/// ------------------------------------------------------------------------------------
public SearchEngine(SearchQuery query)
{
CurrentSearchQuery = query;
_errors.Clear();
try
{
var parser = new PatternParser(App.Project);
SrchItemPatternGroup = parser.Parse(query.SearchItem, EnvironmentType.Item);
EnvBeforePatternGroup = parser.Parse(query.PrecedingEnvironment, EnvironmentType.Before);
EnvAfterPatternGroup = parser.Parse(query.FollowingEnvironment, EnvironmentType.After);
}
catch
{
var error = new SearchQueryValidationError(
string.Format(GetPatternSyntaxErrorMsg(), App.kEmptyDiamondPattern));
_errors.Add(error);
}
m_srchItemStr = query.SearchItem;
m_envBeforeStr = query.PrecedingEnvironment;
m_envAfterStr = query.FollowingEnvironment;
if (_errors != null && _errors.Count > 0)
{
query.Errors.AddRange(_errors);
}
}
/// ------------------------------------------------------------------------------------
public void VerifyDiacriticPlaceholders(string pattern)
{
var match = PatternParser.FindInnerMostSquareBracketPairs(pattern);
while (match.Success)
{
var bracketedText = match.Result("${bracketedText}");
if (bracketedText.Contains(App.DottedCircleC))
{
bracketedText = bracketedText.Replace(App.DottedCircle, string.Empty);
if (bracketedText.Contains("+") && bracketedText.Contains("*"))
{
var error = new SearchQueryValidationError(
LocalizationManager.GetString("PhoneticSearchingMessages.InvalidDiacriticPlaceholderSyntaxMsg",
"The symbols '*' and '+' may not appear between square brackets together " +
"with a diacritic placeholder. One or the other is allowed, but not both."));
error.HelpLinks.AddRange(new[] { "hidSearchPatternsDiacriticPlaceholders", "hidSearchPatternsExamples" });
Errors.Add(error);
}
if (bracketedText.Count(s => s == '+') > 1)
{
var error = new SearchQueryValidationError(
LocalizationManager.GetString("PhoneticSearchingMessages.TooManyOneOrMoreSymbolsInDiacriticPlaceholderMsg",
"The One-Or-More-Diacritics symbol (+) appears too many times with a diacritic " +
"placeholder. Only one is allowed."));
error.HelpLinks.AddRange(new[] { "hidSearchPatternsDiacriticPlaceholders", "hidSearchPatternsExamples" });
Errors.Add(error);
}
if (bracketedText.Count(s => s == '*') > 1)
{
var error = new SearchQueryValidationError(
LocalizationManager.GetString("PhoneticSearchingMessages.TooManyZeroOrMoreSymbolsInDiacriticPlaceholderMsg",
"The Zero-Or-More-Diacritics symbol (*) appears too many times with a " +
"diacritic placeholder. Only one is allowed."));
error.HelpLinks.AddRange(new[] { "hidSearchPatternsDiacriticPlaceholders", "hidSearchPatternsExamples" });
Errors.Add(error);
}
foreach (var symbolInfo in bracketedText.Where(s => s != '+' && s != '*')
.Select(s => App.IPASymbolCache[s]).Where(s => s != null && s.IsBase))
{
var msg = LocalizationManager.GetString("PhoneticSearchingMessages.InvalidSymbolInDiacriticPlaceholderMsg",
"The symbol '{0}' is a base character and was found between square brackets " +
"together with a diacritic placeholder. Base characters are not allowed with " +
"diacritic placeholders.");
var error = new SearchQueryValidationError(string.Format(msg, symbolInfo.Literal));
error.HelpLinks.AddRange(new[] { "hidSearchPatternsDiacriticPlaceholders", "hidSearchPatternsExamples" });
Errors.Add(error);
}
}
match = match.NextMatch();
}
}
/// ------------------------------------------------------------------------------------
private void VerifyDiacriticPlaceholdersAreInAndGroups(string pattern)
{
while (true)
{
var match = PatternParser.FindInnerMostSquareBracketPairs(pattern);
if (!match.Success)
{
break;
}
if (!match.Value.Contains(App.DottedCircleC))
{
pattern = pattern.Replace(match.Value, string.Empty);
}
else
{
_tokenBins[++_token] = match.Value;
pattern = pattern.Replace(match.Value, _token.ToString());
}
}
var fmt = LocalizationManager.GetString("PhoneticSearchingMessages.DiacriticPlaceholderNotInAndGroupMsg",
"The diacritic placeholder pattern '{0}' is not inside an AND group. Diacritic placeholder " +
"patterns must be placed inside AND groups.");
foreach (var error in pattern.Where(c => c > kMinToken)
.Select(c => new SearchQueryValidationError(string.Format(fmt, _tokenBins[c]))))
{
error.HelpLinks.AddRange(new[] { "hidSearchPatternsDiacriticPlaceholders", "hidSearchPatternsAndGroups" });
Errors.Add(error);
}
_tokenBins.Clear();
_token = kMinToken;
}
/// ------------------------------------------------------------------------------------
public void VerifyTextInSquareBrackets(string pattern)
{
var match = PatternParser.FindInnerMostSquareBracketPairs(pattern);
while (match.Success)
{
var bracketedText = match.Result("${bracketedText}");
if (bracketedText != string.Empty &&
!bracketedText.Contains(App.DottedCircle) &&
bracketedText != "C" && bracketedText != "V" &&
!App.AFeatureCache.Keys.Any(f => f == bracketedText) &&
!App.BFeatureCache.Keys.Any(f => f == bracketedText) && !bracketedText.Contains(App.ProportionalToSymbol))
{
var msg = LocalizationManager.GetString("PhoneticSearchingMessages.InvalidTextInSquareBracketsMsg",
"The text '{0}' in square brackets is invalid. Other than surrounding " +
"'AND' groups, square brackets are only used to surround descriptive or " +
"distinctive features, the designators for consonant or vowel classes " +
"('[C]' and '[V]'), or a diacritic placeholder with its diacritics and wildcards.");
var error = new SearchQueryValidationError(string.Format(msg, bracketedText));
error.HelpLinks.AddRange(new[] { "hidSearchPatternsAndGroups", "hidSearchPatternsExamples" });
Errors.Add(error);
}
match = match.NextMatch();
}
}
/// ------------------------------------------------------------------------------------
public void ValidateOrGroups(string item)
{
var orGroupValidator = new OrGroupValidator(_project);
var match = PatternParser.FindInnerMostBracesPair(item);
while (match.Success)
{
orGroupValidator.Verify(match.Value);
if (orGroupValidator.HasErrors)
{
Errors.AddRange(orGroupValidator.Errors);
}
match = match.NextMatch();
}
}
/// ------------------------------------------------------------------------------------
public string StripOutStuffWithValidPlusAndStarSymbols(string pattern)
{
var match = PatternParser.FindInnerMostSquareBracketPairs(pattern);
while (match.Success)
{
if (match.Value.Contains('+') || match.Value.Contains('*'))
{
pattern = pattern.Replace(match.Value, new string('$', match.Value.Length));
}
match = match.NextMatch();
}
return(pattern.Replace("$", string.Empty));
}
/// ------------------------------------------------------------------------------------
public bool Parse()
{
if (!DoesPatternPassBasicChecks())
{
return(false);
}
_parser = new PatternParser(_project, _query);
if (_parser.Parse())
{
return(true);
}
Errors.AddRange(_parser.Errors);
return(false);
}
/// ------------------------------------------------------------------------------------
private string GetPhonesRunInPattern(string pattern)
{
while (true)
{
var match = PatternParser.FindInnerMostParenthesePair(pattern);
if (!match.Success)
{
break;
}
pattern = pattern.Replace(match.Value, string.Empty);
}
while (true)
{
var match = PatternParser.FindInnerMostSquareBracketPairs(pattern);
if (!match.Success)
{
break;
}
pattern = pattern.Replace(match.Value, string.Empty);
}
while (true)
{
var match = PatternParser.FindInnerAngleBracketPairs(pattern);
if (!match.Success)
{
break;
}
pattern = pattern.Replace(match.Value, string.Empty);
}
pattern = pattern.Replace(",", string.Empty);
pattern = pattern.Replace("[", string.Empty);
pattern = pattern.Replace("]", string.Empty);
pattern = pattern.Replace("{", string.Empty);
pattern = pattern.Replace("}", string.Empty);
pattern = pattern.Replace("_", string.Empty);
pattern = pattern.Replace("/", string.Empty);
pattern = pattern.Replace("*", string.Empty);
pattern = pattern.Replace("+", string.Empty);
pattern = pattern.Replace("NOT", string.Empty);
return(pattern.Replace("#", string.Empty));
}
/// ------------------------------------------------------------------------------------
public string TranslateInnerSquareBracketedTextToTokens(string pattern)
{
var match = PatternParser.FindInnerMostSquareBracketPairs(pattern);
while (match.Success)
{
_tokenBins[++_token] = match.Value;
var newPattern = pattern.Substring(0, match.Index) + _token + new string('$', match.Length - 1);
if (match.Index + match.Length < pattern.Length)
{
newPattern += pattern.Substring(match.Index + match.Length);
}
pattern = newPattern;
match = match.NextMatch();
}
return(pattern.Replace("$", string.Empty));
}
/// ------------------------------------------------------------------------------------
public void VerifyOneDiacriticPlaceholderPerAndGroup(string pattern)
{
var originalPattern = pattern;
var match = PatternParser.FindInnerMostSquareBracketPairs(pattern);
while (match.Success)
{
var newPattern = pattern.Substring(0, match.Index) +
(match.Value.Contains(App.DottedCircleC) ? '$' : '@');
newPattern += new string('@', match.Length - 1);
if (match.Index + match.Length < pattern.Length)
{
newPattern += pattern.Substring(match.Index + match.Length);
}
pattern = newPattern;
match = match.NextMatch();
}
pattern = pattern.Replace("@", string.Empty);
match = PatternParser.FindInnerMostSquareBracketPairs(pattern);
while (match.Success)
{
if (match.Value.Count(c => c == '$') > 1)
{
var msg = LocalizationManager.GetString("PhoneticSearchingMessages.TooManyDiacriticPlaceholderMsg",
"The pattern '{0}' contains too many diacritic placeholders in one of the AND groups. " +
"Only one diacritic placeholder is allowed per AND group.");
var error = new SearchQueryValidationError(string.Format(msg, originalPattern));
error.HelpLinks.AddRange(new[] { "hidSearchPatternsDiacriticPlaceholders", "hidSearchPatternsExamples" });
Errors.Add(error);
return;
}
match = match.NextMatch();
}
}
/// ------------------------------------------------------------------------------------
public override void Verify(string pattern)
{
base.Verify(pattern);
VerifyDiacriticPlaceholdersAreInAndGroups(pattern);
pattern = TranslateInnerSquareBracketedTextToTokens(pattern);
while (true)
{
var match = PatternParser.FindInnerMostSquareBracketPairs(pattern);
if (!match.Success)
{
return;
}
LookForMalformedGroup(match.Value);
_tokenBins[++_token] = match.Value;
pattern = pattern.Replace(match.Value, _token.ToString());
}
}
public void TestSetup()
{
_parser = new PatternParser(_prj);
SearchEngine.PhoneCache = _prj.PhoneCache;
App.DottedCircle = "0";
App.DottedCircleC = '0';
App.DiacriticPlaceholder = "[0]";
_prj.PhoneCache["a"] = new PhoneInfo {
Phone = "a", CharType = IPASymbolType.vowel
};
_prj.PhoneCache["d"] = new PhoneInfo {
Phone = "d", CharType = IPASymbolType.consonant
};
((PhoneInfo)_prj.PhoneCache["d"]).BFeatureNames = new List <string> {
"+high", "-cons"
};
((PhoneInfo)_prj.PhoneCache["d"]).AFeatureNames = new List <string> {
"dental"
};
}
/// ------------------------------------------------------------------------------------
public void VerifyTextInAngleBrackets(string pattern)
{
var match = PatternParser.FindInnerAngleBracketPairs(pattern);
while (match.Success)
{
var bracketedText = match.Result("${bracketedText}");
if (!_project.SearchClasses.Any(c => c.Name == bracketedText))
{
var msg = LocalizationManager.GetString("PhoneticSearchingMessages.InvalidTextInAngleBracketsMsg",
"The text '{0}' in angled brackets is not a valid class name.");
var error = new SearchQueryValidationError(string.Format(msg, bracketedText));
error.HelpLinks.Add("hidSearchPatternsExamples");
Errors.Add(error);
}
match = match.NextMatch();
}
}