public void NullOrEmptyStringsDistanceCheck()
{
string s1 = null;
string s2 = "test";
Assert.IsTrue(EditDistance.LevenshteinDistance(s1, s2) == s2.Length);
var temp = s1;
s1 = s2;
s2 = temp;
Assert.IsTrue(EditDistance.LevenshteinDistance(s1, s2) == s1.Length);
s2 = string.Empty;
Assert.IsTrue(EditDistance.LevenshteinDistance(s1, s2) == s1.Length);
temp = s1;
s1 = s2;
s2 = temp;
Assert.IsTrue(EditDistance.LevenshteinDistance(s1, s2) == s2.Length);
s1 = null;
s2 = string.Empty;
Assert.IsTrue(EditDistance.LevenshteinDistance(s1, s2) == 0);
temp = s1;
s1 = s2;
s2 = temp;
Assert.IsTrue(EditDistance.LevenshteinDistance(s1, s2) == 0);
}
public int GoogleSearch(string phrase, string error, out string suggestion)
{
suggestion = "";
int hits = 0;
string url = string.Format("https://www.google.co.id/search?q=\"{0}\"", phrase.Trim());
Random r = new Random();
int interval = r.Next(5000, 9000);
Thread.Sleep(interval);
using (ChromeDriver driver = new ChromeDriver(SeleniumFolder))
{
driver.Navigate().GoToUrl(url);
try
{
int.TryParse(Regex.Replace(driver.FindElement(By.XPath("//div[@id=\"resultStats\"]")).Text, "[^0-9]", ""), out hits);
}
catch { }
try
{
IWebElement noresultNode = driver.FindElement(By.XPath("//div[@id=\"topstuff\"]"));
if (noresultNode.Text.StartsWith("Hasil untuk") && noresultNode.Text.Contains("tidak ditemukan"))
{
hits = 0;
}
}
catch { }
try
{
IWebElement spellNode = driver.FindElement(By.XPath("//a[@class=\"spell\"]"));
var tmpA = spellNode.GetAttribute("innerHTML").Replace(""", "");
string[] arrTmp = Regex.Split(tmpA, "<b><i>");
Dictionary <string, int> candidates = new Dictionary <string, int>();
foreach (string tmp in arrTmp)
{
if (tmp.Trim().Contains("</i></b>"))
{
int posAkhir = tmp.Trim().IndexOf("</i></b>");
string nline = tmp.Trim().Substring(0, posAkhir);
int levenshtein = EditDistance.LevenshteinDistance(nline, error, 4);
if (levenshtein != -1 && levenshtein <= 4)
{
candidates.Add(nline, levenshtein);
}
}
}
if (candidates.Count > 1)
{
var temp = candidates.OrderBy(p => p.Value).ToDictionary(p => p.Key, p => p.Value);
suggestion = temp.First().Key;
}
if (candidates.Count == 1)
{
suggestion = candidates.First().Key;
}
}
catch { }
}
return(hits);
}
public void ReversedStringsDistanceCheck()
{
string s1 = "asdfghjkl";
string s2 = "lkjhgfdsa";
// Expected length - 1, because one char is on the same place (odd number of chars)
Assert.IsTrue(EditDistance.LevenshteinDistance(s1, s2) == s1.Length - 1);
s1 = "qazwsxedcrfv";
s2 = "vfrcdexswzaq";
Assert.IsTrue(EditDistance.LevenshteinDistance(s1, s2) == s1.Length);
}
public void DifferentLengthStringsDistanceCheck()
{
string s1 = "short string";
string s2 = "long string, but not so much";
System.Diagnostics.Trace.WriteLine(EditDistance.LevenshteinDistance(s1, s2));
// Expected length - 7, because we have 7 consecutive chars that are the same in both strings
Assert.IsTrue(EditDistance.LevenshteinDistance(s1, s2) == s2.Length - 7);
s1 = "another one";
s2 = "and another one";
// Expected 4, because 4 chars from the second are not present in the first
Assert.IsTrue(EditDistance.LevenshteinDistance(s1, s2) == 4);
}
public List <CorrectionCandidate> GetCandidatesFromAffixCorrection(int key, string error, out string log)
{
//[prefix-1] + [prefix-2] + root + [suffix] + [possessive] + [particle]
//1.Particles: -lah, -kah, -pun, -tah.
//2.Possessives: -ku, -mu, -nya.
//3.Suffixes: -i, -an, -kan.
//4.Prefixes: meN -, beN -, peN -, teN -, di -, ke -, se -.
log = "";
List <string> Prefixs = new List <string>()
{
"", "di", "ke", "se", "ber", "bel", "be", "te", "ter", "me", "mem", "men", "meng", " menge", "meny", "pe", "per", "pem", "pen", "peng", "penge", "peny", "pel", "memper", "mempel", "menter", "member", "diper", "diter", "dipel", " diber", "keber", "keter"
};
List <string> baseSuffixes = new List <string>()
{
"i", "an", "kan"
};
List <string> possessives = new List <string>()
{
"ku", "mu", "nya"
};
List <string> particles = new List <string>()
{
"lah", "kah", "pun", "tah"
};
//List<string> akhirans = new List<string>() { "", "i", "an", "kan", "ku", "mu", "nja", "lah", "kah", "pun", "tah", "iku", "imu", "inja", "anku", "anmu", "annja", "kanku", "kanmu", "kannja", "ilah", "ikah", "ipun", "itah", "anlah", "ankah", "anpun", "antah", "kanlah", "kankah", "kanpun", "kantah", "kulah", "kukah", "kupun", "kutah", " mulah", "mukah", "mupun", "mutah", " nyalah", "nyakah", "nyapun", "nyatah", "ikulah", "ikukah", "ikupun", "ikutah", "imulah", "imukah", "imupun", "imutah", "inyalah", "inyakah", "inyapun", "inyatah", "ankulah", "ankukah", "ankupun", "ankutah", "anmulah", "anmukah", "anmupun", "anmutah", "annyalah", "annyakah", "annyapun", "annyatah", "kankulah", "kankukah", "kankupun", "kankutah", "kanmulah", "kanmukah", "kanmupun", "kanmutah", "kannjalah", "kannjakah", "kannjapun", "kannjatah" };
List <string> suffixes = new List <string>()
{
""
};
suffixes.AddRange(baseSuffixes);
suffixes.AddRange(possessives);
suffixes.AddRange(particles);
foreach (string s in baseSuffixes)
{
foreach (string po in possessives)
{
suffixes.Add(s + po); // contoh: diperbaikinya
foreach (string pa in particles)
{
suffixes.Add(s + pa); //contoh:dipelukanmulah
}
}
}
foreach (string s in baseSuffixes)
{
foreach (string pa in particles)
{
suffixes.Add(s + pa); // contoh: pertahankanlah
}
}
List <CorrectionCandidate> candidates = new List <CorrectionCandidate>();
Dictionary <string, int> dicCandidates = new Dictionary <string, int>();
Correction correct = new Correction();
string rootWord = GetRootWord(correct.ChangeOldToNewSpell(error));
if (rootWord == "" || rootWord.Length < 3)
{
return(candidates);
}
foreach (string prefix in Prefixs)
{
foreach (string suffix in suffixes)
{
Affixer affixer = new Affixer();
string candidate = correct.ChangeNewToOldSpell(affixer.Affixing(correct.ChangeNewToOldSpell(rootWord), prefix, suffix));
int levenshtein = EditDistance.LevenshteinDistance(candidate, error, 2);
if (levenshtein != -1 && levenshtein <= 2)
{
if (!dicCandidates.ContainsKey(correct.ChangeOldToNewSpell(candidate)))
{
dicCandidates.Add(correct.ChangeOldToNewSpell(candidate), levenshtein);
}
}
}
}
if (dicCandidates.Count == 0)
{
return(candidates);
}
Dictionary <string, int> dicCandidateAndFreq = GetFrequencies(dicCandidates.Keys.ToArray());
foreach (KeyValuePair <string, int> can in dicCandidates)
{
int frequency = 0;
if (dicCandidateAndFreq.ContainsKey(can.Key))
{
frequency = dicCandidateAndFreq[can.Key];
}
CorrectionCandidate corrcandidate = new CorrectionCandidate
{
Key = key,
Error = error,
Candidate = can.Key,
SameBigramAmount = -1,
Frequency = frequency,
LengthDifference = Math.Abs(can.Key.Length - correct.ChangeOldToNewSpell(error).Length),
Levensthein = can.Value
};
candidates.Add(corrcandidate);
log += can.Key + "," + can.Value + "," + frequency.ToString() + ";";
}
if (log.Length > 0)
{
log = "[" + log + "]";
}
return(candidates);
}
public List <CorrectionCandidate> GetCandidates(string spName, int key, string error, string root, string prefix, string suffix, int minSameBigramAmount, int minLengthVariant, int maxLevensthein, out string log)
{
// sample: call getCandidates('depat',2,0,1);
log = "";
List <CorrectionCandidate> lsCandidates = new List <CorrectionCandidate>();
MySqlConnection conn = new MySqlConnection();
conn.ConnectionString = MariaDBConn;
MySqlCommand cmd = new MySqlCommand();
try
{
conn.Open();
cmd.Connection = conn;
cmd.CommandText = spName;
cmd.CommandType = CommandType.StoredProcedure;
cmd.Parameters.AddWithValue("@Word", root);
cmd.Parameters["@Word"].Direction = ParameterDirection.Input;
cmd.Parameters.AddWithValue("@MinSameBigramAmount", minSameBigramAmount);
cmd.Parameters["@MinSameBigramAmount"].Direction = ParameterDirection.Input;
cmd.Parameters.AddWithValue("@MinLengthVariant", minLengthVariant);
cmd.Parameters["@MinLengthVariant"].Direction = ParameterDirection.Input;
cmd.Parameters.AddWithValue("@MaxLevensthein", maxLevensthein);
cmd.Parameters["@MaxLevensthein"].Direction = ParameterDirection.Input;
cmd.Parameters.AddWithValue("@IsLemma", 1);
cmd.Parameters["@IsLemma"].Direction = ParameterDirection.Input;
MySqlDataReader dataReader = cmd.ExecuteReader();
while (dataReader.Read())
{
string stem = dataReader["Unigram"].ToString();
Affixer affixer = new Affixer();
string sCandidate = affixer.Affixing(stem, prefix, suffix);
int levensthein = EditDistance.LevenshteinDistance(sCandidate, error, 2);
if (levensthein != -1 && levensthein <= 2)
{
CorrectionCandidate candidate = new CorrectionCandidate
{
Key = key,
Error = error,
Candidate = sCandidate,
SameBigramAmount = Convert.ToInt32(dataReader["SameBigramAmount"]),
Frequency = 0,
LengthDifference = Convert.ToInt32(dataReader["LengthDifference"]),
Levensthein = levensthein
};
lsCandidates.Add(candidate);
}
}
//close Data Reader
dataReader.Close();
}
catch (MySql.Data.MySqlClient.MySqlException ex)
{
throw new Exception(ex.Message);
}
conn.Close();
// Find Frequency then Update to list:
List <string> sCandidates = new List <string>();;
foreach (CorrectionCandidate cc in lsCandidates)
{
sCandidates.Add(cc.Candidate);
}
Dictionary <string, int> dicCandidateAndFreq = GetFrequencies(sCandidates.ToArray());
foreach (CorrectionCandidate candidate in lsCandidates)
{
if (dicCandidateAndFreq.ContainsKey(candidate.Candidate))
{
candidate.Frequency = dicCandidateAndFreq[candidate.Candidate];
}
log += candidate.Candidate + "," + candidate.Levensthein + "," + candidate.Frequency.ToString() + ";";
}
log = string.Format("[{0},{1},{2}][{3}]", minSameBigramAmount, minLengthVariant, maxLevensthein, log);
return(lsCandidates);
}
