public void Create_Constructor_ExpectSameString()
{
string expect = "Randy Butternubs";
Scrub st = new Scrub(expect);
Assert.IsNotNull(st.CharTransDict);
Assert.IsNotNull(st.RegxTuples);
Assert.IsNotNull(st.StringTransDict);
Assert.AreEqual(expect, st.ToString());
Assert.IsNotNull(st.RegxMatchesDefined);
}
public void Translate_EmptyEverything_ExpectSameString()
{
string expect = "Randy Butternubs";
Scrub st = new Scrub(expect);
st.Strip("").MapChars().MapWords().RegxTranslate().Strip("");
Assert.IsNotNull(st.CharTransDict);
Assert.IsNotNull(st.RegxTuples);
Assert.IsNotNull(st.StringTransDict);
Assert.AreEqual(expect, st.ToString());
}
public void MatchCase_RegxDefined_Matches()
{
string sentence = "wtf does RemoveWTF do? Is WtF Case SeNsItIvE?";
string expectedSentance = "XXX does RemoveWTF do? Is WtF Case SeNsItIvE?";
Scrub st = new Scrub(sentence);
st.Set(sentence);
st.RegxMatchesDefined.Add("RemoveWTF", @"(wtf)|(what the)\s+(hell|$hit)");
st.RegxDefined("RemoveWTF", "XXX");
Assert.AreEqual(expectedSentance, st.ToString());
}
public void SetString_AddStringAfter_ExpectNewString()
{
string expect = "Randy Butternubs";
Scrub st = new Scrub("Haystack Calhoon");
st.Set(expect);
st.Strip("").MapChars().MapWords().RegxTranslate().Strip("");
Assert.IsNotNull(st.CharTransDict);
Assert.IsNotNull(st.RegxTuples);
Assert.IsNotNull(st.StringTransDict);
Assert.AreEqual(expect, st.ToString());
}
public void Predefined_CompactWhitespace_Compacted()
{
string sentence = "¿¡Señor, the Chevrolet guys don't like Dodge guys, and and no one like MaZdA, Ola Senor?! ";
string expectedSentance = "¿¡Señor, the Chevrolet guys don't like Dodge guys, and and no one like MaZdA, Ola Senor?! ";
Scrub st = new Scrub(sentence);
// Compact whitespaces to one space, note does not imply trim!
// overides default empty string replace to replace with single space
// note trailing space at end of string
st.RegxDefined("WhitespaceCompact", " ");
Assert.AreEqual(expectedSentance, st.ToString());
}
public void TestAll()
{
// get most of the mapped accent chars, and their non-accented equiv
// must be 1 to 1 mapping and size of arrays. Easier to do lots of chars this
// way then with lists
string matchChar = "ŠŒŽšœžŸ¥µÀÁÂÃÄÅÆÇÈÉÊËÌÍÎÏÐÑÒÓÔÕÖØÙÚÛÜÝßàáâãäåæçèéêëìíîïðñòóôõöøùúûüýÿ¡¿";
string replaceChar = "SOZsozYYuAAAAAAACEEEEIIIIDNOOOOOOUUUUYsaaaaaaaceeeeiiiionoooooouuuuyy ";
// set up a dictionary, if ignore case, set the dict up with a new comparer
StringComparer comparer = StringComparer.OrdinalIgnoreCase; // default is just Ordinal
Dictionary <string, string> wordDictionary = new Dictionary <string, string>(comparer)
{
{ "chevrolet", "Ford" },
{ "mAzDa", "BMW" },
{ "and and", "and" }, // will never match
};
// Need `System.ValueTuple` package to do this style of init
// on v4.6 and below
List <(string, string)> regxList = new List <(string, string)>
{ // Match, Replace
("BMW", "Fiat"),
(@"\s+", " "), // multi whitespace to 1 space
(@"^\s*|\s*$", "") // trims leading/ending spaces
};
string sentence = "¿¡Señor, the Chevrolet guys don't like Dodge guys, and and no one like MaZdA, Ola Senor?! ";
string expectedSentance = "Magoo the Ford guys don#t like Mercedes guys and and no one like Fiat Ola Magoo?!";
Scrub st = new Scrub(sentence);
// Set dictionary up, case insensitive match
st.SetStringTranslator(wordDictionary, true);
// set up character translators
st.SetCharTranslator(matchChar, replaceChar);
// set up list of regx replaces
st.SetRegxTranslator(regxList);
// add a string translation after the fact
st.StringTransDict.Add("dodge", "Mercedes");
// add a Regx translation after the fact
st.RegxTuples.Add(("Senor", "Mr.Magoo"));
// add a chracter Translation after the fact
st.CharTransDict.Add('\'', '#');
// so all sorts of stuff!
st.Strip("[,]").MapChars().MapWords().RegxTranslate().Strip(@"Mr\.");
Assert.AreEqual(expectedSentance, st.ToString());
}