campo-sirio/include/hashtable.cpp

#include <hashtable.h>

///////////////////////////////////////////////////////////
// Utilities
///////////////////////////////////////////////////////////

static inline unsigned int ELFHash(const char* str)
{
  unsigned int hash = 0;
  unsigned int x    = 0;
  for(const char* c = str; *c; c++)
  {
    hash = (hash << 4) + *c;
    x = hash & 0xF0000000L;
    if (x != 0)
    {
      hash ^= (x >> 24);
      hash &= ~x;
    }
  }
  return (hash & 0x7FFFFFFF);
}

static inline unsigned int JenkinsHash(const char *key)
{
    unsigned int hash = 0;
    for(const char* k = key; *k; k++)
    {
        hash += *k;
        hash += (hash << 10);
        hash ^= (hash >> 6);
    }
    hash += (hash << 3);
    hash ^= (hash >> 11);
    hash += (hash << 15);
    return hash;
}

static bool is_prime(size_t n)
{
  if (n & 1)
  {
    for (size_t d = 3; d < n/2; d+=2)
      if (n % d == 0)
        return false;
    return true;
  }
  return false;
}

static size_t next_size(size_t p)
{
  for (p = p*4-1; !is_prime(p); p += 2);
  return p;
}

///////////////////////////////////////////////////////////
// THash_bucket
///////////////////////////////////////////////////////////

struct THash_entry
{
  char buf[8];
  char* key;
  unsigned int hash;
  TObject* value;
};

struct THash_bucket : public TObject
{
  enum { nMaxEntries = 16 };
  size_t _bloom;
  THash_entry _entry[nMaxEntries];
  int _entries;

private:
  size_t bloom(unsigned int h) const { return 1 << (h%32); }

public:
  int items() const { return _entries; }
  int size() const { return nMaxEntries; }
  
  bool add(const char* key, unsigned int hash, TObject* value);
  void steal(THash_entry& e);

  THash_entry* lookup(const char* key, unsigned int hash);
  THash_entry& operator[](int i) { return _entry[i]; }
  TObject* remove(int i);
  TObject* remove(THash_entry* e);
  THash_bucket() : _bloom(0), _entries(0) {}
  ~THash_bucket();
};

bool THash_bucket::add(const char* key, unsigned int hash, TObject* value)
{
  CHECK(_entries < nMaxEntries, "bucket full");
  _bloom |= bloom(hash);
  THash_entry& e = _entry[_entries++];
  if (strlen(key) < sizeof(e.buf))
  {
    strcpy_s(e.buf, sizeof(e.buf), key);
    e.key = NULL;
  }
  else
  {
    *e.buf = '\0';
    e.key = _strdup(key);
  }
  e.hash = hash;
  e.value = value;
  return _entries > 3*nMaxEntries/4; // Danger!
}

void THash_bucket::steal(THash_entry& i)
{
  CHECK(_entries < nMaxEntries, "bucket full");
  _bloom |= bloom(i.hash);
  THash_entry& e = _entry[_entries++];
  memcpy(&e, &i, sizeof(i));
}

THash_entry* THash_bucket::lookup(const char* key, unsigned int hash)
{
  if (_bloom & bloom(hash))
  {
    for (int p = _entries-1; p >= 0; p--)
    {
      THash_entry& e = _entry[p];
      if (e.hash == hash && strcmp(e.key ? e.key : e.buf, key) == 0)
        return &e;
    }
  }
  return NULL;
}

TObject* THash_bucket::remove(int p)
{
  CHECKD(p >= 0 && p < _entries, "Invalid item index ", p);
  THash_entry& e = _entry[p];
  TObject* obj = e.value;
  if (e.key)
    free(e.key);
  if (p < _entries-1)
    memcpy(&_entry[p], &_entry[_entries-1], sizeof(THash_entry));
  _entries--;
  if (_entries == 0) _bloom = 0;
  return obj;
}

TObject* THash_bucket::remove(THash_entry* e)
{
  for (int p = _entries-1; p >= 0; p--) if (&_entry[p] == e)
    return remove(p);
  return NULL;
}

THash_bucket::~THash_bucket()
{
  for (int p = _entries-1; p >= 0; p--)
    delete remove(p);
}

///////////////////////////////////////////////////////////
// THash_table
///////////////////////////////////////////////////////////

void THash_table::expand() 
{
  TArray* tmp = _buckets;
  const size_t bs = next_size(_buckets->size());
  _buckets = new TArray(bs);

  for (int i = tmp->last(); i >= 0; i = tmp->pred(i))
  {
    THash_bucket& bi = *(THash_bucket*)tmp->objptr(i);
    for (int j = bi.items()-1; j >= 0; j--)
    {
      THash_entry& e = bi[j];
      const size_t buck = e.hash % bs;
      bucket(buck).steal(e);
    }
    bi._entries = 0;
  }

  delete tmp;
}

THash_table::~THash_table()
{
  delete _buckets;
}

THash_entry* THash_table::lookup_ex(const char* key, unsigned int& hash) const
{
  //hash = JenkinsHash(key);
  hash = ELFHash(key);
  const unsigned int buck = hash % _buckets->size();
  THash_bucket* b = (THash_bucket*)_buckets->objptr(buck);
  return b ? b->lookup(key, hash) : NULL;
}

THash_entry* THash_table::lookup(const char* key) const
{
  TObject* obj = NULL;
  unsigned int hash;
  return lookup_ex(key, hash);
}

THash_bucket& THash_table::bucket(size_t n)
{
  THash_bucket* b = (THash_bucket*)_buckets->objptr(n);
  if (b == NULL)
  {
    b = new THash_bucket;
    _buckets->add(b, n);
  }
  return *b;
}

void THash_table::add(TObject* value, const char* key, unsigned int hash) 
{
  const size_t buck = hash % _buckets->size();
  THash_bucket& b = bucket(buck);
  const bool expand_needed = b.add(key, hash, value);
  _items++;
  if (expand_needed)
    expand();
}

bool THash_table::add(const char* key, TObject* value, bool force) 
{
  unsigned int hash;
  THash_entry* current = lookup_ex(key, hash);
  if (current)
  {
    if (force)
    {
      delete current->value;
      current->value = value;
    }
    else
      delete value;
  }
  else
    add(value, key, hash);
  return current != NULL;
}

bool THash_table::add(const char* key, const TObject& value, bool force) 
{
  unsigned int hash;
  THash_entry* current = lookup_ex(key, hash);
  if (current)
  {
    if (force)
    {
      delete current->value;
      current->value = value.dup();
    }
  }
  else
    add(value.dup(), key, hash);
  return current != NULL;
}

TObject* THash_table::get(const char* key) 
{ 
  THash_entry* entry = lookup(key); 
  return entry ? entry->value : NULL; 
}

TObject* THash_table::remove(const char* key) 
{
  TObject* obj = NULL;
  unsigned int hash;
  THash_entry* e = lookup_ex(key, hash);
  if (e)
  {
    const size_t buck = hash % _buckets->size();
    obj = bucket(buck).remove(e);
    _items--;
  }
  return obj;
}

bool THash_table::destroy(const char* key)
{
  TObject* o = remove(key);
  if (o) delete o;
  return o != NULL;
}

THash_table::THash_table(size_t initialCapacity) 
{
  size_t bucketCount = 7;
  if (bucketCount < initialCapacity) 
    bucketCount = next_size(initialCapacity); 
  _buckets = new TArray(bucketCount);
  _items = 0;

  _curr_bucket = _curr_item = 0;
}

TObject* THash_table::succ_item()
{
  while (_curr_bucket >= 0 && _curr_bucket < _buckets->size())
  {
    THash_bucket* b = (THash_bucket*)_buckets->objptr(_curr_bucket);
    if (b)
    {
      if (_curr_item < b->items())
        return (*b)[_curr_item++].value;
    }
    else
      break;
    _curr_bucket = _buckets->succ(_curr_bucket);
    _curr_item = _curr_bucket >= 0 ? 0 : -1;
  }
  return NULL;
}

TObject* THash_table::first_item()
{
  _curr_bucket = _buckets->first();
  _curr_item = _curr_bucket >= 0 ? 0 : -1;
  return succ_item();
}

TObject* THash_table::last_item()
{ return first_item(); }

TObject* THash_table::pred_item()
{ return succ_item(); }