unified/swift: add top-level normalization and fallback scaffold

Author: asgerf

unified/extractor/src/languages/swift/swift.rs

@@ -1,8 +1,20 @@
 use codeql_extractor::extractor::simple;
-use yeast::{build::BuildCtx, rule, DesugaringConfig, PhaseKind};
+use yeast::{rule, DesugaringConfig, PhaseKind};
 
 fn translation_rules() -> Vec<yeast::Rule> {
     vec![
+        // ---- Top-level ----
+        // Capture all top-level statements, including unnamed tokens like `nil`.
+        rule!(
+            (source_file statement: _* @children)
+            =>
+            (top_level
+                body: (block stmt: {..children})
+            )
+        ),
+        // Declarations may be wrapped in local/global wrapper nodes.
+        rule!((global_declaration _ @inner) => {inner}),
+        rule!((local_declaration _ @inner) => {inner}),
         // ---- Fallbacks ----
         rule!(
             (_)

unified/extractor/tests/corpus/swift/closures.txt

@@ -50,6 +50,8 @@ source_file
 
 top_level
   body:
+    block
+      stmt: unsupported_node "let f = { (x: Int) -> Int in x * 2 }"
 
 ===
 Closure with shorthand parameters
@@ -82,6 +84,8 @@ source_file
 
 top_level
   body:
+    block
+      stmt: unsupported_node "let f = { $0 + $1 }"
 
 ===
 Trailing closure
@@ -114,6 +118,8 @@ source_file
 
 top_level
   body:
+    block
+      stmt: unsupported_node "xs.map { $0 * 2 }"
 
 ===
 Closure with capture list
@@ -163,6 +169,8 @@ source_file
 
 top_level
   body:
+    block
+      stmt: unsupported_node "let f = { [weak self] in self?.doThing() }"
 
 ===
 Multi-statement closure
@@ -236,3 +244,5 @@ source_file
 
 top_level
   body:
+    block
+      stmt: unsupported_node "let f = { (x: Int) -> Int in\n  let y = x + 1\n  return y * 2\n}"

unified/extractor/tests/corpus/swift/collections.txt

@@ -28,6 +28,8 @@ source_file
 
 top_level
   body:
+    block
+      stmt: unsupported_node "let xs = [1, 2, 3]"
 
 ===
 Empty array literal with type
@@ -68,6 +70,8 @@ source_file
 
 top_level
   body:
+    block
+      stmt: unsupported_node "let xs: [Int] = []"
 
 ===
 Dictionary literal
@@ -106,6 +110,8 @@ source_file
 
 top_level
   body:
+    block
+      stmt: unsupported_node "let d = [\"a\": 1, \"b\": 2]"
 
 ===
 Set literal
@@ -155,6 +161,8 @@ source_file
 
 top_level
   body:
+    block
+      stmt: unsupported_node "let s: Set<Int> = [1, 2, 3]"
 
 ===
 Tuple literal
@@ -191,6 +199,8 @@ source_file
 
 top_level
   body:
+    block
+      stmt: unsupported_node "let t = (1, \"two\", 3.0)"
 
 ===
 Subscript access
@@ -232,9 +242,8 @@ source_file
 
 top_level
   body:
-    unsupported_node "// TODO: tree-sitter-swift parses `xs[0]` as a call_expression (same shape"
-    unsupported_node "// as `xs(0)`), so the mapping currently produces a call_expr. Update the"
-    unsupported_node "// parser / add a separate subscript_expr node and remap when fixed."
+    block
+      stmt: unsupported_node "let first = xs[0]"
 
 ===
 Dictionary subscript
@@ -276,8 +285,8 @@ source_file
 
 top_level
   body:
-    unsupported_node "// TODO: same parser issue as the array subscript case above —"
-    unsupported_node "// `d[\"key\"]` is parsed as `call_expression(d, (\"key\"))`."
+    block
+      stmt: unsupported_node "let v = d[\"key\"]"
 
 ===
 Tuple member access
@@ -309,3 +318,5 @@ source_file
 
 top_level
   body:
+    block
+      stmt: unsupported_node "let n = t.0"

unified/extractor/tests/corpus/swift/control-flow.txt

@@ -35,6 +35,8 @@ source_file
 
 top_level
   body:
+    block
+      stmt: unsupported_node "if x > 0 {\n  print(x)\n}"
 
 ===
 If-else
@@ -90,6 +92,8 @@ source_file
 
 top_level
   body:
+    block
+      stmt: unsupported_node "if x > 0 {\n  print(x)\n} else {\n  print(-x)\n}"
 
 ===
 If-else-if chain
@@ -165,6 +169,8 @@ source_file
 
 top_level
   body:
+    block
+      stmt: unsupported_node "if x > 0 {\n  print(1)\n} else if x < 0 {\n  print(2)\n} else {\n  print(3)\n}"
 
 ===
 If-let optional binding
@@ -207,6 +213,8 @@ source_file
 
 top_level
   body:
+    block
+      stmt: unsupported_node "if let value = optional {\n  print(value)\n}"
 
 ===
 Guard let
@@ -240,6 +248,8 @@ source_file
 
 top_level
   body:
+    block
+      stmt: unsupported_node "guard let value = optional else { return }"
 
 ===
 Ternary expression
@@ -277,6 +287,8 @@ source_file
 
 top_level
   body:
+    block
+      stmt: unsupported_node "let y = x > 0 ? 1 : -1"
 
 ===
 Switch statement
@@ -357,6 +369,8 @@ source_file
 
 top_level
   body:
+    block
+      stmt: unsupported_node "switch x {\ncase 1:\n  print(\"one\")\ncase 2, 3:\n  print(\"two or three\")\ndefault:\n  print(\"other\")\n}"
 
 ===
 Switch with binding pattern
@@ -445,3 +459,5 @@ source_file
 
 top_level
   body:
+    block
+      stmt: unsupported_node "switch shape {\ncase .circle(let r):\n  print(r)\ncase .square(let s):\n  print(s)\n}"

unified/extractor/tests/corpus/swift/desugar.txt

@@ -17,6 +17,8 @@ source_file
 
 top_level
   body:
+    block
+      stmt: unsupported_node "1 + 2"
 
 ===
 Another additive expression is desugared
@@ -37,3 +39,5 @@ source_file
 
 top_level
   body:
+    block
+      stmt: unsupported_node "foo + bar"

unified/extractor/tests/corpus/swift/functions.txt

@@ -31,6 +31,8 @@ source_file
 
 top_level
   body:
+    block
+      stmt: unsupported_node "func greet() {\n  print(\"hello\")\n}"
 
 ===
 Function with parameters and return type
@@ -93,6 +95,8 @@ source_file
 
 top_level
   body:
+    block
+      stmt: unsupported_node "func add(_ a: Int, _ b: Int) -> Int {\n  return a + b\n}"
 
 ===
 Function with named parameters
@@ -138,6 +142,8 @@ source_file
 
 top_level
   body:
+    block
+      stmt: unsupported_node "func greet(person name: String) {\n  print(name)\n}"
 
 ===
 Function with default parameter value
@@ -185,6 +191,8 @@ source_file
 
 top_level
   body:
+    block
+      stmt: unsupported_node "func greet(name: String = \"world\") {\n  print(name)\n}"
 
 ===
 Variadic function
@@ -249,6 +257,8 @@ source_file
 
 top_level
   body:
+    block
+      stmt: unsupported_node "func sum(_ values: Int...) -> Int {\n  return values.reduce(0, +)\n}"
 
 ===
 Function call
@@ -276,6 +286,8 @@ source_file
 
 top_level
   body:
+    block
+      stmt: unsupported_node "foo(1, 2)"
 
 ===
 Function call with labelled arguments
@@ -306,6 +318,8 @@ source_file
 
 top_level
   body:
+    block
+      stmt: unsupported_node "greet(person: \"Bob\")"
 
 ===
 Method call
@@ -336,6 +350,8 @@ source_file
 
 top_level
   body:
+    block
+      stmt: unsupported_node "list.append(1)"
 
 ===
 Generic function
@@ -387,3 +403,5 @@ source_file
 
 top_level
   body:
+    block
+      stmt: unsupported_node "func identity<T>(_ x: T) -> T {\n  return x\n}"

unified/extractor/tests/corpus/swift/literals.txt

@@ -13,6 +13,8 @@ source_file
 
 top_level
   body:
+    block
+      stmt: unsupported_node "42"
 
 ===
 Negative integer literal
@@ -32,6 +34,8 @@ source_file
 
 top_level
   body:
+    block
+      stmt: unsupported_node "-7"
 
 ===
 Floating-point literal
@@ -48,6 +52,8 @@ source_file
 
 top_level
   body:
+    block
+      stmt: unsupported_node "3.14"
 
 ===
 Boolean literals
@@ -67,6 +73,10 @@ source_file
 
 top_level
   body:
+    block
+      stmt:
+        unsupported_node "true"
+        unsupported_node "false"
 
 ===
 Nil literal
@@ -83,6 +93,8 @@ source_file
 
 top_level
   body:
+    block
+      stmt: unsupported_node "nil"
 
 ===
 String literal
@@ -101,6 +113,8 @@ source_file
 
 top_level
   body:
+    block
+      stmt: unsupported_node "\"hello\""
 
 ===
 String with interpolation
@@ -122,3 +136,5 @@ source_file
 
 top_level
   body:
+    block
+      stmt: unsupported_node "\"hello \\(name)\""