C import: unions (SDL_Event) and their fields; #defines that are expressions, casts or text; typedef'd structs no function mentions; FLAG_A | FLAG_B fits a u64

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

src/cimport.fch

@@ -21,11 +21,13 @@ struct CFunc {
     str unsupported     // non-empty: why Finch can't call it (yet)
 }
 
-struct CConst {         // enum values and simple #define numbers
+struct CConst {         // enum values and #define constants
     bool isFloat
     int i
     float f
     str text            // a float as written, so no digits are lost: 3.14159265358979323846
+    bool isStr          // #define NAME "text"
+    str s
 }
 
 struct CField {
@@ -44,6 +46,7 @@ struct CStruct {
     int align
     str unsupported     // why it can't be used by value (pointers to it still work)
     bool hasHidden
+    bool isUnion        // every field starts at byte 0
 }
 
 struct CImports {
@@ -75,6 +78,7 @@ struct Importer {
     map[str]str recordNames     // record USR -> Finch name (a typedef name wins)
     map[str]bool defining       // USR -> true while its fields are being mapped
     int anon
+    []str exprMacros            // #defines that are expressions: clang works out their values
 }
 
 fn text(CXString s) -> str {
@@ -169,6 +173,7 @@ fn defineRecord(ptr[Importer] imp, CXType type) -> str {
     s := CStruct(name: name, size: size, align: int(clang_Type_getAlignOf(t)))
     if clang_getCursorKind(decl) == CXCursor_UnionDecl {
         s.unsupported = "is a C union"
+        s.isUnion = true
     }
     // the fields are collected by a callback that gets the importer and the struct through one pointer
     ctx := FieldCtx(imp: imp, s: addr(s))
@@ -194,9 +199,10 @@ fn visitField(CXCursor field, ptr data) -> i32 {
     if clang_Cursor_isBitField(field) != 0 {
         c.s.unsupported = "has bit fields"
         f.hidden = true
-    } else if ft.kind == CXType_Record && clang_getCursorKind(clang_getTypeDeclaration(ft)) == CXCursor_UnionDecl {
-        f.hidden = true     // unions are kept as opaque bytes
     } else {
+        if ft.kind == CXType_Record && clang_getCursorKind(clang_getTypeDeclaration(ft)) == CXCursor_UnionDecl {
+            c.s.unsupported = "has a union inside"     // its fields can be used, but not passed by value
+        }
         m := mapType(c.imp, ft, false, false)
         if m.ok {
             f.ty = m.ty
@@ -293,9 +299,6 @@ fn mapType(ptr[Importer] imp, CXType type, bool signature, bool param) -> Mapped
         return ok(ast.ptrTo(inner.ty))
     }
     if k == CXType_Record {
-        if clang_getCursorKind(clang_getTypeDeclaration(t)) == CXCursor_UnionDecl {
-            return no("uses a C union by value")
-        }
         name := defineRecord(imp, t)
         if name.len == 0 {
             return no("uses an incomplete C struct by value")
@@ -366,19 +369,93 @@ fn parseInt(str lit) -> Number {
     return Number(c: CConst(i: n), ok: true)
 }
 
-// Simple #defines: a single number, maybe negative, maybe in parentheses.
+// The tokens after a #define's name; `safe` is false if any of them could not be part of a
-fn macroValue(ptr[Importer] imp, CXCursor c) -> Number {
+// constant expression (braces, semicolons, statements), so it is never handed to clang.
+struct MacroTokens {
+    []str parts
+    bool safe
+}
+
+fn macroTokens(ptr[Importer] imp, CXCursor c) -> MacroTokens {
     ptr[CXToken] tokens = null
     u32 n = 0
     clang_tokenize(imp.tu, clang_getCursorExtent(c), addr(tokens), addr(n))
-    []str parts
+    out := MacroTokens(safe: true)
+    allowed := ["(", ")", "+", "-", "*", "/", "%", "<<", ">>", "|", "&", "^", "~", "!", "<", ">", "<=", ">=", "==", "!=", "&&", "||", "?", ":", ","]
+    words := ["sizeof", "unsigned", "signed", "int", "long", "short", "char", "float", "double", "const", "_Bool"]
+    depth := 0
     for k in 1..int(n) {            // token 0 is the macro's name
-        if clang_getTokenKind(tokens[k]) == CXToken_Comment {
+        kind := clang_getTokenKind(tokens[k])
+        if kind == CXToken_Comment {
             continue
         }
-        parts.push(text(clang_getTokenSpelling(imp.tu, tokens[k])))
+        t := text(clang_getTokenSpelling(imp.tu, tokens[k]))
+        out.parts.push(t)
+        if kind == CXToken_Punctuation {
+            if !allowed.contains(t) {
+                out.safe = false
+            }
+            if t == "(" {
+                depth += 1
+            } else if t == ")" {
+                depth -= 1
+            }
+        } else if kind == CXToken_Keyword && !words.contains(t) {
+            out.safe = false
+        }
     }
     clang_disposeTokens(imp.tu, tokens, n)
+    if depth != 0 {
+        out.safe = false
+    }
+    return out
+}
+
+// "text" as C writes it -> the text: \n \t \" \\ \0 and \xNN; "a" "b" are joined
+fn cString([]str parts) -> Number {
+    out := ""
+    for p in parts {
+        if p.len < 2 || p[0] != '"' || p[p.len - 1] != '"' {
+            return Number()
+        }
+        i := 1
+        while i < p.len - 1 {
+            c := p[i]
+            if c == '\\' && i + 1 < p.len - 1 {
+                i += 1
+                e := p[i]
+                if e == 'n' {
+                    c = '\n'
+                } else if e == 't' {
+                    c = '\t'
+                } else if e == 'r' {
+                    c = '\r'
+                } else if e == '0' {
+                    c = char(0)
+                } else if e == 'x' {
+                    v := 0
+                    while i + 1 < p.len - 1 && "0123456789abcdefABCDEF".contains(str(p[i + 1])) {
+                        i += 1
+                        v = v * 16 + parseInt("0x" + str(p[i])).c.i
+                    }
+                    c = char(v & 255)
+                } else {
+                    c = e
+                }
+            }
+            out = out + str(c)
+            i += 1
+        }
+    }
+    return Number(c: CConst(isStr: true, s: out), ok: true)
+}
+
+// Simple #defines: a single number, maybe negative, maybe in parentheses; or a "text".
+fn macroValue(ptr[Importer] imp, CXCursor c) -> Number {
+    parts := macroTokens(imp, c).parts
+    if parts.len > 0 && parts[0].starts_with("\"") {
+        return cString(parts)
+    }
 
     while parts.len >= 2 && parts[0] == "(" && parts[parts.len - 1] == ")" {
         parts.remove(0)
@@ -438,6 +515,7 @@ fn visit(ptr[Importer] imp, CXCursor c) {
             if !used && !imp.out.structs.has(name) {
                 imp.recordNames[usr] = name
             }
+            defineRecord(imp, u)        // usable even if no function mentions it: Motion m
         }
     } else if kind == CXCursor_FunctionDecl {
         if imp.out.fns.has(name) {
@@ -483,7 +561,7 @@ fn visit(ptr[Importer] imp, CXCursor c) {
         if g.ok {
             imp.out.globals[name] = g.ty
         }
-    } else if kind == CXCursor_StructDecl {
+    } else if kind == CXCursor_StructDecl || kind == CXCursor_UnionDecl {
         if clang_isCursorDefinition(c) != 0 && name.len > 0 && !name.contains(" ") {
             usr := text(clang_getCursorUSR(c))
             if !imp.recordNames.has(usr) {
@@ -500,9 +578,17 @@ fn visit(ptr[Importer] imp, CXCursor c) {
         if clang_Cursor_isMacroBuiltin(c) != 0 || clang_Cursor_isMacroFunctionLike(c) != 0 {
             return
         }
+        if imp.out.consts.has(name) {
+            return
+        }
         v := macroValue(imp, c)
-        if v.ok && !imp.out.consts.has(name) {
+        if v.ok {
             imp.out.consts[name] = v.c
+            return
+        }
+        mt := macroTokens(imp, c)
+        if mt.safe && mt.parts.len > 0 {
+            imp.exprMacros.push(name)        // (1 << 3), (FLAG_A | FLAG_B), ((Uint64)1): asked of clang later
         }
     }
 }
@@ -563,7 +649,60 @@ fn importHeaders([]Header headers, []str dirs, str target, bool windows, str ccF
         imp.header = h.path
         clang_visitChildren(clang_getTranslationUnitCursor(tu), addr(visitTop), addr(imp))
         clang_disposeTranslationUnit(tu)
+        evalMacros(addr(imp), index, h, argv)
     }
     clang_disposeIndex(index)
     return imp.out
 }
+
+// #defines that are expressions: each becomes `static const __auto_type __finch_m_NAME = (NAME);`
+// in a second look at the header, and clang works out the value. One that isn't a constant
+// (a type, a function call) just gives clang an error there, and is left out.
+fn evalMacros(ptr[Importer] imp, ptr index, Header h, []ptr[char] argv) {
+    if imp.exprMacros.len == 0 {
+        return
+    }
+    src := "#include \"" + h.path + "\"\n"
+    for m in imp.exprMacros {
+        src = src + "static const __auto_type __finch_m_" + m + " = (" + m + ");\n"
+    }
+    []str none
+    imp.exprMacros = none
+    file := host.unsavedFile("finch_macros.c", src)
+    ptr tu = null
+    err := clang_parseTranslationUnit2(index, "finch_macros.c", argv.ptr, i32(argv.len), addr(file), 1, u32(CXTranslationUnit_SkipFunctionBodies), addr(tu))
+    if err != CXError_Success || tu == null {
+        return
+    }
+    clang_visitChildren(clang_getTranslationUnitCursor(tu), addr(visitMacroVar), imp)
+    clang_disposeTranslationUnit(tu)
+}
+
+fn visitMacroVar(CXCursor c, CXCursor parent, ptr data) -> i32 {
+    ptr[Importer] imp = data
+    if clang_getCursorKind(c) != CXCursor_VarDecl {
+        return CXChildVisit_Continue
+    }
+    name := text(clang_getCursorSpelling(c))
+    if !name.starts_with("__finch_m_") {
+        return CXChildVisit_Continue
+    }
+    name = name.sub(10, name.len)
+    r := clang_Cursor_Evaluate(c)
+    if r == null || imp.out.consts.has(name) {
+        return CXChildVisit_Continue
+    }
+    k := clang_EvalResult_getKind(r)
+    if k == CXEval_Int {
+        imp.out.consts[name] = CConst(i: int(clang_EvalResult_getAsLongLong(r)))
+    } else if k == CXEval_Float {
+        v := clang_EvalResult_getAsDouble(r)
+        if v == v && v - v == 0.0 {         // not nan or inf
+            imp.out.consts[name] = CConst(isFloat: true, f: v, text: str(v))
+        }
+    } else if k == CXEval_StrLiteral {
+        imp.out.consts[name] = CConst(isStr: true, s: clang_EvalResult_getAsStr(r))
+    }
+    clang_EvalResult_dispose(r)
+    return CXChildVisit_Continue
+}

src/gen.fch

@@ -140,6 +140,7 @@ struct StructDef {
     int index
     int state      // 0 new, 1 resolving, 2 done
     bool owning
+    bool isUnion   // C: a union, every field at byte 0
 }
 
 struct Val {
@@ -1857,6 +1858,10 @@ fn genericCall(ptr[Gen] g, ast.Node c, int ti, int first, str shown) -> Val {
 fn resolveCStruct(ptr[Gen] g, int i) {
     cs := g.ci.structs[g.structs[i].name]
     g.structs[i].state = 1
+    if cs.isUnion {
+        resolveCUnion(g, i, cs)
+        return
+    }
     cur := 0
     []str elems
     for f in cs.fields {
@@ -1890,6 +1895,37 @@ fn resolveCStruct(ptr[Gen] g, int i) {
     g.head.push("%S.C." + cs.name + " = type { " + elems.join(", ") + " }")
 }
 
+// A C union: room for its biggest field, aligned like C aligns it; every field is at byte 0.
+fn resolveCUnion(ptr[Gen] g, int i, cimport.CStruct cs) {
+    for f in cs.fields {
+        if f.hidden {
+            continue
+        }
+        t := resolve(g, f.ty, ast.Node())
+        if t.kind == "struct" {
+            resolveStruct(g, findStruct(g, t.module, t.name))
+        }
+        g.structs[i].fnames.push(f.name)
+        g.structs[i].ftypes.push(t)
+        g.structs[i].lidx.push(0)
+        g.structs[i].offs.push(0)
+    }
+    a := cs.align
+    if a > 8 {
+        a = 8
+    }
+    if a < 1 {
+        a = 1
+    }
+    n := (cs.size + a - 1) / a
+    g.structs[i].unsupported = cs.unsupported
+    g.structs[i].isUnion = true
+    g.structs[i].csize = cs.size
+    g.structs[i].calign = cs.align
+    g.structs[i].state = 2
+    g.head.push("%S.C." + cs.name + " = type { [" + str(n) + " x i" + str(a * 8) + "] }")
+}
+
 fn resolveStruct(ptr[Gen] g, int i) {
     if g.structs[i].state == 2 {
         return
@@ -3567,6 +3603,14 @@ fn arith(ptr[Gen] g, str o, Val l, Val r, ast.Node at) -> Val {
                 panicIf(g, op(g, "icmp eq " + lt + " " + r.v + ", 0"), "division by zero", at)
             }
         }
+        if l.lit && r.lit && l.ty.kind == "int" {
+            // numbers written in the code stay such a number: FLAG_A | FLAG_B fits a u64 or a u32
+            a := int(l.v) or 0
+            b := int(r.v) or 0
+            if (o != "<<" && o != ">>") || (b >= 0 && b < 63) {
+                return Val(v: str(foldInt(o, a, b)), ty: l.ty, lit: true)
+            }
+        }
         return Val(v: op(g, insts[k] + " " + lt + " " + l.v + ", " + r.v), ty: l.ty)
     }
     if isFloat(l.ty) {
@@ -3582,6 +3626,29 @@ fn arith(ptr[Gen] g, str o, Val l, Val r, ast.Node at) -> Val {
     return l
 }
 
+fn foldInt(str o, int a, int b) -> int {
+    if o == "+" {
+        return a + b
+    } else if o == "-" {
+        return a - b
+    } else if o == "*" {
+        return a * b
+    } else if o == "/" {
+        return a / b
+    } else if o == "%" {
+        return a % b
+    } else if o == "&" {
+        return a & b
+    } else if o == "|" {
+        return a | b
+    } else if o == "^" {
+        return a ^ b
+    } else if o == "<<" {
+        return a << b
+    }
+    return a >> b
+}
+
 fn compare(ptr[Gen] g, str o, Val l, Val r, ast.Node at) -> Val {
     eq := o == "==" || o == "!="
     if l.ty.kind == "ptr" || r.ty.kind == "ptr" || l.ty.kind == "null" || r.ty.kind == "null" {
@@ -3747,6 +3814,9 @@ fn ref(ptr[Gen] g, ast.Node e, bool write) -> Ref {
         id := lookup(g, e.name)
         if id < 0 && g.ci.consts.has(e.name) {
             c := g.ci.consts[e.name]
+            if c.isStr {
+                return Ref(isPlace: false, val: strConst(g, c.s))
+            }
             if c.isFloat {
                 return Ref(isPlace: false, val: Val(v: llFloat(c.text), ty: T("float"), lit: true))
             }
@@ -4017,6 +4087,10 @@ fn member(ptr[Gen] g, ast.Node m, bool write) -> Ref {
         fp := op(g, "getelementptr inbounds " + llty(g, t) + ", ptr " + obj.pl.addr + ", i32 0, i32 " + str(sd.lidx[k]))
         return Ref(isPlace: true, pl: Place(addr: fp, ty: ft))
     }
+    if sd.isUnion {
+        u := spill(g, obj.val)
+        return Ref(isPlace: false, val: Val(v: load(g, llty(g, ft), u), ty: ft))
+    }
     fv := op(g, "extractvalue " + llty(g, t) + " " + obj.val.v + ", " + str(sd.lidx[k]))
     out := Val(v: fv, ty: ft)
     if owning(g, ft) && obj.val.fresh {
@@ -5197,6 +5271,21 @@ fn construct(ptr[Gen] g, int si, []ast.Node args, []str names, ast.Node at) -> V
     t := ast.Type(kind: "struct", name: sd.name, module: sd.module)
     st := llty(g, t)
     named := names.len > 0 && names[0].len > 0
+    if sd.isUnion && args.len > 0 {
+        // a union holds one field at a time: Ev(key: k)
+        if args.len != 1 || !named {
+            fail(g, at, sd.name + " is a C union: give one field by name, like " + sd.name + "(" + sd.fnames[0] + ": ...), or none")
+        }
+        k := sd.fnames.find(names[0])
+        if k < 0 {
+            fail(g, args[0], sd.name + " has no field '" + names[0] + "'")
+        }
+        u := alloca(g, st)
+        store(g, st, "zeroinitializer", u)
+        fv := coerce(g, exprWant(g, args[0], sd.ftypes[k]), sd.ftypes[k], args[0], "the field '" + names[0] + "'")
+        store(g, llty(g, sd.ftypes[k]), fv.v, u)
+        return Val(v: load(g, st, u), ty: t)
+    }
     []int given
     for k in 0..sd.fnames.len {
         given.push(-1)

tests/new/c_unions_macros.fch

@@ -0,0 +1,25 @@
+// C unions (SDL_Event-style) and #defines that are expressions, strings or casts
+import "cunion.h"
+link "cunion.c"
+
+fn main() {
+    print(EV_MOTION, EV_KEY, FLAG_BIG, FLAGS_BOTH, HALF, GREETING, LONG_NAME)
+    u64 flags = FLAG_BIG | FLAG_GL
+    u32 small = EV_KEY
+    print(flags, small)
+
+    Event e
+    if poll_event(addr(e)) != 0 && e.type == EV_MOTION {
+        print("motion", e.motion.x, e.motion.y)
+    }
+    k := Event(key: Key(type: EV_KEY, key: 32))
+    print(k.type == EV_KEY, k.key.key)
+    []Event queue
+    queue.push(e)
+    queue.push(k)
+    print(queue[1].key.key, queue.len)
+
+    Tagged t
+    t.as.f = 0.5
+    print(t.as.f)
+}

tests/new/c_unions_macros.out

@@ -0,0 +1,6 @@
+1024 1025 1099511627776 1099511627778 0.5 hi	there ab
+1099511627778 1025
+motion 1.5 -2
+true 32
+32 2
+0.5

tests/new/cunion.c

@@ -0,0 +1,7 @@
+#include "cunion.h"
+int poll_event(Event *e) {
+    e->motion.type = EV_MOTION;
+    e->motion.x = 1.5f;
+    e->motion.y = -2.0f;
+    return 1;
+}

tests/new/cunion.h

@@ -0,0 +1,30 @@
+// C unions and #define constants that are expressions (tests/new/c_unions_macros.fch)
+#include <stdint.h>
+
+typedef struct { uint32_t type; float x, y; } Motion;
+typedef struct { uint32_t type; int32_t key; } Key;
+typedef union {
+    uint32_t type;
+    Motion motion;
+    Key key;
+    uint8_t padding[64];
+} Event;
+
+typedef struct { int kind; union { int i; float f; } as; } Tagged;
+
+#define EV_MOTION 0x400u
+#define EV_KEY (EV_MOTION + 1)
+#define FLAG_BIG ((uint64_t)1 << 40)
+#define FLAG_GL 0x2ULL
+#define FLAGS_BOTH (FLAG_BIG | FLAG_GL)
+#define HALF (1.0 / 2)
+#define GREETING "hi\tthere"
+#define LONG_NAME "a" "b"
+#define NOT_A_CONSTANT do { } while (0)
+
+static inline int fill_key(Event *e, int key) {
+    e->key.type = EV_KEY;
+    e->key.key = key;
+    return 1;
+}
+int poll_event(Event *e);

tests/new/fail/union_two_fields.fch

@@ -0,0 +1,5 @@
+// expect: Event is a C union: give one field by name
+import "../cunion.h"
+fn main() {
+    e := Event(1, 2)
+}