/// <summary>
/// 计算指定文本文档中出现频率最高的字符串，
/// 保存 <see cref="IST{TKey, TValue}.Get(TKey)"/>
/// 和 <see cref="IST{TKey, TValue}.Put(TKey, TValue)"/>
/// 的调用次数以及对应的耗时。
/// </summary>
/// <param name="filename">文件名。</param>
/// <param name="minLength">字符串最小长度。</param>
/// <param name="st">用于计算的符号表。</param>
/// <param name="callIndex">调用次数。</param>
/// <param name="timeRecord">对应耗时。</param>
public static void MostFrequentlyWordAnalysis(string filename, int minLength, IST <string, int> st, out int[] callIndex, out long[] timeRecord)
{
var call = new List <int>();
var time = new List <long>();
var sw = Stopwatch.StartNew();
var callTime = 0;
int distinct = 0, words = 0;
var sr = new StreamReader(File.OpenRead(filename));
var inputs =
sr
.ReadToEnd()
.Split(new char[] { ' ', '\r', '\n' },
StringSplitOptions.RemoveEmptyEntries);
for (var i = 0; i < inputs.Length; i++)
{
if (inputs[i].Length < minLength)
{
continue;
}
words++;
if (st.Contains(inputs[i]))
{
st.Put(inputs[i], st.Get(inputs[i]) + 1);
callTime += 2;
time.Add(sw.ElapsedMilliseconds);
call.Add(callTime);
}
else
{
st.Put(inputs[i], 1);
callTime++;
time.Add(sw.ElapsedMilliseconds);
call.Add(callTime);
distinct++;
}
}
var max = "";
st.Put(max, 0);
callTime++;
time.Add(sw.ElapsedMilliseconds);
call.Add(callTime);
foreach (var s in st.Keys())
{
if (st.Get(s) > st.Get(max))
{
max = s;
}
callTime += 2;
time.Add(sw.ElapsedMilliseconds);
call.Add(callTime);
}
callIndex = call.ToArray();
timeRecord = time.ToArray();
}
/// <summary>
/// 获得指定键中出现频率最高的键。
/// </summary>
/// <typeparam name="TKey">键类型。</typeparam>
/// <param name="st">用于计算的符号表。</param>
/// <param name="keys">所有的键。</param>
/// <returns><paramref name="keys"/> 中出现频率最高的键。</returns>
public static TKey MostFrequentlyKey <TKey>(IST <TKey, int> st, TKey[] keys)
{
foreach (var s in keys)
{
if (st.Contains(s))
{
st.Put(s, st.Get(s) + 1);
}
else
{
st.Put(s, 1);
}
}
var max = keys[0];
foreach (var s in st.Keys())
{
if (st.Get(s) > st.Get(max))
{
max = s;
}
}
return(max);
}
/// <summary>
/// 获得指定文本文档中出现频率最高的所有字符串。
/// </summary>
/// <param name="filename">文件名。</param>
/// <param name="minLength">字符串最小长度。</param>
/// <param name="st">用于计算的符号表。</param>
/// <returns>文本文档出现频率最高的字符串数组。</returns>
public static string[] MostFrequentlyWords(string filename, int minLength, IST <string, int> st)
{
int distinct = 0, words = 0;
var sr = new StreamReader(File.OpenRead(filename));
var inputs =
sr
.ReadToEnd()
.Split(new char[] { ' ', '\r', '\n' },
StringSplitOptions.RemoveEmptyEntries);
foreach (var s in inputs)
{
if (s.Length < minLength)
{
continue;
}
words++;
if (st.Contains(s))
{
st.Put(s, st.Get(s) + 1);
}
else
{
st.Put(s, 1);
distinct++;
}
}
var max = "";
var queue = new Queue <string>();
st.Put(max, 0);
foreach (var s in st.Keys())
{
if (st.Get(s) > st.Get(max))
{
max = s;
queue.Clear();
queue.Enqueue(s);
}
else if (st.Get(s) == st.Get(max))
{
queue.Enqueue(s);
}
}
return(queue.ToArray());
}
/// <summary>
/// 获得指定文本文档中出现频率最高的字符串。
/// </summary>
/// <param name="filename">文件名。</param>
/// <param name="minLength">字符串最小长度。</param>
/// <param name="st">用于计算的符号表。</param>
/// <returns>文本文档出现频率最高的字符串。</returns>
public static string MostFrequentlyWord(string filename, int minLength, IST <string, int> st)
{
int distinct = 0, words = 0;
var sr = new StreamReader(File.OpenRead(filename));
var inputs =
sr
.ReadToEnd()
.Split(new char[] { ' ', '\r', '\n' },
StringSplitOptions.RemoveEmptyEntries);
var lastPut = "";
foreach (var s in inputs)
{
if (s.Length < minLength)
{
continue;
}
words++;
if (st.Contains(s))
{
lastPut = s;
st.Put(s, st.Get(s) + 1);
}
else
{
lastPut = s;
st.Put(s, 1);
distinct++;
}
}
Console.WriteLine("Last Put: " + lastPut + "\t words count: " + words);
var max = "";
st.Put(max, 0);
foreach (var s in st.Keys())
{
if (st.Get(s) > st.Get(max))
{
max = s;
}
}
return(max);
}
/// <summary>
/// 获得指定文本文档中出现频率最高的字符串。
/// </summary>
/// <param name="filename">文件名。</param>
/// <param name="counts">从文件读入的单词数目。</param>
/// <param name="minLength">字符串最小长度。</param>
/// <param name="st">用于计算的符号表。</param>
/// <returns>文本文档出现频率最高的字符串。</returns>
public static string MostFrequentlyWord(string filename, int counts, int minLength, IST <string, int> st)
{
int distinct = 0, words = 0;
var sr = new StreamReader(File.OpenRead(filename));
var inputs =
sr
.ReadToEnd()
.Split(new char[] { ' ', '\r', '\n' },
StringSplitOptions.RemoveEmptyEntries);
for (var i = 0; i < counts && i < inputs.Length; i++)
{
if (inputs[i].Length < minLength)
{
counts++;
continue;
}
words++;
if (st.Contains(inputs[i]))
{
st.Put(inputs[i], st.Get(inputs[i]) + 1);
}
else
{
st.Put(inputs[i], 1);
distinct++;
}
}
var max = "";
st.Put(max, 0);
foreach (var s in st.Keys())
{
if (st.Get(s) > st.Get(max))
{
max = s;
}
}
return(max);
}
/// <summary>
/// 对符号表进行性能测试，先 <see cref="IST{TKey, TValue}.Put(TKey, TValue)"/> <paramref name="n"/> 个字符串，
/// 再进行若干次 <see cref="IST{TKey, TValue}.Get(TKey)"/>，
/// 使得每个元素被平均访问 <paramref name="averageHit"/> 次，
/// 以及同样多的未命中访问。
/// </summary>
/// <param name="st">需要进行性能测试的符号表。</param>
/// <param name="n">需要插入符号表中的字符串数量。</param>
/// <param name="averageHit">平均每个元素被查询的次数。</param>
/// <returns>测试耗时，单位为毫秒。</returns>
public static long Performance(IST <string, int> st, int n, int averageHit)
{
var keys = GetRandomArrayString(n, 2, 50);
var keyNotExist = GetRandomString(51, 52);
var sw = Stopwatch.StartNew();
// 构建
for (var i = 0; i < n; i++)
{
st.Put(keys[i], i);
}
// 查询
for (var i = 0; i < averageHit; i++)
{
for (var j = 0; j < n; j++)
{
st.Get(keys[j]);
st.Get(keyNotExist);
}
}
sw.Stop();
return(sw.ElapsedMilliseconds);
}
static void Test(int n, IST <int, int> st)
{
var data = new int[n];
for (var i = 0; i < n; i++)
{
data[i] = i;
}
Shuffle(data);
var sw = Stopwatch.StartNew();
foreach (var item in data)
{
st.Put(item, item);
}
sw.Stop();
Console.WriteLine("Random Put " + n + ":" + sw.ElapsedMilliseconds + "ms");
Shuffle(data);
sw.Restart();
foreach (var item in data)
{
st.Get(item);
}
sw.Stop();
Console.WriteLine("Random Get " + n + ":" + sw.ElapsedMilliseconds + "ms");
Shuffle(data);
sw.Restart();
foreach (var item in data)
{
st.Delete(item);
}
sw.Stop();
Console.WriteLine("Random Delete " + n + ":" + sw.ElapsedMilliseconds + "ms");
}
