feat: prototype of AST

4 files changed, 201 insertions, 63 deletions

diff --git a/sys/cmd/cc/cc.c b/sys/cmd/cc/cc.c
index cb37af4..fcb9217 100644
--- a/sys/cmd/cc/cc.c
+++ b/sys/cmd/cc/cc.c
@@ -253,6 +253,12 @@ init(void)
 
     C.lxr     = (Lexer){ 0 };
     C.lxr.b   = C.lxr.buf;
+
+    /* predefine macros */
+    dodefine(&C.lxr, "__LINE__");
+    dodefine(&C.lxr, "__FILE__");
+    C.lxr.macline = (uintptr)lookup(&C.lxr.sym, "__LINE__");
+    C.lxr.macfile = (uintptr)lookup(&C.lxr.sym, "__FILE__");
 }
 
 error
diff --git a/sys/cmd/cc/cc.h b/sys/cmd/cc/cc.h
index b45c5fa..144eb4a 100644
--- a/sys/cmd/cc/cc.h
+++ b/sys/cmd/cc/cc.h
@@ -16,7 +16,8 @@ typedef struct Lexer Lexer;
 typedef struct Sym   Sym;
 typedef struct Type  Type;
 
-typedef struct Spec  Spec;
+typedef struct Ptr   Ptr;
+typedef struct Name  Name;
 typedef struct Decl  Decl;
 typedef struct Stmt  Stmt;
 typedef struct Expr  Expr;
@@ -64,7 +65,8 @@ typedef struct Compiler Compiler;
     KEYWORD(Kgoto,"goto")              \
     KEYWORD(Kreturn,"return")          \
     KEYWORD(Ksizeof,"sizeof")          \
-    KEYWORD(Kalignof,"alignof")        
+    KEYWORD(Kalignof,"alignof")        \
+    KEYWORD(Kalignas,"alignas")        
 
 #define KEYWORD(a, b) a,
 enum { KEYWORDS NUM_KEYWORDS };
@@ -250,6 +252,10 @@ struct Lexer
     SymTab sym;
     byte   *b;
     byte    buf[2*1024];
+
+    /* predefined dynamic macros */
+    uintptr macfile;
+    uintptr macline;
 };
 
 /* lex.c functions */
@@ -270,96 +276,206 @@ void puttok(Token);
 // parsing & type resolution
 // tokens -> ast
 
-/* statements */
+struct Node
+{
+    Range   pos;
+    uint32  kind;
+};
+
+/* ast types */
 enum
 {
-    Sbad,
-    Slabel,
+    Nbad,
+    /* labels */
+    Slabel, Scase, 
     Sblock,
     Sexpr,
     Sselect,
-    Sloop,
-    Sjump,
+    /* loops */
+    Sfor, Swhile, Sdo, 
+    /* jumps */
+    Sgoto, Scontin, Sbreak, Sreturn, 
+    /* forks */
+    Sif, Sswitch, 
+
+
+    /* assignments */
+    Xasn, Xmulasn, Xdivasn, Xmodasn, Xsubasn, Xaddasn, 
+    Xlsftasn, Xrsftasn, Xandasn, Xxorasn, Xorasn,
+    /* conditional */
+    Xternary,
+    /* unary prefix ops  */
+    Xref, Xstar, Xplus, Xminus, Xneg, Xnot, Xsizeof, Xalnof, Xpreinc, Xpredec, 
+    Xcast, 
+    /* unary postfix ops */
+    Xpostinc, Xpostdec, Xindex, Xcall, Xsel, Xcmpndlit, 
+    /* binary ops */
+    Xoror, Xandand, Xor, Xxor, Xand, Xeql, Xgt, Xlt, Xgteq, Xlteq, Xlsft, Xrsft, 
+    Xadd, Xsub, Xmul, Xdiv, Xmod,
+    /* primary */
+    Xparen, Xident, Xlit,  
+    /* lists */
+    Xcomma,
+    
+    Dfunc, 
+    Dvar,
 };
 
+/* statements */
 struct Stmt
 {
-    union 
-    {
-
+    struct Node;
+    union {
+        struct {
+            union {
+                string ident;
+                Expr  *x;
+            };
+            Stmt   *stmt;
+        } lbl;
+        struct {
+            long             n;
+            struct Node **item;
+        } blk;
+        Expr *x;
+        struct {
+            union {
+                Expr *x;
+                Decl *d;
+            } init;
+            Expr *cond;
+            Expr *step;
+            Stmt *body;
+        } loop;
+        union{
+            string lbl;
+            Expr  *x;
+        } jmp;
+        struct {
+            Expr *cond;
+            Stmt *body;
+            Stmt *orelse;
+        } sel;
     };
 };
 
 /* expressions */
-enum
+struct Expr
 {
-    Xbad,
-    Xparen,
-    Xident,
-    Xlit,
-    Xassign,
-    Xcomma,
-    Xternary,
-    Xbinary,
-    Xprefix,
-    Xunary,
-    Xpostfix,
-    Xcast,
+    struct Node;
+    union {
+        struct {
+            Expr *l;
+            Expr *r;
+        } asn;
+        struct {
+            Expr *c;
+            Expr *t;
+            Expr *e;
+        } cond;
+        union {
+            Expr *pre;
+            Expr *post;
+        } unary; 
+        struct {
+            uint64 to;
+            Expr   *x;
+        } cast;
+        struct {
+            Expr *l;
+            Expr *r;
+        } binary;
+        struct {
+            Expr *x[2];
+        } comma;
+    };
 };
 
-struct Expr
+/* declarations */
+
+struct Ptr
 {
-    union 
-    {
+    uint64  kind;
+    Ptr    *link;
+};
 
+struct Name 
+{
+    struct Node;
+    union {
+        string        ident;
+        struct Dtor  *paren;
+        struct {
+            uint32 kind;
+            Name  *base;
+            union {
+                Expr *index;
+                /* TODO:
+                 * func params
+                 * variadic arrays
+                 * compound set notation
+                 */
+            };
+        } suffix;
     };
 };
 
-/* declarations */
+struct Dtor
+{
+    Range pos;
+    Ptr   ptr;
+    Name  name;
+};
 
 // specifiers
 enum
 {
-    Mauto   = iota(0), // Corresponds to auto int
+    /* memory */
+    Mauto   = iota(0),
     Mtype   = iota(1),
     Mstatic = iota(2),
     Mreg    = iota(3),
-
-    Qconst  = iota(4),
-    Qrestr  = iota(5),
-    Qvoltl  = iota(6),
-    Finlne  = iota(7),
-
-    Tlong   = iota(8),
-    Tvlong  = iota(9),
-    Tsign   = iota(10),
-    Tunsign = iota(11),
-
-    Tvoid   = iota(12),
-    Tchar   = iota(13),
-    Tshort  = iota(14),
-    Tfloat  = iota(15),
-    Tdouble = iota(16),
-    Tcmplx  = iota(17),
-    Timag   = iota(18),
-
-    Taggr   = iota(19),
-    Tenum   = iota(20),
-    Tname   = iota(21),
-};
-
-enum
-{
-    Dnil,
-    Dbad,
-    Dfunc,
-    Dvar,
+    Mtls    = iota(4),
+
+    /* qualifiers */
+    Qconst  = iota(5),
+    Qrestr  = iota(6),
+    Qvoltl  = iota(7),
+    Qatom   = iota(8),
+
+    Finlne  = iota(9),
+    Fnoret  = iota(10),
+
+    /* types */
+    Tlong   = iota(12),
+    Tvlong  = iota(13),
+    Tsign   = iota(14),
+    Tunsign = iota(15),
+
+    Tvoid   = iota(16),
+    Tchar   = iota(17),
+    Tshort  = iota(18),
+    Tfloat  = iota(19),
+    Tdouble = iota(20),
+    Tcmplx  = iota(21),
+    Timag   = iota(22),
+
+    Tatom   = iota(23),
+
+    Taggr   = iota(24),
+    Tenum   = iota(25),
+    Tname   = iota(26),
+
+    /* alignment */
+    Alnas   = iota(27),
 };
 
 struct Decl
 {
-    Pos  pos;
-    uint kind;
+    struct Node;
+    union {
+
+    };
 };
 
 // -----------------------------------------------------------------------
@@ -426,9 +542,14 @@ struct Compiler
         string  *path;
     } omit;
 
-    string outfile;
-
     Lexer  lxr;
+
+    string outfile;
+    struct {
+        int    cap;
+        int    len;
+        Decl  *decls;
+    } ast;
 }; 
 extern Compiler C;
 
diff --git a/sys/cmd/cc/lex.c b/sys/cmd/cc/lex.c
index 7a08373..e35ea7e 100644
--- a/sys/cmd/cc/lex.c
+++ b/sys/cmd/cc/lex.c
@@ -636,6 +636,16 @@ Dispatch:
 
         sym = lookup(&lx->sym, tok.val.s);
         if (sym && ((uintptr)sym->name != (uintptr)lx->io->path)) {
+            if ((uintptr)sym == lx->macline) {
+                tok.kind  = Alit | Vint;
+                tok.val.i = lx->pos.line;
+                goto Return;
+            }
+            if ((uintptr)sym == lx->macfile) {
+                tok.kind  = Alit | Vstr; 
+                tok.val.s = lx->pos.path;
+                goto Return;
+            }
             io = makeio(sym->name);
             io->rdr.end   += expandmacro(lx, sym, io->b);
             *io->rdr.end++ = EOF;
diff --git a/sys/cmd/cc/pp.c b/sys/cmd/cc/pp.c
index 737e018..30a162b 100644
--- a/sys/cmd/cc/pp.c
+++ b/sys/cmd/cc/pp.c
@@ -1099,7 +1099,8 @@ dodefine(Lexer *lx, string s)
             errorf("redefinition of symbol '%s'", sym->name);
             return 1;
         }
-        sym->macro = "\001";
+        sym = define(&lx->sym, s, Smacro);
+        sym->macro = "\00\02";
     }
 
     return 0;