volksdata/src/codec.c

#include "codec.h"

uint8_t *unescape_unicode (const uint8_t *esc_str, size_t size)
{
    // Output will not be longer than the escaped sequence.
    uint8_t *data = malloc (size + 1);

    size_t len = 0; // Size of output string.
    uint8_t tmp_chr[9];
    for (size_t i = 0; i < size;) {
        int esc_len; // Size of escape sequence.
        if (esc_str[i] == '\\') {
            i++; // Skip over '\\'

            if (esc_str[i] == 'u') {
                // 4-hex (2 bytes) sequence.
                esc_len = 4;
            } else if (esc_str[i] == 'U') {
                // 8-hex (4 bytes) sequence.
                esc_len = 8;
            } else {
                // Unescape other escaped characters.
                data[len++] = unescape_char (esc_str[i++]);
                continue;
            }

            // Continue encoding UTF code point.

            i ++; // Skip over 'u' / 'U'

            // Use tmp_chr to hold the hex string for the code point.
            memcpy (tmp_chr, esc_str + i, esc_len);
            tmp_chr[esc_len] = '\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 cp_len = utf8_encode (tmp_val, tmp_chr);
            if (cp_len == 0) {
                log_error ("Error encoding sequence: %s", tmp_chr);
                return NULL;
            }

            // Copy bytes into destination.
            memcpy (data + len, tmp_chr, cp_len);
#if 0
            // This can generate a LOT of output.
            if (esc_len == 4)
                log_trace ("UC byte value: %2x %2x", data[len], data[len + 1]);
            else
                log_trace (
                    "UC byte value: %2x %2x %2x %2x",
                    data[len], data[len + 1], data[len + 2], data[len + 3]
                );
#endif
            len += cp_len;
            i += esc_len;
        } else {
            data[len++] = esc_str[i++];
        }
    }

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

    return ret;
}


LSUP_Term **
LSUP_obj_list_add (LSUP_Term **ol, LSUP_Term *o)
{
    size_t i = 0;
    while (ol[i++]); // Count includes sentinel.
    LSUP_Term **ret = realloc (ol, sizeof (*ol) * (i + 1));
    if (!ret) return NULL;
    ret[i - 1] = o;
    ret[i] = NULL;

    return ret;
}


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 - 1] = p;
    tmp_p[i] = 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 - 1] = o;
    tmp_o[i] = NULL;
    po->o = tmp_o;

    return LSUP_OK;
}


size_t
LSUP_spo_list_add_triples (
        LSUP_GraphIterator *it, LSUP_Term *s, const LSUP_PredObjList *po)
{
    size_t ct = 0;
    if (!s) {
        log_error ("Subject is NULL!");
        return LSUP_VALUE_ERR;
    }
    if (!po->p) {
        log_error ("Predicate is NULL!");
        return LSUP_VALUE_ERR;
    }
    if (!po->o) {
        log_error ("Object list is NULL!");
        return LSUP_VALUE_ERR;
    }
    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];

            LSUP_rc rc = LSUP_graph_add_iter (it, spo);
            if (rc == LSUP_OK) ct++;
            PRCCK (rc);
        }
    }
    free (spo);

    return ct;
}


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);