symbol interning

sym.c

@@ -0,0 +1,81 @@
+#include <stdlib.h>
+#include <string.h>
+
+#include "wscm.h"
+
+St syms = {0, 0, NULL};
+
+static S *findsym(const char *str, Z len, U32 hash) {
+  if (syms.capacity == 0)
+    return NULL;
+
+  Z ix = hash % syms.capacity;
+  for (Z i = 0; i < syms.capacity; i++) {
+    S *s = syms.data[ix];
+    if (!s)
+      return NULL;
+    if (s->hash == hash && s->len == len)
+      return s;
+    ix = (ix + 1) % syms.capacity;
+  }
+  return NULL;
+}
+
+static void symtabresize(void) {
+  Z cap = syms.capacity;
+  if (cap == 0) {
+    syms.capacity = 16;
+  } else {
+    syms.capacity *= 2;
+  }
+  S **nb = calloc(syms.capacity, sizeof(S *));
+  for (Z i = 0; i < cap; i++) {
+    if (syms.data[i]) {
+      S *s = syms.data[i];
+      Z ix = s->hash % syms.capacity;
+      while (nb[ix])
+        ix = (ix + 1) % syms.capacity;
+      nb[ix] = s;
+    }
+  }
+  if (syms.data != NULL)
+    free(syms.data);
+  syms.data = nb;
+}
+
+U32 hashstring(const char *data, Z len) {
+  U32 hash = 2166136261u;
+  for (Z i = 0; i < len; i++) {
+    hash ^= (uint8_t)data[i];
+    hash *= 16777619u;
+  }
+  return hash;
+}
+
+S *intern(const char *str, I len) {
+  if (len < 0)
+    len = strlen(str);
+
+  U32 hash = hashstring(str, len);
+  S *s = findsym(str, len, hash);
+  if (s)
+    return s;
+
+  s = malloc(sizeof(S));
+  s->data = malloc(len);
+  memcpy(s->data, str, len);
+  s->len = len;
+  s->hash = hash;
+
+  if (syms.count + 1 > syms.capacity)
+    symtabresize();
+
+  Z ix = hash % syms.capacity;
+  while (syms.data[ix])
+    ix = (ix + 1) % syms.capacity;
+
+  syms.data[ix] = s;
+  syms.count++;
+
+  return s;
+}