static void Main(string[] args)
{
Console.WriteLine("PanGu Word Segment");
string input = "", output = "";
if (args.Length >= 2)
{
input = args[0];
output = args[1];
}
else
{
Console.WriteLine("Usage: wordsegment <input> <output> [-pos]");
FormDemo demo = new FormDemo();
demo.ShowDialog();
return;
}
bool showPosition = args.Length >= 3 && args[2] == "-pos";
// init
PanGu.Segment.Init();
PanGu.Match.MatchOptions options = PanGu.Setting.PanGuSettings.Config.MatchOptions;
PanGu.Match.MatchParameter parameters = PanGu.Setting.PanGuSettings.Config.Parameters;
using (System.IO.StreamWriter sw = new System.IO.StreamWriter(output))
{
foreach (string doc in System.IO.File.ReadAllLines(input))
{
Segment segment = new Segment();
ICollection <WordInfo> words = segment.DoSegment(doc, options, parameters);
StringBuilder wordsString = new StringBuilder();
foreach (WordInfo wordInfo in words)
{
if (wordInfo == null)
{
continue;
}
if (showPosition)
{
wordsString.AppendFormat("{0}/({1},{2}) ", wordInfo.Word, wordInfo.Position, wordInfo.Rank);
}
else
{
wordsString.AppendFormat("{0} ", wordInfo.Word);
}
}
sw.WriteLine(wordsString);
}
sw.Flush(); sw.Close();
}
}
private void buttonSaveConfig_Click(object sender, EventArgs e)
{
_Options = PanGu.Setting.PanGuSettings.Config.MatchOptions;
_Parameters = PanGu.Setting.PanGuSettings.Config.Parameters;
UpdateSettings();
PanGu.Setting.PanGuSettings.Save("PanGu.xml");
}
private void buttonSegment_Click(object sender, EventArgs e)
{
_Options = PanGu.Setting.PanGuSettings.Config.MatchOptions.Clone();
_Parameters = PanGu.Setting.PanGuSettings.Config.Parameters.Clone();
UpdateSettings();
if (checkBoxDisplayPosition.Checked)
{
DisplaySegmentAndPostion();
}
else
{
DisplaySegment();
}
}
public Match.MatchParameter GetParameterCopy()
{
Match.MatchParameter parameter = new PanGu.Match.MatchParameter();
parameter.Redundancy = this.Parameters.Redundancy;
parameter.UnknowRank = this.Parameters.UnknowRank;
parameter.BestRank = this.Parameters.BestRank;
parameter.SecRank = this.Parameters.SecRank;
parameter.ThirdRank = this.Parameters.ThirdRank;
parameter.SingleRank = this.Parameters.SingleRank;
parameter.NumericRank = this.Parameters.NumericRank;
parameter.EnglishRank = this.Parameters.EnglishRank;
parameter.SymbolRank = this.Parameters.SymbolRank;
parameter.SimplifiedTraditionalRank = this.Parameters.SimplifiedTraditionalRank;
return(parameter);
}
public static void Init()
{
_Options = new PanGu.Match.MatchOptions();
_Parameters = new PanGu.Match.MatchParameter();
_Options.FrequencyFirst = Convert.ToBoolean(ConfigurationManager.AppSettings["checkBoxFreqFirst"]);
_Options.FilterStopWords = Convert.ToBoolean(ConfigurationManager.AppSettings["FilterStopWords"]);
_Options.ChineseNameIdentify = Convert.ToBoolean(ConfigurationManager.AppSettings["ChineseNameIdentify"]);
_Options.MultiDimensionality = Convert.ToBoolean(ConfigurationManager.AppSettings["MultiDimensionality"]);
_Options.EnglishMultiDimensionality = Convert.ToBoolean(ConfigurationManager.AppSettings["EnglishMultiDimensionality"]);
_Options.ForceSingleWord = Convert.ToBoolean(ConfigurationManager.AppSettings["ForceSingleWord"]);
_Options.TraditionalChineseEnabled = Convert.ToBoolean(ConfigurationManager.AppSettings["TraditionalChineseEnabled"]);
_Options.OutputSimplifiedTraditional = Convert.ToBoolean(ConfigurationManager.AppSettings["OutputSimplifiedTraditional"]);
_Options.UnknownWordIdentify = Convert.ToBoolean(ConfigurationManager.AppSettings["UnknownWordIdentify"]);
_Options.FilterEnglish = Convert.ToBoolean(ConfigurationManager.AppSettings["FilterEnglish"]);
_Options.FilterNumeric = Convert.ToBoolean(ConfigurationManager.AppSettings["FilterNumeric"]);
_Options.IgnoreCapital = Convert.ToBoolean(ConfigurationManager.AppSettings["IgnoreCapital"]);
_Options.EnglishSegment = Convert.ToBoolean(ConfigurationManager.AppSettings["EnglishSegment"]);
_Options.SynonymOutput = Convert.ToBoolean(ConfigurationManager.AppSettings["SynonymOutput"]);
_Options.WildcardOutput = Convert.ToBoolean(ConfigurationManager.AppSettings["WildcardOutput"]);
_Options.WildcardSegment = Convert.ToBoolean(ConfigurationManager.AppSettings["WildcardSegment"]);
_Options.CustomRule = Convert.ToBoolean(ConfigurationManager.AppSettings["CustomRule"]);
_Parameters.Redundancy = Convert.ToInt16(ConfigurationManager.AppSettings["Redundancy"]);
_Parameters.FilterEnglishLength = Convert.ToInt16(ConfigurationManager.AppSettings["FilterEnglishLength"]);
_Parameters.FilterNumericLength = Convert.ToInt16(ConfigurationManager.AppSettings["FilterNumericLength"]);
var config = new Config();
config.ServerHosts = JsonConvert.DeserializeObject <List <ServerHost> >(ConfigurationManager.ConnectionStrings["mongodb"].ToString());
config.BaseDbName = "serch_base";
config.IndexListDbName = "serch_index";
CoreIoc.Register(o => o.RegisterInstance(config).As <Config>().ExternallyOwned());
CoreIoc.Register(o => o.RegisterType <ServerManage>().As <ServerManage>().SingleInstance());
CoreIoc.Register(o => o.RegisterType <DocumentOp>().As <IDocument>().SingleInstance());
CoreIoc.Register(o => o.RegisterInstance(new PanguTokenizer(_Options, _Parameters)).As <ITokenizer>().ExternallyOwned());
CoreIoc.Register(o => o.RegisterType <Index>().As <IIndex>().SingleInstance());
CoreIoc.Register(o => o.RegisterType <Query>().As <Iquery>().SingleInstance());
CoreIoc.Build();
}
private void FormDemo_Load(object sender, EventArgs e)
{
textBoxSource.Text = _InitSource;
PanGu.Segment.Init();
PanGu.Match.MatchOptions options = PanGu.Setting.PanGuSettings.Config.MatchOptions;
checkBoxFreqFirst.Checked = options.FrequencyFirst;
checkBoxFilterStopWords.Checked = options.FilterStopWords;
checkBoxMatchName.Checked = options.ChineseNameIdentify;
checkBoxMultiSelect.Checked = options.MultiDimensionality;
checkBoxEnglishMultiSelect.Checked = options.EnglishMultiDimensionality;
checkBoxForceSingleWord.Checked = options.ForceSingleWord;
checkBoxTraditionalChs.Checked = options.TraditionalChineseEnabled;
checkBoxST.Checked = options.OutputSimplifiedTraditional;
checkBoxUnknownWord.Checked = options.UnknownWordIdentify;
checkBoxFilterEnglish.Checked = options.FilterEnglish;
checkBoxFilterNumeric.Checked = options.FilterNumeric;
checkBoxIgnoreCapital.Checked = options.IgnoreCapital;
checkBoxEnglishSegment.Checked = options.EnglishSegment;
checkBoxSynonymOutput.Checked = options.SynonymOutput;
checkBoxWildcard.Checked = options.WildcardOutput;
checkBoxWildcardSegment.Checked = options.WildcardSegment;
checkBoxCustomRule.Checked = options.CustomRule;
checkBoxDisplayPosition.Checked = options.DisplayPosition;
checkBoxIgnoreSpace.Checked = options.IgnoreSpace;
checkBoxIgnoreEOL.Checked = options.IgnoreEndOfLine;
//if (checkBoxMultiSelect.Checked)
//{
// checkBoxDisplayPosition.Checked = true;
//}
PanGu.Match.MatchParameter parameters = PanGu.Setting.PanGuSettings.Config.Parameters;
numericUpDownRedundancy.Value = parameters.Redundancy;
numericUpDownFilterEnglishLength.Value = parameters.FilterEnglishLength;
numericUpDownFilterNumericLength.Value = parameters.FilterNumericLength;
divider.Text = parameters.CustomDivider;
//str = Microsoft.VisualBasic.Strings.StrConv(str, Microsoft.VisualBasic.VbStrConv.SimplifiedChinese, 0);
}
public static void Init()
{
_Options = new PanGu.Match.MatchOptions();
_Parameters = new PanGu.Match.MatchParameter();
_Options.FrequencyFirst = Convert.ToBoolean(ConfigurationManager.AppSettings["checkBoxFreqFirst"]);
_Options.FilterStopWords = Convert.ToBoolean(ConfigurationManager.AppSettings["FilterStopWords"]);
_Options.ChineseNameIdentify = Convert.ToBoolean(ConfigurationManager.AppSettings["ChineseNameIdentify"]);
_Options.MultiDimensionality = Convert.ToBoolean(ConfigurationManager.AppSettings["MultiDimensionality"]);
_Options.EnglishMultiDimensionality = Convert.ToBoolean(ConfigurationManager.AppSettings["EnglishMultiDimensionality"]);
_Options.ForceSingleWord = Convert.ToBoolean(ConfigurationManager.AppSettings["ForceSingleWord"]);
_Options.TraditionalChineseEnabled = Convert.ToBoolean(ConfigurationManager.AppSettings["TraditionalChineseEnabled"]);
_Options.OutputSimplifiedTraditional = Convert.ToBoolean(ConfigurationManager.AppSettings["OutputSimplifiedTraditional"]);
_Options.UnknownWordIdentify = Convert.ToBoolean(ConfigurationManager.AppSettings["UnknownWordIdentify"]);
_Options.FilterEnglish = Convert.ToBoolean(ConfigurationManager.AppSettings["FilterEnglish"]);
_Options.FilterNumeric = Convert.ToBoolean(ConfigurationManager.AppSettings["FilterNumeric"]);
_Options.IgnoreCapital = Convert.ToBoolean(ConfigurationManager.AppSettings["IgnoreCapital"]);
_Options.EnglishSegment = Convert.ToBoolean(ConfigurationManager.AppSettings["EnglishSegment"]);
_Options.SynonymOutput = Convert.ToBoolean(ConfigurationManager.AppSettings["SynonymOutput"]);
_Options.WildcardOutput = Convert.ToBoolean(ConfigurationManager.AppSettings["WildcardOutput"]);
_Options.WildcardSegment = Convert.ToBoolean(ConfigurationManager.AppSettings["WildcardSegment"]);
_Options.CustomRule = Convert.ToBoolean(ConfigurationManager.AppSettings["CustomRule"]);
_Parameters.Redundancy = Convert.ToInt16(ConfigurationManager.AppSettings["Redundancy"]);
_Parameters.FilterEnglishLength = Convert.ToInt16(ConfigurationManager.AppSettings["FilterEnglishLength"]);
_Parameters.FilterNumericLength = Convert.ToInt16(ConfigurationManager.AppSettings["FilterNumericLength"]);
var config = new Config();
config.ServerHosts = JsonConvert.DeserializeObject<List<ServerHost>>(ConfigurationManager.ConnectionStrings["mongodb"].ToString());
config.BaseDbName = "serch_base";
config.IndexListDbName = "serch_index";
CoreIoc.Register(o => o.RegisterInstance(config).As<Config>().ExternallyOwned());
CoreIoc.Register(o=>o.RegisterType<ServerManage>().As<ServerManage>().SingleInstance());
CoreIoc.Register(o=>o.RegisterType<DocumentOp>().As<IDocument>().SingleInstance());
CoreIoc.Register(o => o.RegisterInstance(new PanguTokenizer(_Options,_Parameters)).As<ITokenizer>().ExternallyOwned());
CoreIoc.Register(o => o.RegisterType<Index>().As<IIndex>().SingleInstance());
CoreIoc.Register(o => o.RegisterType<Query>().As<Iquery>().SingleInstance());
CoreIoc.Build();
}
public PanguTokenizer(PanGu.Match.MatchOptions options, PanGu.Match.MatchParameter parameters)
{
_Options = options;_Parameters = parameters;
segment = new Segment();
}
/// <summary>
/// 根据文章标题智能解析关键字(或标签)
/// </summary>
/// <param name="title"></param>
/// <returns></returns>
private static ICollection<WordInfo> TitleToKeywordWordInfos(string title)
{
PanGu.Segment segment = new Segment();
PanGu.Match.MatchOptions matchOptions = new PanGu.Match.MatchOptions();
//中文人名识别
matchOptions.ChineseNameIdentify = false;
//词频优先
matchOptions.FrequencyFirst = false;
//多元分词
matchOptions.MultiDimensionality = false;
//英文多元分词，这个开关，会将英文中的字母和数字分开
matchOptions.EnglishMultiDimensionality = false;
//过滤停用词
matchOptions.FilterStopWords = true;
//忽略空格、回车、Tab
matchOptions.IgnoreSpace = true;
//强制一元分词
matchOptions.ForceSingleWord = false;
//繁体中文开关
matchOptions.TraditionalChineseEnabled = false;
//同时输出简体和繁体
matchOptions.OutputSimplifiedTraditional = false;
//未登录词识别
matchOptions.UnknownWordIdentify = false;
//过滤英文，这个选项只有在过滤停用词选项生效时才有效
matchOptions.FilterEnglish = true;
//过滤数字，这个选项只有在过滤停用词选项生效时才有效
matchOptions.FilterNumeric = true;
//忽略英文大小写
matchOptions.IgnoreCapital = false;
//英文分词
matchOptions.EnglishSegment = false;
//同义词输出 （同义词输出功能一般用于对搜索字符串的分词，不建议在索引时使用）
matchOptions.SynonymOutput = false;
//通配符匹配输出 （）
matchOptions.WildcardOutput = false;
//对通配符匹配的结果分词
matchOptions.WildcardSegment = false;
PanGu.Match.MatchParameter matchParameter = new PanGu.Match.MatchParameter();
//未登录词权值
matchParameter.UnknowRank = 1;
//最匹配词权值
matchParameter.BestRank = 5;
//次匹配词权值
matchParameter.SecRank = 3;
//再次匹配词权值
matchParameter.ThirdRank = 2;
//强行输出的单字的权值
matchParameter.SingleRank = 1;
//数字的权值
matchParameter.NumericRank = 1;
//英文词汇权值
matchParameter.EnglishRank = 5;
//英文词汇小写的权值
matchParameter.EnglishLowerRank = 3;
//英文词汇词根的权值
matchParameter.EnglishStemRank = 2;
//符号的权值
matchParameter.SymbolRank = 1;
//强制同时输出简繁汉字时，非原来文本的汉字输出权值。 比如原来文本是简体，这里就是输出的繁体字的权值，反之亦然。
matchParameter.SimplifiedTraditionalRank = 1;
//同义词权值
matchParameter.SynonymRank = 1;
//通配符匹配结果的权值
matchParameter.WildcardRank = 1;
//过滤英文选项生效时，过滤大于这个长度的英文
matchParameter.FilterEnglishLength = 0;
//过滤数字选项生效时，过滤大于这个长度的数字
matchParameter.FilterNumericLength = 0;
//用户自定义规则的配件文件名
matchParameter.CustomRuleAssemblyFileName = string.Empty;
//用户自定义规则的类的完整名，即带名字空间的名称
matchParameter.CustomRuleFullClassName = string.Empty;
//冗余度
matchParameter.Redundancy = 0;
return segment.DoSegment(title, matchOptions, matchParameter);
}
private void buttonSaveConfig_Click(object sender, EventArgs e)
{
_Options = PanGu.Setting.PanGuSettings.Config.MatchOptions;
_Parameters = PanGu.Setting.PanGuSettings.Config.Parameters;
UpdateSettings();
PanGu.Setting.PanGuSettings.Save("PanGu.xml");
}
private void buttonSegment_Click(object sender, EventArgs e)
{
_Options = PanGu.Setting.PanGuSettings.Config.MatchOptions.Clone();
_Parameters = PanGu.Setting.PanGuSettings.Config.Parameters.Clone();
UpdateSettings();
if (checkBoxDisplayPosition.Checked)
{
DisplaySegmentAndPostion();
}
else
{
DisplaySegment();
}
}
/// <summary>
/// 根据文章标题智能解析关键字(或标签)
/// </summary>
/// <param name="title"></param>
/// <returns></returns>
private static ICollection <WordInfo> TitleToKeywordWordInfos(string title)
{
PanGu.Segment segment = new Segment();
PanGu.Match.MatchOptions matchOptions = new PanGu.Match.MatchOptions();
//中文人名识别
matchOptions.ChineseNameIdentify = false;
//词频优先
matchOptions.FrequencyFirst = false;
//多元分词
matchOptions.MultiDimensionality = false;
//英文多元分词，这个开关，会将英文中的字母和数字分开
matchOptions.EnglishMultiDimensionality = false;
//过滤停用词
matchOptions.FilterStopWords = true;
//忽略空格、回车、Tab
matchOptions.IgnoreSpace = true;
//强制一元分词
matchOptions.ForceSingleWord = false;
//繁体中文开关
matchOptions.TraditionalChineseEnabled = false;
//同时输出简体和繁体
matchOptions.OutputSimplifiedTraditional = false;
//未登录词识别
matchOptions.UnknownWordIdentify = false;
//过滤英文，这个选项只有在过滤停用词选项生效时才有效
matchOptions.FilterEnglish = true;
//过滤数字，这个选项只有在过滤停用词选项生效时才有效
matchOptions.FilterNumeric = true;
//忽略英文大小写
matchOptions.IgnoreCapital = false;
//英文分词
matchOptions.EnglishSegment = false;
//同义词输出 （同义词输出功能一般用于对搜索字符串的分词，不建议在索引时使用）
matchOptions.SynonymOutput = false;
//通配符匹配输出 （）
matchOptions.WildcardOutput = false;
//对通配符匹配的结果分词
matchOptions.WildcardSegment = false;
PanGu.Match.MatchParameter matchParameter = new PanGu.Match.MatchParameter();
//未登录词权值
matchParameter.UnknowRank = 1;
//最匹配词权值
matchParameter.BestRank = 5;
//次匹配词权值
matchParameter.SecRank = 3;
//再次匹配词权值
matchParameter.ThirdRank = 2;
//强行输出的单字的权值
matchParameter.SingleRank = 1;
//数字的权值
matchParameter.NumericRank = 1;
//英文词汇权值
matchParameter.EnglishRank = 5;
//英文词汇小写的权值
matchParameter.EnglishLowerRank = 3;
//英文词汇词根的权值
matchParameter.EnglishStemRank = 2;
//符号的权值
matchParameter.SymbolRank = 1;
//强制同时输出简繁汉字时，非原来文本的汉字输出权值。 比如原来文本是简体，这里就是输出的繁体字的权值，反之亦然。
matchParameter.SimplifiedTraditionalRank = 1;
//同义词权值
matchParameter.SynonymRank = 1;
//通配符匹配结果的权值
matchParameter.WildcardRank = 1;
//过滤英文选项生效时，过滤大于这个长度的英文
matchParameter.FilterEnglishLength = 0;
//过滤数字选项生效时，过滤大于这个长度的数字
matchParameter.FilterNumericLength = 0;
//用户自定义规则的配件文件名
matchParameter.CustomRuleAssemblyFileName = string.Empty;
//用户自定义规则的类的完整名，即带名字空间的名称
matchParameter.CustomRuleFullClassName = string.Empty;
//冗余度
matchParameter.Redundancy = 0;
return(segment.DoSegment(title, matchOptions, matchParameter));
}
public PanguTokenizer(PanGu.Match.MatchOptions options, PanGu.Match.MatchParameter parameters)
{
_Options = options; _Parameters = parameters;
segment = new Segment();
}
public Match.MatchParameter GetParameterCopy()
{
Match.MatchParameter parameter = new PanGu.Match.MatchParameter();
parameter.Redundancy = this.Parameters.Redundancy;
parameter.UnknowRank = this.Parameters.UnknowRank;
parameter.BestRank = this.Parameters.BestRank;
parameter.SecRank = this.Parameters.SecRank;
parameter.ThirdRank = this.Parameters.ThirdRank;
parameter.SingleRank = this.Parameters.SingleRank;
parameter.NumericRank = this.Parameters.NumericRank;
parameter.EnglishRank = this.Parameters.EnglishRank;
parameter.SymbolRank = this.Parameters.SymbolRank;
parameter.SimplifiedTraditionalRank = this.Parameters.SimplifiedTraditionalRank;
return parameter;
}