public void SaveResults(string wordFrequenciesFilePath, string ngramsFrequenciesFilePath)
{
// Persist word frequencies
using (var writer = new StreamWriter(wordFrequenciesFilePath))
{
var lines = WordFrequencies
.Select(ent => string.Format("{0}{1}{2}", ent.Key, Separator, ent.Value));
foreach (var line in lines)
{
writer.WriteLine(line);
}
}
// Persist ngrams frequencies
using (var writer = new StreamWriter(ngramsFrequenciesFilePath))
{
// First line is the ngrams counter
writer.WriteLine(TotalNgramsCounter);
// Don't try to order NGramsFrequencies since it causes OutOfMemoryExceptions (ordering a dictionary creates an ordered copy in all cases)
foreach (var freq in NGramsFrequencies)
{
var sb = new StringBuilder();
sb.Append(string.Join(Separator.ToString(), freq.Key)).Append(Separator).Append(freq.Value);
writer.WriteLine(sb.ToString());
}
}
}
private void WordFrequencyExplorer()
{
Console.WriteLine($"Unique word count: {UniqueWords.Count:n0}");
Console.WriteLine($"Total word count: {TotalWordCount:n0}");
Console.WriteLine($"Total verse count: {TotalVerseCount:n0}");
Console.WriteLine();
Console.WriteLine();
while (true)
{
Console.Write("Enter word: ");
var line = Console.ReadLine();
var key = line.ToUpperInvariant().Trim();
if (string.IsNullOrWhiteSpace(key))
{
return;
}
if (!WordFrequencies.ContainsKey(key))
{
Console.WriteLine($"{key} was not found in this bible.");
continue;
}
int count = WordFrequencies[key];
Console.WriteLine($"{key} was found ({count:n0}) times in this bible.");
Console.WriteLine();
}
}
public void FirstTry_Basic_Test()
{
string sentence = "This is my sentence and I can write anything one this sentence because this is mine!";
string word = "sentence";
int count = new WordFrequencies().FirstTry(sentence, word);
Assert.Equal(2, count);
}
public void FirstTry_No_Occurrences()
{
string sentence = "This is my sentence and I can write anything one this sentence because this is mine!";
string word = "word";
int count = new WordFrequencies().FirstTry(sentence, word);
Assert.Equal(0, count);
}
public void InvalidNThrowsExceptionTest()
{
// Arrange
var words = StringList("a");
var wordFrequencies = new WordFrequencies(words);
var numToFind = 0;
// Act
wordFrequencies.GetNthMostOccurring(numToFind);
}
public void SecondTry_Word_This_With_3_Occurrences()
{
string sentence = "This is my sentence and I can write anything one this sentence because this is mine!";
string word = "this";
var myWordFrequencies = new WordFrequencies(sentence);
int count = myWordFrequencies.SecondTry(word);
Assert.Equal(3, count);
}
public void SecondTry_Basic_Test()
{
string sentence = "This is my sentence and I can write anything one this sentence because this is mine!";
string word = "sentence";
var myWordFrequencies = new WordFrequencies(sentence);
int count = myWordFrequencies.SecondTry(word);
Assert.Equal(2, count);
}
// Methods ------------------
private void IncreaseWordFreq(string word, long frequency = 1)
{
TotalWordCounter += frequency;
if (WordFrequencies.ContainsKey(word))
{
WordFrequencies[word] += frequency;
}
else
{
WordFrequencies.Add(word, frequency);
}
}
public void OrderingSkipsValuesTest()
{
// Arrange
var words = StringList("a", "a", "b", "b", "c");
var wordFrequencies = new WordFrequencies(words);
var expected = StringList("c");
var numToFind = 3;
// Act
var actual = wordFrequencies.GetNthMostOccurring(numToFind);
// Assert
AssertListsAreEqual(expected, actual);
}
public void TwoWordsSameOccurrencesAreSortedTest()
{
// Arrange
var words = StringList("a", "a", "b", "b");
var wordFrequencies = new WordFrequencies(words);
var expected = StringList("a", "b");
var numToFind = 1;
// Act
var actual = wordFrequencies.GetNthMostOccurring(numToFind);
// Assert
AssertListsAreEqual(expected, actual);
}
public void SingleWordWithMultipleOccurrencesTest()
{
// Arrange
var words = StringList("a", "a");
var wordFrequencies = new WordFrequencies(words);
var expected = StringList("a");
var numToFind = 1;
// Act
var actual = wordFrequencies.GetNthMostOccurring(numToFind);
// Assert
AssertListsAreEqual(expected, actual);
}
public void InBetweenFrequenciesReturnsNothingTest()
{
// Arrange
var words = StringList("a", "a", "b", "b", "c");
var wordFrequencies = new WordFrequencies(words);
List <string> expected = null;
var numToFind = 2;
// Act
var actual = wordFrequencies.GetNthMostOccurring(numToFind);
// Assert
AssertListsAreEqual(expected, actual);
}
private void WordLocationExplorer()
{
while (true)
{
Console.Write("Enter word: ");
var line = Console.ReadLine();
var key = line.ToUpperInvariant().Trim();
if (string.IsNullOrWhiteSpace(key))
{
return;
}
if (!WordFrequencies.ContainsKey(key))
{
Console.WriteLine($"{key} was not found in this bible.");
continue;
}
int count = WordFrequencies[key];
Console.WriteLine();
Console.ForegroundColor = ConsoleColor.DarkBlue;
Console.WriteLine($"{key} was found ({count:n0}) times in this bible.");
Console.ResetColor();
Console.WriteLine();
foreach (var location in WordLocations(key))
{
Console.ForegroundColor = ConsoleColor.Blue;
Console.WriteLine(location.ToString());
Console.ResetColor();
Console.WriteLine(location.Text);
Console.WriteLine();
}
}
}
private void LoadWordFrequencies()
{
foreach (XmlNode book in BibleXml.DocumentElement.ChildNodes)
{
var bookName = book.Attributes["n"].InnerText;
foreach (XmlNode chapter in book.ChildNodes)
{
var chapterName = chapter.Attributes["n"].InnerText;
foreach (XmlNode verse in chapter.ChildNodes)
{
TotalVerseCount++;
var verseNumber = verse.Attributes["n"].InnerText;
string text = verse.InnerText.ToUpperInvariant();
var chars = new List <char>();
for (int i = 0; i < text.Length; i++)
{
var c = text[i];
if (!char.IsLetter(c) && c != ' ' && c != '-')
{
continue;
}
// allow hyphenated words
if (i > 0 && i < text.Length - 1 && c == '-')
{
if (!char.IsLetter(text[i - 1]) || !char.IsLetter(text[i + 1]))
{
continue;
}
}
chars.Add(text[i]);
}
text = string.Concat(chars).ToUpperInvariant();
var words = text.Split(' ', StringSplitOptions.RemoveEmptyEntries);
foreach (var word in words)
{
var tempWord = word.Trim().TrimStart('-').TrimEnd('-');
if (string.IsNullOrWhiteSpace(tempWord))
{
continue;
}
TotalWordCount++;
if (WordFrequencies.ContainsKey(tempWord))
{
WordFrequencies[tempWord]++;
}
else
{
WordFrequencies[tempWord] = 1;
}
UniqueWords.Add(tempWord);
}
}
}
}
}
