1use super::Form;
2use num_bigint::BigInt;
3use num_traits::ToPrimitive;
4use sha2::{Digest, Sha256};
5use std::collections::{HashMap, HashSet};
6
7const MAGIC: &[u8] = b"HALC";
8const LEGACY_MAGIC: &[u8] = b"HIR\0";
9const FORMAT_VERSION: u16 = 1;
10const EXECUTABLE_FOUNDATION_FLAG: u16 = 1;
11const HASH_BYTES: usize = 32;
12const MAX_PAYLOAD_BYTES: usize = 64 * 1024 * 1024;
13const MAX_COLLECTION_ITEMS: i32 = 1_000_000;
14
15const NIL: u8 = 0;
16const FALSE: u8 = 1;
17const TRUE: u8 = 2;
18const LONG: u8 = 3;
19const DOUBLE: u8 = 4;
20const BIG_INTEGER: u8 = 5;
21const STRING: u8 = 6;
22const CHARACTER: u8 = 8;
23const SYMBOL: u8 = 9;
24const KEYWORD: u8 = 10;
25const LIST: u8 = 11;
26const VECTOR: u8 = 12;
27const MAP: u8 = 13;
28const SET: u8 = 14;
29const ORDERED_MAP: u8 = 15;
30const ORDERED_SET: u8 = 16;
31const REGEX: u8 = 17;
32
33#[derive(Debug, Clone)]
34pub struct HalcModule {
35 pub namespace: String,
36 pub resource: String,
37 pub source_hash: Vec<u8>,
38 pub forms: Vec<Form>,
39 pub schemas: HalcSchemaIndex,
40 pub origin: HalcOrigin,
41}
42
43#[derive(Debug, Clone, Default, PartialEq)]
46pub struct HalcSchemaIndex {
47 pub definitions: HashMap<String, Form>,
48 pub functions: HashMap<String, Form>,
49 pub definition_types: HashMap<String, super::SchemaType>,
50 pub function_types: HashMap<String, super::SchemaType>,
51}
52
53impl HalcSchemaIndex {
54 pub fn resolved_function_type(&self, qualified_var: &str) -> Option<&super::SchemaType> {
57 let schema = self.function_types.get(qualified_var)?;
58 match schema {
59 super::SchemaType::Reference(name) => self.definition_types.get(name).or(Some(schema)),
60 _ => Some(schema),
61 }
62 }
63}
64
65#[derive(Debug, Clone, Copy, PartialEq, Eq)]
66pub enum HalcOrigin {
67 Halc,
68 LegacyHir,
69}
70
71pub fn decode_halc(bytes: &[u8]) -> Result<HalcModule, String> {
72 let (payload, origin) = decode_envelope(bytes)?;
73 let mut reader = ByteReader::new(&payload);
74 let namespace = reader.read_string()?;
75 let resource = reader.read_string()?;
76 let source_hash = reader.read_bytes(HASH_BYTES)?;
77 let form_count = reader.read_count()?;
78 let mut forms = Vec::with_capacity(form_count as usize);
79 for _ in 0..form_count {
80 forms.push(reader.read_value()?);
81 }
82 if !reader.is_empty() {
83 return Err("trailing payload bytes".into());
84 }
85 let forms = canonicalize_schema_references(&namespace, forms)?;
86 let schemas = build_schema_index(&namespace, &forms)?;
87 Ok(HalcModule {
88 namespace,
89 resource,
90 source_hash,
91 forms,
92 schemas,
93 origin,
94 })
95}
96
97fn decode_envelope(bytes: &[u8]) -> Result<(Vec<u8>, HalcOrigin), String> {
98 let mut reader = ByteReader::new(bytes);
99 let magic = reader.read_bytes(MAGIC.len())?;
100 let origin = if magic == MAGIC {
101 HalcOrigin::Halc
102 } else if magic == LEGACY_MAGIC {
103 HalcOrigin::LegacyHir
104 } else {
105 return Err("bad magic".into());
106 };
107 let version = reader.read_u16()?;
108 if version != FORMAT_VERSION {
109 return Err(format!("unsupported format version {version}"));
110 }
111 let flags = reader.read_u16()?;
112 if flags != EXECUTABLE_FOUNDATION_FLAG {
113 return Err(format!("unsupported flags {flags}"));
114 }
115 let payload_length = reader.read_u32()? as usize;
116 if payload_length > MAX_PAYLOAD_BYTES {
117 return Err(format!("invalid payload length {payload_length}"));
118 }
119 let expected_hash = reader.read_bytes(HASH_BYTES)?;
120 let payload = reader.read_bytes(payload_length)?;
121 if !reader.is_empty() {
122 return Err("trailing bytes".into());
123 }
124 let actual_hash = Sha256::digest(&payload);
125 if actual_hash[..] != expected_hash[..] {
126 return Err("payload checksum mismatch".into());
127 }
128 Ok((payload, origin))
129}
130
131struct ByteReader<'a> {
132 bytes: &'a [u8],
133 position: usize,
134}
135
136impl<'a> ByteReader<'a> {
137 fn new(bytes: &'a [u8]) -> Self {
138 Self { bytes, position: 0 }
139 }
140
141 fn is_empty(&self) -> bool {
142 self.position >= self.bytes.len()
143 }
144
145 fn remaining(&self) -> usize {
146 self.bytes.len().saturating_sub(self.position)
147 }
148
149 fn read_byte(&mut self) -> Result<u8, String> {
150 if self.position >= self.bytes.len() {
151 return Err("truncated artifact".into());
152 }
153 let byte = self.bytes[self.position];
154 self.position += 1;
155 Ok(byte)
156 }
157
158 fn read_bytes(&mut self, count: usize) -> Result<Vec<u8>, String> {
159 if self.remaining() < count {
160 return Err("truncated artifact".into());
161 }
162 let bytes = self.bytes[self.position..self.position + count].to_vec();
163 self.position += count;
164 Ok(bytes)
165 }
166
167 fn read_u16(&mut self) -> Result<u16, String> {
168 let bytes = self.read_bytes(2)?;
169 Ok(u16::from_be_bytes([bytes[0], bytes[1]]))
170 }
171
172 fn read_u32(&mut self) -> Result<u32, String> {
173 let bytes = self.read_bytes(4)?;
174 Ok(u32::from_be_bytes([bytes[0], bytes[1], bytes[2], bytes[3]]))
175 }
176
177 fn read_i64(&mut self) -> Result<i64, String> {
178 let bytes = self.read_bytes(8)?;
179 Ok(i64::from_be_bytes([
180 bytes[0], bytes[1], bytes[2], bytes[3], bytes[4], bytes[5], bytes[6], bytes[7],
181 ]))
182 }
183
184 fn read_f64(&mut self) -> Result<f64, String> {
185 let bytes = self.read_bytes(8)?;
186 let value = f64::from_be_bytes([
187 bytes[0], bytes[1], bytes[2], bytes[3], bytes[4], bytes[5], bytes[6], bytes[7],
188 ]);
189 if !value.is_finite() {
190 return Err("non-finite number".into());
191 }
192 Ok(value)
193 }
194
195 fn read_string(&mut self) -> Result<String, String> {
196 let length = self.read_u32()? as usize;
197 if length > MAX_PAYLOAD_BYTES {
198 return Err(format!("invalid string length {length}"));
199 }
200 let bytes = self.read_bytes(length)?;
201 String::from_utf8(bytes).map_err(|_| "invalid UTF-8 in string".to_string())
202 }
203
204 fn read_nullable_string(&mut self) -> Result<Option<String>, String> {
205 let present = self.read_byte()? != 0;
206 if present {
207 Ok(Some(self.read_string()?))
208 } else {
209 Ok(None)
210 }
211 }
212
213 fn read_count(&mut self) -> Result<i32, String> {
214 let count = self.read_u32()? as i32;
215 if count < 0 || count > MAX_COLLECTION_ITEMS {
216 return Err(format!("invalid collection count {count}"));
217 }
218 Ok(count)
219 }
220
221 fn read_metadata(&mut self) -> Result<Option<Form>, String> {
222 let present = self.read_byte()? != 0;
223 if present {
224 Ok(Some(self.read_value()?))
225 } else {
226 Ok(None)
227 }
228 }
229
230 fn read_value(&mut self) -> Result<Form, String> {
231 let opcode = self.read_byte()?;
232 match opcode {
233 NIL => Ok(Form::Nil),
234 FALSE => Ok(Form::Bool(false)),
235 TRUE => Ok(Form::Bool(true)),
236 LONG => Ok(Form::Number(self.read_i64()?)),
237 DOUBLE => Ok(Form::Float(self.read_f64()?)),
238 BIG_INTEGER => {
239 let text = self.read_string()?;
240 let value = BigInt::parse_bytes(text.as_bytes(), 10)
241 .ok_or_else(|| "invalid big integer".to_string())?;
242 Ok(match value.to_i64() {
243 Some(value) => Form::Number(value),
244 None => Form::BigInteger(value),
245 })
246 }
247 STRING => Ok(Form::String(self.read_string()?)),
248 CHARACTER => Ok(Form::Character(
249 char::from_u32(self.read_u32()?).ok_or("invalid character code point")?,
250 )),
251 SYMBOL => {
252 let namespace = self.read_nullable_string()?;
253 let name = self.read_string()?;
254 Ok(with_metadata(
255 Form::Symbol(namespaced(namespace, name)),
256 self.read_metadata()?,
257 ))
258 }
259 KEYWORD => {
260 let namespace = self.read_nullable_string()?;
261 let name = self.read_string()?;
262 Ok(with_metadata(
263 Form::Keyword(namespaced(namespace, name)),
264 self.read_metadata()?,
265 ))
266 }
267 LIST => {
268 let count = self.read_count()?;
269 let items = self.read_values(count)?;
270 Ok(with_metadata(Form::List(items), self.read_metadata()?))
271 }
272 VECTOR => {
273 let count = self.read_count()?;
274 let items = self.read_values(count)?;
275 Ok(with_metadata(Form::Vector(items), self.read_metadata()?))
276 }
277 MAP | ORDERED_MAP => {
278 let count = self.read_count()?;
279 let mut entries = Vec::with_capacity(count as usize);
280 for _ in 0..count {
281 let key = self.read_value()?;
282 let value = self.read_value()?;
283 entries.push((key, value));
284 }
285 Ok(with_metadata(Form::Map(entries), self.read_metadata()?))
286 }
287 SET | ORDERED_SET => {
288 let count = self.read_count()?;
289 let items = self.read_values(count)?;
290 Ok(with_metadata(Form::Set(items), self.read_metadata()?))
291 }
292 REGEX => Ok(Form::Regex(self.read_string()?)),
293 _ => Err(format!("unknown value opcode {opcode}")),
294 }
295 }
296
297 fn read_values(&mut self, count: i32) -> Result<Vec<Form>, String> {
298 let mut values = Vec::with_capacity(count as usize);
299 for _ in 0..count {
300 values.push(self.read_value()?);
301 }
302 Ok(values)
303 }
304}
305
306fn with_metadata(value: Form, metadata: Option<Form>) -> Form {
307 match metadata {
308 Some(metadata) => Form::Metadata(Box::new(metadata), Box::new(value)),
309 None => value,
310 }
311}
312
313fn namespaced(namespace: Option<String>, name: String) -> String {
314 match namespace {
315 Some(ns) => format!("{ns}/{name}"),
316 None => name,
317 }
318}
319
320#[cfg(any(test, feature = "halc-encoder"))]
321fn write_string(output: &mut Vec<u8>, value: &str) {
322 output.extend_from_slice(&(value.len() as u32).to_be_bytes());
323 output.extend_from_slice(value.as_bytes());
324}
325
326#[cfg(any(test, feature = "halc-encoder"))]
327fn write_count(output: &mut Vec<u8>, count: i32) {
328 output.extend_from_slice(&count.to_be_bytes());
329}
330
331#[cfg(any(test, feature = "halc-encoder"))]
332fn write_namespaced(output: &mut Vec<u8>, symbol: &str) {
333 if let Some((ns, name)) = symbol.rsplit_once('/') {
334 output.push(1);
335 write_string(output, ns);
336 write_string(output, name);
337 } else {
338 output.push(0);
339 write_string(output, symbol);
340 }
341}
342
343#[cfg(any(test, feature = "halc-encoder"))]
344fn write_values(output: &mut Vec<u8>, values: &[Form]) {
345 write_count(output, values.len() as i32);
346 for value in values {
347 write_value(output, value);
348 }
349}
350
351#[cfg(any(test, feature = "halc-encoder"))]
352fn write_value(output: &mut Vec<u8>, form: &Form) {
353 match form {
354 Form::Metadata(metadata, value) => write_value_with_metadata(output, value, Some(metadata)),
355 _ => write_value_with_metadata(output, form, None),
356 }
357}
358
359#[cfg(any(test, feature = "halc-encoder"))]
360fn write_metadata(output: &mut Vec<u8>, metadata: Option<&Form>) {
361 match metadata {
362 Some(metadata) => {
363 output.push(1);
364 write_value(output, metadata);
365 }
366 None => output.push(0),
367 }
368}
369
370#[cfg(any(test, feature = "halc-encoder"))]
371fn write_value_with_metadata(output: &mut Vec<u8>, form: &Form, metadata: Option<&Form>) {
372 match form {
373 Form::Nil => output.push(NIL),
374 Form::Bool(false) => output.push(FALSE),
375 Form::Bool(true) => output.push(TRUE),
376 Form::Number(n) => {
377 output.push(LONG);
378 output.extend_from_slice(&n.to_be_bytes());
379 }
380 Form::Float(f) => {
381 assert!(f.is_finite(), "non-finite number");
382 output.push(DOUBLE);
383 output.extend_from_slice(&f.to_be_bytes());
384 }
385 Form::BigInteger(s) => {
386 output.push(BIG_INTEGER);
387 write_string(output, &s.to_string());
388 }
389 Form::String(s) => {
390 output.push(STRING);
391 write_string(output, s);
392 }
393 Form::Character(c) => {
394 output.push(CHARACTER);
395 output.extend_from_slice(&(*c as u32).to_be_bytes());
396 }
397 Form::Symbol(s) => {
398 output.push(SYMBOL);
399 write_namespaced(output, s);
400 write_metadata(output, metadata);
401 }
402 Form::Keyword(s) => {
403 output.push(KEYWORD);
404 write_namespaced(output, s);
405 write_metadata(output, metadata);
406 }
407 Form::List(items) => {
408 output.push(LIST);
409 write_values(output, items);
410 write_metadata(output, metadata);
411 }
412 Form::Vector(items) => {
413 output.push(VECTOR);
414 write_values(output, items);
415 write_metadata(output, metadata);
416 }
417 Form::Map(entries) => {
418 output.push(ORDERED_MAP);
419 write_count(output, entries.len() as i32);
420 for (key, value) in entries {
421 write_value(output, key);
422 write_value(output, value);
423 }
424 write_metadata(output, metadata);
425 }
426 Form::Set(items) => {
427 output.push(ORDERED_SET);
428 write_values(output, items);
429 write_metadata(output, metadata);
430 }
431 Form::Regex(s) => {
432 output.push(REGEX);
433 write_string(output, s);
434 }
435 Form::Tagged(_, _) | Form::Metadata(_, _) => {
436 panic!("test encoder does not support tagged/metadata forms")
437 }
438 }
439}
440
441#[cfg(any(test, feature = "halc-encoder"))]
446pub fn encode_halc_module(
447 namespace: &str,
448 resource: &str,
449 source: &str,
450 forms: Vec<Form>,
451) -> Result<Vec<u8>, String> {
452 let forms = canonicalize_schema_references(namespace, forms)?;
453 for form in &forms {
454 validate_finite_form(form)?;
455 }
456 build_schema_index(namespace, &forms)?;
457 let mut payload = Vec::new();
458 write_string(&mut payload, namespace);
459 write_string(&mut payload, resource);
460 payload.extend_from_slice(&Sha256::digest(source.as_bytes()));
461 write_count(&mut payload, forms.len() as i32);
462 for form in forms {
463 write_value(&mut payload, &form);
464 }
465 let mut artifact = Vec::new();
466 artifact.extend_from_slice(MAGIC);
467 artifact.extend_from_slice(&FORMAT_VERSION.to_be_bytes());
468 artifact.extend_from_slice(&EXECUTABLE_FOUNDATION_FLAG.to_be_bytes());
469 artifact.extend_from_slice(&(payload.len() as u32).to_be_bytes());
470 artifact.extend_from_slice(&Sha256::digest(&payload));
471 artifact.extend_from_slice(&payload);
472 Ok(artifact)
473}
474
475#[cfg(any(test, feature = "halc-encoder"))]
476fn validate_finite_form(form: &Form) -> Result<(), String> {
477 match form {
478 Form::Float(value) if !value.is_finite() => Err("non-finite number".into()),
479 Form::Tagged(_, value) => validate_finite_form(value),
480 Form::Metadata(metadata, value) => {
481 validate_finite_form(metadata)?;
482 validate_finite_form(value)
483 }
484 Form::Map(entries) => {
485 for (key, value) in entries {
486 validate_finite_form(key)?;
487 validate_finite_form(value)?;
488 }
489 Ok(())
490 }
491 Form::Set(values) | Form::Vector(values) | Form::List(values) => {
492 for value in values {
493 validate_finite_form(value)?;
494 }
495 Ok(())
496 }
497 _ => Ok(()),
498 }
499}
500
501fn canonicalize_schema_references(
502 namespace: &str,
503 mut forms: Vec<Form>,
504) -> Result<Vec<Form>, String> {
505 let definitions: HashSet<String> = forms
506 .iter()
507 .filter_map(|form| {
508 let Form::List(items) = form else { return None };
509 let Form::Symbol(operator) = items.first()? else {
510 return None;
511 };
512 if !matches!(
513 operator.as_str(),
514 "def" | "defn" | "defmacro" | "defstruct" | "declare"
515 ) {
516 return None;
517 }
518 binding_name(items.get(1)?).map(str::to_owned)
519 })
520 .collect();
521 let schema_values: HashMap<String, usize> = forms
522 .iter()
523 .enumerate()
524 .filter_map(|(index, form)| {
525 let Form::List(items) = form else { return None };
526 if !matches!(items.first(), Some(Form::Symbol(operator)) if operator == "def") {
527 return None;
528 }
529 binding_name(items.get(1)?).map(|name| (name.to_owned(), index))
530 })
531 .collect();
532 let aliases = module_aliases(&forms);
533 let mut schema_roots = Vec::new();
534
535 for form in &mut forms {
536 let Form::List(items) = form else { continue };
537 let Some(Form::Symbol(operator)) = items.first() else {
538 continue;
539 };
540 if operator != "defn" {
541 continue;
542 }
543 let Some(Form::Metadata(metadata, _)) = items.get_mut(1) else {
544 continue;
545 };
546 let Form::Map(entries) = metadata.as_mut() else {
547 continue;
548 };
549 let Some((_, schema)) = entries
550 .iter_mut()
551 .find(|(key, _)| matches!(key, Form::Keyword(name) if name == "schema"))
552 else {
553 continue;
554 };
555 let Form::List(reference) = schema else {
556 continue;
557 };
558 if reference.len() != 2
559 || !matches!(&reference[0], Form::Symbol(operator) if operator == "var")
560 {
561 continue;
562 }
563 let Form::Symbol(target) = &reference[1] else {
564 continue;
565 };
566 let (qualifier, local) = target
567 .rsplit_once('/')
568 .map_or((None, target.as_str()), |(qualifier, local)| {
569 (Some(qualifier), local)
570 });
571 let target_namespace = match qualifier {
572 None | Some("-") => namespace,
573 Some(qualifier) => aliases.get(qualifier).map_or(qualifier, String::as_str),
574 };
575 if target_namespace == namespace && !definitions.contains(local) {
576 return Err(format!("schema Var does not exist: {target}"));
577 }
578 if target_namespace == namespace {
579 schema_roots.push(local.to_owned());
580 }
581 reference[1] = Form::Symbol(format!("{target_namespace}/{local}"));
582 }
583
584 let mut visited = HashSet::new();
585 while let Some(schema_name) = schema_roots.pop() {
586 if !visited.insert(schema_name.clone()) {
587 continue;
588 }
589 let Some(index) = schema_values.get(&schema_name).copied() else {
590 continue;
591 };
592 let Form::List(definition) = &mut forms[index] else {
593 continue;
594 };
595 let Some(schema_value) = definition.get_mut(2) else {
596 continue;
597 };
598 canonicalize_nested_schema_references(
599 schema_value,
600 namespace,
601 &aliases,
602 &definitions,
603 &mut schema_roots,
604 )?;
605 }
606 Ok(forms)
607}
608
609fn canonicalize_nested_schema_references(
610 form: &mut Form,
611 namespace: &str,
612 aliases: &HashMap<String, String>,
613 definitions: &HashSet<String>,
614 local_references: &mut Vec<String>,
615) -> Result<(), String> {
616 if let Form::List(reference) = form {
617 if reference.len() == 2
618 && matches!(&reference[0], Form::Symbol(operator) if operator == "var")
619 {
620 let Form::Symbol(target) = &reference[1] else {
621 return Ok(());
622 };
623 let original = target.clone();
624 let (qualifier, local) = original
625 .rsplit_once('/')
626 .map_or((None, original.as_str()), |(qualifier, local)| {
627 (Some(qualifier), local)
628 });
629 let target_namespace = match qualifier {
630 None | Some("-") => namespace,
631 Some(qualifier) => aliases.get(qualifier).map_or(qualifier, String::as_str),
632 };
633 if target_namespace == namespace {
634 if !definitions.contains(local) {
635 return Err(format!("schema Var does not exist: {original}"));
636 }
637 local_references.push(local.to_owned());
638 }
639 reference[1] = Form::Symbol(format!("{target_namespace}/{local}"));
640 return Ok(());
641 }
642 }
643
644 match form {
645 Form::Tagged(_, value) => canonicalize_nested_schema_references(
646 value,
647 namespace,
648 aliases,
649 definitions,
650 local_references,
651 ),
652 Form::Metadata(metadata, value) => {
653 canonicalize_nested_schema_references(
654 metadata,
655 namespace,
656 aliases,
657 definitions,
658 local_references,
659 )?;
660 canonicalize_nested_schema_references(
661 value,
662 namespace,
663 aliases,
664 definitions,
665 local_references,
666 )
667 }
668 Form::Map(entries) => {
669 for (key, value) in entries {
670 canonicalize_nested_schema_references(
671 key,
672 namespace,
673 aliases,
674 definitions,
675 local_references,
676 )?;
677 canonicalize_nested_schema_references(
678 value,
679 namespace,
680 aliases,
681 definitions,
682 local_references,
683 )?;
684 }
685 Ok(())
686 }
687 Form::Set(values) | Form::Vector(values) | Form::List(values) => {
688 for value in values {
689 canonicalize_nested_schema_references(
690 value,
691 namespace,
692 aliases,
693 definitions,
694 local_references,
695 )?;
696 }
697 Ok(())
698 }
699 _ => Ok(()),
700 }
701}
702
703fn binding_name(form: &Form) -> Option<&str> {
704 match form {
705 Form::Symbol(name) => Some(name),
706 Form::Metadata(_, value) => binding_name(value),
707 _ => None,
708 }
709}
710
711fn module_aliases(forms: &[Form]) -> HashMap<String, String> {
712 let mut aliases = HashMap::new();
713 for form in forms {
714 let Form::List(declaration) = form else {
715 continue;
716 };
717 if !matches!(declaration.first(), Some(Form::Symbol(operator)) if operator == "ns") {
718 continue;
719 }
720 for clause in declaration.iter().skip(2) {
721 let Form::List(clause) = clause else { continue };
722 if !matches!(clause.first(), Some(Form::Keyword(keyword)) if keyword == "require") {
723 continue;
724 }
725 for spec in clause.iter().skip(1) {
726 let Form::Vector(spec) = spec else { continue };
727 let Some(Form::Symbol(target)) = spec.first() else {
728 continue;
729 };
730 for option in spec[1..].chunks(2) {
731 if let [Form::Keyword(key), Form::Symbol(alias)] = option {
732 if key == "as" {
733 aliases.insert(alias.clone(), target.clone());
734 }
735 }
736 }
737 }
738 }
739 }
740 aliases
741}
742
743fn build_schema_index(namespace: &str, forms: &[Form]) -> Result<HalcSchemaIndex, String> {
744 let mut index = HalcSchemaIndex::default();
745 let mut values = HashMap::new();
746 let mut roots = Vec::new();
747
748 for form in forms {
749 let Form::List(items) = form else { continue };
750 let Some(Form::Symbol(operator)) = items.first() else {
751 continue;
752 };
753 let Some(name) = items.get(1).and_then(binding_name) else {
754 continue;
755 };
756 let qualified_name = format!("{namespace}/{name}");
757 if operator == "def" {
758 if let Some(value) = items.get(2) {
759 values.insert(name.to_owned(), value.clone());
760 }
761 continue;
762 }
763 if operator != "defn" {
764 continue;
765 }
766 let Some(Form::Metadata(metadata, _)) = items.get(1) else {
767 continue;
768 };
769 let Form::Map(entries) = metadata.as_ref() else {
770 continue;
771 };
772 let Some(schema) = entries.iter().find_map(|(key, value)| {
773 matches!(key, Form::Keyword(name) if name == "schema").then_some(value)
774 }) else {
775 continue;
776 };
777 index.functions.insert(qualified_name, schema.clone());
778 collect_local_schema_references(schema, namespace, &mut roots);
779 }
780
781 let mut visited = HashSet::new();
782 while let Some(name) = roots.pop() {
783 if !visited.insert(name.clone()) {
784 continue;
785 }
786 let Some(value) = values.get(&name) else {
787 continue;
788 };
789 index
790 .definitions
791 .insert(format!("{namespace}/{name}"), value.clone());
792 collect_local_schema_references(value, namespace, &mut roots);
793 }
794 for (name, schema) in &index.definitions {
795 index.definition_types.insert(
796 name.clone(),
797 super::normalize_schema(schema)
798 .map_err(|error| format!("invalid schema {name}: {error}"))?,
799 );
800 }
801 for (name, schema) in &index.functions {
802 index.function_types.insert(
803 name.clone(),
804 super::normalize_schema(schema)
805 .map_err(|error| format!("invalid function schema {name}: {error}"))?,
806 );
807 }
808 Ok(index)
809}
810
811fn collect_local_schema_references(form: &Form, namespace: &str, output: &mut Vec<String>) {
812 if let Form::List(reference) = form {
813 if reference.len() == 2
814 && matches!(&reference[0], Form::Symbol(operator) if operator == "var")
815 {
816 if let Form::Symbol(target) = &reference[1] {
817 if let Some((qualifier, local)) = target.rsplit_once('/') {
818 if qualifier == namespace {
819 output.push(local.to_owned());
820 }
821 }
822 }
823 return;
824 }
825 }
826 match form {
827 Form::Tagged(_, value) => collect_local_schema_references(value, namespace, output),
828 Form::Metadata(metadata, value) => {
829 collect_local_schema_references(metadata, namespace, output);
830 collect_local_schema_references(value, namespace, output);
831 }
832 Form::Map(entries) => {
833 for (key, value) in entries {
834 collect_local_schema_references(key, namespace, output);
835 collect_local_schema_references(value, namespace, output);
836 }
837 }
838 Form::Set(values) | Form::Vector(values) | Form::List(values) => {
839 for value in values {
840 collect_local_schema_references(value, namespace, output);
841 }
842 }
843 _ => {}
844 }
845}
846
847#[cfg(test)]
848mod tests {
849 use super::*;
850 use crate::kernel::parse;
851
852 fn artifact_payload(forms: Vec<Form>) -> Vec<u8> {
853 encode_halc_module("demo.ns", "demo.hal", "", forms).unwrap()
854 }
855
856 #[test]
857 fn round_trips_primitive_values() {
858 let cases = [
859 "nil",
860 "true",
861 "false",
862 "42",
863 "-7",
864 "3.14",
865 "\"hello\"",
866 "\\x",
867 ":key",
868 ":ns/key",
869 "symbol",
870 "ns/symbol",
871 ];
872 for source in cases {
873 let original = parse(source).unwrap();
874 let bytes = artifact_payload(vec![original.clone()]);
875 let decoded = decode_halc(&bytes).unwrap();
876 assert_eq!(decoded.forms.len(), 1);
877 assert_eq!(decoded.forms[0], original, "{source}");
878 }
879 }
880
881 #[test]
882 fn round_trips_collections() {
883 let original = parse("(do {:a [1 2] :b #{3 4}})").unwrap();
884 let bytes = artifact_payload(vec![original.clone()]);
885 let decoded = decode_halc(&bytes).unwrap();
886 assert_eq!(decoded.forms.len(), 1);
887 assert_eq!(decoded.forms[0], original);
888 }
889
890 #[test]
891 fn schema_var_references_are_checked_and_namespace_canonicalized() {
892 let source = "(ns demo.schema) \
893 (def Customer [:map [:id :int]]) \
894 (defn ^{:schema #'-/Customer} customer-id [customer] customer)";
895 let bytes = encode_halc_module(
896 "demo.schema",
897 "demo/schema.hal",
898 source,
899 crate::kernel::parse_forms(source).unwrap(),
900 )
901 .unwrap();
902 let module = decode_halc(&bytes).unwrap();
903 let Form::List(definition) = &module.forms[2] else {
904 panic!("expected defn");
905 };
906 let Form::Metadata(metadata, _) = &definition[1] else {
907 panic!("expected definition metadata");
908 };
909 let Form::Map(metadata) = metadata.as_ref() else {
910 panic!("expected metadata map");
911 };
912 let schema = metadata
913 .iter()
914 .find_map(|(key, value)| {
915 matches!(key, Form::Keyword(name) if name == "schema").then_some(value)
916 })
917 .unwrap();
918 assert_eq!(
919 schema,
920 &Form::List(vec![
921 Form::Symbol("var".into()),
922 Form::Symbol("demo.schema/Customer".into()),
923 ])
924 );
925
926 let missing = "(ns demo.schema) \
927 (defn ^{:schema #'MissingSchema} invalid [value] value)";
928 assert_eq!(
929 encode_halc_module(
930 "demo.schema",
931 "demo/schema.hal",
932 missing,
933 crate::kernel::parse_forms(missing).unwrap(),
934 )
935 .unwrap_err(),
936 "schema Var does not exist: MissingSchema"
937 );
938 }
939
940 #[test]
941 fn nested_schema_var_references_are_canonicalized_and_checked() {
942 let source = "(ns demo.schema) \
943 (def Address [:map [:street :str]]) \
944 (def Customer [:map [:address #'-/Address]]) \
945 (defn ^{:schema #'Customer} save [customer] customer)";
946 let bytes = encode_halc_module(
947 "demo.schema",
948 "demo/schema.hal",
949 source,
950 crate::kernel::parse_forms(source).unwrap(),
951 )
952 .unwrap();
953 let module = decode_halc(&bytes).unwrap();
954 assert!(module.forms[2]
955 .to_string()
956 .contains("(var demo.schema/Address)"));
957 assert_eq!(module.schemas.functions.len(), 1);
958 assert!(module.schemas.functions.contains_key("demo.schema/save"));
959 assert_eq!(module.schemas.definitions.len(), 2);
960 assert!(module
961 .schemas
962 .definitions
963 .contains_key("demo.schema/Address"));
964 assert!(module
965 .schemas
966 .definitions
967 .contains_key("demo.schema/Customer"));
968 assert!(matches!(
969 module.schemas.resolved_function_type("demo.schema/save"),
970 Some(super::super::SchemaType::Map(fields)) if fields.len() == 1
971 ));
972
973 let missing = "(ns demo.schema) \
974 (def Customer [:map [:address #'MissingAddress]]) \
975 (defn ^{:schema #'Customer} save [customer] customer)";
976 assert_eq!(
977 encode_halc_module(
978 "demo.schema",
979 "demo/schema.hal",
980 missing,
981 crate::kernel::parse_forms(missing).unwrap(),
982 )
983 .unwrap_err(),
984 "schema Var does not exist: MissingAddress"
985 );
986
987 let recursive = "(ns demo.schema) \
988 (def Node [:map [:children [:vector #'Node]]]) \
989 (defn ^{:schema #'Node} walk [node] node)";
990 assert!(encode_halc_module(
991 "demo.schema",
992 "demo/schema.hal",
993 recursive,
994 crate::kernel::parse_forms(recursive).unwrap(),
995 )
996 .is_ok());
997
998 let malformed = "(ns demo.schema) \
999 (def Customer [:map [:name]]) \
1000 (defn ^{:schema #'Customer} save [customer] customer)";
1001 assert_eq!(
1002 encode_halc_module(
1003 "demo.schema",
1004 "demo/schema.hal",
1005 malformed,
1006 crate::kernel::parse_forms(malformed).unwrap(),
1007 )
1008 .unwrap_err(),
1009 "invalid schema demo.schema/Customer: :map schema fields must be [name type] or [name properties type]"
1010 );
1011 }
1012
1013 #[test]
1014 fn round_trips_metadata() {
1015 let original = parse("^:dynamic *value*").unwrap();
1016 let bytes = artifact_payload(vec![original.clone()]);
1017 let decoded = decode_halc(&bytes).unwrap();
1018 assert_eq!(decoded.forms, vec![original]);
1019 }
1020
1021 #[test]
1022 fn rejects_bad_magic() {
1023 let mut bytes = artifact_payload(vec![Form::Nil]);
1024 bytes[0] = 0;
1025 assert!(decode_halc(&bytes).unwrap_err().contains("bad magic"));
1026 }
1027
1028 #[test]
1029 fn rejects_checksum_mismatch() {
1030 let mut bytes = artifact_payload(vec![Form::Nil]);
1031 let last = bytes.len() - 1;
1032 bytes[last] = bytes[last].wrapping_add(1);
1033 assert!(decode_halc(&bytes).unwrap_err().contains("checksum"));
1034 }
1035
1036 #[test]
1037 fn decodes_the_truffle_portable_format_golden_artifact() {
1038 let bytes = hex_bytes(concat!(
1044 "48414c43000100010000013f57211e103028689092d59627fbba64015c289acd1bc5b2e7be27ec53d8bf4c35",
1045 "00000001740000000174e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855",
1046 "0000001100010203000000000000002a044004000000000000050000001e313233343536373839303132",
1047 "333435363738393031323334353637383930060000000668c3a172c3a008000000780901000000056d792e6e73",
1048 "000000066d792d73796d000a00000000026b77000b00000002030000000000000001060000000161000c00000002",
1049 "030000000000000001060000000161000d00000002030000000000000001060000000161030000000000000002",
1050 "060000000162000e00000002030000000000000001030000000000000002000f00000002030000000000000002",
1051 "060000000162030000000000000001060000000161001000000002030000000000000002030000000000000001",
1052 "001100000003612b62",
1053 ));
1054
1055 let module = decode_halc(&bytes).unwrap();
1056 assert_eq!(module.origin, HalcOrigin::Halc);
1057 assert_eq!(module.namespace, "t");
1058 assert_eq!(module.resource, "t");
1059 assert_eq!(module.forms.len(), 17);
1060 assert_eq!(module.forms[0], Form::Nil);
1061 assert_eq!(module.forms[3], Form::Number(42));
1062 assert_eq!(module.forms[6], Form::String("hárà".into()));
1063 assert_eq!(module.forms[7], Form::Character('x'));
1064 assert_eq!(module.forms[8], Form::Symbol("my.ns/my-sym".into()));
1065 assert_eq!(module.forms[9], Form::Keyword("kw".into()));
1066 assert_eq!(module.forms[16], Form::Regex("a+b".into()));
1067 }
1068
1069 #[test]
1070 fn legacy_hir_magic_decodes_but_encoding_always_uses_halc_magic() {
1071 let halc = artifact_payload(vec![Form::Number(42)]);
1072 let mut legacy = halc.clone();
1073 legacy[..4].copy_from_slice(LEGACY_MAGIC);
1074
1075 assert_eq!(decode_halc(&legacy).unwrap().origin, HalcOrigin::LegacyHir);
1076 assert_eq!(&halc[..4], MAGIC);
1077 }
1078
1079 #[test]
1080 fn shared_cross_runtime_goldens_decode() {
1081 let complete = std::fs::read(crate::spec_registry::require(
1082 "01-lang/009-halc/draft/conformance/golden/complete.halc",
1083 ))
1084 .expect("complete HALC golden is readable");
1085 let legacy = std::fs::read(crate::spec_registry::require(
1086 "01-lang/009-halc/draft/conformance/golden/legacy-v1.hir",
1087 ))
1088 .expect("legacy HIR golden is readable");
1089 let current = decode_halc(&complete).unwrap();
1090 assert_eq!(current.origin, HalcOrigin::Halc);
1091 assert_eq!(current.namespace, "halc.conformance.complete");
1092 assert_eq!(current.resource, "conformance/complete.hal");
1093 assert_eq!(decode_halc(&legacy).unwrap().origin, HalcOrigin::LegacyHir);
1094 }
1095
1096 #[test]
1097 fn registry_golden_matches_rust_encoding() {
1098 let source_path = crate::spec_registry::require(
1099 "01-lang/009-halc/draft/conformance/complete.hal",
1100 );
1101 let source = std::fs::read_to_string(source_path).expect("HALC source is readable");
1102 let forms = crate::kernel::parse_forms(&source).expect("HALC source parses");
1103 let encoded = encode_halc_module(
1104 "halc.conformance.complete",
1105 "conformance/complete.hal",
1106 &source,
1107 forms,
1108 )
1109 .expect("HALC source encodes");
1110 let expected = std::fs::read(crate::spec_registry::require(
1111 "01-lang/009-halc/draft/conformance/golden/complete.halc",
1112 ))
1113 .expect("HALC golden is readable");
1114 assert_eq!(expected, encoded);
1115 }
1116
1117 fn hex_bytes(hex: &str) -> Vec<u8> {
1118 (0..hex.len())
1119 .step_by(2)
1120 .map(|index| u8::from_str_radix(&hex[index..index + 2], 16).unwrap())
1121 .collect()
1122 }
1123}