/// <summary>
/// To check the misspelled word in similarity
/// If it is similar to the word in the dictionary, it return true
/// </summary>
/// <param name="dicWord">The word from dictionary</param>
/// <param name="toBeCheckedWord">The mis spelled word needs to be checked</param>
/// <param name="theMissSpelling">The ref of MissSpelling instance</param>
/// <returns></returns>
private bool similarityCheck(string dicWord, string toBeCheckedWord, ref Misspelling theMissSpelling)
{
if (getDistance(dicWord, toBeCheckedWord) <= 2)
{
if (theMissSpelling == null)
{
theMissSpelling = new Misspelling();
}
theMissSpelling.TextPosition = getPosition(dicWord, toBeCheckedWord);
return(true);
}
return(false);
}
/// <summary>
/// To check the char at the same position
/// If the index is less than the threshold value it means it is similar to the word in the dictionary, it return true
/// </summary>
/// <param name="dicWord">The word from dictionary</param>
/// <param name="toBeCheckedWord">The mis spelled word needs to be checked</param>
/// <param name="theMissSpelling">The ref of MissSpelling instance</param>
/// <returns></returns>
private bool projectionCheck(string dicWord, string toBeCheckedWord, ref Misspelling theMissSpelling)
{
int failNum = 0;
Dictionary <int, bool> theTable = new Dictionary <int, bool>();
if (!string.IsNullOrEmpty(dicWord) && !string.IsNullOrEmpty(toBeCheckedWord) && dicWord.Length == toBeCheckedWord.Length)
{
for (int i = 0; i <= dicWord.Length - 1; i++)
{
char left = dicWord[i];
char right = toBeCheckedWord[i];
if (left != right)
{
//the same postion char is different
theTable.Add(i, false);
failNum++;
}
else
{
theTable.Add(i, true);
}
}
if (failNum > 0)
{
//It means there are some char are different
float failFloat = failNum;
float theLegth = dicWord.Length;
float index = failFloat / theLegth;
if (index <= 0.3)
{
//if index less than a value, it means it is simlar to the dic word
//Then suggest the dicWord to change
if (theMissSpelling == null)
{
theMissSpelling = new Misspelling();
}
//It will only store the postion with last matched string
//I think it's better to store every postion for evey matched string
theMissSpelling.TextPosition = theTable.Where(i => i.Value == false).FirstOrDefault().Key;
//indicate the dic word should be return
return(true);
}
}
}
return(false);
}
/// <summary>
/// Main function :
/// given a string of multiple words, return an array of all misspelled words
/// </summary>
/// <param name="text">It is a string which contains several words.The words needs to be checked</param>
/// <returns></returns>
public Misspelling[] CheckText(string text)
{
//get the original word list from text
string[] words = text.Split(new Char[] { '?', ':', ' ', '!', ';', '.', '<', '>' }, StringSplitOptions.RemoveEmptyEntries);
List <Misspelling> theMissSpellingList = new List <Misspelling>();
foreach (var word in words)
{
//remove space and tab
string wordToBeCheck = word.Trim();
//remove 's after a word
if (wordToBeCheck.EndsWith("'s"))
{
wordToBeCheck.Replace("'s", "");
}
Misspelling theMissSpelling = null;
var existedWord = dictionary.Where(s => s == wordToBeCheck).FirstOrDefault();
if (string.IsNullOrEmpty(existedWord))
{
//If can not find the word in the dictionary it means misspelled
theMissSpelling = new Misspelling()
{
Word = wordToBeCheck,
TextPosition = -1
};
var suggestionWords = dictionary.Where(s =>
{
return(isMissSpell(s, wordToBeCheck, ref theMissSpelling));
}).ToArray();
if (theMissSpelling != null)
{
theMissSpelling.Suggestions = suggestionWords;
theMissSpelling.Suggestion = String.Join(",", suggestionWords);
theMissSpellingList.Add(theMissSpelling);
}
}
}
return(theMissSpellingList.ToArray());
}
/// <summary>
/// The predicate for the misspelling checking
/// </summary>
/// <param name="dicWord"></param>
/// <param name="toBeCheckedWord"></param>
/// <param name="theMissSpelling"></param>
/// <returns></returns>
private bool isMissSpell(string dicWord, string toBeCheckedWord, ref Misspelling theMissSpelling)
{
if (dicWord == toBeCheckedWord)
{
//it is correct, so it's not need to suggestion
return(false);
}
//Condition1:
//only some char is wrong but with the same width of string
if (projectionCheck(dicWord, toBeCheckedWord, ref theMissSpelling))
{
return(true);
}
//Confidtion2:
//Maybe there is an additional char inserted into correct word
if (similarityCheck(dicWord, toBeCheckedWord, ref theMissSpelling))
{
return(true);
}
return(false);
}