A human-friendly DSL for ATProto Lexicons
Emit union refs as nsids
authored by
stavola.xyz
and committed by
Tangled
(Apr 15, 2026, 5:58 PM +0300)
0fbec7bc
1ad33d36
+66
-74
+66
-74
mlf-codegen/src/lib.rs
+66
-74
mlf-codegen/src/lib.rs
···
412
412
}
413
413
}
414
414
415
+
/// Resolve a typed reference path to the ATProto NSID string it names
416
+
/// (e.g. `#foo`, `app.bsky.actor.defs#profileViewBasic`, or a bare
417
+
/// `com.atproto.repo.strongRef` for implicit-main references). Used by
418
+
/// both direct `ref` types and union members — any place that needs to
419
+
/// emit the string form of a reference.
420
+
///
421
+
/// When workspace resolution fails (unresolved import, malformed path,
422
+
/// etc.) we fall back to a best-effort rendering rather than erroring:
423
+
/// the semantic check still happens in the resolver, so emitting a
424
+
/// recognisable-looking ref keeps the JSON self-describing for tooling.
425
+
fn resolve_ref_nsid(path: &Path, workspace: &Workspace, current_namespace: &str) -> String {
426
+
if let Some(full_namespace) = workspace.resolve_reference_namespace(path, current_namespace) {
427
+
let last_segment = path.segments.last().unwrap().name.as_str();
428
+
if full_namespace == current_namespace {
429
+
// Sibling def in the current lexicon.
430
+
return format!("#{}", last_segment);
431
+
}
432
+
return format!("{}#{}", full_namespace, last_segment);
433
+
}
434
+
435
+
if path.segments.len() == 1 {
436
+
let name = &path.segments[0].name;
437
+
// Single-segment path that didn't resolve via the workspace —
438
+
// maybe an imported symbol whose original path we can look up.
439
+
let imports = workspace.get_imports(current_namespace);
440
+
if let Some((_, original_path)) = imports.iter().find(|(local, _)| local == name) {
441
+
if original_path.len() > 1 {
442
+
let namespace = original_path[..original_path.len() - 1].join(".");
443
+
let type_name = original_path.last().unwrap();
444
+
return format!("{}#{}", namespace, type_name);
445
+
}
446
+
}
447
+
// Not imported either — assume it names a sibling def.
448
+
return format!("#{}", name);
449
+
}
450
+
451
+
// Multi-segment unresolved path: trust the author's NSID as written.
452
+
let namespace = path.segments[..path.segments.len() - 1]
453
+
.iter()
454
+
.map(|s| s.name.as_str())
455
+
.collect::<Vec<_>>()
456
+
.join(".");
457
+
let def_name = &path.segments.last().unwrap().name;
458
+
format!("{}#{}", namespace, def_name)
459
+
}
460
+
415
461
fn generate_record_json(record: &Record, usage_counts: &HashMap<String, usize>, workspace: &Workspace, current_namespace: &str) -> Value {
416
462
let mut required = Vec::new();
417
463
let mut properties = Map::new();
···
609
655
match ty {
610
656
Type::Primitive { kind, .. } => generate_primitive_json(*kind),
611
657
Type::Reference { path, .. } => {
612
-
// Try to resolve this reference in the workspace
613
-
if let Some(resolved_ty) = workspace.resolve_type_reference(path) {
614
-
// Check if this is an inline type by looking in the workspace
615
-
if workspace.is_inline_type(path) {
616
-
// Inline type: expand it recursively
658
+
// Inline types expand in place; other references resolve to an NSID.
659
+
if workspace.is_inline_type(path) {
660
+
if let Some(resolved_ty) = workspace.resolve_type_reference(path) {
617
661
return generate_type_json(&resolved_ty, usage_counts, workspace, current_namespace);
618
662
}
619
663
}
620
-
621
-
// Not an inline type (or couldn't resolve) - generate a ref
622
-
// First, try to get the fully resolved namespace for this type
623
-
if let Some(full_namespace) = workspace.resolve_reference_namespace(path, current_namespace) {
624
-
// We have the full namespace where this type is defined
625
-
if full_namespace == current_namespace {
626
-
// It's in the current namespace - use local reference
627
-
let type_name = path.segments.last().unwrap().name.as_str();
628
-
json!({
629
-
"type": "ref",
630
-
"ref": format!("#{}", type_name)
631
-
})
632
-
} else {
633
-
// It's in a different namespace - use full reference
634
-
let type_name = path.segments.last().unwrap().name.as_str();
635
-
json!({
636
-
"type": "ref",
637
-
"ref": format!("{}#{}", full_namespace, type_name)
638
-
})
639
-
}
640
-
} else if path.segments.len() == 1 {
641
-
// Couldn't resolve namespace - check if it's an imported type
642
-
let name = &path.segments[0].name;
643
-
let imports = workspace.get_imports(current_namespace);
644
-
645
-
// Look for this name in imports
646
-
if let Some((_local_name, original_path)) = imports.iter().find(|(local, _)| local == name) {
647
-
// Build the full namespace#type reference from the import path
648
-
// original_path is like ["com", "atproto", "label", "defs", "label"]
649
-
// We want "com.atproto.label.defs#label"
650
-
if original_path.len() > 1 {
651
-
let namespace = original_path[..original_path.len() - 1].join(".");
652
-
let type_name = original_path.last().unwrap();
653
-
json!({
654
-
"type": "ref",
655
-
"ref": format!("{}#{}", namespace, type_name)
656
-
})
657
-
} else {
658
-
// Fallback: single-segment import (shouldn't happen but handle it)
659
-
json!({
660
-
"type": "ref",
661
-
"ref": format!("#{}", name)
662
-
})
663
-
}
664
-
} else {
665
-
// Not an import - assume local reference
666
-
json!({
667
-
"type": "ref",
668
-
"ref": format!("#{}", name)
669
-
})
670
-
}
671
-
} else {
672
-
// Multi-segment path ref - use as-is
673
-
let namespace = path.segments[..path.segments.len()-1]
674
-
.iter()
675
-
.map(|s| s.name.as_str())
676
-
.collect::<Vec<_>>()
677
-
.join(".");
678
-
let def_name = &path.segments.last().unwrap().name;
679
-
680
-
json!({
681
-
"type": "ref",
682
-
"ref": format!("{}#{}", namespace, def_name)
683
-
})
684
-
}
664
+
let nsid = resolve_ref_nsid(path, workspace, current_namespace);
665
+
json!({ "type": "ref", "ref": nsid })
685
666
}
686
667
Type::Array { inner, .. } => {
687
668
json!({
···
690
671
})
691
672
}
692
673
Type::Union { types, closed, .. } => {
674
+
// ATProto unions carry an array of NSID *strings*. For ref
675
+
// members we resolve the NSID directly; for non-ref members
676
+
// (e.g. `null` produced by MLF's nullable-via-union pattern)
677
+
// we fall through to the typed form. That fallback is
678
+
// non-spec — ATProto unions strictly contain refs — but
679
+
// preserves the member's semantics until a proper nullable
680
+
// lowering is added upstream.
693
681
let refs: Vec<Value> = types
694
682
.iter()
695
-
.map(|t| generate_type_json(t, usage_counts, workspace, current_namespace))
683
+
.map(|t| match t {
684
+
Type::Reference { path, .. } => {
685
+
Value::String(resolve_ref_nsid(path, workspace, current_namespace))
686
+
}
687
+
_ => generate_type_json(t, usage_counts, workspace, current_namespace),
688
+
})
696
689
.collect();
697
690
let mut union_obj = Map::new();
698
691
union_obj.insert("type".to_string(), json!("union"));
699
-
union_obj.insert("refs".to_string(), json!(refs));
700
-
// Only emit "closed" field if true (closed unions)
701
-
// Open unions omit the field (defaults to false per ATProto spec)
692
+
union_obj.insert("refs".to_string(), Value::Array(refs));
693
+
// Only emit "closed" when true; open unions omit it per spec.
702
694
if *closed {
703
695
union_obj.insert("closed".to_string(), json!(true));
704
696
}