com.questdb.misc.Base64 Maven / Gradle / Ivy

Go to download
Show more of this group Show more artifacts with this name
Show all versions of questdb-core Show documentation
QuestDB is High Performance Time Series Database
There is a newer version: 3.3.3
/*******************************************************************************
 *    ___                  _   ____  ____
 *   / _ \ _   _  ___  ___| |_|  _ \| __ )
 *  | | | | | | |/ _ \/ __| __| | | |  _ \
 *  | |_| | |_| |  __/\__ \ |_| |_| | |_) |
 *   \__\_\\__,_|\___||___/\__|____/|____/
 *
 * Copyright (C) 2014-2016 Appsicle
 *
 * This program is free software: you can redistribute it and/or  modify
 * it under the terms of the GNU Affero General Public License, version 3,
 * as published by the Free Software Foundation.
 *
 * This program is distributed in the hope that it will be useful,
 * but WITHOUT ANY WARRANTY; without even the implied warranty of
 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
 * GNU Affero General Public License for more details.
 *
 * You should have received a copy of the GNU Affero General Public License
 * along with this program.  If not, see .
 *
 ******************************************************************************/

package com.questdb.misc;

public class Base64 {

    private static final byte PADDING = 127;
    private static final byte[] decodeMap = initDecodeMap();

    /**
     * @param text base64Binary data is likely to be long, and decoding requires
     *             each character to be accessed twice (once for counting length, another
     *             for decoding.)
     *             
     *             A benchmark showed that taking {@link String} is faster, presumably
     *             because JIT can inline a lot of string access (with data of 1K chars, it was twice as fast)
     * @return decoded bytes
     */
    public static byte[] parseBase64Binary(String text) {
        final int buflen = guessLength(text);
        final byte[] out = new byte[buflen];
        int o = 0;

        final int len = text.length();
        int i;

        final byte[] quadruplet = new byte[4];
        int q = 0;

        // convert each quadruplet to three bytes.
        for (i = 0; i < len; i++) {
            char ch = text.charAt(i);
            byte v = decodeMap[ch];

            if (v != -1) {
                quadruplet[q++] = v;
            }

            if (q == 4) {
                // quadruplet is now filled.
                out[o++] = (byte) ((quadruplet[0] << 2) | (quadruplet[1] >> 4));
                if (quadruplet[2] != PADDING) {
                    out[o++] = (byte) ((quadruplet[1] << 4) | (quadruplet[2] >> 2));
                }
                if (quadruplet[3] != PADDING) {
                    out[o++] = (byte) ((quadruplet[2] << 6) | (quadruplet[3]));
                }
                q = 0;
            }
        }

        if (buflen == o) // speculation worked out to be OK
        {
            return out;
        }

        // we overestimated, so need to create a new buffer
        byte[] nb = new byte[o];
        System.arraycopy(out, 0, nb, 0, o);
        return nb;
    }

    /**
     * computes the length of binary data speculatively.
     * 

     * 

     * Our requirement is to create byte[] of the exact length to store the binary data.
     * If we do this in a straight-forward way, it takes two passes over the data.
     * Experiments show that this is a non-trivial overhead (35% or so is spent on
     * the first pass in calculating the length.)
     * 

     * 

     * So the approach here is that we compute the length speculatively, without looking
     * at the whole contents. The obtained speculative value is never less than the
     * actual length of the binary data, but it may be bigger. So if the speculation
     * goes wrong, we'll pay the cost of reallocation and buffer copying.
     * 

     * 
     * If the base64 text is tightly packed with no indentation nor illegal char
     * (like what most web services produce), then the speculation of this method
     * will be correct, so we get the performance benefit.
     */
    private static int guessLength(String text) {
        final int len = text.length();

        // compute the tail '=' chars
        int j = len - 1;
        for (; j > -1; j--) {
            byte code = decodeMap[text.charAt(j)];
            if (code != PADDING) {
                if (code == -1) // most likely this base64 text is indented. go with the upper bound
                {
                    return text.length() / 4 * 3;
                }
                break;
            }
        }

        j++;    // text.charAt(j) is now at some base64 char, so +1 to make it the size
        int padSize = len - j;
        if (padSize > 2) // something is wrong with base64. be safe and go with the upper bound
        {
            return text.length() / 4 * 3;
        }

        // so far this base64 looks like it's unindented tightly packed base64.
        // take a chance and create an array with the expected size
        return text.length() / 4 * 3 - padSize;
    }

    private static byte[] initDecodeMap() {
        byte[] map = new byte[128];
        int i;
        for (i = 0; i < 128; i++) {
            map[i] = -1;
        }

        for (i = 'A'; i <= 'Z'; i++) {
            map[i] = (byte) (i - 'A');
        }
        for (i = 'a'; i <= 'z'; i++) {
            map[i] = (byte) (i - 'a' + 26);
        }
        for (i = '0'; i <= '9'; i++) {
            map[i] = (byte) (i - '0' + 52);
        }
        map['+'] = 62;
        map['/'] = 63;
        map['='] = PADDING;

        return map;
    }
}