Add spec submodule and schema-driven compliance tests

.github/workflows/ci.yml

@@ -17,6 +17,8 @@ jobs:
 
     steps:
       - uses: actions/checkout@v6
+        with:
+          submodules: true
 
       - name: Install Rust
         uses: dtolnay/rust-toolchain@stable
@@ -72,6 +74,8 @@ jobs:
     runs-on: ubuntu-latest
     steps:
       - uses: actions/checkout@v6
+        with:
+          submodules: true
 
       - name: Install Rust 1.88
         uses: dtolnay/rust-toolchain@1.88
@@ -104,6 +108,8 @@ jobs:
     runs-on: ubuntu-latest
     steps:
       - uses: actions/checkout@v6
+        with:
+          submodules: true
 
       - name: Install Rust
         uses: dtolnay/rust-toolchain@stable
@@ -122,6 +128,8 @@ jobs:
     runs-on: ubuntu-latest
     steps:
       - uses: actions/checkout@v6
+        with:
+          submodules: true
 
       - name: Install Rust
         uses: dtolnay/rust-toolchain@stable
@@ -243,6 +251,8 @@ jobs:
           - "signatures,encryption"
     steps:
       - uses: actions/checkout@v6
+        with:
+          submodules: true
 
       - name: Install Rust
         uses: dtolnay/rust-toolchain@stable

.github/workflows/spec-check.yml

@@ -0,0 +1,32 @@
+name: Spec Staleness Check
+
+on:
+  schedule:
+    - cron: '0 9 * * 1'  # Weekly, Monday 9am UTC
+  workflow_dispatch: {}
+
+jobs:
+  check:
+    runs-on: ubuntu-latest
+    steps:
+      - uses: actions/checkout@v6
+        with:
+          submodules: true
+
+      - name: Check for upstream spec updates
+        run: |
+          cd spec
+          git fetch origin main
+          LOCAL=$(git rev-parse HEAD)
+          REMOTE=$(git rev-parse origin/main)
+          if [ "$LOCAL" != "$REMOTE" ]; then
+            BEHIND=$(git rev-list --count HEAD..origin/main)
+            echo "::warning::Spec submodule is $BEHIND commits behind upstream"
+            gh issue create --title "Spec submodule is $BEHIND commits behind" \
+              --body "The spec submodule is pinned at \`$LOCAL\` but upstream is at \`$REMOTE\` ($BEHIND commits behind).\n\nRun:\n\`\`\`\ngit submodule update --remote spec\n\`\`\`" \
+              --label "dependencies" || true
+          else
+            echo "Spec submodule is up to date"
+          fi
+        env:
+          GH_TOKEN: ${{ secrets.GITHUB_TOKEN }}

.gitmodules

@@ -0,0 +1,3 @@
+[submodule "spec"]
+	path = spec
+	url = https://github.com/Entrolution/codex-file-format-spec.git

cdx-core/Cargo.toml

@@ -110,6 +110,7 @@ tempfile = "3.14"
 pretty_assertions = "1.4"
 proptest = "1.5"
 criterion = { version = "0.8", features = ["html_reports"] }
+jsonschema = "0.28"
 
 [[bench]]
 name = "document"

cdx-core/tests/conformance.rs

@@ -91,108 +91,9 @@ fn mixed_mark_array_deserializes() {
     assert_eq!(marks[2], Mark::Italic);
 }
 
-#[test]
-fn extension_mark_serializes_without_wrapper() {
-    use cdx_core::content::ExtensionMark;
-
-    let mark = Mark::Extension(ExtensionMark::citation("smith2023"));
-    let json = serde_json::to_string(&mark).unwrap();
-    let val: serde_json::Value = serde_json::from_str(&json).unwrap();
-
-    // Type should be "semantic:citation", not "extension"
-    assert_eq!(val["type"], "semantic:citation");
-    assert_eq!(val["refs"], serde_json::json!(["smith2023"]));
-
-    // Should NOT have wrapper fields
-    assert!(val.get("namespace").is_none());
-    assert!(val.get("markType").is_none());
-    // Should NOT have old singular "ref"
-    assert!(val.get("ref").is_none());
-}
-
-#[test]
-fn extension_mark_deserializes_new_format() {
-    let json = r#"{"type":"semantic:citation","refs":["smith2023"]}"#;
-    let mark: Mark = serde_json::from_str(json).unwrap();
-
-    if let Mark::Extension(ext) = &mark {
-        assert_eq!(ext.namespace, "semantic");
-        assert_eq!(ext.mark_type, "citation");
-        assert_eq!(
-            ext.get_string_array_attribute("refs"),
-            Some(vec!["smith2023"])
-        );
-    } else {
-        panic!("Expected Extension mark, got {mark:?}");
-    }
-}
-
-#[test]
-fn extension_mark_deserializes_old_format() {
-    // Backward compat: old "extension" wrapper format
-    let json = r#"{"type":"extension","namespace":"semantic","markType":"citation","attributes":{"ref":"smith2023"}}"#;
-    let mark: Mark = serde_json::from_str(json).unwrap();
-
-    if let Mark::Extension(ext) = &mark {
-        assert_eq!(ext.namespace, "semantic");
-        assert_eq!(ext.mark_type, "citation");
-        // Old format preserves "ref" as-is in attributes; use get_citation_refs for compat
-        assert_eq!(ext.get_citation_refs(), Some(vec!["smith2023"]));
-    } else {
-        panic!("Expected Extension mark, got {mark:?}");
-    }
-}
-
-#[test]
-fn citation_mark_backward_compat_singular_ref() {
-    // Old format with singular "ref" string
-    let json = r#"{"type":"semantic:citation","ref":"smith2023"}"#;
-    let mark: Mark = serde_json::from_str(json).unwrap();
-
-    if let Mark::Extension(ext) = &mark {
-        assert_eq!(ext.get_citation_refs(), Some(vec!["smith2023"]));
-    } else {
-        panic!("Expected Extension mark, got {mark:?}");
-    }
-}
-
-#[test]
-fn citation_mark_multi_refs_roundtrip() {
-    use cdx_core::content::ExtensionMark;
-
-    let refs = vec!["smith2023".into(), "jones2024".into()];
-    let mark = Mark::Extension(ExtensionMark::multi_citation(&refs));
-
-    let json = serde_json::to_string(&mark).unwrap();
-    let val: serde_json::Value = serde_json::from_str(&json).unwrap();
-    assert_eq!(val["refs"], serde_json::json!(["smith2023", "jones2024"]));
-
-    // Round-trip
-    let parsed: Mark = serde_json::from_str(&json).unwrap();
-    if let Mark::Extension(ext) = &parsed {
-        assert_eq!(
-            ext.get_string_array_attribute("refs"),
-            Some(vec!["smith2023", "jones2024"])
-        );
-    } else {
-        panic!("Expected Extension mark");
-    }
-}
-
-#[test]
-fn citation_mark_normalize_attrs() {
-    use cdx_core::content::ExtensionMark;
-
-    let mut ext = ExtensionMark::new("semantic", "citation")
-        .with_attributes(serde_json::json!({"ref": "smith2023"}));
-    ext.normalize_citation_attrs();
-
-    assert_eq!(
-        ext.get_string_array_attribute("refs"),
-        Some(vec!["smith2023"])
-    );
-    assert!(ext.get_string_attribute("ref").is_none());
-}
+// NOTE: Extension mark serialization, deserialization, and backward-compat
+// tests for citations and glossary have been consolidated into
+// tests/spec_compliance.rs (mark_schema_validation + backward_compatibility).
 
 #[test]
 fn math_mark_uses_source_field() {
@@ -326,32 +227,7 @@ fn spec_example_text_with_bold_string_mark() {
     assert_eq!(output_val, spec_val);
 }
 
-#[test]
-fn spec_example_text_with_citation_mark() {
-    // Spec: extension marks use "namespace:markType" as type, attributes flattened
-    let spec_json = r#"{"value":"important claim","marks":[{"type":"semantic:citation","refs":["smith2023"]}]}"#;
-
-    let text: Text = serde_json::from_str(spec_json).unwrap();
-    assert_eq!(text.value, "important claim");
-    assert_eq!(text.marks.len(), 1);
-
-    if let Mark::Extension(ext) = &text.marks[0] {
-        assert_eq!(ext.namespace, "semantic");
-        assert_eq!(ext.mark_type, "citation");
-        assert_eq!(
-            ext.get_string_array_attribute("refs"),
-            Some(vec!["smith2023"])
-        );
-    } else {
-        panic!("Expected Extension mark");
-    }
-
-    // Re-serialize matches spec format
-    let output = serde_json::to_string(&text).unwrap();
-    let output_val: serde_json::Value = serde_json::from_str(&output).unwrap();
-    let spec_val: serde_json::Value = serde_json::from_str(spec_json).unwrap();
-    assert_eq!(output_val, spec_val);
-}
+// NOTE: spec_example_text_with_citation_mark moved to spec_compliance.rs
 
 #[test]
 fn spec_example_extension_block_academic_theorem() {