| 26 | using opencc::UTF8StringSlice; |
| 27 | |
| 28 | void Extract(const std::vector<std::string>& inputFiles, |
| 29 | const std::string& outputFile) { |
| 30 | std::ostringstream buffer; |
| 31 | for (const auto& inputFile : inputFiles) { |
| 32 | std::ifstream ifs(inputFile); |
| 33 | const std::string contents((std::istreambuf_iterator<char>(ifs)), |
| 34 | (std::istreambuf_iterator<char>())); |
| 35 | buffer << contents; |
| 36 | } |
| 37 | const std::string& text = buffer.str(); |
| 38 | PhraseExtract extractor; |
| 39 | extractor.SetWordMaxLength(2); |
| 40 | extractor.SetPrefixSetLength(1); |
| 41 | extractor.SetSuffixSetLength(1); |
| 42 | extractor.Extract(text); |
| 43 | std::ofstream ofs(outputFile); |
| 44 | for (const auto& word : extractor.Words()) { |
| 45 | const PhraseExtract::Signals& signals = extractor.Signal(word); |
| 46 | const double entropy = signals.prefixEntropy + signals.suffixEntropy; |
| 47 | const double logProbablity = extractor.LogProbability(word); |
| 48 | ofs << word << " " << signals.frequency << " " << logProbablity << " " |
| 49 | << signals.cohesion << " " << entropy << " " << signals.prefixEntropy |
| 50 | << " " << signals.suffixEntropy << std::endl; |
| 51 | } |
| 52 | ofs.close(); |
| 53 | } |
| 54 | |
| 55 | int main(int argc, const char* argv[]) { |
| 56 | try { |
no test coverage detected