Merge pull request #640 from pulseengine/fix/577-validate-commit-name-parity

fix(validate): sync artifact-id shape with the commit-trailer parser (REQ-239, #577)

Author: avrabe

artifacts/requirements.yaml

@@ -7579,7 +7579,7 @@ artifacts:
   - id: REQ-239
     type: requirement
     title: validate <-> commit-name parsing parity
-    status: proposed
+    status: verified
     description: "Sync the validate naming checks with commit-name parsing so naming issues are caught early and consistently. #577. v0.23."
     provenance:
       created-by: ai-assisted

rivet-core/src/commits.rs

@@ -322,23 +322,38 @@ pub fn extract_artifact_refs(value: &str) -> (Vec<String>, Vec<String>) {
 /// but is not itself a valid artifact ID. This deliberately excludes
 /// ordinary hyphenated prose (no digit) and bare numbers (no hyphen),
 /// so `rivet commits` flags genuine typos without choking on free text.
-fn looks_like_artifact_id_attempt(token: &str) -> bool {
+///
+/// `pub` because `rivet validate` reuses it (#577): an artifact whose id
+/// looks like a botched numbered id (e.g. a dotted suffix `H-3.2`) validates
+/// fine but can't be referenced in a commit trailer — validate warns early so
+/// the mismatch isn't discovered only at commit time.
+pub fn looks_like_artifact_id_attempt(token: &str) -> bool {
     !is_artifact_id(token) && token.contains('-') && token.chars().any(|c| c.is_ascii_digit())
 }
 
-/// Check whether a string looks like an artifact ID.
+/// Check whether a string has the shape rivet recognises as an artifact ID
+/// in a **commit trailer** (`Implements: <ID>`). This is the single source of
+/// truth for that shape — `rivet validate` calls it too, so a project can't
+/// have IDs that validate but silently fail to trace through commits (#577).
 ///
-/// Matches simple IDs like `REQ-001` and compound-prefix IDs like
-/// `UCA-C-10`.  The last hyphen-separated segment must be all digits;
-/// all preceding segments must be non-empty uppercase ASCII.
-fn is_artifact_id(s: &str) -> bool {
+/// Matches simple IDs like `REQ-001` and compound-prefix IDs like `UCA-C-10`.
+/// The last hyphen-separated segment must be all digits; every preceding
+/// segment must be non-empty, contain at least one uppercase ASCII letter, and
+/// consist only of uppercase ASCII letters or digits — so a digit-bearing
+/// prefix like `MAD1-101` is accepted (#577), while `123-4` (no letter),
+/// `mad1-1` (lowercase), and `H-3.2` (dotted suffix) are not.
+pub fn is_artifact_id(s: &str) -> bool {
     if let Some(pos) = s.rfind('-') {
         let prefix = &s[..pos];
         let suffix = &s[pos + 1..];
         !prefix.is_empty()
-            && prefix
-                .split('-')
-                .all(|seg| !seg.is_empty() && seg.chars().all(|c| c.is_ascii_uppercase()))
+            && prefix.split('-').all(|seg| {
+                !seg.is_empty()
+                    && seg
+                        .chars()
+                        .all(|c| c.is_ascii_uppercase() || c.is_ascii_digit())
+                    && seg.chars().any(|c| c.is_ascii_uppercase())
+            })
             && !suffix.is_empty()
             && suffix.chars().all(|c| c.is_ascii_digit())
     } else {
@@ -1271,6 +1286,30 @@ mod tests {
         assert!(!is_artifact_id("-1"));
     }
 
+    // #577 (REQ-239): a digit-bearing prefix segment (e.g. `MAD1`) is a valid
+    // commit-trailer ref — the parser used to require letter-only prefixes,
+    // forcing a rename loop. Segments must still contain at least one letter
+    // and the suffix must be all digits.
+    // rivet: verifies REQ-239
+    #[test]
+    fn artifact_id_accepts_digit_bearing_prefix() {
+        assert!(is_artifact_id("MAD1-101"), "digit in prefix segment is ok");
+        assert!(is_artifact_id("A1-B2-3"), "digits across compound segments");
+        assert!(is_artifact_id("REQ-001"), "plain case still works");
+        // Still rejected: no letter, lowercase, dotted suffix, non-digit suffix.
+        assert!(!is_artifact_id("123-4"), "prefix segment needs a letter");
+        assert!(!is_artifact_id("mad1-1"), "lowercase prefix rejected");
+        assert!(!is_artifact_id("H-3.2"), "dotted suffix is not all-digits");
+        assert!(!is_artifact_id("REQ-00A"), "non-digit suffix rejected");
+        // …and the parity heuristic flags the botched-but-digit-bearing ones.
+        assert!(looks_like_artifact_id_attempt("H-3.2"));
+        assert!(!looks_like_artifact_id_attempt("MAD1-101"));
+        assert!(
+            !looks_like_artifact_id_attempt("ARCH-CORE-COMMITS"),
+            "a descriptive (no-digit) id is not a botched numbered id"
+        );
+    }
+
     // -- integration: extract_artifact_ids with ranges --
 
     // rivet: verifies REQ-017

rivet-core/src/validate.rs

@@ -636,6 +636,32 @@ pub fn validate_structural_with_externals_and_variant(
         if is_external_artifact(artifact) {
             continue;
         }
+        // #577 (REQ-239): an id that looks like a botched numbered id — a
+        // hyphen and a digit but not a parseable artifact id, e.g. a dotted
+        // suffix `H-3.2` — validates fine but cannot be referenced in a commit
+        // trailer (`Implements: <id>`). Warn early so the mismatch isn't
+        // discovered only when `rivet commit` rejects the trailer. Shares the
+        // exact shape rule with the commit parser (`crate::commits`) so the
+        // two never diverge (the relaxation there now accepts digit-bearing
+        // prefixes like `MAD1-101`, so only genuinely un-referenceable shapes
+        // remain flagged here). Externally-prefixed ids are already skipped.
+        if crate::commits::looks_like_artifact_id_attempt(&artifact.id) {
+            diagnostics.push(Diagnostic {
+                source_file: None,
+                line: None,
+                column: None,
+                severity: Severity::Warning,
+                artifact_id: Some(artifact.id.clone()),
+                rule: "commit-ref-shape".to_string(),
+                message: format!(
+                    "id '{}' can't be used as a commit-trailer reference \
+                     (trailers need an uppercase-alphanumeric prefix and an \
+                     all-digit suffix, e.g. REQ-001); rename it to trace this \
+                     artifact through commits",
+                    artifact.id
+                ),
+            });
+        }
         let type_def = match lookup_type(artifact, schema, externals) {
             TypeLookup::Found(td) => td,
             TypeLookup::Unknown => {
@@ -2719,6 +2745,53 @@ then:
         );
     }
 
+    /// #577 (REQ-239): validate warns when an artifact id can't be used as a
+    /// commit-trailer reference — a dotted suffix (`H-3.2`) is flagged, while a
+    /// now-accepted digit-bearing prefix (`MAD1-101`) is not. Keeps validate
+    /// and the commit parser in sync so naming issues surface at validate time.
+    ///
+    /// rivet: verifies REQ-239
+    #[test]
+    fn validate_warns_on_non_commit_referenceable_id() {
+        let mut file = minimal_schema("test");
+        file.artifact_types = vec![ArtifactTypeDef {
+            name: "requirement".to_string(),
+            description: "REQ".to_string(),
+            fields: vec![],
+            link_fields: vec![],
+            aspice_process: None,
+            common_mistakes: vec![],
+            example: None,
+            yaml_section: None,
+            yaml_sections: vec![],
+            yaml_section_suffix: None,
+            shorthand_links: std::collections::BTreeMap::new(),
+        }];
+        file.traceability_rules = vec![];
+        let schema = Schema::merge(&[file]);
+
+        let mut store = Store::new();
+        store
+            .insert(minimal_artifact("H-3.2", "requirement"))
+            .unwrap();
+        store
+            .insert(minimal_artifact("MAD1-101", "requirement"))
+            .unwrap();
+
+        let graph = LinkGraph::build(&store, &schema);
+        let diags = validate(&store, &schema, &graph);
+        let flagged: Vec<&str> = diags
+            .iter()
+            .filter(|d| d.rule == "commit-ref-shape")
+            .filter_map(|d| d.artifact_id.as_deref())
+            .collect();
+        assert_eq!(
+            flagged,
+            vec!["H-3.2"],
+            "only the dotted-suffix id is un-referenceable; MAD1-101 is now valid"
+        );
+    }
+
     /// Issue #349: `required-backlink` written as the INVERSE link-type
     /// name (e.g. `supported-by`, the convention used by
     /// `schemas/safety-case.yaml`) was never matched against the stored