public static FuzzyMatcher Create(string query, MatchOption opt)
{
return(new FuzzyMatcher(query, opt));
}
/// <summary>
/// Current method:
/// Character matching + substring matching;
/// 1. Query search string is split into substrings, separator is whitespace.
/// 2. Check each query substring's characters against full compare string,
/// 3. if a character in the substring is matched, loop back to verify the previous character.
/// 4. If previous character also matches, and is the start of the substring, update list.
/// 5. Once the previous character is verified, move on to the next character in the query substring.
/// 6. Move onto the next substring's characters until all substrings are checked.
/// 7. Consider success and move onto scoring if every char or substring without whitespaces matched
/// </summary>
public MatchResult FuzzyMatch(string query, string stringToCompare, MatchOption opt)
{
if (string.IsNullOrEmpty(stringToCompare) || string.IsNullOrEmpty(query))
{
return(new MatchResult(false, UserSettingSearchPrecision));
}
query = query.Trim();
if (_alphabet != null)
{
query = _alphabet.Translate(query);
stringToCompare = _alphabet.Translate(stringToCompare);
}
var fullStringToCompareWithoutCase = opt.IgnoreCase ? stringToCompare.ToLower() : stringToCompare;
var queryWithoutCase = opt.IgnoreCase ? query.ToLower() : query;
var querySubstrings = queryWithoutCase.Split(new[] { ' ' }, StringSplitOptions.RemoveEmptyEntries);
int currentQuerySubstringIndex = 0;
var currentQuerySubstring = querySubstrings[currentQuerySubstringIndex];
var currentQuerySubstringCharacterIndex = 0;
var firstMatchIndex = -1;
var firstMatchIndexInWord = -1;
var lastMatchIndex = 0;
bool allQuerySubstringsMatched = false;
bool matchFoundInPreviousLoop = false;
bool allSubstringsContainedInCompareString = true;
var indexList = new List <int>();
List <int> spaceIndices = new List <int>();
for (var compareStringIndex = 0; compareStringIndex < fullStringToCompareWithoutCase.Length; compareStringIndex++)
{
// To maintain a list of indices which correspond to spaces in the string to compare
// To populate the list only for the first query substring
if (fullStringToCompareWithoutCase[compareStringIndex].Equals(' ') && currentQuerySubstringIndex == 0)
{
spaceIndices.Add(compareStringIndex);
}
if (fullStringToCompareWithoutCase[compareStringIndex] != currentQuerySubstring[currentQuerySubstringCharacterIndex])
{
matchFoundInPreviousLoop = false;
continue;
}
if (firstMatchIndex < 0)
{
// first matched char will become the start of the compared string
firstMatchIndex = compareStringIndex;
}
if (currentQuerySubstringCharacterIndex == 0)
{
// first letter of current word
matchFoundInPreviousLoop = true;
firstMatchIndexInWord = compareStringIndex;
}
else if (!matchFoundInPreviousLoop)
{
// we want to verify that there is not a better match if this is not a full word
// in order to do so we need to verify all previous chars are part of the pattern
var startIndexToVerify = compareStringIndex - currentQuerySubstringCharacterIndex;
if (AllPreviousCharsMatched(startIndexToVerify, currentQuerySubstringCharacterIndex, fullStringToCompareWithoutCase, currentQuerySubstring))
{
matchFoundInPreviousLoop = true;
// if it's the beginning character of the first query substring that is matched then we need to update start index
firstMatchIndex = currentQuerySubstringIndex == 0 ? startIndexToVerify : firstMatchIndex;
indexList = GetUpdatedIndexList(startIndexToVerify, currentQuerySubstringCharacterIndex, firstMatchIndexInWord, indexList);
}
}
lastMatchIndex = compareStringIndex + 1;
indexList.Add(compareStringIndex);
currentQuerySubstringCharacterIndex++;
// if finished looping through every character in the current substring
if (currentQuerySubstringCharacterIndex == currentQuerySubstring.Length)
{
// if any of the substrings was not matched then consider as all are not matched
allSubstringsContainedInCompareString = matchFoundInPreviousLoop && allSubstringsContainedInCompareString;
currentQuerySubstringIndex++;
allQuerySubstringsMatched = AllQuerySubstringsMatched(currentQuerySubstringIndex, querySubstrings.Length);
if (allQuerySubstringsMatched)
{
break;
}
// otherwise move to the next query substring
currentQuerySubstring = querySubstrings[currentQuerySubstringIndex];
currentQuerySubstringCharacterIndex = 0;
}
}
// proceed to calculate score if every char or substring without whitespaces matched
if (allQuerySubstringsMatched)
{
var nearestSpaceIndex = CalculateClosestSpaceIndex(spaceIndices, firstMatchIndex);
var score = CalculateSearchScore(query, stringToCompare, firstMatchIndex - nearestSpaceIndex - 1, lastMatchIndex - firstMatchIndex, allSubstringsContainedInCompareString);
return(new MatchResult(true, UserSettingSearchPrecision, indexList, score));
}
return(new MatchResult(false, UserSettingSearchPrecision));
}
private FuzzyMatcher(string query, MatchOption opt)
{
this.query = query.Trim();
this.opt = opt;
}
/// <summary>
/// Current method has two parts, Acronym Match and Fuzzy Search:
///
/// Acronym Match:
/// Charater listed below will be considered as acronym
/// 1. Character on index 0
/// 2. Character appears after a space
/// 3. Character that is UpperCase
/// 4. Character that is number
///
/// Acronym Match will succeed when all query characters match with acronyms in stringToCompare.
/// If any of the characters in the query isn't matched with stringToCompare, Acronym Match will fail.
/// Score will be calculated based the percentage of all query characters matched with total acronyms in stringToCompare.
///
/// Fuzzy Search:
/// Character matching + substring matching;
/// 1. Query search string is split into substrings, separator is whitespace.
/// 2. Check each query substring's characters against full compare string,
/// 3. if a character in the substring is matched, loop back to verify the previous character.
/// 4. If previous character also matches, and is the start of the substring, update list.
/// 5. Once the previous character is verified, move on to the next character in the query substring.
/// 6. Move onto the next substring's characters until all substrings are checked.
/// 7. Consider success and move onto scoring if every char or substring without whitespaces matched
/// </summary>
public MatchResult FuzzyMatch(string query, string stringToCompare, MatchOption opt)
{
if (string.IsNullOrEmpty(stringToCompare) || string.IsNullOrEmpty(query))
{
return(new MatchResult(false, UserSettingSearchPrecision));
}
query = query.Trim();
TranslationMapping translationMapping;
(stringToCompare, translationMapping) = _alphabet?.Translate(stringToCompare) ?? (stringToCompare, null);
var currentAcronymQueryIndex = 0;
var acronymMatchData = new List <int>();
int acronymsTotalCount = 0;
int acronymsMatched = 0;
var fullStringToCompareWithoutCase = opt.IgnoreCase ? stringToCompare.ToLower() : stringToCompare;
var queryWithoutCase = opt.IgnoreCase ? query.ToLower() : query;
var querySubstrings = queryWithoutCase.Split(new[] { ' ' }, StringSplitOptions.RemoveEmptyEntries);
int currentQuerySubstringIndex = 0;
var currentQuerySubstring = querySubstrings[currentQuerySubstringIndex];
var currentQuerySubstringCharacterIndex = 0;
var firstMatchIndex = -1;
var firstMatchIndexInWord = -1;
var lastMatchIndex = 0;
bool allQuerySubstringsMatched = false;
bool matchFoundInPreviousLoop = false;
bool allSubstringsContainedInCompareString = true;
var indexList = new List <int>();
List <int> spaceIndices = new List <int>();
for (var compareStringIndex = 0; compareStringIndex < fullStringToCompareWithoutCase.Length; compareStringIndex++)
{
// If acronyms matching successfully finished, this gets the remaining not matched acronyms for score calculation
if (currentAcronymQueryIndex >= query.Length && acronymsMatched == query.Length)
{
if (IsAcronymCount(stringToCompare, compareStringIndex))
{
acronymsTotalCount++;
}
continue;
}
if (currentAcronymQueryIndex >= query.Length ||
currentAcronymQueryIndex >= query.Length && allQuerySubstringsMatched)
{
break;
}
// To maintain a list of indices which correspond to spaces in the string to compare
// To populate the list only for the first query substring
if (fullStringToCompareWithoutCase[compareStringIndex] == ' ' && currentQuerySubstringIndex == 0)
{
spaceIndices.Add(compareStringIndex);
}
// Acronym Match
if (IsAcronym(stringToCompare, compareStringIndex))
{
if (fullStringToCompareWithoutCase[compareStringIndex] ==
queryWithoutCase[currentAcronymQueryIndex])
{
acronymMatchData.Add(compareStringIndex);
acronymsMatched++;
currentAcronymQueryIndex++;
}
}
if (IsAcronymCount(stringToCompare, compareStringIndex))
{
acronymsTotalCount++;
}
if (allQuerySubstringsMatched || fullStringToCompareWithoutCase[compareStringIndex] !=
currentQuerySubstring[currentQuerySubstringCharacterIndex])
{
matchFoundInPreviousLoop = false;
continue;
}
if (firstMatchIndex < 0)
{
// first matched char will become the start of the compared string
firstMatchIndex = compareStringIndex;
}
if (currentQuerySubstringCharacterIndex == 0)
{
// first letter of current word
matchFoundInPreviousLoop = true;
firstMatchIndexInWord = compareStringIndex;
}
else if (!matchFoundInPreviousLoop)
{
// we want to verify that there is not a better match if this is not a full word
// in order to do so we need to verify all previous chars are part of the pattern
var startIndexToVerify = compareStringIndex - currentQuerySubstringCharacterIndex;
if (AllPreviousCharsMatched(startIndexToVerify, currentQuerySubstringCharacterIndex,
fullStringToCompareWithoutCase, currentQuerySubstring))
{
matchFoundInPreviousLoop = true;
// if it's the beginning character of the first query substring that is matched then we need to update start index
firstMatchIndex = currentQuerySubstringIndex == 0 ? startIndexToVerify : firstMatchIndex;
indexList = GetUpdatedIndexList(startIndexToVerify, currentQuerySubstringCharacterIndex,
firstMatchIndexInWord, indexList);
}
}
lastMatchIndex = compareStringIndex + 1;
indexList.Add(compareStringIndex);
currentQuerySubstringCharacterIndex++;
// if finished looping through every character in the current substring
if (currentQuerySubstringCharacterIndex == currentQuerySubstring.Length)
{
// if any of the substrings was not matched then consider as all are not matched
allSubstringsContainedInCompareString =
matchFoundInPreviousLoop && allSubstringsContainedInCompareString;
currentQuerySubstringIndex++;
allQuerySubstringsMatched =
AllQuerySubstringsMatched(currentQuerySubstringIndex, querySubstrings.Length);
if (allQuerySubstringsMatched)
{
continue;
}
// otherwise move to the next query substring
currentQuerySubstring = querySubstrings[currentQuerySubstringIndex];
currentQuerySubstringCharacterIndex = 0;
}
}
// return acronym match if all query char matched
if (acronymsMatched > 0 && acronymsMatched == query.Length)
{
int acronymScore = acronymsMatched * 100 / acronymsTotalCount;
if (acronymScore >= (int)UserSettingSearchPrecision)
{
acronymMatchData = acronymMatchData.Select(x => translationMapping?.MapToOriginalIndex(x) ?? x).Distinct().ToList();
return(new MatchResult(true, UserSettingSearchPrecision, acronymMatchData, acronymScore));
}
}
// proceed to calculate score if every char or substring without whitespaces matched
if (allQuerySubstringsMatched)
{
var nearestSpaceIndex = CalculateClosestSpaceIndex(spaceIndices, firstMatchIndex);
var score = CalculateSearchScore(query, stringToCompare, firstMatchIndex - nearestSpaceIndex - 1,
lastMatchIndex - firstMatchIndex, allSubstringsContainedInCompareString);
var resultList = indexList.Select(x => translationMapping?.MapToOriginalIndex(x) ?? x).Distinct().ToList();
return(new MatchResult(true, UserSettingSearchPrecision, resultList, score));
}
return(new MatchResult(false, UserSettingSearchPrecision));
}
