internal void InitSpecialRuleRef(Backend backend, ParseElementCollection parent)
{
Rule rule = null;
switch (_type)
{
case SpecialRuleRefType.Null:
parent.AddArc(backend.EpsilonTransition(1f));
break;
case SpecialRuleRefType.Void:
rule = backend.FindRule("VOID");
if (rule == null)
{
rule = backend.CreateRule("VOID", (SPCFGRULEATTRIBUTES)0);
((IElement)rule).PostParse((IElement)parent);
}
parent.AddArc(backend.RuleTransition(rule, parent._rule, 1f));
break;
case SpecialRuleRefType.Garbage:
{
OneOf oneOf = new OneOf(parent._rule, backend);
oneOf.AddArc(backend.RuleTransition(CfgGrammar.SPRULETRANS_WILDCARD, parent._rule, 0.5f));
oneOf.AddArc(backend.EpsilonTransition(0.5f));
((IElement)oneOf).PostParse((IElement)parent);
break;
}
}
}
/// <summary>
/// Add transition corresponding to Special or Uri.
/// </summary>
internal RuleRef(ParseElementCollection parent, Backend backend, Uri uri, List <Rule> undefRules, string semanticKey, string initParameters)
: base(parent._rule)
{
string id = uri.OriginalString;
Rule ruleRef = null;
int posPound = id.IndexOf('#');
// Get the initial state for the RuleRef.
if (posPound == 0)
{
// Internal RuleRef. Get InitialState of RuleRef.
// GetRuleRef() may temporarily create a Rule placeholder for later resolution.
ruleRef = GetRuleRef(backend, id.Substring(1), undefRules);
}
else
{
// External RuleRef. Build URL:GrammarUri#RuleName
StringBuilder sbExternalRuleUri = new("URL:");
// Add the parameters to initialize a rule
if (!string.IsNullOrEmpty(initParameters))
{
// look for the # and insert the parameters
sbExternalRuleUri.Append(posPound > 0 ? id.Substring(0, posPound) : id);
sbExternalRuleUri.Append('>');
sbExternalRuleUri.Append(initParameters);
if (posPound > 0)
{
sbExternalRuleUri.Append(id.Substring(posPound));
}
}
else
{
sbExternalRuleUri.Append(id);
}
// Get InitialState of external RuleRef.
string sExternalRuleUri = sbExternalRuleUri.ToString();
ruleRef = backend.FindRule(sExternalRuleUri);
if (ruleRef == null)
{
ruleRef = backend.CreateRule(sExternalRuleUri, SPCFGRULEATTRIBUTES.SPRAF_Import);
}
}
Arc rulerefArc = backend.RuleTransition(ruleRef, _rule, 1.0f);
#pragma warning disable 0618
if (!string.IsNullOrEmpty(semanticKey))
{
CfgGrammar.CfgProperty propertyInfo = new();
propertyInfo._pszName = "SemanticKey";
propertyInfo._comValue = semanticKey;
propertyInfo._comType = VarEnum.VT_EMPTY;
backend.AddPropertyTag(rulerefArc, rulerefArc, propertyInfo);
}
#pragma warning restore 0618
parent.AddArc(rulerefArc);
}
public Subset(ParseElementCollection parent, Backend backend, string text, MatchMode mode)
: base(parent._rule)
{
char[] achTrimChars = Helpers._achTrimChars;
foreach (char c in achTrimChars)
{
if (c != ' ' && text.IndexOf(c) >= 0)
{
text = text.Replace(c, ' ');
}
}
parent.AddArc(backend.SubsetTransition(text, mode));
}
/// <summary>
/// Returns the initial state of a special rule.
/// For each type of special rule we make a rule with a numeric id and return a reference to it.
/// </summary>
internal void InitSpecialRuleRef(Backend backend, ParseElementCollection parent)
{
Rule rule = null;
// Create a transition corresponding to Special or Uri
switch (_type)
{
case SpecialRuleRefType.Null:
parent.AddArc(backend.EpsilonTransition(1.0f));
break;
case SpecialRuleRefType.Void:
rule = backend.FindRule(szSpecialVoid);
if (rule == null)
{
rule = backend.CreateRule(szSpecialVoid, 0);
// Rule with no transitions is a void rule.
((IRule)rule).PostParse(parent);
}
parent.AddArc(backend.RuleTransition(rule, parent._rule, 1.0f));
break;
case SpecialRuleRefType.Garbage:
// Garbage transition is optional whereas Wildcard is not. So we need additional epsilon transition.
OneOf oneOf = new(parent._rule, backend);
// Add the garbage transition
oneOf.AddArc(backend.RuleTransition(CfgGrammar.SPRULETRANS_WILDCARD, parent._rule, 0.5f));
// Add a parallel epsilon path
oneOf.AddArc(backend.EpsilonTransition(0.5f));
((IOneOf)oneOf).PostParse(parent);
break;
default:
System.Diagnostics.Debug.Assert(false, "Unknown special ruleref type");
break;
}
}
internal RuleRef(ParseElementCollection parent, Backend backend, Uri uri, List <Rule> undefRules, string semanticKey, string initParameters)
: base(parent._rule)
{
string originalString = uri.OriginalString;
Rule rule = null;
int num = originalString.IndexOf('#');
if (num == 0)
{
rule = GetRuleRef(backend, originalString.Substring(1), undefRules);
}
else
{
StringBuilder stringBuilder = new StringBuilder("URL:");
if (!string.IsNullOrEmpty(initParameters))
{
stringBuilder.Append((num > 0) ? originalString.Substring(0, num) : originalString);
stringBuilder.Append('>');
stringBuilder.Append(initParameters);
if (num > 0)
{
stringBuilder.Append(originalString.Substring(num));
}
}
else
{
stringBuilder.Append(originalString);
}
string text = stringBuilder.ToString();
rule = backend.FindRule(text);
if (rule == null)
{
rule = backend.CreateRule(text, SPCFGRULEATTRIBUTES.SPRAF_Import);
}
}
Arc arc = backend.RuleTransition(rule, _rule, 1f);
if (!string.IsNullOrEmpty(semanticKey))
{
backend.AddPropertyTag(arc, arc, new CfgGrammar.CfgProperty
{
_pszName = "SemanticKey",
_comValue = semanticKey,
_comType = VarEnum.VT_EMPTY
});
}
parent.AddArc(arc);
}
/// <summary>
/// Process the 'subset' element.
/// </summary>
public Subset(ParseElementCollection parent, Backend backend, string text, MatchMode mode)
: base(parent._rule)
{
// replace tab, cr, lf with spaces
foreach (char ch in Helpers._achTrimChars)
{
if (ch == ' ')
{
continue;
}
if (text.IndexOf(ch) >= 0)
{
text = text.Replace(ch, ' ');
}
}
// Add transition to the new state with normalized token.
parent.AddArc(backend.SubsetTransition(text, mode));
}
private void ParseToken(ParseElementCollection parent, string sToken, string pronunciation, string display, float reqConfidence)
{
int requiredConfidence = parent?._confidence ?? 0;
sToken = Backend.NormalizeTokenWhiteSpace(sToken);
if (string.IsNullOrEmpty(sToken))
{
return;
}
parent._confidence = 0;
if (reqConfidence < 0f || reqConfidence.Equals(0.5f))
{
parent._confidence = 0;
}
else if ((double)reqConfidence < 0.5)
{
parent._confidence = -1;
}
else
{
parent._confidence = 1;
}
if (pronunciation != null || display != null)
{
string text = EscapeToken(sToken);
string text2 = (display == null) ? text : EscapeToken(display);
if (pronunciation != null)
{
OneOf oneOf = (pronunciation.IndexOf(';') >= 0) ? new OneOf(parent._rule, _backend) : null;
int num = 0;
int num2 = 0;
while (num < pronunciation.Length)
{
num2 = pronunciation.IndexOf(';', num);
if (num2 == -1)
{
num2 = pronunciation.Length;
}
string text3 = pronunciation.Substring(num, num2 - num);
string text4 = null;
switch (_backend.Alphabet)
{
case AlphabetType.Sapi:
text4 = PhonemeConverter.ConvertPronToId(text3, _grammar.Backend.LangId);
break;
case AlphabetType.Ipa:
text4 = text3;
PhonemeConverter.ValidateUpsIds(text4);
break;
case AlphabetType.Ups:
text4 = PhonemeConverter.UpsConverter.ConvertPronToId(text3);
break;
}
string sWord = string.Format(CultureInfo.InvariantCulture, "/{0}/{1}/{2};", new object[3]
{
text2,
text,
text4
});
if (oneOf != null)
{
oneOf.AddArc(_backend.WordTransition(sWord, 1f, requiredConfidence));
}
else
{
parent.AddArc(_backend.WordTransition(sWord, 1f, requiredConfidence));
}
num = num2 + 1;
}
((IElement)oneOf)?.PostParse((IElement)parent);
}
else
{
string sWord2 = string.Format(CultureInfo.InvariantCulture, "/{0}/{1};", new object[2]
{
text2,
text
});
parent.AddArc(_backend.WordTransition(sWord2, 1f, requiredConfidence));
}
}
else
{
parent.AddArc(_backend.WordTransition(sToken, 1f, requiredConfidence));
}
}
// Disable parameter validation check
/// <summary>
/// Add transition representing the normalized token.
///
/// White Space Normalization - Trim leading/trailing white spaces.
/// Collapse white space sequences to a single ' '.
/// Restrictions - Normalized token cannot be empty.
/// Normalized token cannot contain double-quote.
///
/// If (Parent == Token) And (Parent.SAPIPron.Length > 0) Then
/// Escape normalized token. "/" -> "\/", "\" -> "\\"
/// Build /D/L/P; form from the escaped token and SAPIPron.
///
/// SAPIPron may be a semi-colon delimited list of pronunciations.
/// In this case, a transition for each of the pronunciations will be added.
///
/// AddTransition(NormalizedToken, Parent.EndState, NewState)
/// Parent.EndState = NewState
/// </summary>
private void ParseToken(ParseElementCollection parent, string sToken, string pronunciation, string display, float reqConfidence)
{
int requiredConfidence = (parent != null) ? parent._confidence : CfgGrammar.SP_NORMAL_CONFIDENCE;
// Performs white space normalization in place
sToken = Backend.NormalizeTokenWhiteSpace(sToken);
if (string.IsNullOrEmpty(sToken))
{
return;
}
// "sapi:reqconf" Attribute
parent._confidence = CfgGrammar.SP_NORMAL_CONFIDENCE; // Default to normal
if (reqConfidence < 0 || reqConfidence.Equals(0.5f))
{
parent._confidence = CfgGrammar.SP_NORMAL_CONFIDENCE; // Default to normal
}
else if (reqConfidence < 0.5)
{
parent._confidence = CfgGrammar.SP_LOW_CONFIDENCE;
}
else
{
parent._confidence = CfgGrammar.SP_HIGH_CONFIDENCE;
}
// If SAPIPron is specified, use /D/L/P; as the transition text, for each of the pronunciations.
if (pronunciation != null || display != null)
{
// Escape normalized token. "/" -> "\/", "\" -> "\\"
string sEscapedToken = EscapeToken(sToken);
string sDisplayToken = display == null ? sEscapedToken : EscapeToken(display);
if (pronunciation != null)
{
// Garbage transition is optional whereas Wildcard is not. So we need additional epsilon transition.
OneOf oneOf = pronunciation.IndexOf(';') >= 0 ? new OneOf(parent._rule, _backend) : null;
for (int iCurPron = 0, iDeliminator = 0; iCurPron < pronunciation.Length; iCurPron = iDeliminator + 1)
{
// Find semi-colon delimiter and replace with null
iDeliminator = pronunciation.IndexOf(';', iCurPron);
if (iDeliminator == -1)
{
iDeliminator = pronunciation.Length;
}
string pron = pronunciation.Substring(iCurPron, iDeliminator - iCurPron);
string sSubPron = null;
switch (_backend.Alphabet)
{
case AlphabetType.Sapi:
sSubPron = PhonemeConverter.ConvertPronToId(pron, _grammar.Backend.LangId);
break;
case AlphabetType.Ipa:
sSubPron = pron;
PhonemeConverter.ValidateUpsIds(sSubPron);
break;
case AlphabetType.Ups:
sSubPron = PhonemeConverter.UpsConverter.ConvertPronToId(pron);
break;
}
// Build /D/L/P; form for this pronunciation.
string sDLP = string.Format(CultureInfo.InvariantCulture, "/{0}/{1}/{2};", sDisplayToken, sEscapedToken, sSubPron);
// Add /D/L/P; transition to the new state.
if (oneOf != null)
{
oneOf.AddArc(_backend.WordTransition(sDLP, 1.0f, requiredConfidence));
}
else
{
parent.AddArc(_backend.WordTransition(sDLP, 1.0f, requiredConfidence));
}
}
if (oneOf != null)
{
((IOneOf)oneOf).PostParse(parent);
}
}
else
{
// Build /D/L; form for this pronunciation.
string sDLP = string.Format(CultureInfo.InvariantCulture, "/{0}/{1};", sDisplayToken, sEscapedToken);
// Add /D/L; transition to the new state.
parent.AddArc(_backend.WordTransition(sDLP, 1.0f, requiredConfidence));
}
}
else
{
// Add transition to the new state with normalized token.
parent.AddArc(_backend.WordTransition(sToken, 1.0f, requiredConfidence));
}
}
