//private static Dictionary<string, IndexSet> nameIndexDict = new Dictionary<string, IndexSet>();
#endregion
#region "搜索基本设置"
public static void SetSearchSettings(string configFileName,bool isXmlFile)
{
if (initSettings) return;
try
{
List<Source> sourceList ;
List<IndexSet> indexList;
if (isXmlFile)
{
Config config = (Config)SupportClass.FileUtil.GetObjectFromXmlFile(configFileName, typeof(Config));
sourceList = config.SourceList;
indexList = config.IndexList;
searchSet = config.SearchSet;
dictSet = config.DictionarySet;
fileSet = config.FileIndexSet;
}
else
{
List<string> srcList = SupportClass.FileUtil.GetFileText(configFileName);
sourceList = Source.GetSourceList(srcList);
indexList = IndexSet.GetIndexList(srcList);
searchSet = SearchSet.GetSearchSet(srcList);
dictSet = DictionarySet.GetDictionarySet(srcList);
}
searchIndexList.AddRange(indexList);
ISUtils.CSegment.Segment.SetPaths(dictSet.BasePath, dictSet.NamePath, dictSet.NumberPath, dictSet.FilterPath, dictSet.CustomPaths);
ISUtils.CSegment.Segment.SetDefaults(new ISUtils.CSegment.DictionaryLoader.TextDictionaryLoader(), new ISUtils.CSegment.ForwardMatchSegment());
if (indexDict == null)
indexDict = new Dictionary<IndexSet, Source>();
foreach (IndexSet set in indexList)
{
foreach (Source source in sourceList)
{
if (source.SourceName == set.SourceName)
{
if(indexDict.ContainsKey(set)==false)
indexDict.Add(set, source);
break;
}
}
}
SetBasicDict();
initSettings = true;
}
catch (Exception e)
{
#if DEBUG
System.Console.WriteLine(e.StackTrace.ToString());
#endif
throw e;
}
}
public static void SetIndexSettings(string configFileName,bool isXmlFile)
{
if (initSettings) return;
initSettings = true;
try
{
List<Source> sourceList;
List<IndexSet> indexList;
if (isXmlFile)
{
Config config = (Config)SupportClass.FileUtil.GetObjectFromXmlFile(configFileName, typeof(Config));
sourceList = config.SourceList;
indexList = config.IndexList;
indexerSet = config.IndexerSet;
dictSet = config.DictionarySet;
fileSet = config.FileIndexSet;
}
else
{
List<string> srcList = SupportClass.FileUtil.GetFileText(configFileName);
sourceList = Source.GetSourceList(srcList);
indexList = IndexSet.GetIndexList(srcList);
indexerSet = IndexerSet.GetIndexer(srcList);
dictSet = DictionarySet.GetDictionarySet(srcList);
}
if (indexDict == null)
indexDict = new Dictionary<IndexSet, Source>();
foreach (IndexSet set in indexList)
{
foreach (Source source in sourceList)
{
if (source.SourceName == set.SourceName)
{
if(indexDict.ContainsKey(set)==false)
indexDict.Add(set, source);
break;
}
}
}
}
catch (Exception e)
{
#if DEBUG
System.Console.WriteLine(e.StackTrace.ToString());
#endif
throw e;
}
}
public SearchMaker(string filename)
{
try
{
Config parser = new Config(filename,true);
searchd = parser.GetSearchd();
sourceList = parser.GetSourceList();
indexList = parser.GetIndexList();
dictSet = parser.GetDictionarySet();
fileSet = parser.FileIndexSet;
}
catch (Exception ex)
{
#if DEBUG
Console.WriteLine(string.Format("Exception for open file {0},{1}", filename, ex.ToString()));
#endif
throw;
}
}
public IndexMaker(string filename)
{
try
{
Config parser = new Config(filename,true);
sourceList = parser.GetSourceList();
indexList = parser.GetIndexList();
indexer = parser.GetIndexer();
dictSet = parser.GetDictionarySet();
fileSet = parser.FileIndexSet;
}
catch (Exception ex)
{
#if DEBUG
Console.WriteLine(string.Format("IndexMaker,发生异常，文件名:\t{0},\t异常:\t{1}", filename, ex.ToString()));
#endif
throw;
}
Init();
}
public static void SetIndexSettings(List<Source> sourceList, List<IndexSet> indexList, DictionarySet dictSet, IndexerSet indexerSet)
{
if (initSettings) return;
initSettings = true;
IndexUtil.indexerSet = indexerSet;
IndexUtil.dictSet = dictSet;
if (indexDict == null)
indexDict = new Dictionary<IndexSet, Source>();
foreach (IndexSet set in indexList)
{
foreach (Source source in sourceList)
{
if (source.SourceName == set.SourceName)
{
if(indexDict.ContainsKey(set)==false)
indexDict.Add(set, source);
break;
}
}
}
}
public static void WriteToFile(ref System.IO.StreamWriter sw, DictionarySet dictSet)
{
sw.WriteLine("#############################################");
sw.WriteLine("#dictionary settings");
sw.WriteLine("#############################################");
sw.WriteLine(Config.DictionaryFlag.ToLower());
sw.WriteLine(Config.Prefix);
sw.WriteLine("\t#基本词库");
sw.WriteLine("\t" + DictionarySet.BasePathFlag.ToLower() + "=" + dictSet.BasePath);
sw.WriteLine();
sw.WriteLine("\t#姓氏词库");
sw.WriteLine("\t" + DictionarySet.NamePathFlag.ToLower() + "=" + dictSet.NamePath);
sw.WriteLine();
sw.WriteLine("\t#数值词库");
sw.WriteLine("\t" + DictionarySet.NumberPathFlag.ToLower() + "=" + dictSet.NumberPath);
sw.WriteLine();
sw.WriteLine("\t#用户词库");
foreach(string s in dictSet.CustomPaths)
sw.WriteLine("\t" + DictionarySet.CustomPathFlag.ToLower() + "=" + s);
sw.WriteLine();
sw.WriteLine("\t#过滤词库");
sw.WriteLine("\t" + DictionarySet.FilterPathFlag.ToLower() + "=" + dictSet.FilterPath);
sw.WriteLine(Config.Suffix);
}
/**/
/// <summary>
/// 获取字符串列表中的索引列表
/// </summary>
/// <param name="srcList">字符串列表</param>
/// <returns>返回类型</returns>
public static DictionarySet GetDictionarySet(List<string> srcList)
{
DictionarySet dictSet = new DictionarySet();
List<string> pathList = new List<string>();
bool findDict = false, dictStart = false;
foreach (string s in srcList)
{
if (SupportClass.String.StartsWithNoCase(SupportClass.String.FormatStr(s), Config.DictionaryFlag))
{
findDict = true;
continue;
}
if (findDict && SupportClass.String.FormatStr(s).StartsWith(Config.Prefix))
{
dictStart = true;
continue;
}
if (findDict && dictStart && SupportClass.String.FormatStr(s).StartsWith(Config.Suffix))
{
findDict = false;
dictStart = false;
break;
}
if (SupportClass.String.FormatStr(s).StartsWith(Config.Ignore))
continue;
if (findDict && dictStart && SupportClass.String.StartsWithNoCase(SupportClass.String.FormatStr(s), DictionarySet.BasePathFlag))
{
string format = SupportClass.String.FormatStr(s);
string[] split = SupportClass.String.Split(format,Config.Devider);
#if DEBUG
Console.WriteLine(format);
foreach (string a in split)
Console.WriteLine(a);
#endif
if (split.Length >= 2)
dictSet.BasePath = split[1];
else
dictSet.BasePath = "";
continue;
}
if (findDict && dictStart && SupportClass.String.StartsWithNoCase(SupportClass.String.FormatStr(s), DictionarySet.CustomPathFlag))
{
string format = SupportClass.String.FormatStr(s);
string[] split = SupportClass.String.Split(format, Config.Devider);
#if DEBUG
Console.WriteLine(format);
foreach (string a in split)
Console.WriteLine(a);
#endif
if (split.Length >= 2)
pathList.Add(split[1]);
continue;
}
if (findDict && dictStart && SupportClass.String.StartsWithNoCase(SupportClass.String.FormatStr(s), DictionarySet.FilterPathFlag))
{
string format = SupportClass.String.FormatStr(s);
string[] split = SupportClass.String.Split(format, Config.Devider);
#if DEBUG
Console.WriteLine(format);
foreach (string a in split)
Console.WriteLine(a);
#endif
if (split.Length >= 2)
dictSet.FilterPath = split[1];
else
dictSet.FilterPath = "";
continue;
}
if (findDict && dictStart && SupportClass.String.StartsWithNoCase(SupportClass.String.FormatStr(s), DictionarySet.NamePathFlag))
{
string format = SupportClass.String.FormatStr(s);
string[] split = SupportClass.String.Split(format, Config.Devider);
#if DEBUG
Console.WriteLine(format);
foreach (string a in split)
Console.WriteLine(a);
#endif
if (split.Length >= 2)
dictSet.NamePath = split[1];
else
dictSet.NamePath = "";
continue;
}
if (findDict && dictStart && SupportClass.String.StartsWithNoCase(SupportClass.String.FormatStr(s), DictionarySet.NumberPathFlag))
{
string format = SupportClass.String.FormatStr(s);
string[] split = SupportClass.String.Split(format, Config.Devider);
#if DEBUG
Console.WriteLine(format);
foreach (string a in split)
Console.WriteLine(a);
#endif
if (split.Length >= 2)
dictSet.NumberPath = split[1];
else
dictSet.NumberPath = "";
continue;
}
}
dictSet.CustomPaths = pathList;
return dictSet;
}
public static void WriteConfigFile(string path, List<Source> sourceList, List<IndexSet> indexList,FileIndexSet fileSet, DictionarySet dictSet, IndexerSet indexerSet, SearchSet searchSet)
{
try
{
if (IsFileExists(path) != false)
{
System.IO.File.Delete(path);
}
FileStream fs = new FileStream(path, FileMode.CreateNew);
StreamWriter sw = new StreamWriter(fs,Encoding.GetEncoding("gb2312"));
if(sourceList !=null)
foreach (Source s in sourceList)
Source.WriteToFile(ref sw, s);
if (indexList!=null)
foreach (IndexSet i in indexList)
IndexSet.WriteToFile(ref sw, i);
if (fileSet != null)
FileIndexSet.WriteToFile(ref sw, fileSet);
if (dictSet !=null)
DictionarySet.WriteToFile(ref sw, dictSet);
if(indexerSet !=null)
IndexerSet.WriteToFile(ref sw, indexerSet);
if(searchSet !=null)
SearchSet.WriteToFile(ref sw, searchSet);
sw.Flush();
sw.Close();
fs.Close();
}
catch (Exception e)
{
throw e;
}
}
public static void Execute(List<Source> sourceList, DictionarySet dictSet, IndexerSet indexer, bool create, ref Message msg)
{
try
{
DateTime allStart = DateTime.Now;
msg.AddInfo("All Start at :" + allStart.ToLocalTime());
Utils.IndexUtil.SetIndexSettings(dict, dictSet, indexer);
//由于中文分词结果随中文词库的变化而变化，为了使索引不需要根据中文词库的变化而变化，
//故采用默认的Analyzer来进行分词，即StandardAnalyzer
//Utils.IndexUtil.UseDefaultChineseAnalyzer(true);
Utils.IndexUtil.Index(create);
msg.AddInfo("All End at :"+DateTime.Now.ToLocalTime());
TimeSpan allSpan=DateTime.Now -allStart;
msg.AddInfo(string.Format("All Spend {0} millionseconds.",allSpan.TotalMilliseconds));
msg.Success =true;
msg.Result="Execute Success.";
}
catch (Exception e)
{
#if DEBUG
Console.WriteLine("Execute Indexer Error.Reason:"+e.Message);
#endif
msg.AddInfo("Write Index Error.Reason:"+e.StackTrace.ToString());
msg.Success=false;
msg.ExceptionOccur = true;
throw e;
}
}
public static void SetIndexSettings(List<Source> sourceList, FileIndexSet fileIndexSet, DictionarySet dictSet, IndexerSet indexerSet)
{
if (initSettings) return;
initSettings = true;
if (dict != null)
indexDict = dict;
else
indexDict = new Dictionary<IndexSet, Source>();
IndexUtil.dictSet = dictSet;
IndexUtil.indexerSet = indexerSet;
IndexUtil.fileSet = fileIndexSet;
}
public static void SetSearchSettings(List<Source> sourceList, List<IndexSet> indexList, DictionarySet dictSet, SearchSet searchSet)
{
if (initSettings) return;
SearchUtil.searchSet = searchSet;
SearchUtil.dictSet = dictSet;
ISUtils.CSegment.Segment.SetPaths(dictSet.BasePath, dictSet.NamePath, dictSet.NumberPath, dictSet.FilterPath, dictSet.CustomPaths);
ISUtils.CSegment.Segment.SetDefaults(new ISUtils.CSegment.DictionaryLoader.TextDictionaryLoader(), new ISUtils.CSegment.ForwardMatchSegment());
searchIndexList.AddRange(indexList);
if (indexDict == null)
indexDict = new Dictionary<IndexSet, Source>();
foreach (IndexSet set in indexList)
{
foreach (Source source in sourceList)
{
if (source.SourceName == set.SourceName)
{
if(indexDict.ContainsKey(set)==false)
indexDict.Add(set, source);
break;
}
}
}
SetBasicDict();
initSettings = true;
}
public static void SetSearchSettings(List<Source> sourceList, FileIndexSet fileIndexSet, DictionarySet dictSet, SearchSet searchSet)
{
if (initSettings) return;
if (dict != null)
indexDict = dict;
else
indexDict = new Dictionary<IndexSet, Source>();
searchIndexList.AddRange(indexDict.Keys);
SearchUtil.dictSet = dictSet;
SearchUtil.searchSet = searchSet;
SearchUtil.fileSet = fileIndexSet;
ISUtils.CSegment.Segment.SetPaths(dictSet.BasePath, dictSet.NamePath, dictSet.NumberPath, dictSet.FilterPath, dictSet.CustomPaths);
ISUtils.CSegment.Segment.SetDefaults(new ISUtils.CSegment.DictionaryLoader.TextDictionaryLoader(), new ISUtils.CSegment.ForwardMatchSegment());
SetBasicDict();
initSettings = true;
}
