public void SeveralWordsTest()
{
var lookingWords = new List <string>()
{
"text", "search"
};
var fsa = new SearchFsa <char>();
foreach (var word in lookingWords)
{
fsa.Add(word);
}
fsa.Prepare();
var words = new List <string>();
var ep = fsa.GetEndPoint((index, context) =>
{
var currentWord = context.ToString();
words.Add(currentWord);
});
var data = "the search text example";
foreach (var ch in data)
{
ep.ProcessIncome(ch);
}
foreach (var word in lookingWords)
{
Assert.AreEqual(words.Contains(word), true);
}
}
public void LoremSampleTest()
{
var lookingWords = new List <string>()
{
"vitae", "magna", "tempus", "lectus", "varius", "libero", "sed", "nunc"
};
var fsa = new SearchFsa <char>();
foreach (var word in lookingWords)
{
fsa.Add(word);
}
fsa.Prepare();
var words = new List <string>();
var ep = fsa.GetEndPoint((index, context) =>
{
var currentWord = context.ToString();
words.Add(currentWord);
});
foreach (var ch in Lorem5)
{
ep.ProcessIncome(ch);
}
foreach (var word in lookingWords)
{
Assert.AreEqual(words.Contains(word), true);
}
}
public void OverlappingTextTest()
{
var lookingWords = new List <string>()
{
"test", "tester", "eres", "este", "est", "ste"
};
var fsa = new SearchFsa <char>();
foreach (var word in lookingWords)
{
fsa.Add(word);
}
fsa.Prepare();
var words = new List <string>();
var ep = fsa.GetEndPoint((index, context) =>
{
var currentWord = context.ToString();
words.Add(currentWord);
});
var data = "tetetestesterest";
foreach (var ch in data)
{
ep.ProcessIncome(ch);
}
foreach (var word in lookingWords)
{
Assert.AreEqual(words.Contains(word), true);
}
}
public List <IToken> IntGetTokens(string sourceText, IFileSource fileSource, OffsetsStore offsetsStore)
{
var clearText = ClearSourceText(sourceText, offsetsStore);
SearchFsa <char> fsa = PrepareFsa();
var tokens = new List <IToken>();
var ep = fsa.GetEndPoint((index, context) =>
{
var currentWord = context.ToString();
var token = new CsToken()
{
Code = CSharpKeywords.Keywords[currentWord],
Position = index,
FileSource = fileSource,
StringLength = currentWord.Length
};
tokens.Add(token);
});
foreach (var ch in sourceText)
{
ep.ProcessIncome(ch);
}
return(tokens);
}
private static SearchFsa <char> PrepareFsa()
{
if (_fsa == null)
{
_fsa = new SearchFsa <char>();
foreach (var word in CSharpKeywords.Keywords)
{
_fsa.Add(word.Key);
}
_fsa.Prepare();
}
return(_fsa);
}
public void SingleWordTest()
{
var fsa = new SearchFsa <char>();
fsa.Add("text");
fsa.Prepare();
var correct = false;
var ep = fsa.GetEndPoint((index, context) =>
{
correct = true;
});
var data = "the search text example";
foreach (var ch in data)
{
ep.ProcessIncome(ch);
}
Assert.AreEqual(correct, true);
}
public void OverlappingTextTest2()
{
var lookingWords = new List <string>()
{
"telega", "eleg"
};
var fsa = new SearchFsa <char>();
foreach (var word in lookingWords)
{
fsa.Add(word);
}
fsa.Prepare();
var words = new Dictionary <string, int>();
var ep = fsa.GetEndPoint((index, context) =>
{
var currentWord = context.ToString();
if (!words.ContainsKey(currentWord))
{
words[currentWord] = 0;
}
words[currentWord]++;
});
var data = "telegrammaotelega";
foreach (var ch in data)
{
ep.ProcessIncome(ch);
}
var keys = words.Keys.ToArray();
Func <string, int, bool> checkWord = (word, cnt) => keys.Contains(word) && (words[word] == cnt);
Assert.AreEqual(checkWord("telega", 1), true);
Assert.AreEqual(checkWord("eleg", 2), true);
}
