Skip to content

Solution for long term memory for agent coding CLIs and to facilitate handoff between different agent vendors

Rust

fix(consolidate): key a non-Latin rule name on the folded title

AkitaOnRails1756694f7c32
diff --git a/CHANGELOG.md b/CHANGELOG.md
index db537eb1..289bd063 100644
--- a/CHANGELOG.md
+++ b/CHANGELOG.md
@@ -12,7 +12,10 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
and others) overwriting each other: multi-page consolidation named every such
rule page `_rules/rule.md`, so only the last survived in the live rules list.
Each rule now gets a `_rules/rule-<hash>.md` name derived from its title. Rule
- titles in Latin scripts are named as before. (#1191)
+ titles in Latin scripts are named as before. A restated non-Latin title
+ (case, spacing or punctuation aside) updates its own page. An existing
+ `_rules/rule.md` is left in place; delete it once its rule is restated.
+ (#1191)
- Fixed generated TypeScript integrations (OpenCode, OpenCode 2, OMP,
Pi, OpenClaw) flashing a console window on Windows.
`discoverRemoteIdentity` spawned `git config --get remote.<name>.url`
diff --git a/crates/ai-memory-consolidate/src/consolidator.rs b/crates/ai-memory-consolidate/src/consolidator.rs
index 813709de..9be9d776 100644
--- a/crates/ai-memory-consolidate/src/consolidator.rs
+++ b/crates/ai-memory-consolidate/src/consolidator.rs
@@ -1732,9 +1732,19 @@ fn slugify_for_rule(title: &str) -> String {
// A title with no ASCII letter or digit (Cyrillic, CJK, ...) would
// share one name with every other such title, and each rule would
// overwrite the last. Name it after the title instead. A title with
- // no letter or digit at all has nothing to name it by.
- if decomposed.chars().any(char::is_alphanumeric) {
- let digest = Sha256::digest(decomposed.trim().as_bytes());
+ // no letter or digit at all has nothing to name it by. The hash takes
+ // the same fold as the ASCII slug (lower-cased words, separators
+ // collapsed) so a restated rule updates its own page; combining
+ // marks stay with their letter so `й` and `и` remain distinct.
+ let in_word = |c: char| c.is_alphanumeric() || ('\u{0300}'..='\u{036F}').contains(&c);
+ let key = decomposed
+ .split(|c: char| !in_word(c))
+ .filter(|word| !word.is_empty())
+ .map(str::to_lowercase)
+ .collect::<Vec<_>>()
+ .join("-");
+ if key.chars().any(char::is_alphanumeric) {
+ let digest = Sha256::digest(key.as_bytes());
let hex: String = digest
.iter()
.take(4)
@@ -3255,6 +3265,15 @@ mod tests {
slugify_for_rule("\u{439}\u{43e}"),
slugify_for_rule("\u{438}\u{306}\u{43e}")
);
+ // A restated title updates its own page, as an ASCII one does: case,
+ // spacing and punctuation do not name a new rule.
+ assert_eq!(slugify_for_rule("не коммитить секреты!"), russian);
+ assert_eq!(slugify_for_rule("НЕ КОММИТИТЬ, СЕКРЕТЫ."), russian);
+ // A combining mark is part of its letter, not a separator.
+ assert_ne!(
+ slugify_for_rule("\u{439}\u{43e}"),
+ slugify_for_rule("\u{438} \u{43e}")
+ );
for slug in [chinese, russian] {
let path = format!("_rules/{slug}.md");
let path = PagePath::new(path).unwrap();