From ed50beb4282bfd4240bbb44042d9d3e17c937607 Mon Sep 17 00:00:00 2001
From: Jochen Hoenle <173445474+hoe-jo@users.noreply.github.com>
Date: Tue, 21 Jul 2026 11:00:06 +0200
Subject: [PATCH] [validator] check correct diagram schema before opening

---
 .../component/component_diagram.fbs           |  3 +++
 .../component/component_serializer.rs         |  2 +-
 .../core/src/readers/class_diagram_reader.rs  |  8 +++++++
 .../src/readers/component_diagram_reader.rs   | 22 ++++++-------------
 .../src/readers/sequence_diagram_reader.rs    |  8 +++++++
 5 files changed, 27 insertions(+), 16 deletions(-)

diff --git a/tools/serialization/flatbuffers/component/component_diagram.fbs b/tools/serialization/flatbuffers/component/component_diagram.fbs
index ce0afae8..e320b426 100644
--- a/tools/serialization/flatbuffers/component/component_diagram.fbs
+++ b/tools/serialization/flatbuffers/component/component_diagram.fbs
@@ -84,3 +84,6 @@ table ComponentGraph {
 }
 
 root_type ComponentGraph;
+
+file_identifier "COMD";
+file_extension "comd";
diff --git a/tools/serialization/flatbuffers/component/component_serializer.rs b/tools/serialization/flatbuffers/component/component_serializer.rs
index 85d9dafc..526cdaee 100644
--- a/tools/serialization/flatbuffers/component/component_serializer.rs
+++ b/tools/serialization/flatbuffers/component/component_serializer.rs
@@ -123,7 +123,7 @@ impl ComponentSerializer {
         // --------------------------
         // 4) finish
         // --------------------------
-        builder.finish(root, None);
+        builder.finish(root, Some("COMD"));
 
         builder.finished_data().to_vec()
     }
diff --git a/validation/core/src/readers/class_diagram_reader.rs b/validation/core/src/readers/class_diagram_reader.rs
index b93b29c4..af6f9c42 100644
--- a/validation/core/src/readers/class_diagram_reader.rs
+++ b/validation/core/src/readers/class_diagram_reader.rs
@@ -304,6 +304,14 @@ impl Reader for ClassDiagramReader {
         for path in input {
             let data = fs::read(path).map_err(|e| format!("Failed to read {path}: {e}"))?;
 
+            // The PlantUML parser auto-detects diagram kind; a `.puml` may end up parsed
+            // as a different type (e.g. Component). Skip mismatched buffers instead of
+            // letting the verifier fail on misaligned data.
+            if !fb_class::class_diagram_buffer_has_identifier(&data) {
+                log::warn!("{path}: not a class-diagram, skipping validation");
+                continue;
+            }
+
             let diagram = flatbuffers::root::<fb_class::ClassDiagram>(&data)
                 .map_err(|e| format!("Failed to parse class FlatBuffer {path}: {e}"))?;
 
diff --git a/validation/core/src/readers/component_diagram_reader.rs b/validation/core/src/readers/component_diagram_reader.rs
index 357c0b2b..239329bd 100644
--- a/validation/core/src/readers/component_diagram_reader.rs
+++ b/validation/core/src/readers/component_diagram_reader.rs
@@ -90,27 +90,19 @@ impl ComponentDiagramReader {
     /// Read all `Component` and `Package` entities from the given FlatBuffers
     /// binary files.
     ///
-    /// Files that carry a known non-component file identifier (e.g. `CLSD` for
-    /// class diagrams or `SEQD` for sequence diagrams) are silently skipped so
-    /// that callers can pass all architectural-design FlatBuffers without
-    /// pre-filtering by diagram type.
+    /// Files that don't carry the component-diagram file identifier (`COMD`) —
+    /// e.g. because the PlantUML source was auto-detected as a class, sequence,
+    /// or other diagram type — are silently skipped so that callers can pass
+    /// all architectural-design FlatBuffers without pre-filtering by diagram type.
     pub fn read(paths: &[String]) -> Result<ComponentDiagramInputs, String> {
         let mut out = Vec::new();
 
         for path in paths {
             let data = fs::read(path).map_err(|e| format!("Failed to read {path}: {e}"))?;
 
-            // FlatBuffers stores the file identifier at bytes 4-7 when one is
-            // present.  Component diagrams are written without an identifier
-            // (builder.finish(root, None)), so bytes 4-7 are regular data.
-            // Class diagrams ("CLSD") and sequence diagrams ("SEQD") carry an
-            // explicit identifier.  Skip such files here; they are not
-            // component diagrams and must not be parsed with the component schema.
-            if data.len() >= 8 {
-                let file_id = &data[4..8];
-                if file_id == b"CLSD" || file_id == b"SEQD" {
-                    continue;
-                }
+            if !fb_component::component_graph_buffer_has_identifier(&data) {
+                log::warn!("{path}: not a component-diagram, skipping validation");
+                continue;
             }
 
             let graph = flatbuffers::root::<fb_component::ComponentGraph>(&data)
diff --git a/validation/core/src/readers/sequence_diagram_reader.rs b/validation/core/src/readers/sequence_diagram_reader.rs
index 68400ae8..aa4f24da 100644
--- a/validation/core/src/readers/sequence_diagram_reader.rs
+++ b/validation/core/src/readers/sequence_diagram_reader.rs
@@ -36,6 +36,14 @@ impl Reader for SequenceDiagramReader {
         for path in input {
             let data = fs::read(path).map_err(|e| format!("Failed to read {path}: {e}"))?;
 
+            // See the analogous check in class_diagram_reader.rs: skip a buffer that
+            // doesn't match this schema instead of letting the verifier fail on
+            // misaligned data.
+            if !fb_sequence::sequence_diagram_buffer_has_identifier(&data) {
+                log::warn!("{path}: not a sequence-diagram, skipping validation");
+                continue;
+            }
+
             let diagram = flatbuffers::root::<fb_sequence::SequenceDiagram>(&data)
                 .map_err(|e| format!("Failed to parse sequence FlatBuffer {path}: {e}"))?;