Sha256: 53932e2e251dc46d10f409d4426c45625313ae7e8a0513c0c8de4cfb7a9d7030

Contents?: true

Size: 1008 Bytes

Versions: 11

Compression:

Stored size: 1008 Bytes

Contents

#ifndef CPPJIEBA_SEGMENTBASE_H
#define CPPJIEBA_SEGMENTBASE_H

#include "limonp/Logging.hpp"
#include "PreFilter.hpp"
#include <cassert>


namespace cppjieba {

const char* const SPECIAL_SEPARATORS = " \t\n\xEF\xBC\x8C\xE3\x80\x82";

using namespace limonp;

class SegmentBase {
 public:
  SegmentBase() {
    XCHECK(ResetSeparators(SPECIAL_SEPARATORS));
  }
  virtual ~SegmentBase() {
  }

  virtual void Cut(const string& sentence, vector<string>& words) const = 0;

  bool ResetSeparators(const string& s) {
    symbols_.clear();
    RuneStrArray runes;
    if (!DecodeRunesInString(s, runes)) {
      XLOG(ERROR) << "decode " << s << " failed";
      return false;
    }
    for (size_t i = 0; i < runes.size(); i++) {
      if (!symbols_.insert(runes[i].rune).second) {
        XLOG(ERROR) << s.substr(runes[i].offset, runes[i].len) << " already exists";
        return false;
      }
    }
    return true;
  }
 protected:
  unordered_set<Rune> symbols_;
}; // class SegmentBase

} // cppjieba

#endif

Version data entries

11 entries across 11 versions & 2 rubygems

Version Path
cppjieba_rb-0.4.4 ext/cppjieba/include/cppjieba/SegmentBase.hpp
cppjieba_rb-0.4.3 ext/cppjieba/include/cppjieba/SegmentBase.hpp
cppjieba_rb-0.4.2 ext/cppjieba/include/cppjieba/SegmentBase.hpp
cppjieba_rb-0.4.1 ext/cppjieba/include/cppjieba/SegmentBase.hpp
jieba-rb-5.0.0 ext/cppjieba/include/cppjieba/SegmentBase.hpp
cppjieba_rb-0.3.3 ext/cppjieba/include/cppjieba/SegmentBase.hpp
cppjieba_rb-0.3.1 ext/cppjieba/include/cppjieba/SegmentBase.hpp
cppjieba_rb-0.3.0 ext/cppjieba/include/cppjieba/SegmentBase.hpp
cppjieba_rb-0.2.3 ext/cppjieba/include/cppjieba/SegmentBase.hpp
cppjieba_rb-0.2.2 ext/cppjieba/include/cppjieba/SegmentBase.hpp
cppjieba_rb-0.2.1 ext/cppjieba/include/cppjieba/SegmentBase.hpp