feat(mir): Loop Canonicalizer Phase 3 - skip_whitespace pattern recognition

## Summary
skip_whitespace パターンを Skeleton→Decision で認識可能に。
dev-only 観測で chosen=Pattern3IfPhi / missing_caps=[] を固定。

## Changes
- src/mir/loop_canonicalizer/mod.rs:
  - try_extract_skip_whitespace_pattern() 追加
    - loop(cond) { ... if check { p = p + 1 } else { break } } パターン認識
    - carrier name, delta, body statements を抽出
  - canonicalize_loop_expr() 拡張(skip_whitespace 対応)
    - Pattern3IfPhi 成功時は RoutingDecision::success 返却
    - Skeleton に HeaderCond, Body, Update ステップ追加
    - CarrierSlot に Counter role 設定
    - ExitContract に has_break=true 設定
  - Phase 3 unit tests 追加
    - test_skip_whitespace_pattern_recognition: 基本パターン
    - test_skip_whitespace_with_body_statements: body 付きパターン
    - test_skip_whitespace_fails_without_else: else なし失敗
    - test_skip_whitespace_fails_with_wrong_delta: 減算パターン失敗
  - Phase 2 obsolete tests 削除
- src/mir/builder/control_flow/joinir/routing.rs:
  - Debug 出力拡張(chosen pattern 表示)

## Tests
- cargo test --release --lib loop_canonicalizer::tests: PASS(11 tests)
- cargo test --release --lib: PASS(1044 tests, 退行なし)
- HAKO_JOINIR_DEBUG=1 test_pattern3_skip_whitespace.hako:
  - chosen=Pattern3IfPhi 
  - missing_caps=[] 

## Validation
-  dev-only 観測(HAKO_JOINIR_DEBUG=1)のときだけログ出力
-  フラグ OFF 時は完全不変
-  skip_whitespace パターンで SUCCESS 固定
-  unit tests で全パターン固定

Phase 137-3 complete
This commit is contained in:
nyash-codex
2025-12-16 05:38:18 +09:00
parent e8d93f107c
commit a0009d474d
7 changed files with 405 additions and 108 deletions

View File

@ -28,7 +28,7 @@
- **Phase 134 完了**: Plugin loader best-effort loading決定的順序 + failure 集約 + 継続)を導入。 - **Phase 134 完了**: Plugin loader best-effort loading決定的順序 + failure 集約 + 継続)を導入。
- **Phase 135 完了**: ConditionLoweringBox allocator SSOTP0: 根治修正 + P1: contract_checks Fail-Fast 強化)。 - **Phase 135 完了**: ConditionLoweringBox allocator SSOTP0: 根治修正 + P1: contract_checks Fail-Fast 強化)。
- **Phase 136 完了**: MirBuilder Context SSOT 化(+ ValueId allocator 掃討)。 - **Phase 136 完了**: MirBuilder Context SSOT 化(+ ValueId allocator 掃討)。
- **Phase 137-1 完了**: Loop Canonicalizer型/語彙の SSOTを導入routing/lowering には未介入)。 - **Phase 137-2 完了**: Loop Canonicalizerdev-only 観測)まで完了(既定挙動は不変)。
- **Phase 88 完了**: continue + 可変ステップi=i+const 差分)を dev-only fixture で固定、StepCalculator Box 抽出。 - **Phase 88 完了**: continue + 可変ステップi=i+const 差分)を dev-only fixture で固定、StepCalculator Box 抽出。
- **Phase 89 完了**: P0ContinueReturn detector+ P1lowering 実装)完了。 - **Phase 89 完了**: P0ContinueReturn detector+ P1lowering 実装)完了。
- **Phase 90 完了**: ParseStringComposite + `Null` literal + ContinueReturn同一値の複数 return-ifを dev-only fixture で固定。 - **Phase 90 完了**: ParseStringComposite + `Null` literal + ContinueReturn同一値の複数 return-ifを dev-only fixture で固定。
@ -53,15 +53,15 @@
## 次の指示書(優先順位) ## 次の指示書(優先順位)
### P0: Loop Canonicalizer の Phase 2dev-only 観測の導入 ### P0: Loop Canonicalizer の Phase 3Skeleton→Decision の精密化
**状態**: ✅ 設計 + Phase 1完了、Phase 2 **状態**: ✅ Phase 2 まで完了、Phase 3
**目的**: `LoopSkeleton` / `Capability` / `RoutingDecision` を実際のループ入口で “観測できる” 状態にして、組み合わせ爆発の手前で理由付き Fail-Fast ができる基盤を作る(ただし既定挙動は不変)。 **目的**: `LoopSkeleton` から `RoutingDecision`chosen/missing_caps/notesを安定に計算し、代表ケース`skip_whitespace`)で “期待する選択” をテストで固定する(既定挙動は不変)。
SSOT: SSOT:
- `docs/development/current/main/design/loop-canonicalizer.md` - `docs/development/current/main/design/loop-canonicalizer.md`
実装Phase 1: 実装:
- `src/mir/loop_canonicalizer/mod.rs` - `src/mir/loop_canonicalizer/mod.rs`
**次に触るSSOT**: **次に触るSSOT**:

View File

@ -42,7 +42,7 @@ JoinIR の箱構造と責務、ループ/if の lowering パターンを把握
- `docs/development/current/main/design/joinir-design-map.md` - `docs/development/current/main/design/joinir-design-map.md`
5. Loop Canonicalizer設計 SSOT 5. Loop Canonicalizer設計 SSOT
- `docs/development/current/main/design/loop-canonicalizer.md` - `docs/development/current/main/design/loop-canonicalizer.md`
- 実装Phase 137-1: `src/mir/loop_canonicalizer/mod.rs` - 実装Phase 137-2: `src/mir/loop_canonicalizer/mod.rs`
6. MIR BuilderContext 分割の入口) 6. MIR BuilderContext 分割の入口)
- `src/mir/builder/README.md` - `src/mir/builder/README.md`
7. Scope/BindingIdshadowing・束縛同一性の段階移行 7. Scope/BindingIdshadowing・束縛同一性の段階移行

View File

@ -27,11 +27,11 @@
- Phase 136: MirBuilder の Context 分割を完了し、状態の SSOT を Context に一本化。 - Phase 136: MirBuilder の Context 分割を完了し、状態の SSOT を Context に一本化。
- 詳細: `docs/development/current/main/phases/phase-136/README.md` - 詳細: `docs/development/current/main/phases/phase-136/README.md`
## 20251216Phase 1371(短報) ## 20251216Phase 1372(短報)
- Loop Canonicalizer の Phase 1型/語彙の SSOTを導入routing/lowering には未介入)。 - Loop Canonicalizer の Phase 2dev-only 観測)まで完了(既定挙動は不変)。
- 設計 SSOT: `docs/development/current/main/design/loop-canonicalizer.md` - 設計 SSOT: `docs/development/current/main/design/loop-canonicalizer.md`
- 実装: `src/mir/loop_canonicalizer/mod.rs` - 実装: `src/mir/loop_canonicalizer/mod.rs`+ 観測: `src/mir/builder/control_flow/joinir/routing.rs`
## 20251214現状サマリ ## 20251214現状サマリ

View File

@ -1,6 +1,6 @@
# Loop Canonicalizer設計 SSOT # Loop Canonicalizer設計 SSOT
Status: Phase 1 done型定義まで) Status: Phase 2 donedev-only 観測まで)
Scope: ループ形の組み合わせ爆発を抑えるための “前処理” の設計fixture/shape guard/fail-fast と整合) Scope: ループ形の組み合わせ爆発を抑えるための “前処理” の設計fixture/shape guard/fail-fast と整合)
Related: Related:
- SSOT (契約/不変条件): `docs/development/current/main/joinir-architecture-overview.md` - SSOT (契約/不変条件): `docs/development/current/main/joinir-architecture-overview.md`
@ -160,12 +160,12 @@ pub enum CarrierRole {
## 実装の入口(現状) ## 実装の入口(現状)
Phase 1(型定義のみ)の実装はここ: 実装(Phase 12はここ:
- `src/mir/loop_canonicalizer/mod.rs` - `src/mir/loop_canonicalizer/mod.rs`
注意: 注意:
- ここは「型と語彙の SSOT」を置く場所で、routing/lowering にはまだ介入しない - Phase 2 で `canonicalize_loop_expr(...) -> Result<(LoopSkeleton, RoutingDecision), String>` を導入し、JoinIR ループ入口で dev-only 観測できるようにした(既定挙動は不変)
- Phase 2 以降で `canonicalize(loop_ast) -> (LoopSkeleton, RoutingDecision)` を導入し、dev-only で観測から始める。 - 観測ポイントJoinIR ループ入口): `src/mir/builder/control_flow/joinir/routing.rs``joinir_dev_enabled()` 配下)
## Capability の語彙Fail-Fast reason タグ) ## Capability の語彙Fail-Fast reason タグ)
@ -197,7 +197,7 @@ Canonicalizer の判定結果は `RoutingDecision` に集約し、以下に流
```rust ```rust
pub struct RoutingDecision { pub struct RoutingDecision {
/// 選択された PatternNone = Fail-Fast /// 選択された PatternNone = Fail-Fast
pub chosen: Option<LoopPattern>, pub chosen: Option<LoopPatternKind>,
/// 不足している Capability のリスト /// 不足している Capability のリスト
pub missing_caps: Vec<&'static str>, pub missing_caps: Vec<&'static str>,
@ -206,7 +206,7 @@ pub struct RoutingDecision {
pub notes: Vec<String>, pub notes: Vec<String>,
/// error_tags への追記contract_checks 用) /// error_tags への追記contract_checks 用)
pub error_tags: Vec<ErrorTag>, pub error_tags: Vec<String>,
} }
``` ```

View File

@ -1,7 +1,7 @@
# Phase 137: Loop Canonicalizer前処理 SSOT # Phase 137: Loop Canonicalizer前処理 SSOT
## Status ## Status
- 状態: 🔶 進行中Phase 1 完了) - 状態: 🔶 進行中Phase 2 完了)
## Goal ## Goal
- ループ形の組み合わせ爆発を抑えるため、`AST → LoopSkeleton → (capability/routing)` の前処理を SSOT 化する。 - ループ形の組み合わせ爆発を抑えるため、`AST → LoopSkeleton → (capability/routing)` の前処理を SSOT 化する。
@ -15,11 +15,17 @@
- `RoutingDecision` / capability tags`CAP_MISSING_*` - `RoutingDecision` / capability tags`CAP_MISSING_*`
- 注意: Phase 1 は「型と語彙」のみ。routing/lowering にはまだ介入しない。 - 注意: Phase 1 は「型と語彙」のみ。routing/lowering にはまだ介入しない。
## Phase 2: dev-only 観測の導入 ## Phase 2完了: dev-only 観測の導入
- 目標: ループ入口で `LoopSkeleton``RoutingDecision` を生成し、dev-only で観測できるようにする。 - 入口:
- 既定挙動: 不変dev-only の観測から開始 - Canonicalize: `src/mir/loop_canonicalizer/mod.rs``canonicalize_loop_expr`
- スイッチ: 新しい env を増やさず、`joinir_dev_enabled()`(既存)配下で有効化する。 - 観測ポイント: `src/mir/builder/control_flow/joinir/routing.rs``joinir_dev_enabled()` 配下)
- 既定挙動: 不変dev-only 観測のみ)
## Phase 3: Pattern 検出Skeleton→Decision の精密化)
- 目標: `skip_whitespace` を Skeleton から安定に識別し、`RoutingDecision.chosen``missing_caps` を期待通りに固定する。
- 注意: routing/lowering の変更は dev-only の観測結果が固まってから。
## SSOT ## SSOT

View File

@ -142,9 +142,12 @@ impl MirBuilder {
eprintln!("[loop_canonicalizer] Has exits: {}", skeleton.exits.has_any_exit()); eprintln!("[loop_canonicalizer] Has exits: {}", skeleton.exits.has_any_exit());
eprintln!("[loop_canonicalizer] Decision: {}", eprintln!("[loop_canonicalizer] Decision: {}",
if decision.is_success() { "SUCCESS" } else { "FAIL_FAST" }); if decision.is_success() { "SUCCESS" } else { "FAIL_FAST" });
if let Some(pattern) = decision.chosen {
eprintln!("[loop_canonicalizer] Chosen pattern: {:?}", pattern);
}
eprintln!("[loop_canonicalizer] Missing caps: {:?}", decision.missing_caps);
if decision.is_fail_fast() { if decision.is_fail_fast() {
eprintln!("[loop_canonicalizer] Reason: {}", decision.notes.join("; ")); eprintln!("[loop_canonicalizer] Reason: {}", decision.notes.join("; "));
eprintln!("[loop_canonicalizer] Missing caps: {:?}", decision.missing_caps);
} }
} }
Err(e) => { Err(e) => {

View File

@ -326,14 +326,116 @@ impl std::fmt::Display for CarrierRole {
} }
// ============================================================================ // ============================================================================
// Phase 2: Canonicalization Entry Point // Phase 3: Pattern Recognition Helpers
// ============================================================================ // ============================================================================
/// Canonicalize a loop AST into LoopSkeleton (Phase 2: Minimal Implementation) /// Try to extract skip_whitespace pattern from loop
///
/// Pattern structure:
/// ```
/// loop(cond) {
/// // ... optional body statements (Body)
/// if check_cond {
/// carrier = carrier + const
/// } else {
/// break
/// }
/// }
/// ```
///
/// Returns (carrier_name, delta, body_stmts) if pattern matches.
fn try_extract_skip_whitespace_pattern(
body: &[ASTNode],
) -> Option<(String, i64, Vec<ASTNode>)> {
if body.is_empty() {
return None;
}
// Last statement must be if-else with break
let last_stmt = &body[body.len() - 1];
let (then_body, else_body) = match last_stmt {
ASTNode::If {
then_body,
else_body: Some(else_body),
..
} => (then_body, else_body),
_ => return None,
};
// Then branch must be single assignment: carrier = carrier + const
if then_body.len() != 1 {
return None;
}
let (carrier_name, delta) = match &then_body[0] {
ASTNode::Assignment { target, value, .. } => {
// Extract target variable name
let target_name = match target.as_ref() {
ASTNode::Variable { name, .. } => name.clone(),
_ => return None,
};
// Value must be: target + const
match value.as_ref() {
ASTNode::BinaryOp {
operator: crate::ast::BinaryOperator::Add,
left,
right,
..
} => {
// Left must be same variable
let left_name = match left.as_ref() {
ASTNode::Variable { name, .. } => name,
_ => return None,
};
if left_name != &target_name {
return None;
}
// Right must be integer literal
let delta = match right.as_ref() {
ASTNode::Literal {
value: crate::ast::LiteralValue::Integer(n),
..
} => *n,
_ => return None,
};
(target_name, delta)
}
_ => return None,
}
}
_ => return None,
};
// Else branch must be single break
if else_body.len() != 1 {
return None;
}
match &else_body[0] {
ASTNode::Break { .. } => {
// Success! Extract body statements (all except last if)
let body_stmts = body[..body.len() - 1].to_vec();
Some((carrier_name, delta, body_stmts))
}
_ => None,
}
}
// ============================================================================
// Phase 3: Canonicalization Entry Point
// ============================================================================
/// Canonicalize a loop AST into LoopSkeleton (Phase 3: skip_whitespace pattern recognition)
/// ///
/// Currently supports only the skip_whitespace pattern: /// Currently supports only the skip_whitespace pattern:
/// ``` /// ```
/// loop(cond) { /// loop(cond) {
/// // ... optional body statements
/// if check_cond { /// if check_cond {
/// carrier = carrier + step /// carrier = carrier + step
/// } else { /// } else {
@ -363,66 +465,55 @@ pub fn canonicalize_loop_expr(
_ => return Err(format!("Expected Loop node, got: {:?}", loop_expr)), _ => return Err(format!("Expected Loop node, got: {:?}", loop_expr)),
}; };
// Phase 2: Minimal implementation - detect skip_whitespace pattern only // Phase 3: Try to extract skip_whitespace pattern
// Pattern: loop(cond) { if check { update } else { break } } if let Some((carrier_name, delta, body_stmts)) = try_extract_skip_whitespace_pattern(body) {
// Build skeleton for skip_whitespace pattern
// Check for minimal pattern: single if-else with break
if body.len() != 1 {
return Ok((
LoopSkeleton::new(span),
RoutingDecision::fail_fast(
vec![capability_tags::CAP_MISSING_SINGLE_BREAK],
format!("Phase 2: Only single-statement loops supported (got {} statements)", body.len()),
),
));
}
// Check if it's an if-else statement
let _if_stmt = match &body[0] {
ASTNode::If {
condition: _if_cond,
then_body: _then_body,
else_body,
..
} => {
// Must have else branch
if else_body.is_none() {
return Ok((
LoopSkeleton::new(span),
RoutingDecision::fail_fast(
vec![capability_tags::CAP_MISSING_SINGLE_BREAK],
"Phase 2: If statement must have else branch".to_string(),
),
));
}
// Phase 2: Just validate structure, don't extract components yet
()
}
_ => {
return Ok((
LoopSkeleton::new(span),
RoutingDecision::fail_fast(
vec![capability_tags::CAP_MISSING_SINGLE_BREAK],
"Phase 2: Loop body must be single if-else statement".to_string(),
),
));
}
};
// Build minimal skeleton
let mut skeleton = LoopSkeleton::new(span); let mut skeleton = LoopSkeleton::new(span);
// Add header condition // Step 1: Header condition
skeleton.steps.push(SkeletonStep::HeaderCond { skeleton.steps.push(SkeletonStep::HeaderCond {
expr: Box::new(condition.clone()), expr: Box::new(condition.clone()),
}); });
// For now, just mark as unsupported - full pattern detection will come in Phase 3 // Step 2: Body statements (if any)
if !body_stmts.is_empty() {
skeleton.steps.push(SkeletonStep::Body {
stmts: body_stmts,
});
}
// Step 3: Update step
skeleton.steps.push(SkeletonStep::Update {
carrier_name: carrier_name.clone(),
update_kind: UpdateKind::ConstStep { delta },
});
// Add carrier slot
skeleton.carriers.push(CarrierSlot {
name: carrier_name,
role: CarrierRole::Counter,
update_kind: UpdateKind::ConstStep { delta },
});
// Set exit contract
skeleton.exits = ExitContract {
has_break: true,
has_continue: false,
has_return: false,
break_has_value: false,
};
// Success! Return Pattern3WithIfPhi
let decision = RoutingDecision::success(LoopPatternKind::Pattern3IfPhi);
return Ok((skeleton, decision));
}
// Pattern not recognized - fail fast
Ok(( Ok((
skeleton, LoopSkeleton::new(span),
RoutingDecision::fail_fast( RoutingDecision::fail_fast(
vec![capability_tags::CAP_MISSING_CONST_STEP], vec![capability_tags::CAP_MISSING_CONST_STEP],
"Phase 2: Pattern detection not yet implemented".to_string(), "Phase 3: Loop does not match skip_whitespace pattern".to_string(),
), ),
)) ))
} }
@ -534,22 +625,60 @@ mod tests {
assert!(result.unwrap_err().contains("Expected Loop node")); assert!(result.unwrap_err().contains("Expected Loop node"));
} }
#[test] // ============================================================================
fn test_canonicalize_minimal_loop_structure() { // Phase 3: skip_whitespace Pattern Tests
use crate::ast::LiteralValue; // ============================================================================
// Build minimal loop: loop(true) { if true { } else { break } } #[test]
fn test_skip_whitespace_pattern_recognition() {
use crate::ast::{BinaryOperator, LiteralValue};
// Build skip_whitespace pattern: loop(p < len) { if is_ws == 1 { p = p + 1 } else { break } }
let loop_node = ASTNode::Loop { let loop_node = ASTNode::Loop {
condition: Box::new(ASTNode::Literal { condition: Box::new(ASTNode::BinaryOp {
value: LiteralValue::Bool(true), operator: BinaryOperator::Less,
left: Box::new(ASTNode::Variable {
name: "p".to_string(),
span: Span::unknown(),
}),
right: Box::new(ASTNode::Variable {
name: "len".to_string(),
span: Span::unknown(),
}),
span: Span::unknown(), span: Span::unknown(),
}), }),
body: vec![ASTNode::If { body: vec![ASTNode::If {
condition: Box::new(ASTNode::Literal { condition: Box::new(ASTNode::BinaryOp {
value: LiteralValue::Bool(true), operator: BinaryOperator::Equal,
left: Box::new(ASTNode::Variable {
name: "is_ws".to_string(),
span: Span::unknown(), span: Span::unknown(),
}), }),
then_body: vec![], right: Box::new(ASTNode::Literal {
value: LiteralValue::Integer(1),
span: Span::unknown(),
}),
span: Span::unknown(),
}),
then_body: vec![ASTNode::Assignment {
target: Box::new(ASTNode::Variable {
name: "p".to_string(),
span: Span::unknown(),
}),
value: Box::new(ASTNode::BinaryOp {
operator: BinaryOperator::Add,
left: Box::new(ASTNode::Variable {
name: "p".to_string(),
span: Span::unknown(),
}),
right: Box::new(ASTNode::Literal {
value: LiteralValue::Integer(1),
span: Span::unknown(),
}),
span: Span::unknown(),
}),
span: Span::unknown(),
}],
else_body: Some(vec![ASTNode::Break { else_body: Some(vec![ASTNode::Break {
span: Span::unknown(), span: Span::unknown(),
}]), }]),
@ -562,37 +691,106 @@ mod tests {
assert!(result.is_ok()); assert!(result.is_ok());
let (skeleton, decision) = result.unwrap(); let (skeleton, decision) = result.unwrap();
// Should have header condition step
assert_eq!(skeleton.steps.len(), 1); // Verify success
assert!(decision.is_success());
assert_eq!(decision.chosen, Some(LoopPatternKind::Pattern3IfPhi));
assert_eq!(decision.missing_caps.len(), 0);
// Verify skeleton structure
assert_eq!(skeleton.steps.len(), 2); // HeaderCond + Update
assert!(matches!( assert!(matches!(
skeleton.steps[0], skeleton.steps[0],
SkeletonStep::HeaderCond { .. } SkeletonStep::HeaderCond { .. }
)); ));
assert!(matches!(
skeleton.steps[1],
SkeletonStep::Update { .. }
));
// Phase 2: Should fail-fast (pattern detection not implemented) // Verify carrier
assert!(decision.is_fail_fast()); assert_eq!(skeleton.carriers.len(), 1);
assert!(decision.notes[0].contains("Pattern detection not yet implemented")); assert_eq!(skeleton.carriers[0].name, "p");
assert_eq!(skeleton.carriers[0].role, CarrierRole::Counter);
match &skeleton.carriers[0].update_kind {
UpdateKind::ConstStep { delta } => assert_eq!(*delta, 1),
_ => panic!("Expected ConstStep update"),
}
// Verify exit contract
assert!(skeleton.exits.has_break);
assert!(!skeleton.exits.has_continue);
assert!(!skeleton.exits.has_return);
assert!(!skeleton.exits.break_has_value);
} }
#[test] #[test]
fn test_canonicalize_rejects_multi_statement_loop() { fn test_skip_whitespace_with_body_statements() {
use crate::ast::LiteralValue; use crate::ast::{BinaryOperator, LiteralValue};
// Build loop with 2 statements // Build pattern with body statements before the if:
// loop(p < len) {
// local ch = get_char(p)
// if is_ws { p = p + 1 } else { break }
// }
let loop_node = ASTNode::Loop { let loop_node = ASTNode::Loop {
condition: Box::new(ASTNode::Literal { condition: Box::new(ASTNode::BinaryOp {
value: LiteralValue::Bool(true), operator: BinaryOperator::Less,
left: Box::new(ASTNode::Variable {
name: "p".to_string(),
span: Span::unknown(),
}),
right: Box::new(ASTNode::Variable {
name: "len".to_string(),
span: Span::unknown(),
}),
span: Span::unknown(), span: Span::unknown(),
}), }),
body: vec![ body: vec![
ASTNode::Print { // Body statement
expression: Box::new(ASTNode::Literal { ASTNode::Assignment {
value: LiteralValue::String("test".to_string()), target: Box::new(ASTNode::Variable {
name: "ch".to_string(),
span: Span::unknown(),
}),
value: Box::new(ASTNode::FunctionCall {
name: "get_char".to_string(),
arguments: vec![ASTNode::Variable {
name: "p".to_string(),
span: Span::unknown(),
}],
span: Span::unknown(), span: Span::unknown(),
}), }),
span: Span::unknown(), span: Span::unknown(),
}, },
ASTNode::Break { // The if-else pattern
ASTNode::If {
condition: Box::new(ASTNode::Variable {
name: "is_ws".to_string(),
span: Span::unknown(),
}),
then_body: vec![ASTNode::Assignment {
target: Box::new(ASTNode::Variable {
name: "p".to_string(),
span: Span::unknown(),
}),
value: Box::new(ASTNode::BinaryOp {
operator: BinaryOperator::Add,
left: Box::new(ASTNode::Variable {
name: "p".to_string(),
span: Span::unknown(),
}),
right: Box::new(ASTNode::Literal {
value: LiteralValue::Integer(1),
span: Span::unknown(),
}),
span: Span::unknown(),
}),
span: Span::unknown(),
}],
else_body: Some(vec![ASTNode::Break {
span: Span::unknown(),
}]),
span: Span::unknown(), span: Span::unknown(),
}, },
], ],
@ -602,16 +800,36 @@ mod tests {
let result = canonicalize_loop_expr(&loop_node); let result = canonicalize_loop_expr(&loop_node);
assert!(result.is_ok()); assert!(result.is_ok());
let (_, decision) = result.unwrap(); let (skeleton, decision) = result.unwrap();
assert!(decision.is_fail_fast());
assert!(decision.notes[0].contains("Only single-statement loops supported")); // Verify success
assert!(decision.is_success());
assert_eq!(decision.chosen, Some(LoopPatternKind::Pattern3IfPhi));
// Verify skeleton has Body step
assert_eq!(skeleton.steps.len(), 3); // HeaderCond + Body + Update
assert!(matches!(
skeleton.steps[0],
SkeletonStep::HeaderCond { .. }
));
assert!(matches!(skeleton.steps[1], SkeletonStep::Body { .. }));
assert!(matches!(
skeleton.steps[2],
SkeletonStep::Update { .. }
));
// Verify body contains 1 statement
match &skeleton.steps[1] {
SkeletonStep::Body { stmts } => assert_eq!(stmts.len(), 1),
_ => panic!("Expected Body step"),
}
} }
#[test] #[test]
fn test_canonicalize_rejects_if_without_else() { fn test_skip_whitespace_fails_without_else() {
use crate::ast::LiteralValue; use crate::ast::{BinaryOperator, LiteralValue};
// Build loop with if (no else) // Build pattern without else branch (should fail)
let loop_node = ASTNode::Loop { let loop_node = ASTNode::Loop {
condition: Box::new(ASTNode::Literal { condition: Box::new(ASTNode::Literal {
value: LiteralValue::Bool(true), value: LiteralValue::Bool(true),
@ -622,7 +840,25 @@ mod tests {
value: LiteralValue::Bool(true), value: LiteralValue::Bool(true),
span: Span::unknown(), span: Span::unknown(),
}), }),
then_body: vec![], then_body: vec![ASTNode::Assignment {
target: Box::new(ASTNode::Variable {
name: "p".to_string(),
span: Span::unknown(),
}),
value: Box::new(ASTNode::BinaryOp {
operator: BinaryOperator::Add,
left: Box::new(ASTNode::Variable {
name: "p".to_string(),
span: Span::unknown(),
}),
right: Box::new(ASTNode::Literal {
value: LiteralValue::Integer(1),
span: Span::unknown(),
}),
span: Span::unknown(),
}),
span: Span::unknown(),
}],
else_body: None, // No else branch else_body: None, // No else branch
span: Span::unknown(), span: Span::unknown(),
}], }],
@ -634,6 +870,58 @@ mod tests {
let (_, decision) = result.unwrap(); let (_, decision) = result.unwrap();
assert!(decision.is_fail_fast()); assert!(decision.is_fail_fast());
assert!(decision.notes[0].contains("must have else branch")); assert!(decision
.notes[0]
.contains("does not match skip_whitespace pattern"));
} }
#[test]
fn test_skip_whitespace_fails_with_wrong_delta() {
use crate::ast::{BinaryOperator, LiteralValue};
// Build pattern with wrong update (p = p - 1 instead of p = p + 1)
let loop_node = ASTNode::Loop {
condition: Box::new(ASTNode::Literal {
value: LiteralValue::Bool(true),
span: Span::unknown(),
}),
body: vec![ASTNode::If {
condition: Box::new(ASTNode::Literal {
value: LiteralValue::Bool(true),
span: Span::unknown(),
}),
then_body: vec![ASTNode::Assignment {
target: Box::new(ASTNode::Variable {
name: "p".to_string(),
span: Span::unknown(),
}),
value: Box::new(ASTNode::BinaryOp {
operator: BinaryOperator::Subtract, // Wrong operator
left: Box::new(ASTNode::Variable {
name: "p".to_string(),
span: Span::unknown(),
}),
right: Box::new(ASTNode::Literal {
value: LiteralValue::Integer(1),
span: Span::unknown(),
}),
span: Span::unknown(),
}),
span: Span::unknown(),
}],
else_body: Some(vec![ASTNode::Break {
span: Span::unknown(),
}]),
span: Span::unknown(),
}],
span: Span::unknown(),
};
let result = canonicalize_loop_expr(&loop_node);
assert!(result.is_ok());
let (_, decision) = result.unwrap();
assert!(decision.is_fail_fast());
}
} }