static void SerializerTest(Corpora cops)
{
XmlSerializer ser = new XmlSerializer(typeof(Corpora));
string filename = cops.Language + ".xml";
TextWriter writer = new StreamWriter(filename);
ser.Serialize(writer, cops);
writer.Close();
}
private static void IrrigateCorporaWithBigrams(Corpora cops, List <Tuple <char, char> > bigrams)
{
foreach (Tuple <char, char> tuple_char_char in bigrams)
{
char first_char = tuple_char_char.Item1;
char second_char = tuple_char_char.Item2;
cops.IrrigateBigram(first_char, second_char);
}
}
static void CorporaIrrigationTest()
{
string[] arr_filenames = { @"E:\Dropbox\Codes Hub\C#\Corpus\text samples\english.txt", @"E:\Dropbox\Codes Hub\C#\Corpus\text samples\french.txt", @"E:\Dropbox\Codes Hub\C#\Corpus\text samples\italian.txt" };
foreach (string filename in arr_filenames)
{
Corpora cops = CorpAnalyzer.ExtractCorporaFromFile(filename);
SerializerTest(cops);
}
}
static void DeserializerTest()
{
Corpora cops = new Corpora();
XmlSerializer ser = new XmlSerializer(typeof(Corpora));
string filename = "data.xml";
TextReader reader = new StreamReader(filename);
cops = ser.Deserialize(reader) as Corpora;
reader.Close();
}
static void SerializerTest()
{
string language = "test";
Corpora cops = new Corpora();
cops.Init(language);
XmlSerializer ser = new XmlSerializer(typeof(Corpora));
string filename = "data.xml";
TextWriter writer = new StreamWriter(filename);
ser.Serialize(writer, cops);
writer.Close();
}
static List <Corpora> FuncDeserializeCorporaFiles(List <string> lst_filepaths)
{
List <Corpora> lst_cops = new List <Corpora>();
XmlSerializer ser = new XmlSerializer(typeof(Corpora));
foreach (string filename in lst_filepaths)
{
TextReader reader = new StreamReader(filename);
Corpora cops = new Corpora();
cops = ser.Deserialize(reader) as Corpora;
lst_cops.Add(cops);
reader.Close();
}
return(lst_cops);
}
static List <Corpora> DeserializerTest(int any_number)
{
List <Corpora> lst_cops = new List <Corpora>();
XmlSerializer ser = new XmlSerializer(typeof(Corpora));
string[] arr_filenames = { @"english.xml", @"french.xml", @"italian.xml" };
foreach (string filename in arr_filenames)
{
TextReader reader = new StreamReader(filename);
Corpora cops = new Corpora();
cops = ser.Deserialize(reader) as Corpora;
lst_cops.Add(cops);
reader.Close();
}
return(lst_cops);
}
static void FuncBuildCorporas()
{
OpenFileDialog openTextFileDialog = new OpenFileDialog();
openTextFileDialog.Filter = "txt files (*.txt)|*.txt|All files (*.*)|*.*";
openTextFileDialog.RestoreDirectory = true;
openTextFileDialog.Multiselect = true;
openTextFileDialog.Title = "Please Select Text File(s)";
List <Corpora> lst_cops = new List <Corpora>();
if (openTextFileDialog.ShowDialog() == DialogResult.OK)
{
foreach (string str_path in openTextFileDialog.FileNames)
{
Corpora cops = CorpAnalyzer.ExtractCorporaFromFile(str_path);
lst_cops.Add(cops);
}
FuncSerializeCorporaFiles(lst_cops);
}
}
public static Corpora ExtractCorporaFromFile(string filename)
{
string language = Path.GetFileName(filename).Replace(".txt", "");
Corpora cops = new Corpora();
cops.Init(language);
StreamReader file = new StreamReader(filename);
string line;
while ((line = file.ReadLine()) != null)
{
string str_in_process = TrimNoneAlphabetChars(line);
if (2 <= str_in_process.Length)
{
List <Tuple <char, char> > bigrams = BuildBigramsFromString(str_in_process);
IrrigateCorporaWithBigrams(cops, bigrams);
}
}
file.Close();
return(cops);
}
