示例#1
0
        /// <summary>
        /// Global method for entities generation
        /// </summary>
        /// <param name="inputPath">The input path</param>
        /// <param name="output">Output stream</param>
        static void GenerateEntities(string inputPath, TextWriter output, string language)
        {
            output.WriteLine("<wis>");

            var jarRoot = StanfordEnv.GetStanfordHome();
            var classifiersDirectory = jarRoot + StanfordEnv.CLASIFIERS;

            string[] fileEntries = FilesUtils.GetFiles(inputPath);

            foreach (var document in fileEntries)
            {
                string text = FilesUtils.FileToText(document);
                // XXX: Better a NullObject, but string can't be inherited I think.
                if (text == null)
                {
                    var stderr = new StreamWriter(Console.OpenStandardError());
                    stderr.WriteLine($"The file '{document}' is not supported");
                    stderr.Close();
                    continue;
                }

                var classifier = CRFClassifiers.GetClassifierByLang(language);                 //CRFClassifier.getClassifierNoExceptions(classifiersDirectory + StanfordEnv.GetNerLanguageFiles(language));

                output.WriteLine(classifier.classifyToString(text, "xml", true));
            }
            output.WriteLine("</wis>");
        }
示例#2
0
 public static bool CheckLangFiles(String language)
 {
     return(FilesUtils.ExistsModels(StanfordEnv.GetStanfordHome() +
                                    StanfordEnv.CLASIFIERS + @"/" + language
                                    + ".ancora.distsim.s512.crf.ser.gz"));
 }