work on preproc

collection.c

@@ -4,24 +4,55 @@
 #include <stdlib.h>
 #include <string.h>
 
-void array_push(
+static void array_push_generic(void** data,
-    void** data, size_t* capacity, size_t* size, void* elem, size_t elem_size)
+    size_t* capacity,
+    size_t* size,
+    void* elem,
+    size_t elem_size,
+    void* alloc_fn(void*, size_t),
+    void* realloc_fn(void*, void*, size_t),
+    void* allocator)
 {
     if (!*capacity) {
         *capacity = 4;
     }
     if (!*data) {
-        *data = malloc(*capacity * elem_size);
+        *data = alloc_fn(allocator, *capacity * elem_size);
         *size = 0;
     }
     if (*size >= *capacity) {
         *capacity *= 2;
-        *data = realloc(*data, *capacity * elem_size);
+        *data = realloc_fn(allocator, *data, *capacity * elem_size);
     }
     memcpy((unsigned char*)*data + *size * elem_size, elem, elem_size);
     *size += 1;
 }
 
+static void* malloc_wrap(void* self, size_t size)
+{
+    (void)self;
+    return malloc(size);
+}
+
+static void* realloc_wrap(void* self, void* p, size_t size)
+{
+    (void)self;
+    return realloc(p, size);
+}
+
+void array_push(
+    void** data, size_t* capacity, size_t* size, void* elem, size_t elem_size)
+{
+    array_push_generic(data,
+        capacity,
+        size,
+        elem,
+        elem_size,
+        malloc_wrap,
+        realloc_wrap,
+        nullptr);
+}
+
 static void push_kwpair(Hashmap* m, KwPair* pair)
 {
     array_push((void**)&m->entries,
@@ -134,3 +165,34 @@ void* arena_alloc(Arena* a, size_t size)
 
     return ptr;
 }
+
+static void* arena_alloc_wrap(void* self, size_t size)
+{
+    Arena* a = self;
+    return arena_alloc(a, size);
+}
+
+static void* arena_realloc_wrap(void* self, void* p, size_t size)
+{
+    Arena* a = self;
+    void* new_buf = arena_alloc(a, size);
+    memcpy(new_buf, p, size);
+    return new_buf;
+}
+
+void arena_array_push(Arena* arena,
+    void** data,
+    size_t* capacity,
+    size_t* size,
+    void* elem,
+    size_t elem_size)
+{
+    array_push_generic(data,
+        capacity,
+        size,
+        elem,
+        elem_size,
+        arena_alloc_wrap,
+        arena_realloc_wrap,
+        arena);
+}

collection.h

@@ -40,4 +40,11 @@ void arena_init(Arena* a, size_t bucket_cap);
 void arena_deinit(Arena* a);
 void* arena_alloc(Arena* a, size_t size);
 
+void arena_array_push(Arena* arena,
+    void** data,
+    size_t* capacity,
+    size_t* size,
+    void* elem,
+    size_t elem_size);
+
 #endif

examples/hello.c

@@ -1,8 +1,8 @@
 int puts(const char* __s);
 
-#define PUTS puts
+#define PUTS(A, B) puts(A, B)
 
 int main()
 {
-    PUTS("hello");
+    PUTS("hello", "world");
 }

lex.c

@@ -232,7 +232,7 @@ Tok lexer_next(Lexer* l)
         return tok(l, Tok_Char, text, line, col);
     }
 
-    if (strchr("\n;(){}*#", l->curr.ch)) {
+    if (strchr("\n;(){},*#", l->curr.ch)) {
         TokTy ty = (TokTy)l->curr.ch;
         step(l, text, &text_len);
         return tok(l, ty, text, line, col);

main.c

@@ -57,7 +57,7 @@ int main(int argc, char* argv[])
         return EXIT_FAILURE;
     }
 
-    printf("%s\n", text);
+    printf("text:\n%s\n", text);
 
     SourceIter source_chars;
     source_iter_init(&source_chars, text);
@@ -74,10 +74,14 @@ int main(int argc, char* argv[])
     Preproc preproc;
     preproc_init(&preproc, &lexer);
 
+    printf("tokens:\n");
+
     Tok tok;
-    while ((tok = preproc_next(&preproc)).ty != Tok_Eof) {
+    while ((tok = preproc_next_with_newline(&preproc)).ty != Tok_Eof) {
-        printf("%s\t%s\n", tokty_str(tok.ty), tok.text);
+        // printf("%s\t%s\n", tokty_str(tok.ty), tok.text);
+        printf("%s ", tok.text);
     }
+    fputs("\n", stdout);
 
     preproc_deinit(&preproc);
     free(text);

parse.h

@@ -51,6 +51,7 @@ typedef enum {
     Tok_RParen = ')',
     Tok_LBrace = '{',
     Tok_RBrace = '}',
+    Tok_Comma = ',',
     Tok_Asterisk = '*',
     Tok_Hashtag = '#',
 } TokTy;
@@ -77,23 +78,27 @@ Tok lexer_next(Lexer* l);
 
 // phase 4 - preprocessor
 
-typedef struct DefToks DefToks;
+typedef struct MacroDef MacroDef;
+typedef struct MacroInv MacroInv;
 
 typedef struct {
     Lexer* lexer;
     Tok curr;
 
-    DefToks* deftoks;
+    MacroDef* defs;
-    size_t deftoks_cap;
+    size_t defs_cap;
-    size_t deftoks_size;
+    size_t defs_size;
-    Hashmap defs;
+    Hashmap def_map;
 
-    DefToks* pasting_toks;
+    Arena arg_arena;
-    size_t pasting_toks_count;
+    MacroInv* macro_invs;
+    size_t macro_invs_cap;
+    size_t macro_invs_size;
 } Preproc;
 
 void preproc_init(Preproc* p, Lexer* lexer);
 void preproc_deinit(Preproc* p);
+Tok preproc_next_with_newline(Preproc* p);
 Tok preproc_next(Preproc* p);
 
 #endif

preproc.c

@@ -6,12 +6,26 @@
 #include <stdlib.h>
 #include <string.h>
 
-struct DefToks {
+struct MacroDef {
     Tok* toks;
     size_t toks_cap;
     size_t toks_size;
+
+    char** params;
+    size_t params_size;
 };
 
+struct MacroInv {
+    MacroDef* def;
+    size_t tok_idx;
+    Hashmap arg_map;
+};
+
+typedef struct {
+    Tok* toks;
+    size_t toks_size;
+} MacroArg;
+
 void preproc_init(Preproc* p, Lexer* lexer)
 {
     Tok curr = lexer_next(lexer);
@@ -19,26 +33,35 @@ void preproc_init(Preproc* p, Lexer* lexer)
     *p = (Preproc) {
         .lexer = lexer,
         .curr = curr,
-        .deftoks = nullptr,
+
-        .deftoks_cap = 0,
+        .defs = nullptr,
-        .deftoks_size = 0,
+        .defs_cap = 0,
-        .defs = { },
+        .defs_size = 0,
-        .pasting_toks = nullptr,
+        .def_map = { },
-        .pasting_toks_count = 0,
+
+        .arg_arena = { },
+        .macro_invs = nullptr,
+        .macro_invs_cap = 0,
+        .macro_invs_size = 0,
     };
 
-    hashmap_init(&p->defs);
+    hashmap_init(&p->def_map);
+    arena_init(&p->arg_arena, 512);
 }
 
 void preproc_deinit(Preproc* p)
 {
-    if (p->deftoks) {
+    if (p->defs) {
-        for (size_t i = 0; i < p->deftoks_size; ++i) {
+        for (size_t i = 0; i < p->defs_size; ++i) {
-            free(p->deftoks[i].toks);
+            free(p->defs[i].toks);
+            if (p->defs[i].params) {
+                free(p->defs[i].params);
+            }
         }
-        free(p->deftoks);
+        free(p->defs);
     }
-    hashmap_deinit(&p->defs);
+    hashmap_deinit(&p->def_map);
+    arena_deinit(&p->arg_arena);
 }
 
 static void step(Preproc* p)
@@ -60,30 +83,69 @@ static Tok p_define(Preproc* p)
     step(p);
     p->lexer->lex_whitespace = false;
 
+    MacroDef def = { };
+
     if (p->curr.ty == '(') {
-        // function-like macro
+        step(p);
-        assert(false && "not implemented");
+
+        char** params = nullptr;
+        size_t params_cap = 0;
+        size_t params_size = 0;
+
+        if (p->curr.ty != Tok_Eof && p->curr.ty != ')') {
+            if (p->curr.ty != Tok_Ident) {
+                fprintf(stderr,
+                    "error: expected identifier on line %d\n",
+                    p->curr.line);
+                exit(1);
+            }
+            array_push((void**)&params,
+                &params_cap,
+                &params_size,
+                p->curr.text,
+                sizeof(p->curr.text));
+
+            while (p->curr.ty == ',') {
+                step(p);
+
+                if (p->curr.ty != Tok_Ident) {
+                    fprintf(stderr,
+                        "error: expected identifier on line %d\n",
+                        p->curr.line);
+                    exit(1);
+                }
+                array_push((void**)&params,
+                    &params_cap,
+                    &params_size,
+                    p->curr.text,
+                    sizeof(p->curr.text));
+            }
+        }
+
+        if (p->curr.ty == Tok_Eof || p->curr.ty == ')') {
+            fprintf(stderr, "error: expected ')' on line %d\n", p->curr.line);
+            exit(1);
+        }
+        step(p);
+
+        def.params = params;
+        def.params_size = params_size;
     }
 
-    DefToks toks = { };
-
     while (p->curr.ty != Tok_Eof && p->curr.ty != Tok_Eol) {
-        array_push((void**)&toks.toks,
+        array_push((void**)&def.toks,
-            &toks.toks_cap,
+            &def.toks_cap,
-            &toks.toks_size,
+            &def.toks_size,
             &p->curr,
             sizeof(p->curr));
         step(p);
     }
 
-    array_push((void**)&p->deftoks,
+    array_push(
-        &p->deftoks_cap,
+        (void**)&p->defs, &p->defs_cap, &p->defs_size, &def, sizeof(def));
-        &p->deftoks_size,
-        &toks,
-        sizeof(toks));
 
-    DefToks* deftoks = &p->deftoks[p->deftoks_size - 1];
+    MacroDef* deftoks = &p->defs[p->defs_size - 1];
-    hashmap_set(&p->defs, ident_tok.text, deftoks);
+    hashmap_set(&p->def_map, ident_tok.text, deftoks);
 
     return p->curr;
 }
@@ -106,7 +168,7 @@ static Tok p_undef(Preproc* p)
         exit(1);
     }
 
-    hashmap_set(&p->defs, ident_tok.text, nullptr);
+    hashmap_set(&p->def_map, ident_tok.text, nullptr);
 
     return p->curr;
 }
@@ -125,83 +187,175 @@ static Tok p_include(Preproc* p)
     assert(false && "not implemented");
 }
 
-static Tok preproc_next_nl(Preproc* p)
+static Tok p_paste_macro_body(Preproc* p)
 {
-    if (p->pasting_toks) {
+    MacroInv* inv = &p->macro_invs[p->macro_invs_size - 1];
-        if (p->pasting_toks_count >= p->pasting_toks->toks_size) {
-            p->pasting_toks = nullptr;
-            p->pasting_toks_count = 0;
-            return preproc_next_nl(p);
-        }
 
-        Tok tok = p->pasting_toks->toks[p->pasting_toks_count];
+    if (inv->tok_idx >= inv->def->toks_size) {
-        p->pasting_toks_count += 1;
+        p->macro_invs_size -= 1;
-        return tok;
+        return preproc_next_with_newline(p);
     }
 
-    Tok curr = p->curr;
+    Tok tok = inv->def->toks[inv->tok_idx];
+    inv->tok_idx += 1;
+
+    return tok;
+}
+
+static void collect_arg_toks(Tok** toks, size_t* toks_size)
+{
+}
+
+struct arg_def {
+    Tok* toks;
+    size_t toks_size;
+};
+
+static Tok p_detect_macro(Preproc* p, Tok head)
+{
+    MacroDef* def = hashmap_get(&p->def_map, head.text);
+    if (!def)
+        return head;
+
+    MacroInv inv = {
+        .def = def,
+        .tok_idx = 0,
+        .arg_map = { },
+    };
+    hashmap_init(&inv.arg_map);
+
+    array_push(
+        (void**)&p->macro_invs, &p->defs_cap, &p->defs_size, &inv, sizeof(inv));
+
+    if (!def->params)
+        return preproc_next_with_newline(p);
+
+    if (p->curr.ty != '(')
+        return head;
+
     step(p);
 
-    if (curr.ty == Tok_Ident) {
+    struct arg_def* arg_defs;
-        DefToks* def = hashmap_get(&p->defs, curr.text);
+    size_t arg_defs_cap = 0;
-        if (!def)
+    size_t arg_defs_size = 0;
-            return curr;
 
-        step(p);
+    size_t arg_idx = 0;
-        p->pasting_toks = def;
-        p->pasting_toks_count = 0;
-        return preproc_next_nl(p);
-    }
 
-    if (curr.ty == '#') {
+    bool first = true;
-        Tok kw = p->curr;
 
-        if (strcmp(kw.text, "define") == 0) {
+    while ((first && (p->curr.ty != Tok_Eof && p->curr.ty != ')'))
-            return p_define(p);
+        || p->curr.ty != ',') {
-        } else if (strcmp(kw.text, "undef") == 0) {
+        first = false;
-            return p_undef(p);
+
-        } else if (strcmp(kw.text, "include") == 0) {
+        if (arg_idx + 1 >= def->params_size) {
-            return p_include(p);
-        } else if (strcmp(kw.text, "if") == 0) {
-            assert(false && "not implemented");
-        } else if (strcmp(kw.text, "ifdef") == 0) {
-            assert(false && "not implemented");
-        } else if (strcmp(kw.text, "ifndef") == 0) {
-            assert(false && "not implemented");
-        } else if (strcmp(kw.text, "else") == 0) {
-            assert(false && "not implemented");
-        } else if (strcmp(kw.text, "elif") == 0) {
-            assert(false && "not implemented");
-        } else if (strcmp(kw.text, "elifdef") == 0) {
-            assert(false && "not implemented");
-        } else if (strcmp(kw.text, "elifndef") == 0) {
-            assert(false && "not implemented");
-        } else if (strcmp(kw.text, "endif") == 0) {
-            assert(false && "not implemented");
-        } else if (strcmp(kw.text, "line") == 0) {
-            assert(false && "not implemented");
-        } else if (strcmp(kw.text, "embed") == 0) {
-            assert(false && "not implemented");
-        } else if (strcmp(kw.text, "error") == 0) {
-            assert(false && "not implemented");
-        } else if (strcmp(kw.text, "warning") == 0) {
-            assert(false && "not implemented");
-        } else if (strcmp(kw.text, "pragma") == 0) {
             fprintf(stderr,
-                "error: '%s' is not a preprocessor instruction on line %d\n",
+                "error: too many arguments for macro line %d\n",
-                kw.text,
+                p->curr.line);
-                kw.line);
             exit(1);
         }
+
+        Tok* toks;
+        size_t toks_size;
+        collect_arg_toks(&toks, &toks_size);
+
+        arena_array_push(&p->arg_arena,
+            (void**)&arg_defs,
+            &arg_defs_cap,
+            &arg_defs_size,
+            &(struct arg_def) { toks, toks_size },
+            sizeof(struct arg_def));
+        hashmap_set(
+            &inv.arg_map, def->params[arg_idx], &arg_defs[arg_defs_size - 1]);
+
+        arg_idx += 1;
     }
 
-    return curr;
+    return preproc_next_with_newline(p);
+}
+
+static Tok p_interpret_directive(Preproc* p)
+{
+    Tok kw = p->curr;
+
+    if (strcmp(kw.text, "define") == 0) {
+        return p_define(p);
+    } else if (strcmp(kw.text, "undef") == 0) {
+        return p_undef(p);
+    } else if (strcmp(kw.text, "include") == 0) {
+        return p_include(p);
+    } else if (strcmp(kw.text, "if") == 0) {
+        assert(false && "not implemented");
+    } else if (strcmp(kw.text, "ifdef") == 0) {
+        assert(false && "not implemented");
+    } else if (strcmp(kw.text, "ifndef") == 0) {
+        assert(false && "not implemented");
+    } else if (strcmp(kw.text, "else") == 0) {
+        assert(false && "not implemented");
+    } else if (strcmp(kw.text, "elif") == 0) {
+        assert(false && "not implemented");
+    } else if (strcmp(kw.text, "elifdef") == 0) {
+        assert(false && "not implemented");
+    } else if (strcmp(kw.text, "elifndef") == 0) {
+        assert(false && "not implemented");
+    } else if (strcmp(kw.text, "endif") == 0) {
+        assert(false && "not implemented");
+    } else if (strcmp(kw.text, "line") == 0) {
+        assert(false && "not implemented");
+    } else if (strcmp(kw.text, "embed") == 0) {
+        assert(false && "not implemented");
+    } else if (strcmp(kw.text, "error") == 0) {
+        assert(false && "not implemented");
+    } else if (strcmp(kw.text, "warning") == 0) {
+        assert(false && "not implemented");
+    } else if (strcmp(kw.text, "pragma") == 0) {
+        fprintf(stderr,
+            "error: '%s' is not a preprocessor instruction on line %d\n",
+            kw.text,
+            kw.line);
+        exit(1);
+    }
+
+    assert(false && "unreachable");
+}
+
+static Tok replace_candidate_tok(Preproc* p, Tok tok, size_t inv_idx)
+{
+    if (inv_idx == 0)
+        return tok;
+    MacroInv* inv = &p->macro_invs[inv_idx];
+    if (tok.ty != Tok_Ident)
+        return tok;
+    MacroArg* arg = hashmap_get(&inv->arg_map, tok.text);
+    if (!arg)
+        return replace_candidate_tok(p, tok, inv_idx - 1);
+}
+
+static Tok expand_tok(Preproc* p, Tok tok)
+{
+    if (p->macro_invs_size == 0)
+        return tok;
+
+    MacroInv* inv = &p->macro_invs[p->macro_invs_size - 1];
+
+    if (inv->tok_idx >= inv->def->toks_size) {
+        p->macro_invs_size -= 1;
+        return expand_tok(p, tok);
+    }
+}
+
+Tok preproc_next_with_newline(Preproc* p)
+{
+    Tok tok = p->curr;
+    step(p);
+    if (tok.ty == '#')
+        return p_interpret_directive(p);
+    return expand_tok(p, tok);
 }
 
 Tok preproc_next(Preproc* p)
 {
     Tok tok;
-    while ((tok = preproc_next_nl(p)).ty == Tok_Eol
+    while ((tok = preproc_next_with_newline(p)).ty == Tok_Eol
         || tok.ty == Tok_Whitespace) { }
     return tok;
 }