| 1 | use crate::{Draft, Embedding, Error, Result}; |
| 2 | use regex::RegexSet; |
| 3 | use sha2::{Digest, Sha256}; |
| 4 | use std::sync::OnceLock; |
| 5 | |
| 6 | pub const MAX_BODY_BYTES: usize = 8192; |
| 7 | pub const MAX_FRAME_BYTES: usize = 64 * 1024; |
| 8 | |
| 9 | pub fn sha256(bytes: &[u8]) -> String { |
| 10 | Sha256::digest(bytes) |
| 11 | .iter() |
| 12 | .map(|b| format!("{b:02x}")) |
| 13 | .collect() |
| 14 | } |
| 15 | pub fn content_hash(draft: &Draft) -> Result<String> { |
| 16 | // Exclude timestamps and provenance so replaying the same normalized content |
| 17 | // with a new observation time does not bypass exact-content tombstones. |
| 18 | Ok(sha256(&serde_json::to_vec(&( |
| 19 | draft.kind, |
| 20 | &draft.title, |
| 21 | &draft.body, |
| 22 | &draft.key, |
| 23 | &draft.tags, |
| 24 | ))?)) |
| 25 | } |
| 26 | pub fn ensure_no_secret(text: &str) -> Result<()> { |
| 27 | static SECRETS: OnceLock<RegexSet> = OnceLock::new(); |
| 28 | let rules = SECRETS.get_or_init(|| RegexSet::new([ |
| 29 | r"-----BEGIN (?:RSA |EC |OPENSSH |DSA )?PRIVATE KEY-----", |
| 30 | r"\b(?:sk-(?:proj-|ant-)?[A-Za-z0-9_-]{16,}|gh[pousr]_[A-Za-z0-9_]{20,}|github_pat_[A-Za-z0-9_]{20,})", |
| 31 | r"\b(?:AKIA|ASIA)[A-Z0-9]{16}\b", |
| 32 | r"\bxox[baprs]-[A-Za-z0-9-]{16,}", |
| 33 | r"(?i)\bbearer\s+[A-Za-z0-9._~+/-]{16,}", |
| 34 | r#"(?i)\b(?:api[_-]?key|password|passwd|client[_-]?secret|access[_-]?token)\s*[=:]\s*["']?[A-Za-z0-9+/_=.-]{8,}"#, |
| 35 | r"\beyJ[A-Za-z0-9_-]{8,}\.[A-Za-z0-9_-]{8,}\.[A-Za-z0-9_-]{8,}", |
| 36 | ]).expect("static secret patterns are valid")); |
| 37 | if rules.is_match(text) { |
| 38 | Err(Error::SecretDetected) |
| 39 | } else { |
| 40 | Ok(()) |
| 41 | } |
| 42 | } |
| 43 | pub fn bounded(value: &str, name: &str, max: usize, nonempty: bool) -> Result<()> { |
| 44 | if value.len() > max || (nonempty && value.trim().is_empty()) || value.contains('\0') { |
| 45 | return Err(Error::Invalid(format!( |
| 46 | "{name} is empty, contains NUL, or exceeds its size limit" |
| 47 | ))); |
| 48 | } |
| 49 | Ok(()) |
| 50 | } |
| 51 | pub fn validate_draft(d: &Draft, now: i64) -> Result<()> { |
| 52 | d.scope.validate()?; |
| 53 | if d.valid_from.is_some_and(|t| t < 0) |
| 54 | || d.valid_until.is_some_and(|t| t < 0) |
| 55 | || matches!((d.valid_from,d.valid_until),(Some(a),Some(b)) if a>=b) |
| 56 | { |
| 57 | return Err(Error::Invalid("invalid valid-time interval".into())); |
| 58 | } |
| 59 | bounded(&d.title, "title", 256, true)?; |
| 60 | bounded(&d.body, "body", MAX_BODY_BYTES, true)?; |
| 61 | if let Some(k) = &d.key { |
| 62 | bounded(k, "semantic key", 256, true)?; |
| 63 | } |
| 64 | for v in [d.confidence, d.importance] { |
| 65 | if !v.is_finite() || !(0.0..=1.0).contains(&v) { |
| 66 | return Err(Error::Invalid( |
| 67 | "confidence and importance must be finite values in [0,1]".into(), |
| 68 | )); |
| 69 | } |
| 70 | } |
| 71 | if d.evidence.is_empty() |
| 72 | || d.evidence.len() > 16 |
| 73 | || d.tags.len() > 32 |
| 74 | || d.dependencies.len() > 64 |
| 75 | || d.parent_ids.len() > 32 |
| 76 | { |
| 77 | return Err(Error::Invalid( |
| 78 | "evidence required; collection limits exceeded".into(), |
| 79 | )); |
| 80 | } |
| 81 | for tag in &d.tags { |
| 82 | bounded(tag, "tag", 96, true)?; |
| 83 | } |
| 84 | if d.expires_at.is_some_and(|t| t <= now) { |
| 85 | return Err(Error::Invalid("expiry must be in the future".into())); |
| 86 | } |
| 87 | if let Some(rev) = &d.repository_revision { |
| 88 | bounded(rev, "repository revision", 256, true)?; |
| 89 | } |
| 90 | for e in &d.evidence { |
| 91 | bounded(&e.uri, "evidence URI", 1024, true)?; |
| 92 | bounded(&e.locator, "evidence locator", 256, false)?; |
| 93 | if e.observed_at < 0 || e.observed_at > now.saturating_add(300) { |
| 94 | return Err(Error::Invalid("invalid observation time".into())); |
| 95 | } |
| 96 | if let Some(h) = &e.sha256 { |
| 97 | validate_digest(h)?; |
| 98 | } |
| 99 | } |
| 100 | for (path, digest) in &d.dependencies { |
| 101 | crate::workspace::validate_relative_path(path)?; |
| 102 | validate_digest(digest)?; |
| 103 | } |
| 104 | if (!d.dependencies.is_empty() || d.repository_revision.is_some()) |
| 105 | && d.scope.workspace.is_none() |
| 106 | { |
| 107 | return Err(Error::Invalid( |
| 108 | "repository-bound memories require a workspace scope".into(), |
| 109 | )); |
| 110 | } |
| 111 | for id in &d.parent_ids { |
| 112 | bounded(id, "parent id", 128, true)?; |
| 113 | } |
| 114 | ensure_no_secret(&serde_json::to_string(d)?)?; |
| 115 | Ok(()) |
| 116 | } |
| 117 | pub fn validate_digest(h: &str) -> Result<()> { |
| 118 | if h.len() != 64 |
| 119 | || !h |
| 120 | .bytes() |
| 121 | .all(|c| c.is_ascii_hexdigit() && !c.is_ascii_uppercase()) |
| 122 | { |
| 123 | return Err(Error::Invalid( |
| 124 | "SHA-256 digests must be 64 lowercase hexadecimal characters".into(), |
| 125 | )); |
| 126 | } |
| 127 | Ok(()) |
| 128 | } |
| 129 | pub fn normalize_embedding(e: &Embedding) -> Result<Vec<f32>> { |
| 130 | bounded(&e.model, "embedding model identity", 256, true)?; |
| 131 | if e.vector.is_empty() || e.vector.len() > 8192 || e.vector.iter().any(|v| !v.is_finite()) { |
| 132 | return Err(Error::Invalid( |
| 133 | "invalid embedding dimensions or values".into(), |
| 134 | )); |
| 135 | } |
| 136 | let norm = e |
| 137 | .vector |
| 138 | .iter() |
| 139 | .map(|v| (*v as f64).powi(2)) |
| 140 | .sum::<f64>() |
| 141 | .sqrt(); |
| 142 | if norm <= f64::EPSILON || !norm.is_finite() { |
| 143 | return Err(Error::Invalid( |
| 144 | "embedding norm must be finite and nonzero".into(), |
| 145 | )); |
| 146 | } |
| 147 | Ok(e.vector.iter().map(|v| (*v as f64 / norm) as f32).collect()) |
| 148 | } |
| 149 | /// All MATCH operators become literal text. Never accept raw FTS syntax from a model. |
| 150 | pub fn fts_query(text: &str) -> Result<Option<String>> { |
| 151 | bounded(text, "query", 1024, false)?; |
| 152 | let tokens: Vec<_> = text |
| 153 | .split(|c: char| !c.is_alphanumeric()) |
| 154 | .filter(|s| !s.is_empty()) |
| 155 | .take(32) |
| 156 | .collect(); |
| 157 | if tokens.is_empty() { |
| 158 | return Ok(None); |
| 159 | } |
| 160 | Ok(Some( |
| 161 | tokens |
| 162 | .iter() |
| 163 | .map(|s| format!("\"{}\"", s.replace('"', "\"\""))) |
| 164 | .collect::<Vec<_>>() |
| 165 | .join(" OR "), |
| 166 | )) |
| 167 | } |
| 168 | pub fn excerpt(text: &str, chars: usize) -> String { |
| 169 | text.chars().take(chars).collect() |
| 170 | } |
| 171 |