public static MatchResult Match(string source, string target, bool pinyin = false)
{
if (string.IsNullOrEmpty(source) || string.IsNullOrEmpty(target))
return new MatchResult {Score = 0};
var matcher = FuzzyMatcher.Create(target);
var result = matcher.Evaluate(source);
if (pinyin)
{
// does pinyin score better?
var pinyinScore = ScoreForPinyin(source, target);
if (pinyinScore > result.Score)
result = new MatchResult() {Score = pinyinScore};
}
return result;
}
public MatchResult FuzzyMatch(string query, string stringToCompare)
{
query = query.Trim();
if (string.IsNullOrEmpty(stringToCompare) || string.IsNullOrEmpty(query))
{
return(new MatchResult(false, UserSettingSearchPrecision));
}
var queryWithoutCase = query.ToLower();
string translated = _alphabet.Translate(stringToCompare);
var fullStringToCompareWithoutCase = translated.ToLower();
string key = $"{queryWithoutCase}|{fullStringToCompareWithoutCase}";
MatchResult match = _cache[key] as MatchResult;
if (match == null)
{
match = FuzzyMatchInternal(queryWithoutCase, fullStringToCompareWithoutCase);
CacheItemPolicy policy = new CacheItemPolicy();
policy.SlidingExpiration = new TimeSpan(12, 0, 0);
_cache.Set(key, match, policy);
}
return(match);
}
/// <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 static MatchResult FuzzySearch(string query, string stringToCompare, MatchOption opt)
{
if (string.IsNullOrEmpty(stringToCompare) || string.IsNullOrEmpty(query))
{
return new MatchResult {
Success = false
}
}
;
query = query.Trim();
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>();
for (var compareStringIndex = 0; compareStringIndex < fullStringToCompareWithoutCase.Length; 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 score = CalculateSearchScore(query, stringToCompare, firstMatchIndex, lastMatchIndex - firstMatchIndex, allSubstringsContainedInCompareString);
var pinyinScore = ScoreForPinyin(stringToCompare, query);
var result = new MatchResult
{
Success = true,
MatchData = indexList,
RawScore = Math.Max(score, pinyinScore)
};
return(result);
}
return(new MatchResult {
Success = false
});
}
public MatchResult FuzzyMatchRecurrsive(
string query, string stringToCompare, int queryCurrentIndex, int stringCurrentIndex, List <int> sourceMatchData
)
{
if (queryCurrentIndex == query.Length || stringCurrentIndex == stringToCompare.Length)
{
return(new MatchResult(false, UserSettingSearchPrecision));
}
bool recursiveMatch = false;
List <int> bestRecursiveMatchData = new List <int>();
int bestRecursiveScore = 0;
List <int> matchs = new List <int>();
if (sourceMatchData.Count > 0)
{
foreach (var data in sourceMatchData)
{
matchs.Add(data);
}
}
while (queryCurrentIndex < query.Length && stringCurrentIndex < stringToCompare.Length)
{
char queryLower = char.ToLower(query[queryCurrentIndex]);
char stringToCompareLower = char.ToLower(stringToCompare[stringCurrentIndex]);
if (queryLower == stringToCompareLower)
{
MatchResult match = FuzzyMatchRecurrsive(
query, stringToCompare, queryCurrentIndex, stringCurrentIndex + 1, matchs
);
if (match.Success)
{
if (!recursiveMatch || match.RawScore > bestRecursiveScore)
{
bestRecursiveMatchData = new List <int>();
foreach (int data in match.MatchData)
{
bestRecursiveMatchData.Add(data);
}
bestRecursiveScore = match.Score;
}
recursiveMatch = true;
}
matchs.Add(stringCurrentIndex);
queryCurrentIndex += 1;
}
stringCurrentIndex += 1;
}
bool matched = queryCurrentIndex == query.Length;
int outScore;
if (matched)
{
outScore = 100;
int penality = 3 * matchs[0];
outScore = outScore - penality;
int unmatched = stringToCompare.Length - matchs.Count;
outScore = outScore - (5 * unmatched);
int consecutiveMatch = 0;
for (int i = 0; i < matchs.Count; i++)
{
int indexCurent = matchs[i];
if (i > 0)
{
int indexPrevious = matchs[i - 1];
if (indexCurent == indexPrevious + 1)
{
consecutiveMatch += 1;
outScore += 10 * consecutiveMatch;
}
else
{
consecutiveMatch = 0;
}
}
char current = stringToCompare[indexCurent];
bool currentUpper = char.IsUpper(current);
if (indexCurent > 0)
{
char neighbor = stringToCompare[indexCurent - 1];
if (currentUpper && char.IsLower(neighbor))
{
outScore += 30;
}
bool isNeighbourSeparator = neighbor == '_' || neighbor == ' ';
if (isNeighbourSeparator)
{
outScore += 50;
if (currentUpper)
{
outScore += 50;
}
}
}
else
{
outScore += 50;
if (currentUpper)
{
outScore += 50;
}
}
}
}
else
{
outScore = 0;
}
if (recursiveMatch && (!matched || bestRecursiveScore > outScore))
{
matchs = new List <int>();
foreach (int data in bestRecursiveMatchData)
{
matchs.Add(data);
}
outScore = bestRecursiveScore;
return(new MatchResult(true, UserSettingSearchPrecision, matchs, outScore));
}
else if (matched)
{
return(new MatchResult(true, UserSettingSearchPrecision, matchs, outScore));
}
else
{
return(new MatchResult(false, UserSettingSearchPrecision));
}
}
/// <summary>
/// refer to https://github.com/mattyork/fuzzy
/// </summary>
public static MatchResult FuzzySearch(string query, string stringToCompare, MatchOption opt)
{
if (string.IsNullOrEmpty(stringToCompare) || string.IsNullOrEmpty(query))
{
return new MatchResult {
Success = false
}
}
;
query = query.Trim();
var fullStringToCompareWithoutCase = opt.IgnoreCase ? stringToCompare.ToLower() : stringToCompare;
var queryWithoutCase = opt.IgnoreCase ? query.ToLower() : query;
int currentQueryToCompareIndex = 0;
var queryToCompareSeparated = queryWithoutCase.Split(' ');
var currentQueryToCompare = queryToCompareSeparated[currentQueryToCompareIndex];
var patternIndex = 0;
var firstMatchIndex = -1;
var firstMatchIndexInWord = -1;
var lastMatchIndex = 0;
bool allMatched = false;
bool isFullWordMatched = false;
bool allWordsFullyMatched = true;
var indexList = new List <int>();
for (var index = 0; index < fullStringToCompareWithoutCase.Length; index++)
{
var ch = stringToCompare[index];
if (fullStringToCompareWithoutCase[index] == currentQueryToCompare[patternIndex])
{
if (firstMatchIndex < 0)
{ // first matched char will become the start of the compared string
firstMatchIndex = index;
}
if (patternIndex == 0)
{ // first letter of current word
isFullWordMatched = true;
firstMatchIndexInWord = index;
}
else if (!isFullWordMatched)
{ // 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
int startIndexToVerify = index - patternIndex;
bool allMatch = true;
for (int indexToCheck = 0; indexToCheck < patternIndex; indexToCheck++)
{
if (fullStringToCompareWithoutCase[startIndexToVerify + indexToCheck] !=
currentQueryToCompare[indexToCheck])
{
allMatch = false;
}
}
if (allMatch)
{ // update to this as a full word
isFullWordMatched = true;
if (currentQueryToCompareIndex == 0)
{ // first word so we need to update start index
firstMatchIndex = startIndexToVerify;
}
indexList.RemoveAll(x => x >= firstMatchIndexInWord);
for (int indexToCheck = 0; indexToCheck < patternIndex; indexToCheck++)
{ // update the index list
indexList.Add(startIndexToVerify + indexToCheck);
}
}
}
lastMatchIndex = index + 1;
indexList.Add(index);
// increase the pattern matched index and check if everything was matched
if (++patternIndex == currentQueryToCompare.Length)
{
if (++currentQueryToCompareIndex >= queryToCompareSeparated.Length)
{ // moved over all the words
allMatched = true;
break;
}
// otherwise move to the next word
currentQueryToCompare = queryToCompareSeparated[currentQueryToCompareIndex];
patternIndex = 0;
if (!isFullWordMatched)
{ // if any of the words was not fully matched all are not fully matched
allWordsFullyMatched = false;
}
}
}
else
{
isFullWordMatched = false;
}
}
// return rendered string if we have a match for every char or all substring without whitespaces matched
if (allMatched)
{
// check if all query string was contained in string to compare
bool containedFully = lastMatchIndex - firstMatchIndex == queryWithoutCase.Length;
var score = CalculateSearchScore(query, stringToCompare, firstMatchIndex, lastMatchIndex - firstMatchIndex, containedFully, allWordsFullyMatched);
var pinyinScore = ScoreForPinyin(stringToCompare, query);
var result = new MatchResult
{
Success = true,
MatchData = indexList,
RawScore = Math.Max(score, pinyinScore)
};
return(result);
}
return(new MatchResult {
Success = false
});
}
public static int ScoreAfterSearchPrecisionFilter(this MatchResult matchResult)
{
return(matchResult.IsSearchPrecisionScoreMet() ? matchResult.Score : 0);
}