add parser

Makefile

@@ -1,5 +1,5 @@
 
-larpcc: main.c lex.c preproc.c collection.c
+larpcc: main.c lex.c preproc.c collection.c parse.c syms.c
 	gcc $^ -o $@ -std=c23 -pedantic -Wall -Wextra -g -fsanitize=address
 
 

ast.h

@@ -8,57 +8,101 @@ typedef struct {
     int col;
 } Loc;
 
+typedef struct AstStmt AstStmt;
+typedef struct AstDecl AstDecl;
+typedef struct AstDeclar AstDeclar;
+typedef struct AstExpr AstExpr;
+
+typedef struct AstStmtBlock AstStmtBlock;
+
+struct AstStmtBlock {
+    ARRAY_FIELD(AstStmt*, stmts);
+    Loc loc;
+};
+
+typedef struct AstDeclBlock AstDeclBlock;
+
+struct AstDeclBlock {
+    ARRAY_FIELD(AstDecl*, decls);
+    Loc loc;
+};
+
+typedef enum {
+    AstStmt_Decl,
+    AstStmt_Expr,
+} AstStmtTy;
+
+struct AstStmt {
+    AstStmtTy ty;
+    Loc loc;
+    union {
+        AstDecl* decl;
+        AstExpr* expr;
+    };
+};
+
 typedef enum : unsigned int {
-    AstSpec_Typedef = 1 << 0,
+    AstSpec_Int = 1 << 0,
-    // AstSpec_Constexpr = 1 << 1,
+    AstSpec_Char = 1 << 1,
-    // AstSpec_Auto = 1 << 2,
+    AstSpec_DefTy = 1 << 6,
-    // AstSpec_Register = 1 << 3,
+    AstSpec_Unsigned = 1 << 7,
-    AstSpec_Static = 1 << 4,
+    AstSpec_Static = 1 << 8,
-    // AstSpec_Extern = 1 << 5,
+    AstSpec_Const = 1 << 9,
-    // AstSpec_ThreadLocal = 1 << 6,
+    AstSpec_Inline = 1 << 10,
-    AstSpec_Const = 1 << 7,
+    AstSpec_Typedef = 1 << 11,
-    // AstSpec_Volatile = 1 << 8,
+    AstSpec_Enum = 1 << 12,
-    // AstSpec_Restrict = 1 << 9,
+    AstSpec_Struct = 1 << 13,
-    // AstSpec_Atomic = 1 << 10,
+    AstSpec_Union = 1 << 14,
-    AstSpec_Inline = 1 << 11,
 } AstSpec;
 
+struct AstDecl {
+    AstSpec spec;
+    ARRAY_FIELD(AstDeclar*, declars);
+    AstStmtBlock* function_body;
+    Loc loc;
+};
+
 typedef enum {
     AstDeclar_Ident,
     AstDeclar_Pointer,
     AstDeclar_Array,
+    AstDeclar_Function,
 } AstDeclarTy;
 
-typedef struct AstDeclar AstDeclar;
-
 struct AstDeclar {
     AstDeclarTy ty;
     Loc loc;
+    AstExpr* init_expr;
     union {
         char* ident;
         struct {
             AstDeclar* declar;
             AstSpec spec;
         } pointer;
+        struct {
+            AstDeclar* declar;
+            AstExpr* expr;
+        } array;
+        struct {
+            ARRAY_FIELD(AstDecl, params);
+        } function;
     };
 };
 
-typedef struct {
-    AstSpec spec;
-    unsigned long type_id;
-    ARRAY_FIELD(AstDeclar, declars);
-    Loc loc;
-} AstDecl;
-
 typedef enum {
-    AstStmt_Decl,
+    AstExpr_Str,
-} AstStmtTy;
+    AstExpr_Call,
+} AstExprTy;
 
-typedef struct {
+struct AstExpr {
-    AstStmtTy ty;
+    AstExprTy ty;
     union {
-        AstDecl decl;
+        char* str_val;
+        struct {
+            AstExpr* expr;
+            ARRAY_FIELD(AstExpr*, args);
+        } call;
+    };
 };
-} AstStmt;
 
 #endif

collection.c

@@ -225,3 +225,10 @@ void arena_array_push(Arena* arena,
         arena_realloc_wrap,
         arena);
 }
+
+char* arena_strdup(Arena* a, const char* str)
+{
+    char* p = arena_alloc(a, strlen(str) + 1);
+    strcpy(p, str);
+    return p;
+}

collection.h

@@ -74,4 +74,6 @@ void arena_array_push(Arena* arena,
     void* elem,
     size_t elem_size);
 
+char* arena_strdup(Arena* a, const char* str);
+
 #endif

lex.c

@@ -96,8 +96,7 @@ void source_iter_init(SourceIter* r, const char* text)
 Char source_iter_next(SourceIter* r)
 {
     char ch = r->text[r->i];
-    int line = r->line;
+    Loc loc = { r->line, r->col };
-    int col = r->col;
 
     if (r->text[r->i] != '\0') {
         if (ch == '\n') {
@@ -109,7 +108,7 @@ Char source_iter_next(SourceIter* r)
         r->i += 1;
     }
 
-    return (Char) { ch, line, col };
+    return (Char) { ch, loc };
 }
 
 void lsl_iter_init(LslIter* r, SourceIter* source_char_iter)
@@ -149,13 +148,13 @@ void lexer_init(Lexer* r, LslIter* lsl_char_iter, Arena* arena)
     };
 }
 
-static Tok tok(Lexer* l, TokTy ty, const char* text, int line, int col)
+static Tok tok(Lexer* l, TokTy ty, const char* text, Loc loc)
 {
     char* text_dup = arena_alloc(l->arena, strlen(text) + 1);
     text_dup[0] = '\0';
     strcpy(text_dup, text);
 
-    return (Tok) { ty, text_dup, line, col };
+    return (Tok) { ty, text_dup, loc };
 }
 
 static bool done(Lexer* l)
@@ -176,14 +175,13 @@ static void step(Lexer* l, char text[], size_t* text_len)
 
 Tok lexer_next(Lexer* l)
 {
-    int line = l->curr.line;
+    Loc loc = l->curr.loc;
-    int col = l->curr.col;
 
     char text[max_token_text_len] = "";
     size_t text_len = 0;
 
     if (done(l)) {
-        return tok(l, Tok_Eof, text, line, col);
+        return tok(l, Tok_Eof, text, loc);
     }
 
     bool matched = false;
@@ -193,7 +191,7 @@ Tok lexer_next(Lexer* l)
     }
     if (matched) {
         if (l->lex_whitespace) {
-            return tok(l, Tok_Whitespace, text, line, col);
+            return tok(l, Tok_Whitespace, text, loc);
         }
         return lexer_next(l);
     }
@@ -206,7 +204,7 @@ Tok lexer_next(Lexer* l)
         step(l, text, &text_len);
     }
     if (matched) {
-        return tok(l, Tok_Ident, text, line, col);
+        return tok(l, Tok_Ident, text, loc);
     }
 
     while ((l->curr.ch >= '1' && l->curr.ch <= '9')
@@ -215,7 +213,7 @@ Tok lexer_next(Lexer* l)
         step(l, text, &text_len);
     }
     if (matched) {
-        return tok(l, Tok_Int, text, line, col);
+        return tok(l, Tok_Int, text, loc);
     }
 
     if (l->lex_headername && (l->curr.ch == '"' || l->curr.ch == '<')) {
@@ -228,11 +226,11 @@ Tok lexer_next(Lexer* l)
             fprintf(stderr,
                 "error: malformed header name, expected '%c' on line %d\n",
                 ter,
-                l->curr.line);
+                l->curr.loc.line);
             return lexer_next(l);
         }
         step(l, text, &text_len);
-        return tok(l, Tok_HeaderName, text, line, col);
+        return tok(l, Tok_HeaderName, text, loc);
     }
 
     if (l->curr.ch == '"') {
@@ -248,11 +246,11 @@ Tok lexer_next(Lexer* l)
         if (done(l) || l->curr.ch != '"') {
             fprintf(stderr,
                 "error: malformed string literal on line %d\n",
-                l->curr.line);
+                l->curr.loc.line);
             return lexer_next(l);
         }
         step(l, text, &text_len);
-        return tok(l, Tok_Str, text, line, col);
+        return tok(l, Tok_Str, text, loc);
     }
 
     if (l->curr.ch == '\'') {
@@ -266,22 +264,22 @@ Tok lexer_next(Lexer* l)
         if (done(l) && l->curr.ch != '\'') {
             fprintf(stderr,
                 "error: malformed character literal on line %d\n",
-                l->curr.line);
+                l->curr.loc.line);
             return lexer_next(l);
         }
         step(l, text, &text_len);
-        return tok(l, Tok_Char, text, line, col);
+        return tok(l, Tok_Char, text, loc);
     }
 
     if (strchr("\n;(){},*#", l->curr.ch)) {
         TokTy ty = (TokTy)l->curr.ch;
         step(l, text, &text_len);
-        return tok(l, ty, text, line, col);
+        return tok(l, ty, text, loc);
     }
     fprintf(stderr,
         "error: illegal character '%c' on line %d\n",
         l->curr.ch,
-        l->curr.line);
+        l->curr.loc.line);
     step(l, text, &text_len);
     return lexer_next(l);
 }

main.c

@@ -30,26 +30,26 @@ int main(int argc, char* argv[])
 
     printf("--- tokens ---\n");
 
-    int line = 1;
+    Loc loc = { 1, 1 };
-    int col = 1;
     Tok tok;
     while ((tok = tokstream_next(&tokstream)).ty != Tok_Eof) {
         // printf("%s\t%s\n", tokty_str(tok.ty), tok.text);
-        if (line > tok.line) {
+        Loc tok_loc = tok.loc;
-            line = 0;
+        if (loc.line > tok_loc.line) {
+            loc.line = 0;
         }
-        while (line < tok.line) {
+        while (loc.line < tok_loc.line) {
             printf("\n");
-            line += 1;
+            loc.line += 1;
-            col = 1;
+            loc.col = 1;
         }
-        while (col < tok.col) {
+        while (loc.col < tok_loc.col) {
             printf(" ");
-            col += 1;
+            loc.col += 1;
         }
         printf("%s", tok.text);
-        line = tok.line;
+        loc.line = tok_loc.line;
-        col = tok.col + strlen(tok.text);
+        loc.col = tok_loc.col + strlen(tok.text);
     }
     fputs("\n", stdout);
 

parse.c

@@ -1,5 +1,7 @@
 #include "parse.h"
+#include "ast.h"
 #include "collection.h"
+#include "syms.h"
 #include <assert.h>
 #include <stddef.h>
 #include <stdio.h>
@@ -41,11 +43,11 @@ static KwNodeOpt* kwmatcher_create_path(
             .idx = new_idx,
         };
         arena_array_push(&m->arena,
-            (void**)&(m)->nodes,
+            (void**)&m->nodes,
-            &(m)->nodes_cap,
+            &m->nodes_cap,
-            &(m)->nodes_size,
+            &m->nodes_size,
-            (&new_node),
+            &new_node,
-            sizeof(*(&new_node)));
+            sizeof(new_node));
         KwNode* node = &m->nodes[new_idx];
 
         KwNodeOpt new_opt = {
@@ -57,8 +59,8 @@ static KwNodeOpt* kwmatcher_create_path(
             (void**)&node->opts,
             &node->opts_cap,
             &node->opts_size,
-            (&new_opt),
+            &new_opt,
-            sizeof(*(&new_opt)));
+            sizeof(new_opt));
 
         opt = &node->opts[node->opts_size - 1];
     }
@@ -147,7 +149,7 @@ static Tok kwmatcher_next(KwMatcher* m)
         return tok;
     TokTy ty = kwmatcher_match(m, tok.text);
     if (ty != Tok_Ident)
-        printf("info: matched '%s' on line %d\n", tok.text, tok.line);
+        printf("info: matched '%s' on line %d\n", tok.text, tok.loc.line);
     tok.ty = ty;
     return tok;
 }
@@ -155,6 +157,11 @@ static Tok kwmatcher_next(KwMatcher* m)
 struct Parser {
     KwMatcher kwmatcher;
     Arena* arena;
+    Arena sym_arena;
+
+    Syms* ord_ns;
+    Syms* label_ns;
+    Syms* tag_ns;
 
     Tok curr;
 };
@@ -166,6 +173,11 @@ Parser* parser_new(TokStream* tokstream, Arena* arena)
     *p = (Parser) {
         .kwmatcher = { },
         .arena = arena,
+        .sym_arena = { },
+
+        .ord_ns = syms_new(),
+        .label_ns = syms_new(),
+        .tag_ns = syms_new(),
 
         .curr = { },
     };
@@ -181,3 +193,311 @@ void parser_free(Parser* p)
     kwmatcher_deinit(&p->kwmatcher);
     free(p);
 }
+
+static void step(Parser* p)
+{
+    p->curr = kwmatcher_next(&p->kwmatcher);
+}
+
+#define DEFINE_NEW(TYPE, FN_NAME)                                              \
+    [[maybe_unused]] static TYPE* new_##FN_NAME(Parser* p, TYPE init)          \
+    {                                                                          \
+        TYPE* val = arena_alloc(p->arena, sizeof(TYPE));                       \
+        *val = init;                                                           \
+        return val;                                                            \
+    }
+
+DEFINE_NEW(AstStmt, stmt)
+DEFINE_NEW(AstDecl, decl)
+DEFINE_NEW(AstDeclar, declar)
+
+static bool is_decl_start_kw(TokTy ty)
+{
+    switch (ty) {
+        case Tok_KwInt:
+        case Tok_KwChar:
+        case Tok_KwConst:
+            return true;
+        default:
+            return false;
+    }
+}
+
+static bool is_defty_ident(Parser* p, Tok tok)
+{
+    if (tok.ty != Tok_Ident)
+        return false;
+    Sym* sym = syms_get(p->ord_ns, tok.text);
+    return sym && sym->ty == Sym_Typedef;
+}
+
+AstStmt* parser_parse_stmt(Parser* p)
+{
+    Loc loc = p->curr.loc;
+
+    if (is_decl_start_kw(p->curr.ty) || is_defty_ident(p, p->curr)) {
+        AstDecl* decl = parser_parse_decl(p);
+
+        return new_stmt(p,
+            (AstStmt) {
+                .ty = AstStmt_Decl,
+                .loc = loc,
+                .decl = decl,
+            });
+    }
+
+    AstExpr* expr = parser_parse_expr(p);
+
+    return new_stmt(p,
+        (AstStmt) {
+            .ty = AstStmt_Expr,
+            .loc = loc,
+            .expr = expr,
+        });
+}
+
+static AstDeclar* parse_declar(Parser* p);
+static AstSpec parse_decl_spec(Parser* p);
+
+static AstStmtBlock* parse_stmt_block(Parser* p);
+
+AstDecl* parser_parse_decl(Parser* p)
+{
+    Loc loc = p->curr.loc;
+
+    AstSpec spec = parse_decl_spec(p);
+
+    ARRAY_VAR(AstDeclar*, declars);
+
+    AstDeclar* first = parse_declar(p);
+
+    array_push((void**)&declars,
+        &declars_cap,
+        &declars_size,
+        (&first),
+        sizeof(*(&first)));
+
+    if (first->ty == AstDeclar_Function && p->curr.ty == '{') {
+        AstStmtBlock* block = parse_stmt_block(p);
+        return new_decl(p,
+            (AstDecl) {
+                .spec = spec,
+                .declars = declars,
+                .declars_cap = declars_cap,
+                .declars_size = declars_size,
+                .function_body = block,
+                .loc = loc,
+            });
+    }
+
+    while (p->curr.ty == ',') {
+        step(p);
+
+        AstDeclar* declar = parse_declar(p);
+
+        array_push((void**)&declars,
+            &declars_cap,
+            &declars_size,
+            (&declar),
+            sizeof(*(&declar)));
+    }
+
+    return new_decl(p,
+        (AstDecl) {
+            .spec = spec,
+            .declars = declars,
+            .declars_cap = declars_cap,
+            .declars_size = declars_size,
+            .function_body = nullptr,
+            .loc = loc,
+        });
+}
+
+static AstDeclar* parse_declar_l1(Parser* p);
+static AstDeclar* parse_declar_l2(Parser* p);
+static AstDeclar* parse_declar_l3(Parser* p);
+
+static AstDeclar* parse_declar(Parser* p)
+{
+    AstDeclar* declar = parse_declar_l1(p);
+
+    if (p->curr.ty == '=') {
+        step(p);
+        AstExpr* expr = parser_parse_expr(p);
+        declar->init_expr = expr;
+    }
+
+    return declar;
+}
+
+static AstDeclar* parse_declar_l1(Parser* p)
+{
+    Loc loc = p->curr.loc;
+
+    if (p->curr.ty == '*') {
+        step(p);
+
+        AstSpec spec = parse_decl_spec(p);
+
+        AstDeclar* declar = parse_declar_l1(p);
+        return new_declar(p,
+            (AstDeclar) {
+                .ty = AstDeclar_Pointer,
+                .loc = loc,
+                .pointer = {
+                    .declar = declar,
+                    .spec = spec,
+                },
+            });
+    }
+
+    return parse_declar_l2(p);
+}
+
+static AstDeclar* parse_declar_l2(Parser* p)
+{
+    Loc loc = p->curr.loc;
+
+    AstDeclar* declar = parse_declar_l3(p);
+
+    while (p->curr.ty == '[') {
+        step(p);
+        AstExpr* expr = nullptr;
+        if (p->curr.ty != ']') {
+            expr = parser_parse_expr(p);
+        }
+        if (p->curr.ty != ']') {
+            fprintf(
+                stderr, "error: expected ']' on line %d\n", p->curr.loc.line);
+            exit(1);
+        }
+        step(p);
+        declar = new_declar(p,
+            (AstDeclar) {
+                .ty = AstDeclar_Array,
+                .loc = loc,
+                .array = {
+                    .declar = declar,
+                    .expr = expr,
+                },
+            });
+    }
+
+    return declar;
+}
+
+static AstDeclar* parse_declar_l3(Parser* p)
+{
+    Loc loc = p->curr.loc;
+
+    if (p->curr.ty == Tok_Ident) {
+        char* ident = arena_strdup(p->arena, p->curr.text);
+        step(p);
+        return new_declar(p,
+            (AstDeclar) {
+                .ty = AstDeclar_Array,
+                .loc = loc,
+                .ident = ident,
+            });
+    }
+
+    if (p->curr.ty == '(') {
+        step(p);
+
+        AstDeclar* declar = parse_declar_l1(p);
+
+        if (p->curr.ty != ')') {
+            fprintf(
+                stderr, "error: expected ')' on line %d\n", p->curr.loc.line);
+            exit(1);
+        }
+        step(p);
+
+        if (p->curr.ty != '(') {
+            return declar;
+        }
+        step(p);
+
+        ARRAY_VAR(AstDecl, params);
+
+        while (true) {
+            AstDecl* d = parser_parse_decl(p);
+            arena_array_push(p->arena,
+                (void**)&params,
+                &params_cap,
+                &params_size,
+                (&d),
+                sizeof(*(&d)));
+
+            if (p->curr.ty != ',')
+                break;
+            step(p);
+        }
+
+        if (p->curr.ty != '(') {
+            fprintf(
+                stderr, "error: expected '(' on line %d\n", p->curr.loc.line);
+            exit(1);
+        }
+        step(p);
+
+        return new_declar(p,
+            (AstDeclar) {
+                .ty = AstDeclar_Function,
+                .loc = loc,
+                .function = {
+                    .params = params,
+                    .params_cap = params_cap,
+                    .params_size = params_size,
+                },
+            });
+    }
+
+    fprintf(
+        stderr, "error: expected declarator on line %d\n", p->curr.loc.line);
+    exit(1);
+}
+
+static AstSpec parse_decl_spec(Parser* p)
+{
+    AstSpec spec = 0;
+
+    while (true) {
+        switch (p->curr.ty) {
+            case Tok_Ident: {
+                Sym* sym = syms_get(p->ord_ns, p->curr.text);
+                if (sym->ty != Sym_Typedef)
+                    break;
+                spec |= AstSpec_DefTy;
+                step(p);
+                continue;
+            }
+            case Tok_KwInt:
+                spec |= AstSpec_Int;
+                step(p);
+                continue;
+            case Tok_KwChar:
+                spec |= AstSpec_Char;
+                step(p);
+                continue;
+            case Tok_KwConst:
+                spec |= AstSpec_Const;
+                step(p);
+                continue;
+            default:
+        }
+
+        break;
+    }
+
+    return spec;
+}
+
+static AstStmtBlock* parse_stmt_block(Parser* p)
+{
+    Loc loc = p->curr.loc;
+}
+
+AstExpr* parser_parse_expr(Parser* p)
+{
+}

parse.h

@@ -1,7 +1,9 @@
 #ifndef PARSE_H
 #define PARSE_H
 
+#include "ast.h"
 #include "collection.h"
+#include "syms.h"
 #include <stddef.h>
 
 char* read_textfile(const char* filename);
@@ -13,8 +15,7 @@ char* read_textfile(const char* filename);
 
 typedef struct {
     char ch;
-    int line;
+    Loc loc;
-    int col;
 } Char;
 
 typedef struct {
@@ -70,8 +71,7 @@ const char* tokty_str(TokTy ty);
 typedef struct {
     TokTy ty;
     char* text;
-    int line;
+    Loc loc;
-    int col;
 } Tok;
 
 typedef struct {
@@ -108,5 +108,8 @@ typedef struct Parser Parser;
 
 Parser* parser_new(TokStream* tokstream, Arena* arena);
 void parser_free(Parser* p);
+AstStmt* parser_parse_stmt(Parser* p);
+AstDecl* parser_parse_decl(Parser* p);
+AstExpr* parser_parse_expr(Parser* p);
 
 #endif

preproc.c

@@ -8,19 +8,19 @@
 
 typedef struct {
     ARRAY_FIELD(Hashmap, maps);
-} Syms;
+} PPSyms;
 
-static void syms_push_layer(Syms* s);
+static void ppsyms_push_layer(PPSyms* s);
 
-static void syms_init(Syms* s)
+static void ppsyms_init(PPSyms* s)
 {
-    *s = (Syms) {
+    *s = (PPSyms) {
         ARRAY_FIELD_INIT(maps),
     };
-    syms_push_layer(s);
+    ppsyms_push_layer(s);
 }
 
-static void syms_deinit(Syms* s)
+static void ppsyms_deinit(PPSyms* s)
 {
     if (s->maps) {
         for (size_t i = 0; i < s->maps_size; ++i) {
@@ -30,14 +30,14 @@ static void syms_deinit(Syms* s)
     }
 }
 
-static void syms_push_layer(Syms* s)
+static void ppsyms_push_layer(PPSyms* s)
 {
     Hashmap m = { };
     hashmap_init(&m, sizeof(size_t));
     array_push((void**)&s->maps, &s->maps_cap, &s->maps_size, &m, sizeof(m));
 }
 
-static void syms_pop_layer(Syms* s)
+static void ppsyms_pop_layer(PPSyms* s)
 {
     assert(s->maps_size > 0);
 
@@ -45,19 +45,19 @@ static void syms_pop_layer(Syms* s)
     s->maps_size -= 1;
 }
 
-static Hashmap* syms_current(Syms* s)
+static Hashmap* ppsyms_current(PPSyms* s)
 {
     assert(s->maps_size > 0);
     return &s->maps[s->maps_size - 1];
 }
 
-static void syms_define(Syms* s, const char* ident, size_t def_idx)
+static void ppsyms_define(PPSyms* s, const char* ident, size_t def_idx)
 {
-    Hashmap* m = syms_current(s);
+    Hashmap* m = ppsyms_current(s);
     hashmap_set(m, ident, &def_idx);
 }
 
-static size_t* syms_get(Syms* s, const char* ident)
+static size_t* ppsyms_get(PPSyms* s, const char* ident)
 {
     for (size_t i = 0; i < s->maps_size; ++i) {
         Hashmap* m = &s->maps[s->maps_size - i - 1];
@@ -114,7 +114,7 @@ struct Preproc {
     Tok curr;
 
     ARRAY_FIELD(Def, defs);
-    Syms syms;
+    PPSyms syms;
 
     CondStack condstack;
 
@@ -159,7 +159,7 @@ Preproc* preproc_new(char* filename, char* text, Arena* arena)
     // null-definition at idx 0
     ARRAY_FIELD_PUSH(p, defs, &(Def) { });
 
-    syms_init(&p->syms);
+    ppsyms_init(&p->syms);
 
     return p;
 }
@@ -180,7 +180,7 @@ void preproc_free(Preproc* p)
         }
         free(p->defs);
     }
-    syms_deinit(&p->syms);
+    ppsyms_deinit(&p->syms);
     free(p);
 }
 
@@ -206,7 +206,7 @@ static void step(Preproc* p)
     p->tok_stack_size -= 1;
 
     if (p->curr.ty == Tok_EndOfMacroInvocation) {
-        syms_pop_layer(&p->syms);
+        ppsyms_pop_layer(&p->syms);
         step(p);
         return;
     }
@@ -217,7 +217,7 @@ static void collect_arg(Preproc* p, size_t def_idx, size_t arg_idx)
     if (arg_idx >= p->defs[def_idx].params_size) {
         fprintf(stderr,
             "error: too many arguments for macro on line %d\n",
-            p->curr.line);
+            p->curr.loc.line);
         exit(1);
     }
 
@@ -248,8 +248,9 @@ static void collect_arg(Preproc* p, size_t def_idx, size_t arg_idx)
 
     if (p->curr.ty != ',' && p->curr.ty != ')') {
         assert(p->curr.ty == Tok_Eof);
-        fprintf(
+        fprintf(stderr,
-            stderr, "error: expected ',' or ')' on line %d\n", p->curr.line);
+            "error: expected ',' or ')' on line %d\n",
+            p->curr.loc.line);
         exit(1);
     }
 
@@ -260,7 +261,7 @@ static void collect_arg(Preproc* p, size_t def_idx, size_t arg_idx)
             .toks_size = toks_size,
         }));
 
-    syms_define(&p->syms, p->defs[def_idx].params[arg_idx], p->defs_size - 1);
+    ppsyms_define(&p->syms, p->defs[def_idx].params[arg_idx], p->defs_size - 1);
 }
 
 #define A(B, C) B C
@@ -282,7 +283,7 @@ static void collect_args(Preproc* p, size_t def_idx)
         }
     }
     if (p->curr.ty != ')') {
-        fprintf(stderr, "error: expected ')' on line %d\n", p->curr.line);
+        fprintf(stderr, "error: expected ')' on line %d\n", p->curr.loc.line);
         exit(1);
     }
 
@@ -290,7 +291,7 @@ static void collect_args(Preproc* p, size_t def_idx)
     if (idx > p->defs[def_idx].params_size) {
         fprintf(stderr,
             "error: too few arguments for macro on line %d\n",
-            p->curr.line);
+            p->curr.loc.line);
         exit(1);
     }
 }
@@ -299,7 +300,7 @@ static Tok expand_tok(Preproc* p)
 {
     Tok tok = p->curr;
     size_t* def_idx;
-    if (tok.ty != Tok_Ident || !(def_idx = syms_get(&p->syms, tok.text))) {
+    if (tok.ty != Tok_Ident || !(def_idx = ppsyms_get(&p->syms, tok.text))) {
         step(p);
         return tok;
     }
@@ -308,8 +309,8 @@ static Tok expand_tok(Preproc* p)
     // when seeing the same macro a second time
     // in the invocation tree, treat it as a normal token.
     // undefine it at this layer.
-    syms_push_layer(&p->syms);
+    ppsyms_push_layer(&p->syms);
-    syms_define(&p->syms, tok.text, 0);
+    ppsyms_define(&p->syms, tok.text, 0);
 
     if (p->defs[*def_idx].params) {
         step(p);
@@ -323,8 +324,7 @@ static Tok expand_tok(Preproc* p)
     for (size_t i = 0; i < p->defs[*def_idx].toks_size; ++i) {
         Def* def = &p->defs[*def_idx];
         Tok expanded_tok = def->toks[def->toks_size - i - 1];
-        expanded_tok.line = tok.line;
+        expanded_tok.loc = tok.loc;
-        expanded_tok.col = tok.col;
         ARRAY_FIELD_PUSH(p, tok_stack, &expanded_tok);
     }
 
@@ -338,8 +338,9 @@ static Tok do_define(Preproc* p)
 {
     step(p);
     if (p->curr.ty != Tok_Ident) {
-        fprintf(
+        fprintf(stderr,
-            stderr, "error: expected identifier on line %d\n", p->curr.line);
+            "error: expected identifier on line %d\n",
+            p->curr.loc.line);
         exit(1);
     }
     Tok ident_tok = p->curr;
@@ -374,7 +375,8 @@ static Tok do_define(Preproc* p)
         }
 
         if (p->curr.ty != ')') {
-            fprintf(stderr, "error: expected ')' on line %d\n", p->curr.line);
+            fprintf(
+                stderr, "error: expected ')' on line %d\n", p->curr.loc.line);
             exit(1);
         }
         step(p);
@@ -402,12 +404,13 @@ static Tok do_define(Preproc* p)
     def.toks_size = toks_size;
 
     ARRAY_FIELD_PUSH(p, defs, &def);
-    syms_define(&p->syms, ident_tok.text, p->defs_size - 1);
+    ppsyms_define(&p->syms, ident_tok.text, p->defs_size - 1);
 
     return p->curr;
 
 error_expected_identifier:
-    fprintf(stderr, "error: expected identifier on line %d\n", p->curr.line);
+    fprintf(
+        stderr, "error: expected identifier on line %d\n", p->curr.loc.line);
     exit(1);
 }
 
@@ -416,8 +419,9 @@ static Tok do_undef(Preproc* p)
     step(p);
 
     if (p->curr.ty != Tok_Ident) {
-        fprintf(
+        fprintf(stderr,
-            stderr, "error: expected identifier on line %d\n", p->curr.line);
+            "error: expected identifier on line %d\n",
+            p->curr.loc.line);
         exit(1);
     }
     Tok ident_tok = p->curr;
@@ -425,11 +429,11 @@ static Tok do_undef(Preproc* p)
     if (p->curr.ty != Tok_Eof && p->curr.ty != Tok_Eol) {
         fprintf(stderr,
             "error: unexpected extra tokens on line %d\n",
-            p->curr.line);
+            p->curr.loc.line);
         exit(1);
     }
 
-    syms_define(&p->syms, ident_tok.text, 0);
+    ppsyms_define(&p->syms, ident_tok.text, 0);
 
     return p->curr;
 }
@@ -445,8 +449,9 @@ static Tok do_include(Preproc* p)
     curr_file->lexer.lex_headername = false;
 
     if (p->curr.ty != Tok_HeaderName) {
-        fprintf(
+        fprintf(stderr,
-            stderr, "error: expected header name on line %d\n", p->curr.line);
+            "error: expected header name on line %d\n",
+            p->curr.loc.line);
         exit(1);
     }
 
@@ -456,7 +461,7 @@ static Tok do_include(Preproc* p)
     if (p->curr.ty != Tok_Eof && p->curr.ty != Tok_Eol) {
         fprintf(stderr,
             "error: unexpected extra tokens on line %d\n",
-            p->curr.line);
+            p->curr.loc.line);
         exit(1);
     }
 
@@ -506,12 +511,13 @@ static Tok interpret_directive(Preproc* p);
 
 static Tok do_ifndef(Preproc* p)
 {
-    int original_line = p->curr.line;
+    int original_line = p->curr.loc.line;
 
     step(p);
     if (p->curr.ty != Tok_Ident) {
-        fprintf(
+        fprintf(stderr,
-            stderr, "error: expected identifier on line %d\n", p->curr.line);
+            "error: expected identifier on line %d\n",
+            p->curr.loc.line);
         exit(1);
     }
     Tok ident = p->curr;
@@ -520,11 +526,11 @@ static Tok do_ifndef(Preproc* p)
     if (p->curr.ty != Tok_Eof && p->curr.ty != Tok_Eol) {
         fprintf(stderr,
             "error: unexpected extra tokens on line %d\n",
-            p->curr.line);
+            p->curr.loc.line);
         exit(1);
     }
 
-    bool cond = syms_get(&p->syms, ident.text) == nullptr;
+    bool cond = ppsyms_get(&p->syms, ident.text) == nullptr;
 
     condstack_push(&p->condstack, cond);
 
@@ -561,7 +567,7 @@ static Tok do_endif(Preproc* p)
     if (p->curr.ty != Tok_Eof && p->curr.ty != Tok_Eol) {
         fprintf(stderr,
             "error: unexpected extra tokens on line %d\n",
-            p->curr.line);
+            p->curr.loc.line);
         exit(1);
     }
 
@@ -576,8 +582,9 @@ static Tok interpret_directive(Preproc* p)
         return p->curr;
 
     if (p->curr.ty != Tok_Ident) {
-        fprintf(
+        fprintf(stderr,
-            stderr, "error: expected identifier on line %d\n", p->curr.line);
+            "error: expected identifier on line %d\n",
+            p->curr.loc.line);
         exit(1);
     }
 
@@ -617,7 +624,7 @@ static Tok interpret_directive(Preproc* p)
     fprintf(stderr,
         "error: '%s' is not a preprocessor instruction on line %d\n",
         kw.text,
-        kw.line);
+        kw.loc.line);
     exit(1);
 }
 
@@ -659,7 +666,7 @@ Tok tokstream_next(TokStream* s)
         text[0] = '\0';
         strncat(text, curr.text, strlen(curr.text) - 1);
         strcat(text, next.text + 1);
-        s->curr = (Tok) { .ty = Tok_Str, text, curr.line, curr.col };
+        s->curr = (Tok) { .ty = Tok_Str, text, curr.loc };
         return tokstream_next(s);
     }
 

syms.c

@@ -0,0 +1,52 @@
+#include "syms.h"
+#include "collection.h"
+#include <assert.h>
+#include <stdlib.h>
+
+struct Syms {
+    ARRAY_FIELD(Hashmap, maps);
+};
+
+Syms* syms_new()
+{
+    Syms* s = malloc(sizeof(Syms));
+    *s = (Syms) {
+        ARRAY_FIELD_INIT(maps),
+    };
+
+    ARRAY_FIELD_PUSH(s, maps, &(Hashmap) { });
+    hashmap_init(&s->maps[0], sizeof(Sym*));
+
+    return s;
+}
+
+void syms_free(Syms* s)
+{
+    free(s);
+}
+
+void syms_push_scope(Syms* s)
+{
+    ARRAY_FIELD_PUSH(s, maps, &(Hashmap) { });
+    hashmap_init(&s->maps[s->maps_size - 1], sizeof(Sym*));
+}
+
+void syms_pop_scope(Syms* s)
+{
+    assert(s->maps_size > 0);
+    hashmap_deinit(&s->maps[s->maps_size - 1]);
+    s->maps_size -= 1;
+}
+
+void syms_define(Syms* s, const char* ident, Sym* sym)
+{
+    Hashmap* m = &s->maps[s->maps_size - 1];
+    hashmap_set(m, ident, &sym);
+}
+
+Sym* syms_get(Syms* s, const char* ident)
+{
+    Hashmap* m = &s->maps[s->maps_size - 1];
+    Sym** res = hashmap_get(m, ident);
+    return res ? *res : nullptr;
+}

syms.h

@@ -0,0 +1,23 @@
+#ifndef SYMS_H
+#define SYMS_H
+
+typedef enum {
+    Sym_Typedef,
+} SymTy;
+
+typedef struct {
+    SymTy ty;
+} Sym;
+
+typedef struct Syms Syms;
+
+Syms* syms_new();
+void syms_free(Syms* s);
+
+void syms_push_scope(Syms* s);
+void syms_pop_scope(Syms* s);
+
+void syms_define(Syms* s, const char* ident, Sym* sym);
+Sym* syms_get(Syms* s, const char* ident);
+
+#endif