systemprompt_security/policy/secrets/
entropy.rs1use base64::Engine as _;
15use base64::engine::general_purpose::{STANDARD_NO_PAD, URL_SAFE_NO_PAD};
16use regex::Regex;
17
18pub const DEFAULT_MIN_LEN: usize = 32;
19
20pub const DEFAULT_THRESHOLD: f64 = 0.80;
29
30const TOKEN_DELIMITERS: &str = "\"'`()[]{}<>,;:";
31const TOKEN_CHARSET_EXTRA: &str = "+/=_-";
32const ENTROPY_CEILING_SYMBOLS: usize = 64;
33
34const MIN_STRUCTURED_LEN: usize = 16;
35const MAX_FIELD_NUMBER: u64 = 64;
36const MAX_NESTING_DEPTH: u32 = 4;
37const MIN_NESTED_PAYLOAD_LEN: usize = 4;
38const TEXT_RATIO_NUMERATOR: usize = 9;
39const TEXT_RATIO_DENOMINATOR: usize = 10;
40
41#[derive(Debug, Clone)]
45pub struct EntropyConfig {
46 pub enabled: bool,
47 pub min_len: usize,
48 pub threshold: f64,
49 pub allowlist: Vec<Regex>,
50}
51
52impl Default for EntropyConfig {
53 fn default() -> Self {
54 Self {
55 enabled: true,
56 min_len: DEFAULT_MIN_LEN,
57 threshold: DEFAULT_THRESHOLD,
58 allowlist: Vec::new(),
59 }
60 }
61}
62
63#[must_use]
64pub fn find_high_entropy_token<'a>(text: &'a str, config: &EntropyConfig) -> Option<&'a str> {
65 if !config.enabled {
66 return None;
67 }
68 text.split(|c: char| c.is_whitespace() || TOKEN_DELIMITERS.contains(c))
69 .find(|token| is_credential_shaped(token, config))
70}
71
72fn is_credential_shaped(token: &str, config: &EntropyConfig) -> bool {
73 token.len() >= config.min_len
74 && token
75 .chars()
76 .all(|c| c.is_ascii_alphanumeric() || TOKEN_CHARSET_EXTRA.contains(c))
77 && token.chars().any(|c| c.is_ascii_uppercase())
78 && token.chars().any(|c| c.is_ascii_lowercase())
79 && token.chars().any(|c| c.is_ascii_digit())
80 && entropy_ratio(token) >= config.threshold
81 && !config.allowlist.iter().any(|re| re.is_match(token))
82 && !is_structured_payload(token)
83}
84
85fn shannon_entropy(s: &str) -> f64 {
86 let mut counts = [0u32; 256];
87 let bytes = s.as_bytes();
88 for &b in bytes {
89 counts[usize::from(b)] += 1;
90 }
91 let len = bytes.len() as f64;
92 counts
93 .iter()
94 .filter(|&&c| c > 0)
95 .map(|&c| {
96 let p = f64::from(c) / len;
97 -p * p.log2()
98 })
99 .sum()
100}
101
102fn entropy_ratio(s: &str) -> f64 {
103 let symbols = s.len().min(ENTROPY_CEILING_SYMBOLS);
104 let ceiling = (symbols as f64).log2();
105 if ceiling <= 0.0 {
106 return 0.0;
107 }
108 shannon_entropy(s) / ceiling
109}
110
111fn is_structured_payload(token: &str) -> bool {
112 decode_base64(token).is_some_and(|bytes| {
113 bytes.len() >= MIN_STRUCTURED_LEN && (is_mostly_text(&bytes) || is_protobuf(&bytes, 0))
114 })
115}
116
117fn decode_base64(token: &str) -> Option<Vec<u8>> {
118 let body = token.trim_end_matches('=');
119 STANDARD_NO_PAD
120 .decode(body)
121 .or_else(|_| URL_SAFE_NO_PAD.decode(body))
122 .ok()
123}
124
125fn is_mostly_text(bytes: &[u8]) -> bool {
126 if bytes.is_empty() {
127 return false;
128 }
129 let printable = bytes
130 .iter()
131 .filter(|&&b| matches!(b, b'\t' | b'\n' | b'\r') || (0x20..0x7f).contains(&b))
132 .count();
133 printable * TEXT_RATIO_DENOMINATOR >= bytes.len() * TEXT_RATIO_NUMERATOR
134}
135
136fn is_protobuf(bytes: &[u8], depth: u32) -> bool {
142 if depth > MAX_NESTING_DEPTH || bytes.len() < MIN_STRUCTURED_LEN {
143 return false;
144 }
145 parse_protobuf(bytes, depth).is_some_and(|parsed| parsed.fields >= 2 && parsed.nested_structure)
146}
147
148struct ParsedMessage {
149 fields: usize,
150 nested_structure: bool,
151}
152
153fn parse_protobuf(bytes: &[u8], depth: u32) -> Option<ParsedMessage> {
154 let mut cursor = 0usize;
155 let mut parsed = ParsedMessage {
156 fields: 0,
157 nested_structure: false,
158 };
159 while cursor < bytes.len() {
160 let (tag, after_tag) = read_varint(bytes, cursor)?;
161 let field_number = tag >> 3;
162 if field_number == 0 || field_number > MAX_FIELD_NUMBER {
163 return None;
164 }
165 cursor = match tag & 7 {
166 0 => read_varint(bytes, after_tag)?.1,
167 1 => advance(bytes, after_tag, 8)?,
168 5 => advance(bytes, after_tag, 4)?,
169 2 => {
170 let (len, after_len) = read_varint(bytes, after_tag)?;
171 let len = usize::try_from(len).ok()?;
172 let end = advance(bytes, after_len, len)?;
173 let payload = bytes.get(after_len..end)?;
174 if payload.len() >= MIN_NESTED_PAYLOAD_LEN
175 && (is_mostly_text(payload) || is_protobuf(payload, depth + 1))
176 {
177 parsed.nested_structure = true;
178 }
179 end
180 },
181 _ => return None,
182 };
183 parsed.fields += 1;
184 }
185 Some(parsed)
186}
187
188fn advance(bytes: &[u8], cursor: usize, by: usize) -> Option<usize> {
189 let end = cursor.checked_add(by)?;
190 (end <= bytes.len()).then_some(end)
191}
192
193fn read_varint(bytes: &[u8], start: usize) -> Option<(u64, usize)> {
194 let mut value = 0u64;
195 let mut shift = 0u32;
196 let mut cursor = start;
197 loop {
198 let byte = *bytes.get(cursor)?;
199 cursor += 1;
200 value |= u64::from(byte & 0x7f) << shift;
201 if byte & 0x80 == 0 {
202 return Some((value, cursor));
203 }
204 shift += 7;
205 if shift >= 64 {
206 return None;
207 }
208 }
209}