volksdata/src/codec.c

#include "codec.h"

uint8_t *unescape_unicode (const uint8_t *esc_str)
{
    size_t size = strlen ((char *)esc_str);
    uint8_t *data = malloc (size + 1);

    size_t j = 0;
    uint8_t tmp_chr[5];
    for (size_t i = 0; i < size;) {
        if (esc_str[i] == '\\') {
            i++; // Skip over '\\'

            // 4-hex sequence.
            if (esc_str[i] == 'u') {
                i ++; // Skip over 'u'

                // Use tmp_chr to hold the hex string for the code point.
                memcpy (tmp_chr, esc_str + i, sizeof (tmp_chr) - 1);
                tmp_chr[4] = '\0';

                uint32_t tmp_val = strtol ((char*)tmp_chr, NULL, 16);
                log_debug ("tmp_val: %d", tmp_val);

                // Reuse tmp_chr to hold the byte values for the code point.
                int nbytes = utf8_encode (tmp_val, tmp_chr);
                if (nbytes == 0)
                    log_error ("Error encoding sequence: \\u%s", tmp_chr);

                // Copy bytes into destination.
                memcpy (data + j, tmp_chr, nbytes);
                log_trace ("UC byte value: %x %x", data[j], data[j + 1]);

                j += nbytes;
                i += 4;

            // 8-hex sequence.
            } else if (esc_str[i] == 'U') {
                i ++; // Skip over 'U'
                log_error ("4-byte sequence encoding is not yet implemented.");
                return NULL; // TODO encode 4-byte sequences

            // Unescape other escaped characters.
            } else data[j++] = unescape_char (esc_str[i++]);
        } else {
            // Copy ASCII char verbatim.
            data[j++] = esc_str[i++];
        }
    }

    data[j++] = '\0';
    uint8_t *tmp = realloc (data, j); // Compact result.
    if (UNLIKELY (!tmp)) return NULL;
    data = tmp;

    return data;
}


LSUP_Term **
LSUP_obj_list_add (LSUP_Term **ol, LSUP_Term *o)
{
    size_t i = 0;
    while (ol[i++]);
    LSUP_Term **tmp = realloc (ol, sizeof (*ol) * (i + 1));
    if (!tmp) return NULL;
    tmp[i] = o;
    tmp[i + 1] = NULL;

    return ol;
}


LSUP_PredObjList *
LSUP_pred_obj_list_new (void)
{
    /*
     * Init state:
     * {p: [NULL], o: [NULL]}
     */
    LSUP_PredObjList *po;
    MALLOC_GUARD (po, NULL);
    // Set sentinels.
    CALLOC_GUARD (po->p, NULL);
    CALLOC_GUARD (po->o, NULL);

    return po;
}


void
LSUP_pred_obj_list_free (LSUP_PredObjList *po)
{
    for (size_t i = 0; po->p[i]; i++) {
        LSUP_term_free (po->p[i]);
    }

    for (size_t i = 0; po->o[i]; i++) {
        for (size_t j = 0; po->o[i][j]; j++) {
            LSUP_term_free (po->o[i][j]);
        }
    }

    free (po);
}


LSUP_rc
LSUP_pred_obj_list_add (LSUP_PredObjList *po, LSUP_Term *p, LSUP_Term **o)
{
    size_t i;

    i = 0;
    while (po->p[i++]);  // Count includes sentinel.
    LSUP_Term **tmp_p = realloc (po->p, sizeof (*po->p) * (i + 1));
    if (!tmp_p) return LSUP_MEM_ERR;
    tmp_p[i] = p;
    tmp_p[i + 1] = NULL;
    po->p = tmp_p;

    i = 0;
    while (po->o[i++]);
    LSUP_Term ***tmp_o = realloc (po->o, sizeof (*po->o) * (i + 1));
    if (!tmp_o) return LSUP_MEM_ERR;
    tmp_o[i] = o;
    tmp_o[i + 1] = NULL;
    po->o = tmp_o;

    return LSUP_OK;
}


LSUP_rc
LSUP_spo_list_add_triples (
        LSUP_GraphIterator *it, LSUP_Term *s, const LSUP_PredObjList *po)
{
    LSUP_Triple *spo = LSUP_triple_new (s, NULL, NULL);
    for (size_t i = 0; po->p[i]; i++) {
        spo->p = po->p[i];
        for (size_t j = 0; po->o[i][j]; j++) {
            spo->o = po->o[i][j];

            PRCCK (LSUP_graph_add_iter (it, spo));
        }
    }
    free (spo);

    return LSUP_OK;
}


LSUP_Term *
LSUP_bnode_add_collection (LSUP_GraphIterator *it, LSUP_Term **ol)
{
    LSUP_NSMap *nsm = LSUP_graph_namespace (LSUP_graph_iter_graph (it));
    LSUP_Term
        *s = LSUP_term_new (LSUP_TERM_BNODE, NULL, NULL),
        *rdf_first = LSUP_iriref_new ("rdf:first", nsm),
        *rdf_rest = LSUP_iriref_new ("rdf:rest", nsm),
        *rdf_nil = LSUP_iriref_new ("rdf:nil", nsm),
        *link;

    LSUP_Triple *spo = TRP_DUMMY;
    link = s;
    for (size_t i = 0; ol[i]; i++) {
        spo->s = link;
        spo->p = rdf_first;
        spo->o = ol[i];
        PRCNL (LSUP_graph_add_iter (it, spo));

        spo->p = rdf_rest;
        spo->o = (
                ol[i + 1] ? LSUP_term_new (LSUP_TERM_BNODE, NULL, NULL)
                : rdf_nil);

        PRCNL (LSUP_graph_add_iter (it, spo));

        if (link != s) LSUP_term_free (link);
        // Current object becomes next subject. Irrelevant for last item.
        link = spo->o;
    }

    LSUP_term_free (rdf_first);
    LSUP_term_free (rdf_rest);
    LSUP_term_free (rdf_nil);
    free (spo);

    return s;
}


/*
 * Extern inline functions.
 */

char unescape_char (const char c);
uint8_t *uint8_dup (const uint8_t *str);
uint8_t *uint8_ndup (const uint8_t *str, size_t size);