/// <summary>
/// Parse the document searching for sentences where the entity found.
/// Returns a csv line with the file, the entity the sentence and the sintax analisis of the sentences
/// </summary>
/// <param name="text">Document text</param>
/// <param name="entity">Entity.</param>
/// <param name="origFile">Original file.</param>
public static List <string[]> Parse(string text, string entity, string origFile, string language)
{
var results = new List <string[]>();
//Load spanish models.
var modelsDirectory = StanfordEnv.PARSER_MODELS;
var lexparserDirectory = modelsDirectory + StanfordEnv.GetParserLanguageFiles(language);
var lp = LexicalizedParser.loadModel(lexparserDirectory);
string[] splittedText = SplitText(text);
List <string> entityLines = GetEntitiesLines(splittedText, entity);
foreach (var line in entityLines)
{
//Parser sentence.
var tokenizerFactory = PTBTokenizer.factory(new CoreLabelTokenFactory(), "");
var sent2Reader = new java.io.StringReader(line);
var rawWords2 = tokenizerFactory.getTokenizer(sent2Reader).tokenize();
sent2Reader.close();
var tree2 = lp.apply(rawWords2);
results.Add(new string[] { origFile, entity, line, tree2.ToString() });
}
return(results);
}
/// <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>");
}
public static CRFClassifier GetClassifierByLang(string lang)
{
if (!classifiers.ContainsKey(lang))
{
classifiers.Add(lang,
CRFClassifier.getClassifierNoExceptions(classifiersDirectory + StanfordEnv.GetNerLanguageFiles(lang)));
}
return(classifiers[lang]);
}
public static bool CheckLangFiles(String language)
{
return(FilesUtils.ExistsModels(StanfordEnv.GetStanfordHome() +
StanfordEnv.CLASIFIERS + @"/" + language
+ ".ancora.distsim.s512.crf.ser.gz"));
}