当前位置: 首页>>代码示例>>C++>>正文


C++ PBowDocBs::SaveTxtStat方法代码示例

本文整理汇总了C++中PBowDocBs::SaveTxtStat方法的典型用法代码示例。如果您正苦于以下问题:C++ PBowDocBs::SaveTxtStat方法的具体用法?C++ PBowDocBs::SaveTxtStat怎么用?C++ PBowDocBs::SaveTxtStat使用的例子?那么, 这里精选的方法代码示例或许可以为您提供帮助。您也可以进一步了解该方法所在PBowDocBs的用法示例。


在下文中一共展示了PBowDocBs::SaveTxtStat方法的1个代码示例,这些例子默认根据受欢迎程度排序。您可以为喜欢或者感觉有用的代码点赞,您的评价将有助于系统推荐出更棒的C++代码示例。

示例1: main


//.........这里部分代码省略.........
  // bag-of-words to create
  PBowDocBs BowDocBs;

  // load input data
  if (!InFPath.Empty()){ // directory-files
    // prepare stop-words
    PSwSet SwSet=TSwSet::GetSwSet(SwSetTypeNm);
    if (!SwSetFNm.Empty()) { SwSet->LoadFromFile(SwSetFNm); }
    // prepare stemmer
    PStemmer Stemmer=TStemmer::GetStemmer(StemmerTypeNm);
    // load bow
    BowDocBs=TBowFl::LoadHtmlTxt(InFPath, RecurseDirP, Recs,
     SwSet, Stemmer, MxNGramLen, MnNGramFq, SaveDocP);
  } else
  if (!InMtxFNm.Empty()){ // matrix-file
    PBowSimMtx BowSimMtx=TBowSimMtx::LoadTxt(InMtxFNm);
    BowDocBs=TBowFl::LoadFromSimMtx(BowSimMtx);
  } else
  if (!InTabFNm.Empty()){ // tab-file
    BowDocBs=TBowFl::LoadTabTxt(InTabFNm, Recs);
  } else
  if (!InTsactFNm.Empty()){ // transaction-file
    BowDocBs=TBowFl::LoadTsactTxt(InTsactFNm, Recs);
  } else
  if (!InSparseFNm.Empty()){ // sparse-file
    TStr DocDefFNm=TStr::PutFExt(InSparseFNm, ".tup");
    TStr WordDefFNm=TStr::PutFExt(InSparseFNm, ".var");
    TStr TrainDataFNm=InSparseFNm;
    BowDocBs=TBowFl::LoadSparseTxt(DocDefFNm, WordDefFNm, TrainDataFNm, Recs);
  } else
  if (!InSvmLightFNm.Empty()){ // SvmLight-file
    TStr DocDefFNm=TStr::PutFExt(InSparseFNm, ".tup");
    TStr WordDefFNm=TStr::PutFExt(InSparseFNm, ".var");
    TStr TrainDataFNm=InSvmLightFNm;
    BowDocBs=TBowFl::LoadSvmLightTxt(DocDefFNm, WordDefFNm, TrainDataFNm, "", Recs);
  } else
  if (!InTBsFNm.Empty()){ // Text-Base-file
    //BowDocBs=TBowFl::LoadTBsTxt(InTBsFNm, Recs);
  } else
  if (!InCpdFNm.Empty()){ // Compact-Doc-file
    BowDocBs=TBowFl::LoadCpdTxt(InCpdFNm, Recs,
     SwSetTypeNm, StemmerTypeNm, MxNGramLen, MnNGramFq);
  } else
  if (!InLnDocFNm.Empty()){ // Line-Documents-file
    BowDocBs=TBowFl::LoadLnDocTxt(InLnDocFNm, false, Recs,
     SwSetTypeNm, StemmerTypeNm, MxNGramLen, MnNGramFq, SaveDocP);
  } else
  if (!InNmLnDocFNm.Empty()){ // Named-Line-Documents-file
    BowDocBs=TBowFl::LoadLnDocTxt(InNmLnDocFNm, true, Recs,
     SwSetTypeNm, StemmerTypeNm, MxNGramLen, MnNGramFq, SaveDocP);
  } else
  if (!InReuters21578FPath.Empty()){ // Reuters-21578-file
    BowDocBs=TBowFl::LoadReuters21578Txt(InReuters21578FPath, Recs,
     SwSetTypeNm, StemmerTypeNm, MxNGramLen, MnNGramFq);
  } else 
  if (!InCiaWFBFPath.Empty()){ // CIA-World-Fact-Book
    BowDocBs=TBowFl::LoadCiaWFBTxt(InCiaWFBFPath, Recs,
     SwSetTypeNm, StemmerTypeNm, MxNGramLen, MnNGramFq);
  } else 
  if (!InDaxFNm.Empty()) { // DocumentAtlasXml-File
    PSwSet SwSet=TSwSet::GetSwSet(SwSetTypeNm);
    PStemmer Stemmer=TStemmer::GetStemmer(StemmerTypeNm);
    BowDocBs=TVizMapXmlDocBs::LoadBowDocBs(InDaxFNm, 
     SwSet, Stemmer, MxNGramLen, MnNGramFq);
  } else {
    TExcept::Throw("No Input-File specified!");
  }

  // save bow-file
  if (!OutBowFNm.Empty()){
    TStr::PutFExtIfEmpty(OutBowFNm, ".Bow");
    printf("Saving Bow to '%s' ...", OutBowFNm.CStr());
    BowDocBs->SaveBin(OutBowFNm);
    printf(" Done.\n");
  }

  // save statistics
  if ((OutStatP)&&(!OutBowFNm.Empty())){
    // save bow-statistics
    TStr OutBowStatFNm=TStr::PutFExt(OutBowFNm, ".Bow.Txt");
    TStr::PutFExt(OutBowStatFNm, ".Bow.Txt");
    if (!OutBowStatFNm.Empty()){
      printf("Saving Bow-Statistics to '%s' ...", OutBowStatFNm.CStr());
      BowDocBs->SaveTxtStat(OutBowStatFNm);
      printf(" Done.\n");
    }
    // save ngram-statistics
    PNGramBs NGramBs=BowDocBs->GetNGramBs();
    TStr OutNGramStatFNm=TStr::PutFExt(OutBowFNm, ".NGram.Txt");
    if (!NGramBs.Empty()){
      printf("Saving NGram-Statistics to '%s' ...", OutNGramStatFNm.CStr());
      NGramBs->SaveTxt(OutNGramStatFNm, true);
      printf(" Done.\n");
    }
  }

  return 0;
  Catch;
  return 1;
}
开发者ID:,项目名称:,代码行数:101,代码来源:


注:本文中的PBowDocBs::SaveTxtStat方法示例由纯净天空整理自Github/MSDocs等开源代码及文档管理平台,相关代码片段筛选自各路编程大神贡献的开源项目,源码版权归原作者所有,传播和使用请参考对应项目的License;未经允许,请勿转载。