volksdata/src/term.c

#include "tpl.h"

#include "term.h"

/*
 * tpl packing format for the term structure.
 */
#define TERM_PACK_FMT "S(suc)"

/*
 * Extern variables.
 */

IDCache *LSUP_id_cache = NULL;
uint32_t LSUP_default_dtype_key = 0;
regex_t *LSUP_uri_ptn;


/*
 * Static variables.
 */

// Characters not allowed in a URI string.
static const char *invalid_uri_chars = "<>\" {}|\\^`";


/*
 * API functions.
 */

LSUP_Term *
LSUP_term_new (
        LSUP_TermType type, const char *data, const char *metadata)
{
    LSUP_Term *term;
    CALLOC_GUARD (term, NULL);

    // If undefined, just set the type.
    if (type == LSUP_TERM_UNDEFINED) term->type = type;

    else if (UNLIKELY (LSUP_term_init (
                    term, type, data, metadata) != LSUP_OK)) {
        free (term);
        return NULL;
    }

    return term;
}


LSUP_Term *
LSUP_term_new_from_buffer (const LSUP_Buffer *sterm)
{
    if (UNLIKELY (!sterm)) return NULL;

    LSUP_Term *term;
    MALLOC_GUARD (term, NULL);

    tpl_node *tn;

    tn = tpl_map (TERM_PACK_FMT, term);
    if (UNLIKELY (!tn)) goto fail;

    if (UNLIKELY (tpl_load (tn, TPL_MEM, sterm->addr, sterm->size) < 0))
        goto fail;

    if (UNLIKELY (tpl_unpack (tn, 0) < 0)) goto fail;

    tpl_free (tn);

    return term;

fail:
    tpl_free (tn);
    free (term);

    return NULL;
}


LSUP_Buffer *
LSUP_buffer_new_from_term (const LSUP_Term *term)
{
    if (UNLIKELY (!term)) return NULL;

    LSUP_Buffer *sterm;
    MALLOC_GUARD (sterm, NULL);

    int rc = tpl_jot (
            TPL_MEM, &sterm->addr, &sterm->size, TERM_PACK_FMT, term);
    if (rc != 0) {
        free (sterm);
        return NULL;
    }

    return sterm;
}


LSUP_rc
LSUP_term_init(
        LSUP_Term *term, LSUP_TermType type,
        const char *data, const char *metadata)
{
    // This can never be LSUP_TERM_UNDEFINED.
    if (!data) return LSUP_VALUE_ERR;
    term->type = type;

    // Validate URI.
    if (term->type == LSUP_TERM_IRIREF) {
        if (strpbrk (data, invalid_uri_chars) != NULL) {
            log_error (
                    "Characters %s are not allowed. Got: %s\n",
                    invalid_uri_chars, data);

            return LSUP_VALUE_ERR;
        }

        if (regexec (LSUP_uri_ptn, data, 0, NULL, 0) != 0) {
            fprintf (stderr, "Error matching URI pattern.\n");

            return LSUP_VALUE_ERR;
        }
    }

    char *data_tmp = realloc (term->data, strlen (data) + 1);
    if (UNLIKELY (!data_tmp)) return LSUP_MEM_ERR;
    term->data = data_tmp;
    strcpy (term->data, data);

    if (term->type == LSUP_TERM_LT_LITERAL) {
        term->lang = XXH64 (metadata, strlen (metadata) + 1, HASH_SEED);
        LSUP_tcache_add_id (term->lang, metadata);

    } else if (metadata && strcmp (metadata, DEFAULT_DTYPE) != 0) {
        term->datatype = XXH64 (metadata, strlen (metadata) + 1, HASH_SEED);
        LSUP_tcache_add_id (term->datatype, metadata);
    }

    return LSUP_OK;
}


LSUP_rc
LSUP_uri_init (LSUP_Term *term, const char *data)
{
    if (!data) {
        uuid_t uuid;
        uuid_generate_random (uuid);

        uuid_str_t uuid_str;
        uuid_unparse_lower (uuid, uuid_str);

        char uri[UUIDSTR_SIZE + 10];
        sprintf (uri, "urn:uuid4:%s", uuid_str);

        data = uri;
    }

    return LSUP_term_init (term, LSUP_TERM_IRIREF, data, NULL);
}


bool LSUP_term_equals (const LSUP_Term *term1, const LSUP_Term *term2)
{
    if (term1->type != term2->type)
        return false;

    if (strcmp (term1->data, term2->data) != 0)
        return false;

    if (term1->type == LSUP_TERM_LITERAL)
        return term1->datatype == term2->datatype;

    if (term1->type == LSUP_TERM_LT_LITERAL)
        return term1->lang == term2->lang;

    return true;
}


void LSUP_term_done (LSUP_Term *term)
{
    free (term->data);
    term->data = NULL;
}


void LSUP_term_free (LSUP_Term *term)
{
    if (LIKELY (term != NULL)) {
        free (term->data);
        free (term);
    }
}


LSUP_rc
LSUP_tcache_add_id (const uint32_t key, const char *data)
{
    struct id_cache_t *entry;

    HASH_FIND_INT (LSUP_id_cache, &key, entry);
    if (entry) return LSUP_NOACTION;

    MALLOC_GUARD (entry, LSUP_MEM_ERR);
    entry->key = key;
    entry->data = strdup (data);
    HASH_ADD_INT (LSUP_id_cache, key, entry);

    return LSUP_OK;
}


const char *
LSUP_tcache_get_id (const uint32_t key)
{
    struct id_cache_t *entry;

    HASH_FIND_INT (LSUP_id_cache, &key, entry);
    if (entry) log_trace ("Id found for key %d: %s", key, entry->data);
    else log_trace ("No ID found for key %d.", key);

    return (entry) ? entry->data : NULL;
}


// Extern inline functions.

LSUP_Key LSUP_term_hash (const LSUP_Term *term);
LSUP_Term *LSUP_uri_new (const char *data);
LSUP_rc LSUP_uri_init (LSUP_Term *term, const char *data);