE substitution performed before matching does not collapse the pattern. TweetTokenizer is intended for tokenizing untrusted social-media text, so any service that applies it, or the module-level casual_tokenize, to submitted text can be stalled per request without authentication. Version 3.10.1 bounds the label repetition.",
"evidence": "osv:CVE-2026-72818",
"products": [],
"severity": "high",
"structured": {
"id": "CVE-2026-72818",
"aliases": [
"CVE-2026-72818",
"GHSA-qx2g-xrx7-vfh8",
"PYSEC-2026-3870"
],
"metrics": [
{
"type": "CVSS_V4",
"score": "CVSS:4.0/AV:N/AC:L/AT:N/PR:N/UI:N/VC:N/VI:N/VA:H/SC:N/SI:N/SA:N"
}
],
"affected": [
{
"ranges": [
{
"repo": "https://github.com/nltk/nltk",
"type": "GIT",
"events": [
{
"introduced": "0"
},
{
"fixed": "35813c85af3f13d4f7196085854eafb3d0c5db02"
}
],
"database_specific": {
"extracted_events": [
{
"range": [
{
"introduced": "0"
},
{
"fixed": "3.10.1"
}
],
"source": "AFFECTED_FIELD",
"strategy": "StandardRange"
},
{
"range": [
{
"fixed": "35813c85af3f13d4f7196085854eafb3d0c5db02"
}
],
"source": "REFERENCES_TAG",
"original_tag": "v3.10.1"
}
]
}
}
],
"versions": [
"v3.10.1-rc2",
"v3.10.0-rc2",
"v3.10.0",
"v3.10.1-rc1",
"v3.10.0-rc1",
"3.9.4",
"3.9.3",
"v3.9.2",
"3.9.2",
"3.9.1",
"3.9",
"3.8.2",
"3.8.1",
"3.8",
"3.7",
"3.6.7",
"3.6.6",
"3.6.5",
"3.6.2",
"3.6.1",
"3.6",
"3.5",
"3.5b1",
"3.4.4",
"3.4.3",
"3.4.1",
"3.4",
"3.3",
"3.2.5",
"3.2.4",
"3.2.3",
"3.2.2",
"3.2.1",
"3.2",
"3.1",
"3.0.5",
"3.0.4",
"3.0.3",
"3.0.2",
"3.0.0b1",
"3.0a4",
"2.0.1rc4",
"2.0.1rc3",
"2.0.1rc2",
"2.0.1rc1"
],
"database_specific": {
"source": "https://storage.googleapis.com/cve-osv-conversion/osv-output/CVE-2026-72818.json"
}
}
],
"modified": "2026-10-28T02:30:52.414636100Z",
"provider": "osv",
"references": [
"https://pypi.org",
"https://github.com/CVEProject/cvelistV5/tree/main/cves/2026/72xxx/CVE-2026-72818.json",
"https://github.com/nltk/nltk/releases/tag/v3.10.1",
"https://nvd.nist.gov/vuln/detail/CVE-2026-72818",
"https://www.vulncheck.com/advisories/nltk-tweettokenizer-url-pattern-backtracks-catastrophically-on-naked-domain-like-input",
"https://github.com/nltk/nltk/issues/3704",
"https://github.com/nltk/nltk",
"https://github.com/nltk/nltk/blob/3.9.4/nltk/tokenize/casual.py"
],
"provider_metadata": {
"cwe_ids": [
"CWE-1333"
],
"cna_assigner": "VulnCheck",
"osv_generated_from": "https://github.com/CVEProject/cvelistV5/tree/main/cves/2026/72xxx/CVE-2026-72818.json"
}
},
"related_cves": [
{
"value": "CVE-2026-72818",
"evidence": "/id"
}
],
"security_relevant": true,
"severity_evidence": "/severity/0/score"
},
"source_id": "core-osv",
"document_id": 521291,
"published_at": "2026-08-20T21:57:34.23+00:00"
}