Support for unions

README.md

@@ -14,8 +14,8 @@ Most of the functionality is contained within the `toast` binary that is built w
 
 The nimterop feature set is still limited when compared with c2nim. Supported language constructs include:
 - `#define NAME VALUE` where `VALUE` is a number (int, float, hex)
-- `struct X`, `typedef struct`, `enum X`, `typedef enum`
+- `struct X`, `typedef struct`, `enum X`, `typedef enum`, `union X`, `typedef union`
-- Functions with primitive types, structs, enums and typedef structs/enums as params and return values
+- Functions with primitive types, structs, enums, unions and typedef structs/enums/unions as params and return values
 - Pointers to data types
 
 Given the simplicity and success of this approach so far, it seems feasible to continue on for more complex code. The goal is to make interop seamless so nimterop will focus on wrapping headers and not the outright conversion of C/C++ implementation.

nimterop/ast.nim

@@ -84,7 +84,7 @@ proc searchAst(root: TSNode) =
       if name in gStateRT.ast:
         for ast in gStateRT.ast[name]:
           if searchAstForNode(ast, node):
-            ast.tonim()
+            ast.tonim(ast, node)
             break
         gStateRT.data = @[]
     else:

nimterop/globals.nim

@@ -2,6 +2,8 @@ import tables
 
 import regex
 
+import treesitter/runtime
+
 type
   Kind* = enum
     exactlyOne
@@ -13,7 +15,7 @@ type
     name*: string
     kind*: Kind
     children*: seq[ref Ast]
-    tonim*: proc () {.closure, locks: 0.}
+    tonim*: proc (ast: ref Ast, node: TSNode) {.closure, locks: 0.}
     regex*: Regex
 
   State* = object
@@ -28,7 +30,7 @@ type
 
     ast*: Table[string, seq[ref Ast]]
     data*: seq[tuple[name, val: string]]
-    grammar*: seq[tuple[grammar: string, call: proc() {.locks: 0.}]]
+    grammar*: seq[tuple[grammar: string, call: proc(ast: ref Ast, node: TSNode) {.locks: 0.}]]
 
 var
   gStateCT* {.compiletime.}: State

nimterop/grammar.nim

@@ -1,7 +1,9 @@
-import strformat, tables
+import strformat, strutils, tables
 
 import regex
 
+import treesitter/runtime
+
 import "."/[getters, globals, lisp]
 
 proc initGrammar() =
@@ -12,7 +14,7 @@ proc initGrammar() =
     (preproc_arg)
    )
   """,
-    proc () {.closure, locks: 0.} =
+    proc (ast: ref Ast, node: TSNode) {.closure, locks: 0.} =
       let
         name = gStateRT.data[0].val.getIdentifier()
         val = gStateRT.data[1].val.getLit()
@@ -41,7 +43,7 @@ proc initGrammar() =
     )
    )
   """,
-    proc () {.closure, locks: 0.} =
+    proc (ast: ref Ast, node: TSNode) {.closure, locks: 0.} =
       var
         name = gStateRT.data[1].val.getIdentifier()
         typ = gStateRT.data[0].val.getIdentifier()
@@ -51,12 +53,31 @@ proc initGrammar() =
         gStateRT.typeStr &= &"  {name}* = {typ}\n"
   ))
 
-  proc pStructCommon(name: string, fstart, fend: int, prefix="") =
+  proc pStructCommon(ast: ref Ast, node: TSNode, name: string, fstart, fend: int) =
-    let
+    var
       nname = name.getIdentifier()
+      prefix = ""
+      union = ""
+
+    case $node.tsNodeType():
+      of "struct_specifier":
+        prefix = "struct "
+      of "union_specifier":
+        prefix = "union "
+        union = " {.union.}"
+      of "type_definition":
+        if node.getTSNodeNamedChildCountSansComments() != 0:
+          for i in 0 .. node.tsNodeNamedChildCount()-1:
+            let
+              nchild = $node.tsNodeNamedChild(i).tsNodeType()
+            if nchild != "comment":
+              if nchild == "union_specifier":
+                union = " {.union.}"
+              break
+
     if nname notin gStateRT.types:
       gStateRT.types.add(nname)
-      gStateRT.typeStr &= &"  {nname}* {{.importc: \"{prefix}{name}\", header: {gStateRT.currentHeader}, bycopy.}} = object\n"
+      gStateRT.typeStr &= &"  {nname}* {{.importc: \"{prefix}{name}\", header: {gStateRT.currentHeader}, bycopy.}} = object{union}\n"
 
       var
         i = fstart
@@ -69,13 +90,13 @@ proc initGrammar() =
 
   # struct X {}
   gStateRT.grammar.add(("""
-   (struct_specifier
+   (struct_specifier|union_specifier
     (type_identifier)
     (field_declaration_list
      (field_declaration+
       (primitive_type|type_identifier?)
       (sized_type_specifier?
-       (primitive_type)
+       (primitive_type?)
       )
       (struct_specifier?
        (type_identifier)
@@ -88,14 +109,14 @@ proc initGrammar() =
     )
    )
   """,
-    proc () {.closure, locks: 0.} =
+    proc (ast: ref Ast, node: TSNode) {.closure, locks: 0.} =
-      pStructCommon(gStateRT.data[0].val, 1, 0, "struct ")
+      pStructCommon(ast, node, gStateRT.data[0].val, 1, 1)
   ))
 
   # typedef struct X {}
   gStateRT.grammar.add(("""
    (type_definition
-    (struct_specifier
+    (struct_specifier|union_specifier
      (field_declaration_list
       (field_declaration+
        (primitive_type|type_identifier?)
@@ -115,11 +136,11 @@ proc initGrammar() =
     (type_identifier)
    )
   """,
-    proc () {.closure, locks: 0.} =
+    proc (ast: ref Ast, node: TSNode) {.closure, locks: 0.} =
-      pStructCommon(gStateRT.data[^1].val, 0, 1)
+      pStructCommon(ast, node, gStateRT.data[^1].val, 0, 1)
   ))
 
-  proc pEnumCommon(name: string, fstart, fend: int, prefix="") =
+  proc pEnumCommon(ast: ref Ast, node: TSNode, name: string, fstart, fend: int) =
     let
       nname = name.getIdentifier()
     if nname notin gStateRT.types:
@@ -151,8 +172,8 @@ proc initGrammar() =
     )
    )
   """,
-    proc () {.closure, locks: 0.} =
+    proc (ast: ref Ast, node: TSNode) {.closure, locks: 0.} =
-      pEnumCommon(gStateRT.data[0].val, 1, 0)
+      pEnumCommon(ast, node, gStateRT.data[0].val, 1, 0)
   ))
 
   # typedef enum {} X
@@ -169,8 +190,8 @@ proc initGrammar() =
     (type_identifier)
    )
   """,
-    proc () {.closure, locks: 0.} =
+    proc (ast: ref Ast, node: TSNode) {.closure, locks: 0.} =
-      pEnumCommon(gStateRT.data[^1].val, 0, 1)
+      pEnumCommon(ast, node, gStateRT.data[^1].val, 0, 1)
   ))
 
   # typ function(typ param1, ...)
@@ -181,7 +202,7 @@ proc initGrammar() =
     (sized_type_specifier?
      (primitive_type?)
     )
-    (struct_specifier?
+    (struct_specifier|union_specifier?
      (type_identifier)
     )
     (function_declarator?
@@ -193,7 +214,7 @@ proc initGrammar() =
        (sized_type_specifier?
         (primitive_type?)
        )
-       (struct_specifier?
+       (struct_specifier|union_specifier?
         (type_identifier)
        )
        (enum_specifier?
@@ -216,7 +237,7 @@ proc initGrammar() =
         (sized_type_specifier?
          (primitive_type?)
         )
-        (struct_specifier?
+        (struct_specifier|union_specifier?
          (type_identifier)
         )
         (enum_specifier?
@@ -232,7 +253,7 @@ proc initGrammar() =
     )
    )
   """,
-    proc () {.closure, locks: 0.} =
+    proc (ast: ref Ast, node: TSNode) {.closure, locks: 0.} =
       let
         ftyp = gStateRT.data[0].val.getIdentifier()
         fname = gStateRT.data[1].val
@@ -277,10 +298,11 @@ proc parseGrammar*() =
 
     ast.tonim = gStateRT.grammar[i].call
     ast.initRegex()
-    if ast.name notin gStateRT.ast:
+    for n in ast.name.split("|"):
-      gStateRT.ast[ast.name] = @[ast]
+      if n notin gStateRT.ast:
-    else:
+        gStateRT.ast[n] = @[ast]
-      gStateRT.ast[ast.name].add(ast)
+      else:
+        gStateRT.ast[n].add(ast)
 
 proc printGrammar*() =
   for name in gStateRT.ast.keys():

tests/include/test.c

@@ -5,7 +5,7 @@ int test_call_int() {
 }
 
 #ifdef FORCE
-struct STRUCT1 _test_call_int_param_(int param1) {
+struct STRUCT1 _test_call_param_(int param1) {
 	struct STRUCT1 s;
 
 	s.field1 = param1;
@@ -14,7 +14,7 @@ struct STRUCT1 _test_call_int_param_(int param1) {
 }
 #endif
 
-STRUCT2 test_call_int_param2(int param1, STRUCT2 param2) {
+STRUCT2 test_call_param2(int param1, STRUCT2 param2) {
 	STRUCT2 s;
 
 	s.field1 = param1 + param2.field1;
@@ -22,7 +22,7 @@ STRUCT2 test_call_int_param2(int param1, STRUCT2 param2) {
 	return s;
 }
 
-STRUCT2 test_call_int_param3(int param1, struct STRUCT1 param2) {
+STRUCT2 test_call_param3(int param1, struct STRUCT1 param2) {
 	STRUCT2 s;
 
 	s.field1 = param1 + param2.field1;
@@ -30,6 +30,22 @@ STRUCT2 test_call_int_param3(int param1, struct STRUCT1 param2) {
 	return s;
 }
 
-ENUM2 test_call_int_param4(enum ENUM param1) {
+ENUM2 test_call_param4(enum ENUM param1) {
 	return enum4;
 }
+
+union UNION1 test_call_param5(float param1) {
+	union UNION1 u;
+
+	u.field2 = param1;
+
+	return u;
+}
+
+unsigned char test_call_param6(UNION2 param1) {
+	return param1.field2;
+}
+
+int test_call_param7(union UNION1 param1) {
+	return param1.field1;
+}

tests/include/test.h

@@ -36,8 +36,21 @@ typedef struct {
 	int *field;
 } STRUCT4;
 
+union UNION1 {
+	int field1;
+	float field2;
+};
+
+typedef union {
+	double field1;
+	unsigned char field2;
+} UNION2;
+
 int test_call_int();
-struct STRUCT1 _test_call_int_param_(int param1);
+struct STRUCT1 _test_call_param_(int param1);
-STRUCT2 test_call_int_param2(int param1, STRUCT2 param2);
+STRUCT2 test_call_param2(int param1, STRUCT2 param2);
-STRUCT2 test_call_int_param3(int param1, struct STRUCT1 param2);
+STRUCT2 test_call_param3(int param1, struct STRUCT1 param2);
-ENUM2 test_call_int_param4(enum ENUM param1);
+ENUM2 test_call_param4(enum ENUM param1);
+union UNION1 test_call_param5(float param1);
+unsigned char test_call_param6(UNION2 param1);
+int test_call_param7(union UNION1 param1);

tests/tnimterop_c.nim

@@ -28,6 +28,9 @@ var
   vptr: VOIDPTR
   iptr: INTPTR
 
+  u: UNION1
+  u2: UNION2
+
 pt = 3
 ct = 4
 
@@ -38,10 +41,16 @@ s3.field1 = 7
 e = enum1
 e2 = enum4
 
+u2.field2 = 'c'
+
 check test_call_int() == 5
-check test_call_int_param(5).field1 == 5
+check test_call_param(5).field1 == 5
-check test_call_int_param2(5, s2).field1 == 11
+check test_call_param2(5, s2).field1 == 11
-check test_call_int_param3(5, s).field1 == 10
+check test_call_param3(5, s).field1 == 10
-check test_call_int_param4(e) == e2
+check test_call_param4(e) == e2
+check test_call_param5(5.0).field2 == 5.0
+check test_call_param6(u2) == 'c'
+u.field1 = 4
+check test_call_param7(u) == 4
 
 cAddStdDir()