public void KeywordValidator_passRightAndWringStrings_validateResults()
{
var sut = new KeywordValidator();
Assert.True(sut.Validate("word"), "Simple word should pass");
Assert.True(sut.Validate("Word"), "Words with different cases should pass");
Assert.True(sut.Validate("PhD"), "Words such as abbreviations should pass");
Assert.True(sut.Validate("One two three"), "Composed words acceptable");
Assert.True(sut.Validate("français"), "All UTF8 word symbols should pass");
Assert.True(sut.Validate("français русский übersetzen"), "All UTF8 word symbols should pass");
Assert.True(sut.Validate("dad's"), "' should pass");
Assert.True(sut.Validate("ABC."), "dot is acceptable");
Assert.True(sut.Validate("data-structure"), "dash is ok");
Assert.True(sut.Validate("1"), "Numbers are ok You can find something like 3-d say in MOBY examples");
Assert.True(sut.Validate("test1"), "Numbers are ok You can find something like 3-d say in MOBY examples");
Assert.True(sut.Validate("$100-a-plate dinner"), "Numbers are ok You can find something like $100-a-plate dinner say in MOBY examples");
Assert.False(sut.Validate("five%"), "% is not a word symbol");
Assert.False(sut.Validate("#1"), "# is not a word symbol");
Assert.False(sut.Validate("synonym #4"), "# is not a word symbol");
Assert.False(sut.Validate("contains\n"), "terminal symbols should rise error");
Assert.False(sut.Validate("\n"), "terminal symbols should rise error");
Assert.False(sut.Validate("False!"), "Sentence marks are not ok");
Assert.False(sut.Validate("True?"), "Sentence marks are not ok");
}
/// <summary>
/// Reads file with synonyms from Ripicts format (word, syn1;raiing1,syn2;raiting2 etc.)
/// </summary>
/// <param name="filename">The filename.</param>
/// <param name="datastructure">The in memory storage structure.</param>
/// <exception cref="System.ArgumentNullException">data structure must me initialized and could not be null</exception>
/// <exception cref="System.IO.FileNotFoundException">Passed filename with keywords not found.</exception>
/// <exception cref="System.IO.IOException">Failed to read file.</exception>
public void ReadFile(string filename, ref IinMemoryStorageStructure datastructure)
{
if (datastructure == null)
{
throw new ArgumentNullException("datastructure must me initialized and could not be null");
}
if (!File.Exists(filename))
{
throw new FileNotFoundException("Passed filename with keywords not found.", filename);
}
int stringCounter = 0;
using (StreamReader reader = new System.IO.StreamReader(filename))
{
KeywordValidator validator = new KeywordValidator();
while (!reader.EndOfStream)
{
stringCounter++;
var input = reader.ReadLine();
// empty string ignore them
if (!input.Trim().Equals(string.Empty))
{
string keyword;
var synRelevanceCollection = new List<Tuple<string, int>>();
if (input.IndexOf(',') > 0)
{
keyword = input.Substring(0, input.IndexOf(","));
var collection = input.Substring(input.IndexOf(",") + 1).Trim().Split(',').Select(s => s.Trim()).ToArray();
for (int i = 0; i < collection.Length; i++)
{
var element = collection[i];
var pair = element.Trim().Split(';');
if (pair.Length > 0 && validator.Validate(pair[0]))
{
var synvalue = pair[0].Trim();
int relValue = 0;
if (pair.Length > 1)
{
int.TryParse(pair[1], out relValue);
}
if (!synvalue.Equals(string.Empty))
{
synRelevanceCollection.Add(new Tuple<string, int>(synvalue, relValue));
}
}
else
{
throw new FormatException(string.Format("Format on the line {0} is not right. Missing ';' symbol or value '{1}' illegal.", stringCounter, element));
}
}
}
else
{
// means that this is just a word without synonyms;
keyword = input.Trim();
}
try
{
datastructure.AddSynonyms(keyword, synRelevanceCollection.ToArray());
}
catch (Exception ex)
{
throw new Exception(string.Format("Format on the line {0}.", stringCounter), ex);
}
}
}
}
}