lsup_rdf/include/buffer.h

#ifndef _LSUP_BUFFER_H
#define _LSUP_BUFFER_H

#include "xxhash.h"

#include "core.h"

#ifndef HASH_SEED
/** @brief Seed used for all hashing. Compile-time configurable.
 */
#define HASH_SEED 0
#endif

// "NULL" key, a value that is never user-provided. Used to mark special
// values (e.g. deleted records).
#define NULL_KEY 0


/** @brief General-purpose data buffer.
 *
 * The structure is transparently exposed so that the related API only defines
 * few basic helper methods. Other operations, such as appending, may be
 * performed by simply using the addr and size attributes.
 *
 * A buffer can be initialized once and reused multiple times, e.g. in a loop,
 * without being freed between iterations, by using #LSUP_buffer_init.
 */
typedef struct LSUP_Buffer {
    /*@null@*/ void *addr;
    size_t size;
} LSUP_Buffer;


/** @brief Triple of byte buffers.
 *
 * This is a generic data triple. Store implementations should handle this
 * data type rather than RDF terms and triples. Conversion to/from RDF terms
 * and triples is done in the term and triple modules.
 */
typedef struct buffer_triple_t {
    LSUP_Buffer *s;
    LSUP_Buffer *p;
    LSUP_Buffer *o;
} LSUP_BufferTriple;


typedef enum {
    TRP_POS_S = 0,
    TRP_POS_P = 1,
    TRP_POS_O = 2,
} LSUP_TriplePos;


/** Initialize or reuse a buffer handle.
 *
 * The handle must have been created with #LSUP_buffer_new*().
 *
 * The data block is resized without being freed first. The handle must be
 * eventually freed with #LSUP_buffer_done() after use.
 *
 * @param buf[in] A buffer handle obtained with #LSUP_buffer_new or by manual
 *  allocation.
 *
 * @param size[in] New size.
 *
 * @param data[in] If not NULL, data to replace the existing ones. The size
 *  of the data to be copied is determined by the size parameter. If NULL, the
 *  existing data are preserved as with a normal realloc().
 */
LSUP_rc
LSUP_buffer_init (LSUP_Buffer *buf, const size_t size, const void *data);


/** @brief Create a new buffer and optionally populate it with data.
 *
 * To change the buffer size and/or data later call #LSUP_buffer_init.
 *
 * To copy a buffer just do buf2 = LSUP_buffer_new (buf1->size, buf1->addr);
 *
 * @param size[in] Length of the data.
 *
 * @param data[in] Optional data to initially populate the object with. If
 *  NULL, the buffer data are garbage.
 *
 * @return LSUP_Buffer pointer. It must be freed with #LSUP_buffer_free. NULL
 *  on error.
 */
inline LSUP_Buffer *
LSUP_buffer_new (const size_t size, const void *data)
{
    LSUP_Buffer *buf;
    CALLOC_GUARD (buf, NULL);

    if (LSUP_buffer_init (buf, size, data) != LSUP_OK) {
        free (buf->addr);
        free (buf);
        return NULL;
    }

    return buf;
}


/** @brief Dummy buffer to be used with #LSUP_buffer_init.
 */
#define BUF_DUMMY LSUP_buffer_new (0, NULL)


/** @brief Free the content of a buffer.
 */
void LSUP_buffer_done (LSUP_Buffer *buf);


/** @brief Free a buffer.
 */
void LSUP_buffer_free (LSUP_Buffer *buf);


/** @brief Hash a buffer.
 */
inline LSUP_Key
LSUP_buffer_hash (const LSUP_Buffer *buf)
{ return (buf == NULL) ? NULL_KEY : XXH64 (buf->addr, buf->size, HASH_SEED); }


/** @brief Print a byte string of a given length in a human-readable format.
 *
 * The string is printed in Python style: printable characters are output
 * literally, and non-printable ones as hex sequences.
 */
void LSUP_buffer_print (const LSUP_Buffer *buf);


/** @brief Format a buffer into anb ASCII string.
 *
 * The string has non-printable characters escaped as "\xNN".
 *
 * @param buf[in] Buffer to convert.
 *
 * @return Formatted string. It must be freed with free().
 */
char *
LSUP_buffer_as_str (const LSUP_Buffer *buf);


/** @brief Compare two buffers.
 *
 * The return value is the same as memcmp.
 */
inline int LSUP_buffer_cmp (const LSUP_Buffer *buf1, const LSUP_Buffer *buf2)
{
    return memcmp (
            buf1->addr, buf2->addr,
            (buf1->size > buf2->size ? buf1->size : buf2->size));
}


/** @brief Return whether two buffers are equal.
 *
 * This may be faster than #LSUP_buffer_cmp() because it does a size comparison
 * first.
 */
inline bool LSUP_buffer_eq (const LSUP_Buffer *buf1, const LSUP_Buffer *buf2)
{
    if (buf1->size != buf2->size) return false;

    return (LSUP_buffer_cmp (buf1, buf2) == 0) ? true : false;
}


/*
 * Buffer triples.
 */

LSUP_BufferTriple *
LSUP_btriple_new(LSUP_Buffer *s, LSUP_Buffer *p, LSUP_Buffer *o);


/** @brief Initialize internal term pointers in a heap-allocated buffer triple.
 *
 * The triple must be freed with #LSUP_btriple_free().
 *
 * @param sspo[in] Serialized triple pointer to initialize.
 */
LSUP_rc
LSUP_btriple_init (
        LSUP_BufferTriple *sspo, LSUP_Buffer *s, LSUP_Buffer *p, LSUP_Buffer *o);


/** @brief Free the internal pointers of a buffer triple.
 *
 * @param sspo[in] Buffer triple to be freed.
 */
void
LSUP_btriple_done (LSUP_BufferTriple *sspo);


/** @brief Free a buffer triple and all its internal pointers.
 *
 * NOTE: If the buffer pointers are not to be freed (e.g. they are owned by a
 * back end), use a simple free(sspo) instead of this.
 *
 * @param sspo[in] Buffer triple to be freed.
 */
void
LSUP_btriple_free (LSUP_BufferTriple *sspo);


/** @brief Free a buffer triple and its buffer handles but not the buffer data.
 *
 * This is useful when freeing a "dummy" triple whose buffers are owned by the
 * caller but the data the terms point to are owned by the store.
 *
 */
void
LSUP_btriple_free_shallow (LSUP_BufferTriple *sspo);


/** @brief Get serialized triple by term position.
 *
 * Useful for looping over all terms.
 *
 * @param trp[in] Serialized triple pointer.
 *
 * @param n[in] A number between 0÷2.
 *
 * @return Corresponding serialized term or NULL if n is out of range.
 */
inline LSUP_Buffer *
LSUP_btriple_pos (const LSUP_BufferTriple *btrp, LSUP_TriplePos n)
{
    if (n == TRP_POS_S) return btrp->s;
    if (n == TRP_POS_P) return btrp->p;
    if (n == TRP_POS_O) return btrp->o;

    return NULL;
}


/** @brief Hash a buffer triple.
 *
 * TODO This doesn't handle blank nodes correctly.
 */
inline LSUP_Key
LSUP_btriple_hash (const LSUP_BufferTriple *strp)
{
    return XXH64 (
        strp->s->addr, strp->s->size,
        XXH64 (
            strp->p->addr, strp->p->size,
            XXH64 (strp->o->addr, strp->o->size, HASH_SEED)
        )
    );
}


#define STRP_DUMMY LSUP_btriple_new (NULL, NULL, NULL)

#endif