private void ChangeEncoding(string encodingName)
{
var newEncoding = EncodingEx.GetEncodingByName(encodingName);
if (newEncoding == null || ReferenceEquals(newEncoding, encoding) || encoding.Equals(newEncoding))
{
return;
}
decoder = newEncoding.GetDecoder();
encoding = newEncoding;
}
private bool TryHandleParameterizedCommand(string name, string parameters)
{
var commandName = name.ToUpperInvariant();
switch (commandName)
{
case "FLAG": // parse in the try string
return(TrySetFlagMode(parameters));
case "KEY": // parse in the keyboard string
Builder.KeyString = Builder.Dedup(parameters);
return(true);
case "TRY": // parse in the try string
Builder.TryString = Builder.Dedup(parameters);
return(true);
case "SET": // parse in the name of the character set used by the .dict and .aff
var encoding = EncodingEx.GetEncodingByName(parameters);
if (encoding == null)
{
Builder.LogWarning("Failed to get encoding: " + parameters);
return(false);
}
Builder.Encoding = encoding;
return(true);
case "LANG": // parse in the language for language specific codes
Builder.Language = Builder.Dedup(parameters.Trim());
Builder.Culture = GetCultureFromLanguage(Builder.Language);
return(true);
case "SYLLABLENUM": // parse in the flag used by compound_check() method
Builder.CompoundSyllableNum = Builder.Dedup(parameters);
return(true);
case "WORDCHARS": // parse in the extra word characters
Builder.WordChars = CharacterSet.Create(parameters);
return(true);
case "IGNORE": // parse in the ignored characters (for example, Arabic optional diacretics characters)
Builder.IgnoredChars = CharacterSet.Create(parameters);
return(true);
case "COMPOUNDFLAG": // parse in the flag used by the controlled compound words
return(TryParseFlag(parameters, out Builder.CompoundFlag));
case "COMPOUNDMIDDLE": // parse in the flag used by compound words
return(TryParseFlag(parameters, out Builder.CompoundMiddle));
case "COMPOUNDBEGIN": // parse in the flag used by compound words
return(EnumEx.HasFlag(Builder.Options, AffixConfigOptions.ComplexPrefixes)
? TryParseFlag(parameters, out Builder.CompoundEnd)
: TryParseFlag(parameters, out Builder.CompoundBegin));
case "COMPOUNDEND": // parse in the flag used by compound words
return(EnumEx.HasFlag(Builder.Options, AffixConfigOptions.ComplexPrefixes)
? TryParseFlag(parameters, out Builder.CompoundBegin)
: TryParseFlag(parameters, out Builder.CompoundEnd));
case "COMPOUNDWORDMAX": // parse in the data used by compound_check() method
Builder.CompoundWordMax = IntEx.TryParseInvariant(parameters);
return(Builder.CompoundWordMax.HasValue);
case "COMPOUNDMIN": // parse in the minimal length for words in compounds
Builder.CompoundMin = IntEx.TryParseInvariant(parameters);
if (!Builder.CompoundMin.HasValue)
{
Builder.LogWarning("Failed to parse CompoundMin: " + parameters);
return(false);
}
if (Builder.CompoundMin.GetValueOrDefault() < 1)
{
Builder.CompoundMin = 1;
}
return(true);
case "COMPOUNDROOT": // parse in the flag sign compounds in dictionary
return(TryParseFlag(parameters, out Builder.CompoundRoot));
case "COMPOUNDPERMITFLAG": // parse in the flag used by compound_check() method
return(TryParseFlag(parameters, out Builder.CompoundPermitFlag));
case "COMPOUNDFORBIDFLAG": // parse in the flag used by compound_check() method
return(TryParseFlag(parameters, out Builder.CompoundForbidFlag));
case "COMPOUNDSYLLABLE": // parse in the max. words and syllables in compounds
return(TryParseCompoundSyllable(parameters));
case "NOSUGGEST":
return(TryParseFlag(parameters, out Builder.NoSuggest));
case "NONGRAMSUGGEST":
return(TryParseFlag(parameters, out Builder.NoNgramSuggest));
case "FORBIDDENWORD": // parse in the flag used by forbidden words
Builder.ForbiddenWord = TryParseFlag(parameters);
return(Builder.ForbiddenWord.HasValue);
case "LEMMA_PRESENT": // parse in the flag used by forbidden words
return(TryParseFlag(parameters, out Builder.LemmaPresent));
case "CIRCUMFIX": // parse in the flag used by circumfixes
return(TryParseFlag(parameters, out Builder.Circumfix));
case "ONLYINCOMPOUND": // parse in the flag used by fogemorphemes
return(TryParseFlag(parameters, out Builder.OnlyInCompound));
case "PSEUDOROOT": // parse in the flag used by `needaffixs'
case "NEEDAFFIX": // parse in the flag used by `needaffixs'
return(TryParseFlag(parameters, out Builder.NeedAffix));
case "REP": // parse in the typical fault correcting table
return(TryParseStandardListItem(EntryListType.Replacements, parameters, ref Builder.Replacements, TryParseReplacements));
case "ICONV": // parse in the input conversion table
return(TryParseConv(parameters, EntryListType.Iconv, ref Builder.InputConversions));
case "OCONV": // parse in the output conversion table
return(TryParseConv(parameters, EntryListType.Oconv, ref Builder.OutputConversions));
case "PHONE": // parse in the phonetic conversion table
return(TryParseStandardListItem(EntryListType.Phone, parameters, ref Builder.Phone, TryParsePhone));
case "CHECKCOMPOUNDPATTERN": // parse in the checkcompoundpattern table
return(TryParseStandardListItem(EntryListType.CompoundPatterns, parameters, ref Builder.CompoundPatterns, TryParseCheckCompoundPatternIntoCompoundPatterns));
case "COMPOUNDRULE": // parse in the defcompound table
return(TryParseStandardListItem(EntryListType.CompoundRules, parameters, ref Builder.CompoundRules, TryParseCompoundRuleIntoList));
case "MAP": // parse in the related character map table
return(TryParseStandardListItem(EntryListType.Map, parameters, ref Builder.RelatedCharacterMap, TryParseMapEntry));
case "BREAK": // parse in the word breakpoints table
return(TryParseStandardListItem(EntryListType.Break, parameters, ref Builder.BreakPoints, TryParseBreak));
case "VERSION":
Builder.Version = parameters;
return(true);
case "MAXNGRAMSUGS":
Builder.MaxNgramSuggestions = IntEx.TryParseInvariant(parameters);
return(Builder.MaxNgramSuggestions.HasValue);
case "MAXDIFF":
Builder.MaxDifferency = IntEx.TryParseInvariant(parameters);
return(Builder.MaxDifferency.HasValue);
case "MAXCPDSUGS":
Builder.MaxCompoundSuggestions = IntEx.TryParseInvariant(parameters);
return(Builder.MaxCompoundSuggestions.HasValue);
case "KEEPCASE": // parse in the flag used by forbidden words
return(TryParseFlag(parameters, out Builder.KeepCase));
case "FORCEUCASE":
return(TryParseFlag(parameters, out Builder.ForceUpperCase));
case "WARN":
return(TryParseFlag(parameters, out Builder.Warn));
case "SUBSTANDARD":
return(TryParseFlag(parameters, out Builder.SubStandard));
case "PFX":
case "SFX":
var parseAsPrefix = "PFX" == commandName;
if (EnumEx.HasFlag(Builder.Options, AffixConfigOptions.ComplexPrefixes))
{
parseAsPrefix = !parseAsPrefix;
}
return(parseAsPrefix
? TryParseAffixIntoList(parameters, ref Builder.Prefixes)
: TryParseAffixIntoList(parameters, ref Builder.Suffixes));
case "AF":
return(TryParseStandardListItem(EntryListType.AliasF, parameters, ref Builder.AliasF, TryParseAliasF));
case "AM":
return(TryParseStandardListItem(EntryListType.AliasM, parameters, ref Builder.AliasM, TryParseAliasM));
default:
Builder.LogWarning($"Unknown command {commandName} with params: {parameters}");
return(false);
}
}
private void ChangeEncoding(ReadOnlySpan <char> encodingName)
{
var newEncoding = EncodingEx.GetEncodingByName(encodingName);
ChangeEncoding(newEncoding);
}
private bool TryHandleParameterizedCommand(string commandName, string parameters)
{
if (commandName == null || !CommandMap.TryGetValue(commandName, out AffixReaderCommandKind command))
{
Builder.LogWarning($"Unknown command {commandName} with params: {parameters}");
return(false);
}
switch (command)
{
case AffixReaderCommandKind.Flag:
return(TrySetFlagMode(parameters));
case AffixReaderCommandKind.KeyString:
Builder.KeyString = Builder.Dedup(parameters);
return(true);
case AffixReaderCommandKind.TryString:
Builder.TryString = Builder.Dedup(parameters);
return(true);
case AffixReaderCommandKind.SetEncoding:
var encoding = EncodingEx.GetEncodingByName(parameters);
if (encoding == null)
{
Builder.LogWarning("Failed to get encoding: " + parameters);
return(false);
}
Builder.Encoding = encoding;
return(true);
case AffixReaderCommandKind.Language:
Builder.Language = Builder.Dedup(parameters.Trim());
Builder.Culture = GetCultureFromLanguage(Builder.Language);
return(true);
case AffixReaderCommandKind.CompoundSyllableNum:
Builder.CompoundSyllableNum = Builder.Dedup(parameters);
return(true);
case AffixReaderCommandKind.WordChars:
Builder.WordChars = CharacterSet.Create(parameters);
return(true);
case AffixReaderCommandKind.Ignore:
Builder.IgnoredChars = CharacterSet.Create(parameters);
return(true);
case AffixReaderCommandKind.CompoundFlag:
return(TryParseFlag(parameters, out Builder.CompoundFlag));
case AffixReaderCommandKind.CompoundMiddle:
return(TryParseFlag(parameters, out Builder.CompoundMiddle));
case AffixReaderCommandKind.CompoundBegin:
return(EnumEx.HasFlag(Builder.Options, AffixConfigOptions.ComplexPrefixes)
? TryParseFlag(parameters, out Builder.CompoundEnd)
: TryParseFlag(parameters, out Builder.CompoundBegin));
case AffixReaderCommandKind.CompoundEnd:
return(EnumEx.HasFlag(Builder.Options, AffixConfigOptions.ComplexPrefixes)
? TryParseFlag(parameters, out Builder.CompoundBegin)
: TryParseFlag(parameters, out Builder.CompoundEnd));
case AffixReaderCommandKind.CompoundWordMax:
Builder.CompoundWordMax = IntEx.TryParseInvariant(parameters);
return(Builder.CompoundWordMax.HasValue);
case AffixReaderCommandKind.CompoundMin:
Builder.CompoundMin = IntEx.TryParseInvariant(parameters);
if (!Builder.CompoundMin.HasValue)
{
Builder.LogWarning("Failed to parse CompoundMin: " + parameters);
return(false);
}
if (Builder.CompoundMin.GetValueOrDefault() < 1)
{
Builder.CompoundMin = 1;
}
return(true);
case AffixReaderCommandKind.CompoundRoot:
return(TryParseFlag(parameters, out Builder.CompoundRoot));
case AffixReaderCommandKind.CompoundPermitFlag:
return(TryParseFlag(parameters, out Builder.CompoundPermitFlag));
case AffixReaderCommandKind.CompoundForbidFlag:
return(TryParseFlag(parameters, out Builder.CompoundForbidFlag));
case AffixReaderCommandKind.CompoundSyllable:
return(TryParseCompoundSyllable(parameters));
case AffixReaderCommandKind.NoSuggest:
return(TryParseFlag(parameters, out Builder.NoSuggest));
case AffixReaderCommandKind.NoNGramSuggest:
return(TryParseFlag(parameters, out Builder.NoNgramSuggest));
case AffixReaderCommandKind.ForbiddenWord:
Builder.ForbiddenWord = TryParseFlag(parameters);
return(Builder.ForbiddenWord.HasValue);
case AffixReaderCommandKind.LemmaPresent:
return(TryParseFlag(parameters, out Builder.LemmaPresent));
case AffixReaderCommandKind.Circumfix:
return(TryParseFlag(parameters, out Builder.Circumfix));
case AffixReaderCommandKind.OnlyInCompound:
return(TryParseFlag(parameters, out Builder.OnlyInCompound));
case AffixReaderCommandKind.NeedAffix:
return(TryParseFlag(parameters, out Builder.NeedAffix));
case AffixReaderCommandKind.Replacement:
return(TryParseStandardListItem(EntryListType.Replacements, parameters, ref Builder.Replacements, TryParseReplacements));
case AffixReaderCommandKind.InputConversions:
return(TryParseConv(parameters, EntryListType.Iconv, ref Builder.InputConversions));
case AffixReaderCommandKind.OutputConversions:
return(TryParseConv(parameters, EntryListType.Oconv, ref Builder.OutputConversions));
case AffixReaderCommandKind.Phone:
return(TryParseStandardListItem(EntryListType.Phone, parameters, ref Builder.Phone, TryParsePhone));
case AffixReaderCommandKind.CheckCompoundPattern:
return(TryParseStandardListItem(EntryListType.CompoundPatterns, parameters, ref Builder.CompoundPatterns, TryParseCheckCompoundPatternIntoCompoundPatterns));
case AffixReaderCommandKind.CompoundRule:
return(TryParseStandardListItem(EntryListType.CompoundRules, parameters, ref Builder.CompoundRules, TryParseCompoundRuleIntoList));
case AffixReaderCommandKind.Map:
return(TryParseStandardListItem(EntryListType.Map, parameters, ref Builder.RelatedCharacterMap, TryParseMapEntry));
case AffixReaderCommandKind.Break:
return(TryParseStandardListItem(EntryListType.Break, parameters, ref Builder.BreakPoints, TryParseBreak));
case AffixReaderCommandKind.Version:
Builder.Version = parameters;
return(true);
case AffixReaderCommandKind.MaxNgramSuggestions:
Builder.MaxNgramSuggestions = IntEx.TryParseInvariant(parameters);
return(Builder.MaxNgramSuggestions.HasValue);
case AffixReaderCommandKind.MaxDifferency:
Builder.MaxDifferency = IntEx.TryParseInvariant(parameters);
return(Builder.MaxDifferency.HasValue);
case AffixReaderCommandKind.MaxCompoundSuggestions:
Builder.MaxCompoundSuggestions = IntEx.TryParseInvariant(parameters);
return(Builder.MaxCompoundSuggestions.HasValue);
case AffixReaderCommandKind.KeepCase:
return(TryParseFlag(parameters, out Builder.KeepCase));
case AffixReaderCommandKind.ForceUpperCase:
return(TryParseFlag(parameters, out Builder.ForceUpperCase));
case AffixReaderCommandKind.Warn:
return(TryParseFlag(parameters, out Builder.Warn));
case AffixReaderCommandKind.SubStandard:
return(TryParseFlag(parameters, out Builder.SubStandard));
case AffixReaderCommandKind.Prefix:
case AffixReaderCommandKind.Suffix:
var parseAsPrefix = AffixReaderCommandKind.Prefix == command;
if (EnumEx.HasFlag(Builder.Options, AffixConfigOptions.ComplexPrefixes))
{
parseAsPrefix = !parseAsPrefix;
}
return(parseAsPrefix
? TryParseAffixIntoList(parameters, ref Builder.Prefixes)
: TryParseAffixIntoList(parameters, ref Builder.Suffixes));
case AffixReaderCommandKind.AliasF:
return(TryParseStandardListItem(EntryListType.AliasF, parameters, ref Builder.AliasF, TryParseAliasF));
case AffixReaderCommandKind.AliasM:
return(TryParseStandardListItem(EntryListType.AliasM, parameters, ref Builder.AliasM, TryParseAliasM));
default:
Builder.LogWarning($"Unknown parsed command {command}");
return(false);
}
}
private void ChangeEncoding(StringSlice encodingName)
{
var newEncoding = EncodingEx.GetEncodingByName(encodingName);
ChangeEncoding(newEncoding);
}