static void Main(string[] args)
{
string result_folder = @"f:\tmp";
List <string> parsed_sentences = new List <string>();
int MAX_SAMPLE = int.MaxValue;
int MAX_LEN = int.MaxValue;
string dictionary_path = "";
string syntax_templates = "";
string[] filters = { "3" };
#region Command_Line_Options
for (int i = 0; i < args.Length; ++i)
{
if (args[i] == "-parsing")
{
parsed_sentences.Add(args[i + 1]);
i++;
}
else if (args[i] == "-dict")
{
dictionary_path = args[i + 1];
i++;
}
else if (args[i] == "-templates")
{
syntax_templates = args[i + 1];
i++;
}
else if (args[i] == "-output")
{
result_folder = args[i + 1];
i++;
}
else if (args[i] == "-max_samples")
{
MAX_SAMPLE = int.Parse(args[i + 1]);
i++;
}
else if (args[i] == "-skip")
{
nb_skip = int.Parse(args[i + 1]);
i++;
}
else if (args[i] == "-max_len")
{
MAX_LEN = int.Parse(args[i + 1]);
i++;
}
else if (args[i] == "-filter")
{
filters = args[i + 1].Split(',');
i++;
}
else
{
throw new ApplicationException(string.Format("Unknown option {0}", args[i]));
}
}
#endregion Command_Line_Options
preprocessor = new Preprocessor();
if (!string.IsNullOrEmpty(syntax_templates))
{
syntax_checker.LoadTemplates(syntax_templates);
}
Console.WriteLine("Loading dictionary {0}", dictionary_path);
SolarixGrammarEngineNET.GrammarEngine2 gren = new SolarixGrammarEngineNET.GrammarEngine2();
gren.Load(dictionary_path, true);
// Файл для сохранения отобранных предложений-фактов.
wrt_samples = new System.IO.StreamWriter(System.IO.Path.Combine(result_folder, "facts.txt"));
// Предложения, которые не прошли детальную проверку синтаксической структуры
wrt_skipped = new System.IO.StreamWriter(System.IO.Path.Combine(result_folder, "skipped.txt"));
// Фильтр для предиката, с возможными значениями "3", "2s" и "1s"
string filter_verb = filters.Where(z => "3 1s 2s".Split(' ').Contains(z)).FirstOrDefault();
// Фильтр типа предложения. Допустимые значение - пустое или "q"
string filter_sent = (filters.Where(z => "q".Split(' ').Contains(z)).FirstOrDefault()) ?? "";
DateTime start_time = DateTime.Now;
#region Processing_All_Files
foreach (string mask in parsed_sentences)
{
string[] files = null;
if (System.IO.Directory.Exists(mask))
{
files = System.IO.Directory.GetFiles(mask, "*.parsing.txt");
}
else if (mask.IndexOfAny("*?".ToCharArray()) != -1)
{
files = System.IO.Directory.GetFiles(System.IO.Path.GetDirectoryName(mask), System.IO.Path.GetFileName(mask));
}
else
{
files = new string[1] {
mask
};
}
Console.WriteLine("Number of parsing files={0}", files.Length);
foreach (string file in files)
{
if (sample_count >= MAX_SAMPLE)
{
break;
}
Console.WriteLine("Processing {0}...", file);
using (Sentences src = new Sentences(file))
{
while (src.Next() && sample_count < MAX_SAMPLE)
{
Sentence sent = src.GetFetched();
sample_count++;
if (sent.root == null)
{
continue;
}
if (sample_count > 0 && (sample_count % 10000) == 0)
{
Console.Write("{0} samples extracted\r", sample_count);
}
if (sample_count >= nb_skip)
{
//Console.WriteLine("DEBUG [{0}] {1}", sample_count, sent.GetText());
ProcessSentence(gren, sent, MAX_LEN, filter_verb, filter_sent);
}
}
}
}
}
#endregion Processing_All_Files
wrt_samples.Close();
wrt_skipped.Close();
return;
}
