public TextModel(string name, IFlexibleWordCountModel counts)
{
SetName(name);
SetIncludeQuotes(true);
if (counts == null)
{
throw new ArgumentException("counts must not be null.");
}
_counts = counts.Copy();
_length = _counts.GetLength();
}
public ITextModel GetTextModel(string name, IFlexibleWordCountModel counts)
{
if (counts == null)
{
throw new ArgumentException("Counts model must not be null.");
}
if (counts.GetLength() < 1)
{
throw new ArgumentException("Counts model must not have length 0.");
}
return(new TextModel(name, counts));
}
public void Initialize()
{
_modelFactory = new ModelFactory();
int[] countsWithQuotes = new int[] { 0, 3, 1, 4, 2, 5 };
int[] countsWithoutQuotes = new int[] { 6, 9, 7, 10, 8, 11 };
ISingleWordCountModel withQuotes = _modelFactory.GetSingleCountModel(countsWithQuotes);
ISingleWordCountModel withoutQuotes = _modelFactory.GetSingleCountModel(countsWithoutQuotes);
wordCountModel = _modelFactory.GetFlexibleCountModel(withQuotes, withoutQuotes);
nullWordCountModel = null;
model = new TextModel("model", wordCountModel);
}
public ITextModel GetOne(Expression <Func <Text, bool> > criteria)
{
Text text = _db.Texts.FirstOrDefault(criteria);
if (text == null)
{
return(null);
}
IFlexibleWordCountModel counts = GetCountsFromText(text);
ITextModel model = _modelFactory.GetTextModel(text.Name, counts);
model.SetAuthor(text.Author);
model.SetIncludeQuotes(Convert.ToBoolean(text.IncludeQuotes));
return(model);
}
public void Initialize()
{
_modelFactory = new ModelFactory();
int[] countsOne = new int[] { 1, 2, 3, 4, 5 };
singleCountIncludeQuotes = _modelFactory.GetSingleCountModel(countsOne);
int[] countsTwo = new int[] { 1, 14, 3, 28, 5 };
singleCountExcludeQuotes = _modelFactory.GetSingleCountModel(countsTwo);
singleCountNull = null;
int[] countsLengthOne = new int[] { 55 };
singleCountLengthOne = _modelFactory.GetSingleCountModel(countsLengthOne);
int[] countsLengthZeroA = new int[0];
singleCountLengthZeroA = new FakeSingleWordCountModel(countsLengthZeroA);
int[] countsLengthZeroB = new int[0];
singleCountLengthZeroB = new FakeSingleWordCountModel(countsLengthZeroB);
int[] countsWithNegative = new int[] { 1, -2, 3, 4, 5 };
singleCountNegativeCount = new FakeSingleWordCountModel(countsWithNegative);
flexibleWordCountModel = _modelFactory.GetFlexibleCountModel(singleCountIncludeQuotes, singleCountExcludeQuotes);
}
/// <summary>
/// The method to process a text and get the word counts.
/// </summary>
/// <param name="text">The text of the text itself.</param>
/// <param name="model">The flexible word count model to fill with counts.</param>
private void GenerateCounts(TextReader text, IFlexibleWordCountModel model)
{
int arraySize = UniversalConstants.CountSize;
int multiplier = UniversalConstants.ConstantMultiplier;
int[] countsWithQuotes = new int[arraySize];
int[] countsWithoutQuotes = new int[arraySize];
int[] frequencyWithQuotes = new int[arraySize];
int[] frequencyWithoutQuotes = new int[arraySize];
int totalWordCountWithQuotes = 0;
int totalWordCountWithoutQuotes = 0;
string delimPattern = @"\s+";
Regex delim = new Regex(delimPattern);
bool inQuotes = false;
bool continueWord = false;
string firstHalfOfWord = "";
int previousWordLength = 0;
//bool mismatchedQuotationMarks = false;
string line;
while ((line = text.ReadLine()) != null) // read text file line by line until end of line
{
if (line.Length != 0) // skip line if empty
{
line = Regex.Replace(line, "[–—]", " "); // treat em dashes and en dashes as spaces since they don't link words together like hyphens
string[] wordsArray = delim.Split(line.Trim()); // split the line using delimiter
for (int i = 0; i < wordsArray.Length; i++) // iterate through split array
{
string currentWord = wordsArray[i]; // grab a single word to count from split array
if (continueWord) // this conditional handles the case of if the previous line ends with a hyphen
{
if (currentWord[0] >= 'A' && currentWord[0] <= 'Z') // if the first letter of the current word is uppercase it means the previous hyphen was used incorrectly. don't change the current word and don't uncount the previous wordlength
{
firstHalfOfWord = "";
continueWord = false;
}
else // append the previous word to the front of the current word if the hyphen was used correctly
{
currentWord = firstHalfOfWord + currentWord;
firstHalfOfWord = "";
continueWord = false;
if (inQuotes) // uncount previous wordlength from counts with quotes if currently inside of quotations
{
countsWithQuotes[previousWordLength - 1]--;
totalWordCountWithQuotes--;
}
else // uncount previous wordlength for both counts if currently outside of quotations
{
countsWithQuotes[previousWordLength - 1]--;
countsWithoutQuotes[previousWordLength - 1]--;
totalWordCountWithQuotes--;
totalWordCountWithoutQuotes--;
}
}
}
// if the last word of the line ends with a hyphen, store the word in a variable, removing the hyphen
if (i == wordsArray.Length - 1)
{
if (currentWord[currentWord.Length - 1] == '-')
{
firstHalfOfWord = currentWord.Substring(0, currentWord.Length - 1);
continueWord = true;
}
}
// if it locates a starting quotation mark, set as inside quotations
if (currentWord[0] == '"' || currentWord[0] == '“')
{
inQuotes = true;
}
string modifiedCurrentWord = Regex.Replace(currentWord, "[\"]", ""); // remove quotes from the current word
modifiedCurrentWord = Regex.Replace(modifiedCurrentWord, "[^a-zA-Z0-9']+$", ""); // remove non-alphanumeric characters from the end of the word except for apostrophes
// Debug.Print(modifiedCurrentWord);
if (!(modifiedCurrentWord.Length == 0))
{
int wordLength = modifiedCurrentWord.Length;
previousWordLength = wordLength; // variable used in case a wordlength count has to be uncounted when counting the next word
if (!inQuotes) // if outside of quotations, increase count for both the count including and excluding words in quotations
{
totalWordCountWithQuotes++;
totalWordCountWithoutQuotes++;
if (wordLength < countsWithQuotes.Length)
{
countsWithQuotes[wordLength - 1]++;
countsWithoutQuotes[wordLength - 1]++;
}
else
{
countsWithQuotes[countsWithQuotes.Length - 1]++;
countsWithoutQuotes[countsWithoutQuotes.Length - 1]++;
}
}
else // if inside of quotations, increase count for only the count including words in quotations
{
totalWordCountWithQuotes++;
if (wordLength < countsWithQuotes.Length)
{
countsWithQuotes[wordLength - 1]++;
}
else
{
countsWithQuotes[countsWithQuotes.Length - 1]++;
}
}
}
// if it locates an ending quotation mark, set as no longer inside quotations
if (currentWord[currentWord.Length - 1] == '"' || currentWord[currentWord.Length - 1] == '”')
{
inQuotes = false;
}
}
}
}
// calculates frequency per 1000 words
for (int i = 0; i < countsWithQuotes.Length; i++)
{
frequencyWithQuotes[i] = (int)(((double)countsWithQuotes[i] / totalWordCountWithQuotes) * multiplier);
frequencyWithoutQuotes[i] = (int)(((double)countsWithoutQuotes[i] / totalWordCountWithoutQuotes) * multiplier);
}
Debug.Write("\nTotal with quotes: " + totalWordCountWithQuotes);
Debug.Write("\nTotal without quotes: " + totalWordCountWithoutQuotes);
// determines if there are mismatched quotation marks
//if (inQuotes)
//{
// mismatchedQuotationMarks = true;
//}
// set wordlength counts for the model
for (int i = 0; i < frequencyWithQuotes.Length; i++)
{
model.SetAt(true, i, frequencyWithQuotes[i]);
}
for (int i = 0; i < frequencyWithoutQuotes.Length; i++)
{
model.SetAt(false, i, frequencyWithoutQuotes[i]);
}
}
//For test purposes. Comment out later.
public void GenerateCountsTestMethod(TextReader text, IFlexibleWordCountModel model)
{
GenerateCounts(text, model);
}
public void ConstructorInputNegativeCount()
{
flexibleWordCountModel = _modelFactory.GetFlexibleCountModel(singleCountNegativeCount, singleCountExcludeQuotes);
}
public void ConstructorInputUnequalLength()
{
flexibleWordCountModel = _modelFactory.GetFlexibleCountModel(singleCountLengthOne, singleCountExcludeQuotes);
}
public void ConstructorLengthZeroInput()
{
flexibleWordCountModel = _modelFactory.GetFlexibleCountModel(singleCountLengthZeroA, singleCountLengthZeroB);
}
public void ConstructorDuplicateInput()
{
flexibleWordCountModel = _modelFactory.GetFlexibleCountModel(singleCountIncludeQuotes, singleCountIncludeQuotes);
}
public void ConstructorNullInput()
{
flexibleWordCountModel = _modelFactory.GetFlexibleCountModel(singleCountNull, singleCountExcludeQuotes);
}
public void ValidConstruction()
{
flexibleWordCountModel = _modelFactory.GetFlexibleCountModel(singleCountIncludeQuotes, singleCountExcludeQuotes);
}
