Function cbm_ac_scan_batch

internal/cbm/ac.c:367–410

cbm_ac_scan_batch scans multiple NUL-separated names through the automaton. For each name, reports all unique matched pattern IDs. Returns total number of matches written to out_matches. Parameters: ac — automaton names_buf — concatenated names separated by NUL bytes name_offsets — start offset of each name in names_buf name_lengths — length of each name num_names — number of name

//   out_matches  — output buffer for (name_index, pattern_id) pairs
//   max_matches  — capacity of out_matches
int cbm_ac_scan_batch(const CBMAutomaton *ac, const char *names_buf, const int *name_offsets,
                      const int *name_lengths, int num_names, CBMMatchResult *out_matches,
                      int max_matches) {
    int total = 0;
    const int alpha_size = ac->alpha_size;
    const int *go_table = ac->go_table;

    for (int n = 0; n < num_names && total < max_matches; n++) {
        const char *text = names_buf + name_offsets[n];
        int text_len = name_lengths[n];
        int state = 0;

        // Track which patterns matched for this name (deduplicate).
        uint64_t seen = 0;

        for (int i = 0; i < text_len; i++) {
            int c = ac->alpha_map[(unsigned char)text[i]];
            state = go_table[(state * alpha_size) + c];

            // Walk output chain for >64 patterns.
            int s = state;
            while (s > 0 && total < max_matches) {
                // Bitmask fast path for first 64 patterns.
                uint64_t bits = ac->output[s] & ~seen;
                while (bits && total < max_matches) {
                    int pid = __builtin_ctzll(bits);
                    out_matches[total].name_index = n;
                    out_matches[total].pattern_id = pid;
                    total++;
                    seen |= CBM_AC_PATTERN_BIT(pid);
                    bits = CBM_AC_CLEAR_LOW_BIT(bits);
                }

                // Follow output_next for patterns beyond bitmask range.
                int next_state = ac->output_next[s];
                if (next_state == CBM_AC_NO_STATE || next_state == s) {
                    break;
                }
                s = next_state;
            }
        }
    }
    return total;
}

// ─── Info ──────────────────────────────────────────────────────────────────