static void Main(string[] args)
{
XmlDocument doc = new XmlDocument();
doc.Load("PWREE.da3");
XmlNode dict = doc.ChildNodes[1];
XmlNodeList dcList = dict.SelectNodes("单词块");
FileStream fs = new FileStream("result.txt",FileMode.Create);
StreamWriter writer = new StreamWriter(fs);
WordsssDB.WordsssDBManager manager = new WordsssDB.WordsssDBManager();
int i = 0;
foreach (XmlNode dcNode in dcList)
{
XmlNode xgNode = dcNode.SelectSingleNode("单词解释块/基本词义/单词项/相关词");
// if(xgNode != null)
Regex patter1 = new Regex("l{[a-z.A-Z]+}");
MatchCollection matches = patter1.Matches(xgNode.InnerText);
XmlNode wordNode = dcNode.SelectSingleNode("单词");
string base_word = wordNode.InnerText;
writer.WriteLine(wordNode.InnerText);
if(matches.Count != 0)
writer.WriteLine("<1>");
foreach(Match word in matches)
{
string rep = word.ToString().Replace("l{", "");
rep = rep.Replace("}", "");
writer.WriteLine(" " + rep);
if (manager.addAntonym(base_word, rep) == -1)
{
writer.WriteLine("FAILED");
}
}
Regex pattern2 = new Regex("L{[a-z.A-Z]+}");
MatchCollection matches2 = pattern2.Matches(xgNode.InnerText);
if(matches2.Count != 0)
writer.WriteLine("<2>");
foreach (Match word in matches2)
{
string rep = word.ToString().Replace("L{", "");
rep = rep.Replace("}", "");
writer.WriteLine(" " + rep);
if (manager.addSynonym(base_word, rep) == -1)
{
writer.WriteLine("FAILED");
}
}
if(i++ % 100 == 0)
Console.WriteLine(i);
}
writer.Close();
}
static void Main(string[] args)
{
XmlDocument doc = new XmlDocument();
doc.Load("AHD.xml");
FileStream fs = new FileStream("result.txt", FileMode.Create);
StreamWriter writer = new StreamWriter(fs);
XmlNode dictNode = doc.ChildNodes[1];
int MAX_WORD = 100000;
WordsssDB.WordsssDBManager manager = new WordsssDB.WordsssDBManager();
for (int i = 0; i < MAX_WORD && i < dictNode.ChildNodes.Count; i ++ )
{
string word_name = "";
XmlNode ckNode = dictNode.ChildNodes[i];
XmlNode dcNode = ckNode.SelectSingleNode("单词");
if (dcNode != null)
word_name = dcNode.FirstChild.Value;
word_name = word_name.Replace("'", "''");
XmlNode jcNode = ckNode.SelectSingleNode("单词解释块/继承用法");
if (jcNode != null)
{
XmlNodeList jcList = jcNode.SelectNodes("单词项/单词原型");
if (jcList.Count != 0)
{
writer.WriteLine(word_name + " " + jcList.Count);
foreach (XmlNode jc in jcList)
{
string str = jc.FirstChild.Value.Replace("&2{”}", "");
str = str.Replace("&2{“}", "");
str = str.Replace(" 或", "");
str = str.Replace("'", "''");
writer.WriteLine(" " +str);
if (manager.addDerivation(word_name, str) == -1)
{
writer.WriteLine("FAILED");
}
}
}
}
if(i % 1000 == 0)
{
Console.WriteLine(i);
}
}
}
static void Main(string[] args)
{
FileStream fs = new FileStream("1_1_all_fullalpha.txt",FileMode.Open);
StreamReader reader = new StreamReader(fs);
FileStream outFile = new FileStream("out2.txt",FileMode.Create);
StreamWriter writer = new StreamWriter(outFile);
WordsssDB.WordsssDBManager manager = new WordsssDB.WordsssDBManager();
int i = 0;
bool bAddWord = false;
Regex pattern = new Regex("^[a-zA-Z]");
HashSet<string> strHash = new HashSet<string>();
Dictionary<string, Frequency> strDict = new Dictionary<string, Frequency>();
while (!reader.EndOfStream)
{
string strLine = reader.ReadLine();
string[] splitLine = strLine.Split(new char[]{'\t'},StringSplitOptions.RemoveEmptyEntries);
int frequency = int.Parse(splitLine[3]);
int frequency2 = int.Parse(splitLine[4]);
double frequency3 = double.Parse(splitLine[5]);
if (frequency == 0 && splitLine[2] == ":")
{
bAddWord = false;
continue;
}
string str;
if (splitLine[2] == "%" && frequency != 0)
{
bAddWord = true;
continue;
}
else if (splitLine[2] == ":")
{
bAddWord = false;
}
if (splitLine[0] == "@")
str = splitLine[2];
else
str = splitLine[0];
if (!pattern.IsMatch(str))
continue;
if (bAddWord == false && splitLine[2] != ":")
continue;
if (!strHash.Contains(str))
{
strHash.Add(str);
strDict.Add(str, new Frequency(frequency, frequency2, frequency3));
}
else
{
strDict[str].frequency1 += frequency;
}
// Console.WriteLine(manager.addFrequency(current, currentFrequency, currentFrequency2, frequency3));
}
int j = 0;
Console.WriteLine(strHash.Count);
foreach (string str in strHash)
{
//int j = manager.addFrequency(str, strDict[str].frequency1, strDict[str].frequency2, strDict[str].frequency3);
if (j % 100 == 0)
Console.WriteLine(j);
//if (str.Contains('\''))
//{
string strRep = str.Replace("'", "''");
// if (strDict[str].frequency1 == 0)
// { // manager.addFrequency(strRep, strDict[str].frequency1, strDict[str].frequency2, strDict[str].frequency3);
if (manager.addFrequency(strRep, strDict[str].frequency1, strDict[str].frequency2, strDict[str].frequency3) == -1)
{
Console.WriteLine(str);
Console.Read();
}
writer.WriteLine(str + " " + strDict[str].frequency1 + "," + strDict[str].frequency2 + "," + strDict[str].frequency3);
//}
j++;
//}
}
writer.WriteLine(strHash.Count());
writer.WriteLine(j);
manager.CloseManager();
writer.Close();
}
static void Main(string[] args)
{
int MAX_WORD_COUNT = 130000;
int BEGIN_WORD = 0;//未更新
XmlDocument doc = new XmlDocument();
doc.Load("AHD - Copy.xml");
FileStream fs = new FileStream("sound.txt", FileMode.Create);
StreamWriter writer = new StreamWriter(fs);
WordsssDB.WordsssDBManager manager = new WordsssDB.WordsssDBManager();
XmlNode dictNode = doc.ChildNodes[1];
Console.WriteLine(dictNode.ChildNodes.Count);
for (int i = BEGIN_WORD; i < (BEGIN_WORD + MAX_WORD_COUNT) && i < dictNode.ChildNodes.Count; i++)
{
string word_name;
string word_type;
XmlNode ckNode = dictNode.ChildNodes[i];
XmlNode dcNode = ckNode.SelectSingleNode("单词");
if (dcNode == null)
continue;
word_name = dcNode.FirstChild.Value;
word_name = processString(word_name);
if (i % 1000 == 0)
Console.WriteLine(i);
int word_id = -1;
if ((word_id = manager.getWordId(word_name)) == -1)
continue;
//Console.WriteLine(word_name);
XmlNodeList jxNodeList = ckNode.SelectNodes("单词解释块");
foreach (XmlNode jxNode in jxNodeList)
{
XmlNode dxNode = jxNode.SelectSingleNode("基本词义/单词词性");
if (dxNode != null)
{
word_type = dxNode.FirstChild.Value;
}
else
word_type = "";
XmlNode ybNode = jxNode.SelectSingleNode("基本词义/单词音标/国际音标");
if (ybNode == null)
continue;
string word_sound = ybNode.FirstChild.Value;
Regex pattern = new Regex("{([^}]*)}");
if(pattern.IsMatch(word_sound))
word_sound = pattern.Match(word_sound).Groups[1].Value;
word_sound = word_sound.Replace("'","''");
writer.WriteLine(word_name + " " + ybNode.FirstChild.Value + " " + getWordType(word_type));
manager.updateAHDSound(word_id, getWordType(word_type), word_sound);
}
//writer.WriteLine(doc.ChildNodes[1].FirstChild.SelectSingleNode("//JX").FirstChild.Value);
}
writer.Close();
}
static void Main(string[] args)
{
FileStream fs = new FileStream("1_1_all_fullalpha.txt", FileMode.Open);
StreamReader reader = new StreamReader(fs);
FileStream writerFile = new FileStream("result.txt", FileMode.Create);
StreamWriter writer = new StreamWriter(writerFile);
bool hasConversion = false;
Regex pattern = new Regex("^[a-zA-Z]");
int count = 0;
int i = 0;
string current_word = "";
HashSet<string> wordHash = new HashSet<string>();
Dictionary<string, HashSet<string>> conversionDic = new Dictionary<string, HashSet<string>>();
while (!reader.EndOfStream)
{
string strLine = reader.ReadLine();
string[] strSplit = strLine.Split(new char[]{'\t'},StringSplitOptions.RemoveEmptyEntries);
int frequency = int.Parse(strSplit[3]);
if(i++ % 10000 == 0)
Console.WriteLine(i);
if(strSplit[2]== "%" && frequency != 0)
{
if (!pattern.IsMatch(strSplit[0]))
continue;
hasConversion = true;
current_word = strSplit[0];
wordHash.Add(current_word);
// count++;
continue;
}
else if (strSplit[2] == ":")
{
hasConversion = false;
if (frequency != 0 && pattern.IsMatch(strSplit[0]))
{
wordHash.Add(strSplit[0]);
count++;
}
continue;
}
else if (strSplit[2] == "%" && frequency == 0)
{
hasConversion = false;
continue;
}
if (hasConversion == true)
{
count++;
wordHash.Add(strSplit[2]);
if (conversionDic.Keys.Contains(current_word)&& current_word != strSplit[2])
conversionDic[current_word].Add(strSplit[2]);
else if(current_word != strSplit[2]){
conversionDic.Add(current_word, new HashSet<string> { strSplit[2] });
}
}
}
Console.WriteLine(count);
Console.WriteLine(wordHash.Count());
Console.WriteLine(conversionDic.Count());
int countDict = 0;
WordsssDB.WordsssDBManager manager = new WordsssDB.WordsssDBManager();
foreach (string strConversion in wordHash)
{
/// writer.WriteLine(strConversion);
if (conversionDic.Keys.Contains(strConversion))
{
countDict++;
string word_base = strConversion.Replace("'", "''");
writer.WriteLine(countDict + " " + word_base);
foreach (string word in conversionDic[strConversion])
{
string word_conversion = word.Replace("'", "''");
writer.WriteLine(" " + word_conversion);
if (manager.addConversion(word_base,word_conversion) == -1)
{
writer.WriteLine("FAILED");
}
}
}
}
writer.WriteLine(countDict);
writer.WriteLine(conversionDic.Keys.Count());
writer.WriteLine(wordHash.Count());
writer.Close();
}
