public TrieTreeNode AddWord(string word, double frequency, POSType pos)
{
if (word.Length == 1)
{
return(AddWord(word[0], frequency, pos));
}
else
{
char[] chars = word.ToCharArray();
TrieTreeNode node = _root;
for (int i = 0; i < chars.Length; i++)
{
TrieTreeNode newnode = node.AddChild(chars[i]);
node = newnode;
}
if (frequency == double.NaN)
{
node.IncreaseFrequency();
}
else
{
node.Frequency += frequency;
}
node.WordEnded = true;
if (pos != POSType.UNKNOWN && pos != POSType.NEWLINE)
{
node.AddPOS(pos);
}
return(node);
}
}
public static void SetPosType(POSType posType)
{
var log = Mvx.IoCProvider.Resolve <IMvxLog>();
log.Debug("App: Setting Pos type : {0}", posType.ToString());
PosType = posType;
}
private WordDictionaryFile LoadFromTextFile(String fileName)
{
WordDictionaryFile dictFile = new WordDictionaryFile();
dictFile.Dicts = new List <IDataNode>();
using (StreamReader sr = new StreamReader(fileName, Encoding.UTF8))
{
while (!sr.EndOfStream)
{
string line = sr.ReadLine();
string[] strs = line.Split(new char[] { '|' });
if (strs.Length == 3)
{
string word = strs[0].Trim();
POSType pos = (POSType)int.Parse(strs[1].Substring(2, strs[1].Length - 2), System.Globalization.NumberStyles.HexNumber);
double frequency = double.Parse(strs[2]);
WordAttribute dict = new WordAttribute(word, pos, frequency);
dictFile.Dicts.Add(dict);
}
}
}
return(dictFile);
}
public TrieTreeNode AddWord(string word, double frequency, POSType pos)
{
if (word.Length == 1)
{
return AddWord(word[0], frequency, pos);
}
else
{
char[] chars = word.ToCharArray();
TrieTreeNode node = _root;
for (int i = 0; i < chars.Length; i++)
{
TrieTreeNode newnode = node.AddChild(chars[i]);
node = newnode;
}
if (frequency == double.NaN)
node.IncreaseFrequency();
else
node.Frequency += frequency;
node.WordEnded = true;
if (pos != POSType.UNKNOWN && pos != POSType.NEWLINE)
node.AddPOS(pos);
return node;
}
}
private WordDictionaryFile LoadFromBinFile(String fileName, out string verNumStr)
{
WordDictionaryFile dictFile = new WordDictionaryFile();
dictFile.Dicts = new List <IDataNode>();
FileStream fs = new FileStream(fileName, FileMode.Open, FileAccess.Read);
byte[] version = new byte[32];
fs.Read(version, 0, version.Length);
String ver = Encoding.UTF8.GetString(version, 0, version.Length);
int zeroPosition = ver.IndexOf('\0');
if (zeroPosition >= 0)
{
ver = ver.Substring(0, zeroPosition);
}
var matches = verRegex.Matches(ver);
if (matches.Count > 0)
{
verNumStr = matches[0].Value;
}
else
{
verNumStr = null;
}
while (fs.Position < fs.Length)
{
byte[] buf = new byte[sizeof(int)];
fs.Read(buf, 0, buf.Length);
int length = BitConverter.ToInt32(buf, 0);
buf = new byte[length];
fs.Read(buf, 0, buf.Length);
string word = Encoding.UTF8.GetString(buf, 0, length - sizeof(int) - sizeof(double));
POSType pos = (POSType)BitConverter.ToInt32(buf, length - sizeof(int) - sizeof(double));
double frequency = BitConverter.ToDouble(buf, length - sizeof(double));
WordAttribute dict = new WordAttribute(word, pos, frequency);
string.Intern(dict.Word);
dictFile.Dicts.Add(dict);
}
fs.Close();
return(dictFile);
}
internal TrieTreeNode AddWord(char ch, double frequency, bool isWordEnded, POSType pos)
{
TrieTreeNode newnode = _root.AddChild(ch);
if (frequency == double.NaN)
newnode.IncreaseFrequency();
else
newnode.Frequency += frequency;
newnode.WordEnded = true;
if (pos != POSType.UNKNOWN && pos != POSType.NEWLINE)
newnode.AddPOS(pos);
return newnode;
}
/// <summary>
/// create a instance of ParseResult
/// </summary>
/// <param name="text">word text</param>
/// <param name="startIndex">start position</param>
/// <param name="type">POS value</param>
/// <param name="value">.NET value for this word</param>
/// <returns></returns>
public static ParseResult Create(string text, int startIndex, POSType type, object value)
{
if (startIndex < 0)
{
throw new ArgumentOutOfRangeException("startIndex must be bigger than 0.");
}
ParseResult pr = new ParseResult();
pr.Text = text.Trim();
pr.StartPos = startIndex;
pr.Type = type;
pr.Value = value;
return(pr);
}
internal TrieTreeNode AddWord(char ch, double frequency, bool isWordEnded, POSType pos)
{
TrieTreeNode newnode = _root.AddChild(ch);
if (frequency == double.NaN)
{
newnode.IncreaseFrequency();
}
else
{
newnode.Frequency += frequency;
}
newnode.WordEnded = true;
if (pos != POSType.UNKNOWN && pos != POSType.NEWLINE)
{
newnode.AddPOS(pos);
}
return(newnode);
}
void AssertServerResponse(string response, bool expectedBoolean, string expectedWord, POSType expectedPOS)
{
if (response.StartsWith("Server side error"))
Assert.Fail("server error");
string[] data = response.Split(new char[] { SOH });
if (data.Length <= 2)
{
if (expectedBoolean == true)
Assert.Fail("expected 0 but was 1");
}
else
{
if (expectedBoolean == false)
Assert.Fail("expected 1 but was 0");
if(expectedPOS!= POSType.UNKNOWN)
Assert.AreEqual(expectedPOS, (POSType)Int32.Parse(data[3]));
Assert.AreEqual(expectedWord, data[1]);
}
}
/// <summary>
/// create a instance of ParseResult
/// </summary>
/// <param name="text">word text</param>
/// <param name="startIndex">start position</param>
/// <param name="type">POS value</param>
/// <returns></returns>
public static ParseResult Create(string text, int startIndex, POSType type)
{
return(Create(text, startIndex, type, null));
}
public TrieTreeNode AddWord(string word, POSType pos)
{
return AddWord(word, double.NaN, pos);
}
public TrieTreeNode AddWord(char ch, double frequency, POSType pos)
{
return AddWord(ch, frequency, true, pos);
}
public TrieTreeNode AddWord(string word, POSType pos)
{
return(AddWord(word, double.NaN, pos));
}
public TrieTreeNode AddWord(char ch, double frequency, POSType pos)
{
return(AddWord(ch, frequency, true, pos));
}
public static void AssertParseResult(ParseResult pr, string text, int startPos, POSType type)
{
Assert.AreEqual(text, pr.Text);
Assert.AreEqual(startPos, pr.StartPos);
Assert.AreEqual(type, pr.Type);
}
public WordAttribute(string word, POSType pos, double frequency)
{
this.Word = word;
this.POS = pos;
this.Frequency = frequency;
}
public static string GetPOS(POSType pos)
{
return POS[(int)pos];
}
void AssertServerResponse(string response, bool expectedBoolean, string expectedWord, POSType expectedPOS)
{
if (response.StartsWith("Server side error"))
{
Assert.Fail("server error");
}
string[] data = response.Split(new char[] { SOH });
if (data.Length <= 2)
{
if (expectedBoolean == true)
{
Assert.Fail("expected 0 but was 1");
}
}
else
{
if (expectedBoolean == false)
{
Assert.Fail("expected 1 but was 0");
}
if (expectedPOS != POSType.UNKNOWN)
{
Assert.AreEqual(expectedPOS, (POSType)Int32.Parse(data[3]));
}
Assert.AreEqual(expectedWord, data[1]);
}
}
public TrieTreeNode AddWord(char ch, POSType pos)
{
return(AddWord(ch, double.NaN, true, pos));
}
public void AddPOS(POSType pos)
{
_posvalue = _posvalue | (int)pos;
}
public static void AssertParseResult(ParseResult pr, string text, int startPos, POSType type)
{
Assert.AreEqual(text, pr.Text);
Assert.AreEqual(startPos, pr.StartPos);
Assert.AreEqual(type, pr.Type);
}
public void RemovePOS(POSType pos)
{
_posvalue = _posvalue & ~(int)pos;
}
public void RemovePOS(POSType pos)
{
_posvalue = _posvalue & ~(int)pos;
}
private void btnQuery_Click(object sender, EventArgs e)
{
Regex regex = new Regex("dict://([\\d\\.]+):(\\d{3,5})", RegexOptions.Compiled);
string addr = tbDictAddress.Text.Trim();
var matches = regex.Matches(addr);
string serverAddr = matches[0].Groups[1].Value;
int port = Int32.Parse(matches[0].Groups[2].Value);
EndPoint serverAddress = new IPEndPoint(IPAddress.Parse(serverAddr), port);
int posValue = 0;
foreach (var selectedObject in checkedListBox1.CheckedItems)
{
string value = (string)selectedObject;
var values = Enum.GetValues(typeof(POSType));
foreach (var v in values)
{
if (((System.Enum)v).Description() == value)
{
POSType pos = (POSType)v;
posValue |= ((int)pos);
break;
}
}
}
using (DictionaryServiceClient dsc = new DictionaryServiceClient())
{
dsc.Connect(serverAddress);
TrieTreeResult result = null;
if (radioButton1.Checked)
{
result = dsc.MaximumMatch(tbWord.Text.Trim(), posValue);
}
else if (radioButton2.Checked)
{
result = dsc.ReverseMaximumMatch(tbWord.Text.Trim(), posValue);
}
else
{
result = dsc.ExactMatch(tbWord.Text.Trim(), posValue);
}
if (result != null)
{
string resultText = result.Word;
int originalLength1 = 0, originalLength2 = 0;
if (result.Frequency != 0)
{
originalLength1 = resultText.Length;
resultText += ", 频率：" + result.Frequency;
}
if (result.POS > 0)
{
originalLength2 = resultText.Length;
resultText += ", 类型：" + ConvertPOSTypesToString(ConvertValueToPOS(result.POS));
}
richTextBox1.Text = resultText;
richTextBox1.ForeColor = System.Drawing.Color.Black;
if (originalLength1 > 0)
{
richTextBox1.Select(originalLength1 + 1, 4);
richTextBox1.SelectionColor = System.Drawing.Color.Blue;
}
if (originalLength2 > 0)
{
richTextBox1.Select(originalLength2 + 1, 4);
richTextBox1.SelectionColor = System.Drawing.Color.Blue;
}
}
else
{
richTextBox1.ForeColor = System.Drawing.Color.Red;
richTextBox1.Text = "未找到合适词";
}
}
}
public TrieTreeNode AddWord(char ch, POSType pos)
{
return AddWord(ch, double.NaN, true, pos);
}
public void AddPOS(POSType pos)
{
_posvalue = _posvalue | (int)pos;
}