复合词是其中包含2个或更多词的词。但是，我们可以做得更好。我们需要您创建一个包含所有单词的1（无意义）单词。

但是，我们希望这个词尽可能短。我们可以使用重叠的字母来实现此目的。

例如，如果您的单词列表为["cat", "atom", "a"]，则需要返回"catom"。

输入输出

您的程序需要将单词列表作为输入，并返回复合单词作为输出。

根据Google的说法，您将使用的单词列表是英语中排名前10000的单词（如果发现这个列表太简单了，我可以将其更改为更长的单词）。作为参考，只需在每个单词后面附加一个65888的分数即可。

您的分数是您最后一个单词中的字母数，越低越好。抢领带的人去了第一张海报。

C ++，最终字长：38272

（优化版本大约花费了20分钟）

#include <iostream>
#include <string>
#include <vector>

std::size_t calcOverlap(const std::string &a, const std::string &b, std::size_t limit, std::size_t minimal) {
    std::size_t la = a.size();
    for(std::size_t p = std::min(std::min(la, b.size()), limit + 1); -- p > minimal; ) {
        if(a.compare(la - p, p, b, 0, p) == 0) {
            return p;
        }
    }
    return 0;
}

int main() {
    std::vector<std::string> words;

    // Load all words from input
    while(true) {
        std::string word;
        std::getline(std::cin, word);
        if(word.empty()) {
            break;
        }
        words.push_back(word);
    }

    std::cerr
        << "Input word count: " << words.size() << std::endl;

    // Remove all fully subsumed words

    for(auto p = words.begin(); p != words.end(); ) {
        bool subsumed = false;
        for(auto i = words.begin(); i != words.end(); ++ i) {
            if(i == p) {
                continue;
            }
            if(i->find(*p) != std::string::npos) {
                subsumed = true;
                break;
            }
        }
        if(subsumed) {
            p = words.erase(p);
        } else {
            ++ p;
        }
    }

    std::cerr
        << "After subsuming checks: " << words.size()
        << std::endl;

    // Sort words longest-to-shortest (not necessary but doesn't hurt. Makes finding maxlen a tiny bit easier)
    std::sort(words.begin(), words.end(), [](const std::string &a, const std::string &b) {
        return a.size() > b.size();
    });

    std::size_t maxlen = words.front().size();

    // Repeatedly combine most-compatible words until there is only one left
    std::size_t bestPossible = maxlen - 1;
    while(words.size() > 1) {
        auto bestA = words.begin();
        auto bestB = -- words.end();
        std::size_t bestOverlap = 0;
        for(auto p = ++ words.begin(), e = words.end(); p != e; ++ p) {
            if(p->size() - 1 <= bestOverlap) {
                continue;
            }
            for(auto q = words.begin(); q != p; ++ q) {
                std::size_t overlap = calcOverlap(*p, *q, bestPossible, bestOverlap);
                if(overlap > bestOverlap) {
                    bestA = p;
                    bestB = q;
                    bestOverlap = overlap;
                }
                overlap = calcOverlap(*q, *p, bestPossible, bestOverlap);
                if(overlap > bestOverlap) {
                    bestA = q;
                    bestB = p;
                    bestOverlap = overlap;
                }
            }
            if(bestOverlap == bestPossible) {
                break;
            }
        }
        std::string newStr = std::move(*bestA);
        newStr.append(*bestB, bestOverlap, std::string::npos);

        if(bestA == -- words.end()) {
            words.pop_back();
            *bestB = std::move(words.back());
            words.pop_back();
        } else {
            *bestB = std::move(words.back());
            words.pop_back();
            *bestA = std::move(words.back());
            words.pop_back();
        }

        // Remove any words which are now in the result
        for(auto p = words.begin(); p != words.end(); ) {
            if(newStr.find(*p) != std::string::npos) {
                std::cerr << "Now subsumes: " << *p << std::endl;
                p = words.erase(p);
            } else {
                ++ p;
            }
        }

        std::cerr
            << "Words remaining: " << (words.size() + 1)
            << " Latest combination: (" << bestOverlap << ") " << newStr
            << std::endl;

        words.push_back(std::move(newStr));
        bestPossible = bestOverlap; // Merging existing words will never make longer merges possible
    }

    std::string result = words.front();

    std::cout
        << result
        << std::endl;
    std::cerr
        << "Word size: " << result.size()
        << std::endl;
    return 0;
}

验证bash一线：

cat commonwords.txt | while read p; do grep "$p" merged.txt >/dev/null || echo "Not found: $p"; done

它还产生了一些非常酷的进行中的单词。这是我的一些最爱：

• 聚酯日子（合成怀旧）
• 阿富汗（[您不喜欢的政客插入]会说的话）
• Togethernet （友好的互联网）
• Elephantom （大鬼魂）
• 防雷（如“我不知道为什么他们觉得有必要使它防雷，但这让我感到紧张”）

和：

• codecribingo （也许这个网站即将面临挑战？）

最终输出在此处的pastebin上：http : //pastebin.com/j3qYb65b

C ++ 11，38272个字母，证明是最佳的

保证该算法可以提供解决方案的下限。在这种情况下，它可以达到下限并输出最佳的38272字母解决方案。（这与Dave贪婪算法找到的解决方案相匹配。我很惊讶，也为发现它是最佳选择而感到有些失望，但是，确实如此。）

它通过解决如下构建的网络上的最小成本流问题来工作。

• 首先，换句话说，任何单词都是多余的；丢弃它们。
• 对于每个单词w，绘制两个节点w _0和w _1，其中w _0是容量为1的源，而w _1是容量为1的宿。
• 对于任何单词的每个（严格）前缀或后缀a，请绘制一个节点a。
• 对于w的每个后缀a，从w _0到a的容量为1且成本为0 的圆弧。
• 对于w的每个前缀a，从a到w _1 画一条弧，容量为1，成本为length（w）-length（a）。

包含每个单词的长度为n的任何字符串都可以转换为该网络上的流，成本最多为n。因此，此网络上的最小成本流是最短此类字符串的长度的下限。

如果我们很幸运（在这种情况下，我们是幸运的），那么在将进入w _1 的流重定向回w _0之后，我们将找到一个最佳流，该流只有一个连接的组件，并且通过该节点进入空串。如果是这样，它将包含一条从此处开始和结束的欧拉回路。这样的欧拉电路可以作为最佳长度的字符串读出。

如果我们不走运，请在其他连接的组件的空字符串和最短字符串之间添加一些额外的弧，以确保存在欧拉回路。在那种情况下，字符串将不再是最佳的。

我将LEMON库用于其最小成本流和欧拉电路算法。（这是我第一次使用该库，给我留下了深刻的印象-我一定会再次将其用于将来的图形算法需求。）LEMON带有四种不同的最小成本流算法；你可以在这里与试戴--net，--cost，--cap，和--cycle（默认值）。

程序在0.5秒内运行，生成此输出字符串。

#include <iostream>
#include <string>
#include <unordered_map>
#include <unordered_set>
#include <vector>
#include <lemon/core.h>
#include <lemon/connectivity.h>
#include <lemon/euler.h>
#include <lemon/maps.h>
#include <lemon/list_graph.h>
#include <lemon/network_simplex.h>
#include <lemon/cost_scaling.h>
#include <lemon/capacity_scaling.h>
#include <lemon/cycle_canceling.h>

using namespace std;

typedef lemon::ListDigraph G;

struct Word {
    G::Node suffix, prefix;
    G::Node tour_node;
};

struct Edge {
    unordered_map<string, Word>::iterator w;
    G::Arc arc;
};

struct Affix {
    vector<Edge> suffix, prefix;
    G::Node node;
    G::Node tour_node;
};

template<class MCF>
bool solve(const G &net, const G::ArcMap<int> &lowerMap, const G::ArcMap<int> &upperMap, const G::ArcMap<int> &costMap, const G::NodeMap<int> &supplyMap, int &totalCost, G::ArcMap<int> &flowMap)
{
    MCF mcf(net);
    if (mcf.lowerMap(lowerMap).upperMap(upperMap).costMap(costMap).supplyMap(supplyMap).run() != mcf.OPTIMAL)
        return false;
    totalCost = mcf.totalCost();
    mcf.flowMap(flowMap);
    return true;
}

int main(int argc, char **argv)
{
    clog << "Reading dictionary from stdin" << endl;
    unordered_map<string, Affix> affixes;
    unordered_map<string, Word> words;
    unordered_set<string> subwords;
    G net, tour;
    G::ArcMap<int> lowerMap(net), upperMap(net), costMap(net);
    G::NodeMap<int> supplyMap(net);
    string new_word;
    while (getline(cin, new_word)) {
        if (subwords.find(new_word) != subwords.end())
            continue;
        for (auto i = new_word.begin(); i != new_word.end(); ++i) {
            for (auto j = new_word.end(); j != i; --j) {
                string s(i, j);
                words.erase(s);
                subwords.insert(s);
            }
        }
        words.emplace(new_word, Word());
    }
    for (auto w = words.begin(); w != words.end(); ++w) {
        w->second.suffix = net.addNode();
        supplyMap.set(w->second.suffix, 1);
        w->second.prefix = net.addNode();
        supplyMap.set(w->second.prefix, -1);
        for (auto i = w->first.begin(); ; ++i) {
            affixes.emplace(string(w->first.begin(), i), Affix()).first->second.prefix.push_back(Edge {w});
            affixes.emplace(string(i, w->first.end()), Affix()).first->second.suffix.push_back(Edge {w});
            if (i == w->first.end())
                break;
        }
        w->second.tour_node = tour.addNode();
    }
    for (auto a = affixes.begin(); a != affixes.end();) {
        if (a->second.suffix.empty() || a->second.prefix.empty() ||
            (a->second.suffix.size() == 1 && a->second.prefix.size() == 1 &&
             a->second.suffix.begin()->w == a->second.prefix.begin()->w)) {
            affixes.erase(a++);
        } else {
            a->second.node = net.addNode();
            supplyMap.set(a->second.node, 0);
            for (auto &e : a->second.suffix) {
                e.arc = net.addArc(e.w->second.suffix, a->second.node);
                lowerMap.set(e.arc, 0);
                upperMap.set(e.arc, 1);
                costMap.set(e.arc, 0);
            }
            for (auto &e : a->second.prefix) {
                e.arc = net.addArc(a->second.node, e.w->second.prefix);
                lowerMap.set(e.arc, 0);
                upperMap.set(e.arc, 1);
                costMap.set(e.arc, e.w->first.length() - a->first.length());
            }
            a->second.tour_node = lemon::INVALID;
            ++a;
        }
    }

    clog << "Read " << words.size() << " words and found " << affixes.size() << " affixes; ";
    clog << "created network with " << countNodes(net) << " nodes and " << countArcs(net) << " arcs" << endl;

    int totalCost;
    G::ArcMap<int> flowMap(net);
    bool solved;
    if (argc > 1 && string(argv[1]) == "--net") {
        clog << "Using network simplex algorithm" << endl;
        solved = solve<lemon::NetworkSimplex<G>>(net, lowerMap, upperMap, costMap, supplyMap, totalCost, flowMap);
    } else if (argc > 1 && string(argv[1]) == "--cost") {
        clog << "Using cost scaling algorithm" << endl;
        solved = solve<lemon::CostScaling<G>>(net, lowerMap, upperMap, costMap, supplyMap, totalCost, flowMap);
    } else if (argc > 1 && string(argv[1]) == "--cap") {
        clog << "Using capacity scaling algorithm" << endl;
        solved = solve<lemon::CapacityScaling<G>>(net, lowerMap, upperMap, costMap, supplyMap, totalCost, flowMap);
    } else if ((argc > 1 && string(argv[1]) == "--cycle") || true) {
        clog << "Using cycle canceling algorithm" << endl;
        solved = solve<lemon::CycleCanceling<G>>(net, lowerMap, upperMap, costMap, supplyMap, totalCost, flowMap);
    }

    if (!solved) {
        clog << "error: no solution found" << endl;
        return 1;
    }
    clog << "Lower bound: " << totalCost << endl;

    G::ArcMap<string> arcLabel(tour);
    G::Node empty = tour.addNode();
    affixes.find("")->second.tour_node = empty;
    for (auto &a : affixes) {
        for (auto &e : a.second.suffix) {
            if (flowMap[e.arc]) {
                if (a.second.tour_node == lemon::INVALID)
                    a.second.tour_node = tour.addNode();
                arcLabel.set(tour.addArc(e.w->second.tour_node, a.second.tour_node), "");
            }
        }
        for (auto &e : a.second.prefix) {
            if (flowMap[e.arc]) {
                if (a.second.tour_node == lemon::INVALID)
                    a.second.tour_node = tour.addNode();
                arcLabel.set(tour.addArc(a.second.tour_node, e.w->second.tour_node), e.w->first.substr(a.first.length()));
            }
        }
    }

    clog << "Created tour graph with " << countNodes(tour) << " nodes and " << countArcs(tour) << " arcs" << endl;

    G::NodeMap<int> compMap(tour);
    int components = lemon::stronglyConnectedComponents(tour, compMap);
    if (components != 1) {
        vector<unordered_map<string, Affix>::iterator> breaks(components, affixes.end());
        for (auto a = affixes.begin(); a != affixes.end(); ++a) {
            if (a->second.tour_node == lemon::INVALID)
                continue;
            int c = compMap[a->second.tour_node];
            if (c == compMap[empty])
                continue;
            auto &b = breaks[compMap[a->second.tour_node]];
            if (b == affixes.end() || b->first.length() > a->first.length())
                b = a;
        }
        int offset = 0;
        for (auto &b : breaks) {
            if (b != affixes.end()) {
                arcLabel.set(tour.addArc(empty, b->second.tour_node), b->first);
                arcLabel.set(tour.addArc(b->second.tour_node, empty), "");
                offset += b->first.length();
            }
        }
        clog << "warning: Found " << components << " components; solution may be suboptimal by up to " << offset << " letters" << endl;
    }

    if (!lemon::eulerian(tour)) {
        clog << "error: failed to make tour graph Eulerian" << endl;
        return 1;
    }

    for (lemon::DiEulerIt<G> e(tour, empty); e != lemon::INVALID; ++e)
        cout << arcLabel[e];
    cout << endl;

    return 0;
}

Java 8，〜5分钟，长度为39,279

import java.io.BufferedReader;
import java.io.File;
import java.io.FileReader;
import java.util.ArrayList;
import java.util.HashSet;
import java.util.List;
import java.util.Optional;
import java.util.Set;
import java.util.stream.Collectors;

public class Words {

    public static void main(String[] args) throws Throwable {
        File file = new File("words.txt");
        List<String> wordsList = new ArrayList<>();
        BufferedReader reader = new BufferedReader(new FileReader(file));
        String line;
        while ((line = reader.readLine()) != null) {
            wordsList.add(line);
        }
        reader.close();

        Set<String> words = new HashSet<>();

        System.out.println("Finished I/O");

        for (int i = 0; i < wordsList.size(); i++) { //Step 1: remove any words that occur in other words
            boolean in = false;
            for (int j = 0; j < wordsList.size(); j++) {
                if (i != j && wordsList.get(j).contains(wordsList.get(i))) {
                    in = true;
                    break;
                }
            }
            if (!in) {
                words.add(wordsList.get(i));
            }
        }

        System.out.println("Removed direct containers");

        List<String> working = words.stream().sorted((c, b) -> Integer.compare(c.length(), b.length())).collect(Collectors.toList()); //Sort by length, shortest first
        StringBuilder result = new StringBuilder();
        result.append(working.get(0));
        while (!working.isEmpty()) {
            Optional<String> target = working.stream().sorted((c, b) -> Integer.compare(firstLastCommonality(result.toString(), b), firstLastCommonality(result.toString(), c))).findFirst(); //Find the string that has the greatest in common with the end of 'result'
            if(target.isPresent()) { //It really should be present, but just in case
                String s = target.get();
                working.remove(s);
                int commonality = firstLastCommonality(result.toString(), s);
                s = s.substring(commonality);
                result.append(s);
            }
        }

        System.out.println("Finished algorithm");

        String r = result.toString();
        System.out.println("The string: \n" + r);
        System.out.println("Length: \n" + r.length());
        System.out.println("Verified: \n" + !wordsList.stream().filter(s -> !r.contains(s)).findFirst().isPresent());
    }

    private static int firstLastCommonality(String a, String b) {
        int count = 0;
        int len = b.length();
        while (!a.endsWith(b) && !b.equals("")) {
            b = cutLastChar(b);
            count++;
        }
        return len - count;
    }

    private static String cutLastChar(String string) {
        if (string.length() - 1 < 0) {
            return string;
        } else {
            return string.substring(0, string.length() - 1);
        }
    }

}

输入：

• 工作目录中名为“ words.txt”的文件，格式与主帖子中的文件完全相同

输出：

Finished I/O
Removed direct containers
Finished algorithm
The string: 
[Moved to pastebin](http://pastebin.com/iygyR3zL)
Length: 
39279
Verified: 
true

Python 2，39254个字符

在我的计算机上运行需要1-2分钟，通过获取最长的单词然后始终将单词添加到具有最多共同字符串的结果字符串中来工作。（在此之前，应删除作为其他单词的子字符串的所有单词，以防止不必要的添加到字符串中。）

更新：试图从两个方向看，但这并没有任何改善。（也许正在使用可以在以后更好用的单词？）

链接到pastebin上的单词。

前100个字符：

telecommunicationsawayneillegallynnbabyenbcjrxltdxmlbsrcwvtxxxboxespnycdsriconsentencessexyrsslipodc

码：

import re
import urllib

def suffix_dist(w1,w2):
    for i in range(min(map(len,[w1,w2])))[::-1]:
        if w1[-i:]==w2[:i]:
            return i
    return 0

url="https://raw.githubusercontent.com/first20hours/google-10000-english/master/google-10000-english.txt"
s=urllib.urlopen(url).read()
words=s.split()
words.sort(key=len,reverse=True)

## remove words that are substrings of other words anyway
for i in range(len(words))[::-1]:
    if any(words[i] in w for w in words[:i]):
        words.pop(i)

print len(words)

words.sort(key=len)
w1=words.pop(-1)
result=w1
while words:
    ## get the word with longest shared suffix/prefix
    w2=max(words,key=lambda x:suffix_dist(w1,x))
    print w2
    words.pop(words.index(w2))
    if w2 in result:
        break
    result+=w2[suffix_dist(w1,w2):]
    w1=w2


print result[:100]
print len(result)
print "Test:", all(w in result for w in s.split())

Ruby，共39222个字符

在他的Python答案中使用与@KarlKastor类似的方法，但起始字符串是最小的单词之一，而不是最大的单词。另一个优化（我不知道它有多大帮助）是在每次添加之间，它会修剪由于重叠的单词而可能已包含在字符串中的所有单词。

在我的计算机上运行了4分钟多一点，没有计算Web请求检索单词列表的时间，但不是4:20。

Pastebin上的字。

require 'net/http'

puts "Obtaining word list..."
data = Net::HTTP.get(URI'https://raw.githubusercontent.com/first20hours/google-10000-english/master/google-10000-english.txt')
puts "Word list obtained!"

puts "Starting calculation..."
time = Time.now

words = data.split.sort_by(&:size)
words.reject!{|w| words.find{|x| w!=x && x.include?(w)}}

string = words.shift

def merge_dist(prefix, suffix)
    [prefix.size,suffix.size].min.downto(0).find{|i| prefix.end_with?(suffix[0,i])}
end

while words.length > 0
    suffix = words.max_by{|w| merge_dist string, w}
    string += suffix[merge_dist(string, suffix)..-1]
    words.reject!{|w| string.include? w}
end

delta = Time.now - time

puts "Calculation completed in #{delta} seconds!"
puts "Word is #{string.size} chars long."

open("word.txt", 'w') << string

puts "Word saved to word.txt"

PowerShell v2 +，46152个字符

param([System.Collections.ArrayList]$n)$n=$n|sort length -des;while($n){$x=$n.Count;$o+=$n[0];0..$x|%{if($o.IndexOf($n[$_])-ge0){$n.RemoveAt($_)}}}$o

将输入作为列表，将其强制转换为ArrayList（以便我们进行操作）。我们sort也通过length在-descending顺序。然后，while我们的输入数组中仍然有单词，进行循环。每次迭代时，将helper $x设置为等于我们剩下的数量，$o然后将列表中的下一项添加到输出中，然后遍历列表中的所有内容。如果a .IndexOf不等于-1（即，该单词在中找到$o），我们将从剩余单词列表中删除该单词。最后，最后输出$o。

我无权使用Pastebin或类似工具，因此这是临时字词的开头和结尾- telecommunicationscharacterizationresponsibilitiessublimedirectory...fcmxvtwvfxwujmjsuhjjrxjdbkdxqc。我想这已经减少了大约20,000个字符的输入，所以我认为还不错。

我正在努力改进。

PHP 46612字符

这只是一个开始。我希望改善它。到目前为止，我所做的只是删除作为另一个单词的子字符串的任何单词。我正在处理阵列的3个副本，但内存似乎不是问题。

<?php
set_time_limit(3000);

$words=file('https://raw.githubusercontent.com/first20hours/google-10000-english/master/google-10000-english.txt');
$words = array_map('trim', $words);

usort($words, function ($a, $b)
{
    if (strlen($a) == strlen($b) ){
        return 0;
    }
    return ( strlen($a) < strlen($b) )? -1 : 1;
});

$final_array=$words;
$comparison_array=$words;


  foreach($words as $key => $word){
    echo $word."<br>\n";
      foreach($comparison_array as $nestedKey => $nestedWord){
          if (strlen($nestedWord) <= strlen($word)) {
            unset($comparison_array[$nestedKey]);
            continue;
          }
          if( strpos($nestedWord,$word) !== FALSE ){
              unset($final_array[$key]);
              $restart=true;
              break;
          } 
      }    
  }


sort($final_array);
$compound='';
$compound = implode($final_array);
echo $compound;
echo "  <br><br>\n\n". strlen($compound);