Clickhouse通过JDBC直接查询Bitmap

ClickHouseStatement statement = null;
            try {
                statement = connection.createStatement();
                ClickHouseRowBinaryInputStream in = statement.executeQueryClickhouseRowBinaryStream(sql);
                ClickHouseBitmap bit = in.readBitmap(ClickHouseDataType.UInt64);
                Roaring64NavigableMap obj = (Roaring64NavigableMap) bit.unwrap();
                return new ExtRoaringBitmap(obj);
            } catch (Exception e) {
                log.error("查询位图错误 [ip:{}][sql:{}]", ip, sql, e);
                throw e;
            } finally {
                DbUtils.close(statement);
            }

上面的方法直接读取Bitmap会大量占用应用内存，怎么进行优化呢？
我们可以通过Clickhouse把Bitmap转成列，通过流式读取bitmap里的offset，在应用里创建Bitmap

private static String SQL_WRAP = "SELECT bitmap_arr AS offset FROM (SELECT bitmapToArray(aplus_bitmap) AS bitmap_arr  FROM (SELECT ({}) AS aplus_bitmap LIMIT 1)) ARRAY JOIN bitmap_arr";

    public static Roaring64NavigableMap getBitmap(Connection conn, String sql){
        Statement statement = null;
        ResultSet rs = null;
        try {
            statement = conn.createStatement(
                    ResultSet.TYPE_FORWARD_ONLY,
                    ResultSet.CONCUR_READ_ONLY);

            String exeSql = ATool.format(SQL_WRAP, sql);
            rs = statement.executeQuery(exeSql);

            Roaring64NavigableMap bitmap = new Roaring64NavigableMap();
            while (rs.next()) {
                Long offset = rs.getLong(1);
                if(offset != null){
                    bitmap.add(offset);
                }
            }
            return bitmap;
        } catch (Exception e){
            log.error("从CK读取bitmap错误, SQL:{}", sql, e);
        } finally {
            ATool.close(rs, statement);
        }
        return null;
    }

这里我们在创建Statement时设置了ResultSet.TYPE_FORWARD_ONLY这个参数，在从ResultSet（结果集）中读取记录的时，对于访问过的记录就自动释放了内存。
进一步减少了内存的使用。