Redis实现AutoComplete

使用redis的有序集合实现前缀的autocomplete。
1.首先把数据存入redis有续集，
原文件格式如下：

存入redis中格式，完整的名字有*结尾：

11609) "y"
11610) "ya"
11611) "yal"
11612) "yalo"
11613) "yalon"
11614) "yalond"
11615) "yalonda*"
11616) "yas"
11617) "yasm"
11618) "yasme"
11619) "yasmee"
11620) "yasmeen*"
11621) "yasmi"
11622) "yasmin*"
11623) "ye"
11624) "yel"
11625) "yele"
11626) "yelen"
11627) "yelena*"
11628) "yet"
11629) "yett"
11630) "yetta*"
11631) "yetti"
11632) "yettie*"
11633) "yetty*"

2.根据输入的前缀，设置开始查找的起始位置

127.0.0.1:6379> zrank test y
(integer) 11608
127.0.0.1:6379> 

如y开头，从11608开始查找。

3.从起始位置分批（循环）向后查询（zrange），将符合前缀且*结尾的词放入返回列表，直至当列表size大于所需要的数量（maxNeeded）或不符合所输入的前缀时，返回列表。
用jedis实现：

package com.zzhblh.sample;

import redis.clients.jedis.Jedis;
import redis.clients.jedis.JedisPool;
import redis.clients.jedis.JedisPoolConfig;
import redis.clients.jedis.Pipeline;
import java.io.BufferedReader;
import java.io.IOException;
import java.io.InputStreamReader;
import java.net.URL;
import java.util.ArrayList;
import java.util.Collections;
import java.util.List;
import java.util.Set;

public class testJedisAutocomplete {

    private final Jedis redis;
    private final Pipeline p;
    private final String redisKey;
    private long start,end;

    testJedisAutocomplete(final Jedis redis, final Pipeline p, final String redisKey, final URL dictionaryUrl, final boolean populate,
                 final boolean clear) throws IOException {
        this.redis = redis;
        this.p = p;
        this.redisKey = redisKey;

        if(clear) {
            // Truncate existing Dictionary
            clearDictionary();
        }
        if(populate) {
            start = System.currentTimeMillis();
            // Load the entries from dictionaryUrl
            loadDictionary(dictionaryUrl);
            end = System.currentTimeMillis();
            System.out.println("add with pipeline used [" + (end - start) / 1000 + "] seconds ..");
            p.close();
        }
    }

    public List complete(final String prefix, final int count) {
        // 1. Find the position of the prefix in the sorted set in Redis
        if(null == prefix) {
            return Collections.emptyList();
        }
        int prefixLength = prefix.length();
        int start = redis.zrank(redisKey, prefix).intValue();
        if(start < 0 || prefixLength == 0) {
            return Collections.emptyList();
        }
        List results = new ArrayList();
        int found = 0, rangeLength = 50, maxNeeded = count;
        while(found < maxNeeded) {
            Set rangeResults = redis.zrange(redisKey, start, start + rangeLength - 1);
            start += rangeLength;
            if(rangeResults.isEmpty()) {
                break;
            }
            for(final String entry : rangeResults) {
                int minLength = Math.min(entry.length(), prefixLength);
                if(!entry.substring(0, minLength).equalsIgnoreCase(prefix.substring(0, minLength))) {
                    return results;
                }
                if(entry.endsWith("*") && found < maxNeeded) {
                    results.add(entry.substring(0, entry.length() - 1));
                    found = results.size();
                }
            }
        }
        return results;
    }

    private void clearDictionary() {
        redis.del(redisKey);
    }


    private void loadDictionary(final URL dictionaryUrl) throws IOException {
        InputStreamReader inputStreamReader = new InputStreamReader(dictionaryUrl.openStream());
        BufferedReader reader = new BufferedReader(inputStreamReader);
        String word;
        while((word = reader.readLine()) != null) {
            word = word.trim();
            // Add the word if the word does not start with #
            if(!word.isEmpty() && !word.startsWith("#")) {
                addWord(word);
            }
        }
        reader.close();
        inputStreamReader.close();
    }

    private void addWord(final String word) {
        // Add all the possible prefixes of the given word and also the given
        // word with a * suffix.
        p.zadd(redisKey, 0, word + "*");
        for(int index = 1, total = word.length(); index < total; index++) {
            p.zadd(redisKey, 0, word.substring(0, index));
        }

    }

    public static class Builder {

        private final Jedis redis;
        private final Pipeline p;
        private final String redisKey;
        private final URL dictionaryUrl;
        private boolean clear = true;
        private boolean populate = true;

        public Builder(final Jedis redis, final  Pipeline p, final String redisKey, final URL dictionaryUrl) {
            this.redis = redis;
            this.p = p;
            this.redisKey = redisKey;
            this.dictionaryUrl = dictionaryUrl;
        }

        public Builder clear(final boolean clear) {
            this.clear = clear;
            this.populate = true;
            return this;
        }

        public Builder populate(final boolean populate) {
            this.populate = populate;
            return this;
        }

        public testJedisAutocomplete build() throws IOException {
            return new testJedisAutocomplete(redis, p, redisKey, dictionaryUrl, populate, clear);
        }

    }

    public static void main(final String[] args) throws Exception {
        JedisPoolConfig config = new JedisPoolConfig();
        JedisPool pool = new JedisPool(config, "107.170.248.158", 6379);
        Jedis redis = pool.getResource();
        redis.auth("123");
        Pipeline p = redis.pipelined();
        URL url = new URL("http://antirez.com/misc/female-names.txt");

        testJedisAutocomplete autoComplete = new testJedisAutocomplete.Builder(redis, p, "test", url).clear(true).populate(true).build();
        System.out.println(autoComplete.complete("y", 80));
    }

}