寫過.Net或是Java程式的開發人員,或多或少都曾聽過這些程式語言在處理字串時,底層會有個名為String pool的機制,幫我們自動重用已經建立的字串實體,減少記憶體的耗費。

String pool簡單來說就是一個HashTable,其Key值是字串內容,Value是物件實體的位置。其內容值會在程式編譯時期做初始設定,將程式碼中用到的字串加入。因此不同字串變數所存放的靜態字串,若是是一樣的字串,字串會在編譯時期加入String pool,透過String pool的協助兩個變數會指到相同的物件實體。 string str1 = “abcd1234”; string str2 = “abcd1234”; string str3 = str1; string str4 = “abcd” + “1234”;

        Console.WriteLine(string.Format("ReferenceEquals(str1, str2) = {0}", ReferenceEquals(str1, str2)));
        Console.WriteLine(string.Format("ReferenceEquals(str1, str3) = {0}", ReferenceEquals(str1, str3)));
        Console.WriteLine(string.Format("ReferenceEquals(str1, str4) = {0}", ReferenceEquals(str1, str4))); 

若是動態產生的字串,因編譯階段並無法確定其字串為何,故無法在編譯時期將其加入String pool,因此就算字串內容相同也會指派不同的物件實體。

        string temp = "aaaa";
        string str1 = "aaaaaaaa";
        string str2 = temp + "aaaa";
        string str3 = new string('a', 8);

        Console.WriteLine(string.Format("ReferenceEquals(str1, str2) = {0}", ReferenceEquals(str1, str2)));
        Console.WriteLine(string.Format("ReferenceEquals(str1, str3) = {0}", ReferenceEquals(str1, str3))); 

此時我們可視需求使用String.Intern靜態方法將其加入String pool。該方法會回傳String pool中對應的物件實體,若字串不存在於String pool中,會將其加入String pool再回傳。

        string temp = "aaaa";
        string str1 = "aaaaaaaa";
        string str2 = temp + "aaaa";
        string str3 = new string('a', 8);

        str2 = string.Intern(str2);
        str3 = string.Intern(str3);

        Console.WriteLine(string.Format("ReferenceEquals(str1, str2) = {0}", ReferenceEquals(str1, str2)));
        Console.WriteLine(string.Format("ReferenceEquals(str1, str3) = {0}", ReferenceEquals(str1, str3)));

若有需要也可透過String.IsInterned靜態方法判別字串是否已加入String pool。若字串存在於String pool中,會回傳對應的物件實體,反之回傳null。

        string str1 = new string('a', 8);
        Console.WriteLine(string.Format(@"String.IsInterned(str1) = {0}", String.IsInterned(str1) != null));

        str1 = string.Intern(str1);
        
        Console.WriteLine(string.Format(@"String.IsInterned(str1) = {0}", String.IsInterned(str1) != null));

那對於我們開發人員來說,理解了String Pool有甚麼用呢?理解String Pool能清楚的掌握到字串是否是指向同一物件參考,這在做字串比對動作時就是一個不錯的使用時機,因為一般的字串比對底層是以Byte為基礎的方式去比對,當比對的字串很長時,整個處理效能就跟著低落,若是可以善用String pool,我們只需比對兩者是否指到相同的物件實體就可以了,可以獲得較佳的效能。

    static void Main(string[] args)
    {
        Test(1000000000, 10000000);
    }

    private static void Test(int testCount, int stringLength)
    {
        NormalCompare1(string.Empty, string.Empty);
        NormalCompare2(string.Empty, string.Empty);
        ReferenceCompare(string.Empty, string.Empty);

        string str1 = new string('a', stringLength);
        string str2 = str1;

        Console.WriteLine("testCount = " + testCount.ToString());
        Console.WriteLine("stringLength = " + stringLength.ToString());

        Console.WriteLine("NormalCompare1...");
        Stopwatch sw = Stopwatch.StartNew();
        for (int i = 0; i < testCount; i++)
        {
            NormalCompare1(str1, str2);
        }
        Console.WriteLine("Elapsed: " + sw.ElapsedMilliseconds.ToString());

        Console.WriteLine("NormalCompare2...");
        sw.Restart();
        for (int i = 0; i < testCount; i++)
        {
            NormalCompare2(str1, str2);
        }
        Console.WriteLine("Elapsed: " + sw.ElapsedMilliseconds.ToString());

        str1 = string.Intern(str1);
        str2 = string.Intern(str2);
        Console.WriteLine("ReferenceCompare...");
        sw.Restart();
        for (int i = 0; i < testCount; i++)
        {
            ReferenceCompare(str1, str2);
        }
        Console.WriteLine("Elapsed: " + sw.ElapsedMilliseconds.ToString());

        Console.WriteLine(new string('=',50));
    }

    private static Boolean NormalCompare1(string str1, string str2)
    {
        return str1 == str2;
    }

    private static Boolean NormalCompare2(string str1, string str2)
    {
        return str1.Equals(str2);
    }

    private static Boolean ReferenceCompare(string str1, string str2)
    {
        return ReferenceEquals(str1, str2);
    }