ston/ston_ht.h

   1 /*!@file
   2   \brief   STON Hash Tables
   3   \details Aligned general purpose hash functions and memory definitions
   4            whose columns are provided, and whose rows, and sizes, are derived.
   5
   6              ht_size = header.ht_columns << header.ht_2pow;
   7              ht_rows = 0x1 << header.ht_2pow;
   8
   9            All generic hashtables in henge must have a power-of-two number of
  10            rows.  An ht_columns value that is also a power-of-two will result in
  11            a power-of-two sized memory imprint for the structure, making it easy
  12            to page align.
  13
  14            Elements in the columns may be of any arbitrary size.
  15
  16              typedef uint32_t my_ht_type;
  17              ht_bytes = ht_size * sizeof(my_ht_type);
  18
  19            implementation covers only 32-bit unit sizes.
  20
  21   \author  Ken Grimes
  22   \date    Feb 2017
  23   ----------------------------------------------------------------------------*/
  24 #ifndef _STON_HT_T_
  25 #define _STON_HT_T_
  26 /* Define STON_NOSTATIC to expose included function symbols */
  27 #ifndef STON_NOSTATIC
  28 #define STON_FUNC_STATIC static
  29 #else
  30 #define STON_FUNC_STATIC
  31 #endif //STON_NOSTATIC
  32 /* If GNUC is detected, uses attributes to stop inlining */
  33 #ifdef  __GNUC__
  34 #define STON_FUNC_NOINLINE __attribute__ ((noinline))
  35 #else
  36 #define STON_FUNC_NOINLINE
  37 #endif //__GNUC__
  38 /* Define STON_NOINLINE to prevent inline compiler hints */
  39 #ifndef STON_NOINLINE
  40 #define STON_FUNC_INLINE inline
  41 #else
  42 #define STON_FUNC_INLINE
  43 #endif //STON_NOINLINE
  44 /* Define STON_FUNC to override the default STON Function attributes */
  45 #ifndef STON_FUNC
  46 #define STON_FUNC STON_FUNC_STATIC STON_FUNC_INLINE
  47 #endif //STON_FUNC
  48 #ifndef STON_NOSTDIO
  49 #include <stdio.h>
  50 #include <string.h> //memcpy
  51 #include <alloca.h>
  52 #endif //STON_NOSTDIO
  53 #include <stdint.h>
  54 /* STON Hashtable Structure
  55    Hashtables are stored as dynamically sized two dimensional arrays
  56 */
  57 typedef struct ston_ht_header_t
  58 { uint16_t ht_columns;
  59   uint8_t  ht_2pow, ht_flags;
  60 }* ston_ht;
  61
  62 STON_FUNC
  63 size_t    ston_up2pow(size_t);
  64 STON_FUNC_STATIC
  65 STON_FUNC_NOINLINE
  66 ston_ht   ston_ht32_fread(FILE*,long,void*(*)(size_t));
  67 STON_FUNC
  68 ston_ht   ston_ht32_create(uint16_t,size_t,uint8_t,void*(*)(size_t));
  69 STON_FUNC
  70 uint32_t* ston_ht32_row(ston_ht,uint32_t);
  71 STON_FUNC
  72 uint32_t  ston_ht32_insert(ston_ht,uint32_t,uint16_t,uint32_t);
  73
  74 #define   ston_ht32_new(_COL,_N,_F,_FN) ston_ht32_create(_COLS,ston_up2pow(_N << 1),_F,_FN)
  75 #define   ston_ht32_col(_HT,_KEY,_COL) (ston_ht32_row(_HT,_KEY) + _COL)
  76 #define   ston_ht32_insertx(_HT,_KEY,_COL,_VAL) *ston_ht32_col(_HT,_KEY,_COL) = _VAL
  77 #define   ston_ht_size(_HT)            ((_HT)->ht_columns << (_HT)->ht_2pow)
  78 #define   ston_ht_rows(_HT)            (0x1 << (_HT)->ht_2pow)
  79 #define   ston_ht_cols(_HT)            ((_HT)->ht_columns)
  80 #define   ston_ht_start(_HT)           (((uint8_t*)(_HT)) + sizeof(*(_HT)))
  81 #define   ston_ht_keyrow(_HT,_KEY)     ((_KEY) & (ston_ht_rows(ht) - 1))
  82 #define   ston_ht32_start(_HT)         ((uint32_t*)ston_ht_start(_HT))
  83 #define   ston_ht32_size(_HT)          (ston_ht_size(_HT) * sizeof(uint32_t))
  84
  85 /** @see http://graphics.stanford.edu/~seander/bithacks.html#RoundUpPowerOf2 */
  86 STON_FUNC
  87 size_t ston_up2pow
  88 ( size_t val )
  89 { val = (val << 1) - 1;
  90   val |= val >> 1;
  91   val |= val >> 2;
  92   val |= val >> 4;
  93   val |= val >> 8;
  94   val |= val >> 16;
  95   return ++val;
  96 }
  97
  98 /* Creates a new hash table, provided a memory allocation function that takes a
  99    single size_t bytes, a column count, and a row count which determines the
 100    size of the table.
 101
 102    use ston_ht32_new to specify the exact or estimated number of unique keys
 103    held in the table.  With ston_ht32_new, the provided ht_rows is doubled, and
 104    rounded up to the nearest power of two to create a hash table with minimal
 105    collisions.
 106 */
 107 STON_FUNC
 108 ston_ht ston_ht32_create
 109 ( uint16_t ht_columns,
 110   size_t   ht_rows,
 111   uint8_t  ht_flags,
 112   void*    (*alloc_fn)(size_t)
 113 )
 114 { size_t  ht_size = ht_rows * ht_columns * sizeof(uint32_t);
 115   ston_ht ht = (ston_ht) alloc_fn(sizeof(struct ston_ht_header_t) + ht_size);
 116   if (ht != NULL)
 117     { for (ht->ht_2pow = 0; ht_size; ht->ht_2pow++)
 118         ht_size = ht_size >> 1;
 119       ht->ht_columns = ht_columns;
 120       ht->ht_flags = ht_flags;
 121     }
 122   return ht;
 123 }
 124
 125 #ifndef STON_NO_STDIO
 126 /* Reads a 32-bit hash table out of the provided file at the provide fpos, into
 127    a buffer allocated by alloc_fn.  Memory is allocated to the stack until the
 128    entire structure is verified, and all file operations are finished.
 129    Returns NULL with properly set errno on failure.
 130 */
 131 STON_FUNC_STATIC
 132 STON_FUNC_NOINLINE
 133 ston_ht ston_ht32_fread
 134 ( FILE* file,
 135   long  fpos,
 136   void* (*alloc_fn)(size_t)
 137 )
 138 { struct ston_ht_header_t header;
 139   ston_ht                 stack_ht, ht;
 140   long                    fpos_start;
 141   size_t                  table_size, alloc_size;
 142   if ((fpos_start = ftell(file)) == -1)
 143     return NULL;
 144   if (fread(&header, sizeof(header), 1, file) != 1)
 145     return NULL;
 146   table_size = ston_ht32_size(&header);
 147   alloc_size = sizeof(header) + table_size;
 148   stack_ht = (ston_ht) alloca(alloc_size);
 149   memcpy(stack_ht, &header, sizeof(header));
 150   if (fread(stack_ht + sizeof(header), table_size, 1, file) != 1)
 151     return NULL;
 152   if (fseek(file, fpos_start, SEEK_SET) != 0)
 153     return NULL;
 154   ht = (ston_ht) alloc_fn(alloc_size);
 155   if (ht != NULL)
 156     memcpy(ht, stack_ht, alloc_size);
 157   return ht;
 158 }
 159 #endif
 160
 161 /* Returns a pointer to the row of data in the hashtable containing the provided
 162    key, inserts if not found.  Returns NULL on overflow.
 163 */
 164 STON_FUNC_STATIC
 165 uint32_t* ston_ht32_row
 166 ( struct ston_ht_header_t* ht,
 167   uint32_t                key
 168 )
 169 { uint32_t* row,* row_start = ston_ht32_start(ht);
 170   uint16_t  ht_cols = ston_ht_cols(ht);
 171   size_t    row_number = ston_ht_keyrow(ht,key);
 172   size_t    row_max = ston_ht_rows(ht);
 173   uint8_t   looped = 0;
 174  next_row:
 175   row = row_start + (row_number * ht_cols);
 176   if (row[0] != 0)
 177     goto populated;
 178  write_position:
 179   row[0] = key;
 180   return row;
 181  populated:
 182   if (row[0] == key)
 183     goto write_position;
 184   if (row_number < row_max)
 185     row_number++;
 186   else if (looped)
 187       return NULL;
 188   else
 189     { looped++;
 190       row_number = 0;
 191     }
 192   goto next_row;
 193 }
 194
 195 /* Inserts a value into a hashtable at the specified column, returning the
 196    previous value */
 197 STON_FUNC
 198 uint32_t ston_ht32_insert
 199 ( struct ston_ht_header_t* ht,
 200   uint32_t                key,
 201   uint16_t                column,
 202   uint32_t                value
 203 )
 204 { uint32_t* value_location, old_value;
 205   value_location = ston_ht32_col(ht,key,column);
 206   old_value = *value_location;
 207   *value_location = value;
 208   return old_value;
 209 }
 210
 211 /* STON Dynamic Hashtable Structure
 212    A dynamic form of the generic hashtable implementation above which uses
 213    external allocation.
 214 */
 215 typedef struct ston_dht_header_t
 216 { uint16_t ht_columns;
 217   uint8_t  ht_2pow, ht_flags;
 218   void     (*ht_alloc)(size_t);
 219   void*    ht_pages[];
 220 }* ston_dht;
 221
 222 #define   ston_dht_size(_HT)          (ston_ht_size(_HT))
 223 #define   ston_dht_rows(_HT)          (ston_ht_rows(_HT))
 224 #define   ston_dht_cols(_HT)          (ston_ht_cols(_HT))
 225 #define   ston_dht_keyrow(_HT,_KEY)   (ston_ht_keyrow(_HT,_KEY))
 226 #define   ston_dht_start(_HT)         (_HT->ht_pages[0])
 227 #define   ston_dht32_start(_HT)       ((_uint32*)ston_dht_start(_HT))
 228 ston_dht  ston_dht32_create(uint16_t,size_t,void*(*)(size_t));
 229 uint32_t* ston_dht32_row(ston_dht,uint32_t);
 230 #define   ston_dht32_col(_HT,_KEY,_COL) (ston_dht32_row(_HT,_KEY) + _COL)
 231 uint32_t  ston_dht32_insert(ston_dht,uint32_t,uint16_t,uint32_t);
 232 #define   ston_dht32_insertx(_HT,_KEY,_COL,_VAL) *ston_dht32_col(_HT,_KEY,_COL) = _VAL
 233
 234
 235 #endif //_STON_HT_H_