optimize symbol->string

extlib/benz/eval.c

@@ -117,7 +117,7 @@ analyzer_scope_destroy(pic_state *pic, analyze_scope *scope)
 }
 
 static bool
-search_scope(analyze_scope *scope, pic_sym *sym)
+search_scope(pic_state *pic, analyze_scope *scope, pic_sym *sym)
 {
   return kh_get(a, &scope->args, sym) != kh_end(&scope->args) || kh_get(a, &scope->locals, sym) != kh_end(&scope->locals) || scope->depth == 0;
 }
@@ -128,7 +128,7 @@ find_var(pic_state *pic, analyze_scope *scope, pic_sym *sym)
   int depth = 0, ret;
 
   while (scope) {
-    if (search_scope(scope, sym)) {
+    if (search_scope(pic, scope, sym)) {
       if (depth > 0) {
         kh_put(a, &scope->captures, sym, &ret); /* capture! */
       }
@@ -145,7 +145,7 @@ define_var(pic_state *pic, analyze_scope *scope, pic_sym *sym)
 {
   int ret;
 
-  if (search_scope(scope, sym)) {
+  if (search_scope(pic, scope, sym)) {
     if (scope->depth > 0 || pic_reg_has(pic, pic->globals, sym)) {
       pic_warnf(pic, "redefining variable: ~s", pic_obj_value(sym));
     }

extlib/benz/gc.c

@@ -384,6 +384,7 @@ gc_mark_object(pic_state *pic, struct pic_object *obj)
     break;
   }
   case PIC_TT_SYMBOL: {
+    LOOP(obj->u.sym.str);
     break;
   }
   case PIC_TT_REG: {
@@ -559,7 +560,7 @@ gc_finalize_object(pic_state *pic, struct pic_object *obj)
     break;
   }
   case PIC_TT_SYMBOL: {
-    pic_free(pic, (void *)obj->u.sym.cstr);
+    /* TODO: remove this symbol's entry from pic->syms immediately */
     break;
   }
   case PIC_TT_REG: {

extlib/benz/include/picrin.h

@@ -45,7 +45,7 @@ typedef struct pic_state pic_state;
 #include "picrin/read.h"
 #include "picrin/gc.h"
 
-KHASH_DECLARE(s, const char *, pic_sym *)
+KHASH_DECLARE(s, pic_str *, pic_sym *)
 
 typedef struct pic_checkpoint {
   PIC_OBJECT_HEADER

extlib/benz/include/picrin/khash.h

@@ -27,6 +27,8 @@
 #ifndef AC_KHASH_H
 #define AC_KHASH_H
 
+#include <stddef.h>
+
 typedef int khint_t;
 typedef khint_t khiter_t;
 
@@ -41,23 +43,6 @@ typedef khint_t khiter_t;
 #define ac_roundup32(x)                                                 \
   (--(x), (x)|=(x)>>1, (x)|=(x)>>2, (x)|=(x)>>4, (x)|=(x)>>8, (x)|=(x)>>16, ++(x))
 
-PIC_INLINE khint_t ac_X31_hash_string(const char *s)
-{
-  khint_t h = (khint_t)*s;
-  if (h) for (++s ; *s; ++s) h = (h << 5) - h + (khint_t)*s;
-  return h;
-}
-PIC_INLINE khint_t ac_Wang_hash(khint_t key)
-{
-  key += ~(key << 15);
-  key ^=  (key >> 10);
-  key +=  (key << 3);
-  key ^=  (key >> 6);
-  key += ~(key << 11);
-  key ^=  (key >> 16);
-  return key;
-}
-
 #define ac_fsize(m) ((m) < 16? 1 : (m)>>4)
 #define ac_hash_upper(x) ((((x) * 2) * 77 / 100 + 1) / 2)
 
@@ -71,7 +56,7 @@ PIC_INLINE khint_t ac_Wang_hash(khint_t key)
   void kh_init_##name(kh_##name##_t *h);                                \
   void kh_destroy_##name(pic_state *, kh_##name##_t *h);                 \
   void kh_clear_##name(kh_##name##_t *h);                               \
-  khint_t kh_get_##name(const kh_##name##_t *h, khkey_t key);           \
+  khint_t kh_get_##name(pic_state *, const kh_##name##_t *h, khkey_t key); \
   void kh_resize_##name(pic_state *, kh_##name##_t *h, khint_t new_n_buckets); \
   khint_t kh_put_##name(pic_state *, kh_##name##_t *h, khkey_t key, int *ret); \
   void kh_del_##name(kh_##name##_t *h, khint_t x);
@@ -95,8 +80,9 @@ PIC_INLINE khint_t ac_Wang_hash(khint_t key)
       h->size = h->n_occupied = 0;                                      \
     }                                                                   \
   }                                                                     \
-  khint_t kh_get_##name(const kh_##name##_t *h, khkey_t key)            \
+  khint_t kh_get_##name(pic_state *pic, const kh_##name##_t *h, khkey_t key) \
   {                                                                     \
+    (void)pic;                                                          \
     if (h->n_buckets) {                                                 \
       khint_t k, i, last, mask, step = 0;                               \
       mask = h->n_buckets - 1;                                          \
@@ -220,9 +206,6 @@ PIC_INLINE khint_t ac_Wang_hash(khint_t key)
 #define kh_ptr_hash_equal(a, b) ((a) == (b))
 #define kh_int_hash_func(key) (int)(key)
 #define kh_int_hash_equal(a, b) ((a) == (b))
-#define kh_str_hash_func(key) ac_X31_hash_string(key)
-#define kh_str_hash_equal(a, b) (strcmp(a, b) == 0)
-#define kh_int_hash_func2(k) ac_Wang_hash((khint_t)key)
 
 /* --- END OF HASH FUNCTIONS --- */
 
@@ -232,7 +215,7 @@ PIC_INLINE khint_t ac_Wang_hash(khint_t key)
 #define kh_clear(name, h) kh_clear_##name(h)
 #define kh_resize(name, h, s) kh_resize_##name(pic, h, s)
 #define kh_put(name, h, k, r) kh_put_##name(pic, h, k, r)
-#define kh_get(name, h, k) kh_get_##name(h, k)
+#define kh_get(name, h, k) kh_get_##name(pic, h, k)
 #define kh_del(name, h, k) kh_del_##name(h, k)
 
 #define kh_exist(h, x) (!ac_iseither((h)->flags, (x)))

extlib/benz/include/picrin/string.h

@@ -29,6 +29,7 @@ int pic_str_len(pic_str *);
 pic_str *pic_str_cat(pic_state *, pic_str *, pic_str *);
 pic_str *pic_str_sub(pic_state *, pic_str *, int, int);
 int pic_str_cmp(pic_state *, pic_str *, pic_str *);
+int pic_str_hash(pic_state *, pic_str *);
 const char *pic_str_cstr(pic_state *, pic_str *);
 
 pic_str *pic_format(pic_state *, const char *, ...);

extlib/benz/include/picrin/symbol.h

@@ -13,7 +13,7 @@ struct pic_id {
   union {
     struct pic_symbol {
       PIC_OBJECT_HEADER
-      const char *cstr;
+      pic_str *str;
     } sym;
     struct {
       PIC_OBJECT_HEADER

extlib/benz/string.c

@@ -288,6 +288,19 @@ pic_str_cmp(pic_state *pic, pic_str *str1, pic_str *str2)
   return strcmp(pic_str_cstr(pic, str1), pic_str_cstr(pic, str2));
 }
 
+int
+pic_str_hash(pic_state *pic, pic_str *str)
+{
+  const char *s;
+  int h = 0;
+
+  s = pic_str_cstr(pic, str);
+  while (*s) {
+    h = (h << 5) - h + *s++;
+  }
+  return h;
+}
+
 const char *
 pic_str_cstr(pic_state *pic, pic_str *str)
 {

extlib/benz/symbol.c

@@ -4,43 +4,39 @@
 
 #include "picrin.h"
 
-KHASH_DEFINE(s, const char *, pic_sym *, kh_str_hash_func, kh_str_hash_equal)
+#define kh_pic_str_hash(a) (pic_str_hash(pic, (a)))
+#define kh_pic_str_cmp(a, b) (pic_str_cmp(pic, (a), (b)) == 0)
+
+KHASH_DEFINE(s, pic_str *, pic_sym *, kh_pic_str_hash, kh_pic_str_cmp)
 
 pic_sym *
 pic_intern_str(pic_state *pic, pic_str *str)
-{
-  return pic_intern(pic, pic_str_cstr(pic, str));
-}
-
-pic_sym *
-pic_intern(pic_state *pic, const char *cstr)
 {
   khash_t(s) *h = &pic->syms;
   pic_sym *sym;
   khiter_t it;
   int ret;
-  char *copy;
 
-  it = kh_put(s, h, cstr, &ret);
+  it = kh_put(s, h, str, &ret);
   if (ret == 0) {               /* if exists */
     sym = kh_val(h, it);
     pic_gc_protect(pic, pic_obj_value(sym));
     return sym;
   }
 
-  copy = pic_malloc(pic, strlen(cstr) + 1);
-  strcpy(copy, cstr);
-  kh_key(h, it) = copy;
-
-  kh_val(h, it) = pic->sQUOTE; /* insert dummy */
-
   sym = (pic_sym *)pic_obj_alloc(pic, sizeof(pic_sym), PIC_TT_SYMBOL);
-  sym->cstr = copy;
+  sym->str = str;
   kh_val(h, it) = sym;
 
   return sym;
 }
 
+pic_sym *
+pic_intern(pic_state *pic, const char *cstr)
+{
+  return pic_intern_str(pic, pic_make_cstr(pic, cstr));
+}
+
 pic_id *
 pic_make_identifier(pic_state *pic, pic_id *id, struct pic_env *env)
 {
@@ -53,9 +49,9 @@ pic_make_identifier(pic_state *pic, pic_id *id, struct pic_env *env)
 }
 
 const char *
-pic_symbol_name(pic_state PIC_UNUSED(*pic), pic_sym *sym)
+pic_symbol_name(pic_state *pic, pic_sym *sym)
 {
-  return sym->cstr;
+  return pic_str_cstr(pic, sym->str);
 }
 
 const char *
@@ -104,7 +100,7 @@ pic_symbol_symbol_to_string(pic_state *pic)
 
   pic_get_args(pic, "m", &sym);
 
-  return pic_obj_value(pic_make_cstr(pic, sym->cstr));
+  return pic_obj_value(sym->str);
 }
 
 static pic_value