Use charpool to back string_pool

string_pool (string interning) was using STL strings with lots of extra
constructions, when all we need is a const char array. We can use
charpool for that and get all the benefits there of tightly-packed heap
allocations.

charpool.cc

@@ -72,8 +72,8 @@
 
 static char *charpool[16];
 static int currentcharpool;
-static int currentcharpoolsz;
-static char *nextchar;
+static size_t currentcharpoolsz;
+static size_t nexti;
 
 /* Allocated blocks are allocated to multiples of ALIGN_ON. This is the
    definition used by the malloc in Glibc 2.7, which says that it "suffices for
@@ -87,7 +87,8 @@ static int cp_init(void) {
   /* Create our char pool */
   currentcharpool = 0;
   currentcharpoolsz = 16384;
-  nextchar = charpool[0] = (char *) safe_malloc(currentcharpoolsz);
+  nexti = 0;
+  charpool[0] = (char *) safe_malloc(currentcharpoolsz);
   charpool_initialized = 1;
   return 0;
 }
@@ -109,8 +110,8 @@ static inline void cp_grow(void) {
   }
   currentcharpoolsz <<= 1;
 
-  nextchar = charpool[currentcharpool] = (char *)
-    safe_malloc(currentcharpoolsz);
+  nexti = 0;
+  charpool[currentcharpool] = (char *) safe_malloc(currentcharpoolsz);
 }
 
 void *cp_alloc(int sz) {
@@ -122,9 +123,9 @@ void *cp_alloc(int sz) {
   if ((modulus = sz % ALIGN_ON))
     sz += ALIGN_ON - modulus;
 
-  if ((nextchar - charpool[currentcharpool]) + sz <= currentcharpoolsz) {
-    p = nextchar;
-    nextchar += sz;
+  if (nexti + sz <= currentcharpoolsz) {
+    p = charpool[currentcharpool] + nexti;
+    nexti += sz;
     return p;
   }
   /* Doh!  We've got to make room */
@@ -134,32 +135,12 @@ void *cp_alloc(int sz) {
 
 }
 
-char *cp_strdup(const char *src) {
-const char *p;
-char *q;
-/* end points to the first illegal char */
-char *end;
-int modulus;
-
- cp_init();
-
- end = charpool[currentcharpool] + currentcharpoolsz;
- q = nextchar;
- p = src;
- while((nextchar < end) && *p) {
-   *nextchar++ = *p++;
+const char *cp_strndup(const char *src, int len) {
+  char *dst = (char *) cp_alloc(len + 1); // Additional byte for null terminator
+  dst[len] = '\0';
+  return (const char *) memcpy(dst, src, len);
 }
 
- if (nextchar < end) {
-   /* Goody, we have space */
-   *nextchar++ = '\0';
-   if ((modulus = (nextchar - q) % ALIGN_ON))
-     nextchar += ALIGN_ON - modulus;
-   return q;
- }
-
- /* Doh!  We ran out -- need to allocate more */
- cp_grow();
-
- return cp_strdup(src);
+const char *cp_strdup(const char *src) {
+  return cp_strndup(src, strlen(src));
 }

charpool.h

@@ -66,7 +66,9 @@
 #define CHARPOOL_H
 
 void *cp_alloc(int sz);
-char *cp_strdup(const char *src);
+/* len does not include null terminator */
+const char *cp_strndup(const char *src, int len);
+const char *cp_strdup(const char *src);
 
 void cp_free(void);
 

string_pool.cc

@@ -58,9 +58,11 @@
  ***************************************************************************/
 #include "string_pool.h"
 #include <nbase.h>
+#include "charpool.h"
 
 #include <set>
 #include <string>
+#include <cstring>
 #include <utility>
 #include <stdarg.h>
 #include <ctype.h>
@@ -69,19 +71,52 @@
 #undef NDEBUG
 #include <assert.h>
 
+class StringPoolItem {
+  public:
+    const char *str;
+    int len;
+    bool in_cp;
+
+    StringPoolItem(const char *i_str, int i_len) : str(i_str), len(i_len), in_cp(false) {}
+    ~StringPoolItem() {} // charpool allocations are permanent and can't be freed.
+    StringPoolItem(const StringPoolItem& other) {
+      // If the string is already in the charpool, there's no reason we should
+      // be copy-constructed, since that only happens on a successful insert
+      // (new unique item)
+      assert(!other.in_cp);
+      this->len = other.len;
+      this->str = cp_strndup(other.str, other.len);
+      this->in_cp = true;
+    }
+
+// asdfq <> asdf
+    bool operator< (const StringPoolItem& other) const {
+      return this->len < other.len || memcmp(this->str, other.str, other.len) < 0;
+    }
+};
+
+typedef std::set<StringPoolItem> StringPool;
+
+const char *string_pool_insert_len(const char *s, int len)
+{
+  static StringPool pool;
+  StringPoolItem spi (s, len);
+
+  StringPool::iterator it = pool.insert(spi).first;
+  assert(it->in_cp); // We should only be storing charpool-allocated strings
+
+  return it->str;
+}
+
 const char *string_pool_insert(const char *s)
 {
-  static std::set<std::string> pool;
-  static std::pair<std::set<std::string>::iterator, bool> pair;
-
-  pair = pool.insert(s);
-
-  return pair.first->c_str();
+  return string_pool_insert_len(s, strlen(s));
 }
 
 const char *string_pool_substr(const char *s, const char *t)
 {
-  return string_pool_insert(std::string(s, t).c_str());
+  assert(t >= s);
+  return string_pool_insert_len(s, t - s);
 }
 
 const char *string_pool_substr_strip(const char *s, const char *t) {
@@ -133,7 +168,7 @@ const char *string_pool_sprintf(const char *fmt, ...)
       break;
   }
 
-  s = string_pool_insert(buf);
+  s = string_pool_insert_len(buf, n);
   free(buf);
 
   return s;