/// <summary>
/// Get subchain of chainkeys and words that satisfies the passed in condition
/// </summary>
/// <param name="condition"></param>
/// <returns></returns>
internal ChainMap GetSubChain(ChainCondition condition)
{
//If conditions won't be applied return original chains
if (condition == null || condition == _alwaysTrueCondition ||
(condition.WordCondition == null && condition.KeyCondition == null))
{
return(_chains);
}
ChainMap newMap = new ChainMap();
bool valuesFound = false;
foreach (ChainKey key in _chains.GetAllKeys())
{
if (condition.KeyCondition != null)
{
if (condition.KeyCondition.Invoke(key))
{
foreach (string word in _chains.GetValues(key))
{
if (condition.WordCondition != null)
{
if (condition.WordCondition.Invoke(word))
{
newMap.AddToChain(key, word);
valuesFound = true;
}
}
}
}
}
}
if (!valuesFound)
{
throw new Exceptions.NoPossibleElements("Chain map filter filtered everything out");
}
return(newMap);
}
/// <summary>
/// Generate a new sentece that contains a minimum number of non punctuation words
/// This minimum number of words is not guaranteed, but it is a best try
/// </summary>
/// <param name="minLength">Minimum length of sentence. Use 0 to indicate
/// no minimum</param>
/// <returns></returns>
public string GenerateSentence(int minLength)
{
if (minLength < 0)
{
throw new Exceptions.InvalidArguments("Number of words generated must be positive");
}
if (minLength > Defs.MAX_SENTENCE_LENGTH)
{
throw new Exceptions.InvalidArguments("Minimum sentence length greater than default maximum of " +
Defs.MAX_SENTENCE_LENGTH);
}
StringBuilder sb = new StringBuilder();
//Word generator should reset to beginning of a sentence if needed
ChainCondition sentenceStartCondition = new ChainCondition(
ck => IsSentenceEnd(ck.Words[ck.Words.Length - 1]) ||
ck.Words[ck.Words.Length - 1] == null, w => !IsPunctuation(w));
//redundant for clarity
if (!_sentenceGenerator.SubchainsInitialized)
{
_sentenceGenerator.ResetSubchains(sentenceStartCondition, null);
}
else if (IsSentenceEnd(_sentenceGenerator.CurrentWord))
{
//We are already at the beginning of a new sentence
//This is to try to optimize perfomance so that we don't do a reset more than necessary
//A reset is an expensive operation as all chains need to be parsed
_sentenceGenerator.GetNextWord(sentenceStartCondition.WordCondition, Defs.MAX_ITERATIONS);
}
if (!(sentenceStartCondition.KeyCondition(_sentenceGenerator.CurrKey) &&
(sentenceStartCondition.WordCondition(_sentenceGenerator.CurrentWord))))
{
_sentenceGenerator.ResetReadOnly();
}
int countWords;
for (countWords = 0; countWords < Defs.MAX_SENTENCE_LENGTH; countWords++)
{
string nextWord = _sentenceGenerator.CurrentWord;
if (!(countWords == 0 || IsPunctuation(nextWord)))
{
//prepend space if not first word or punctuation
sb.Append(" ");
sb.Append(nextWord);
}
else if (countWords == 0)
{
//capitalize first word
if (!string.IsNullOrEmpty(nextWord))
{
sb.Append(nextWord.Capitalize());
}
}
else
{
sb.Append(nextWord);
}
//If we get to sentence end, we are done
if (IsSentenceEnd(nextWord))
{
break;
}
if (countWords < minLength)
{
_sentenceGenerator.GetNextWord(w => !IsSentenceEnd(w), Defs.MAX_ITERATIONS);
}
else
{
_sentenceGenerator.GetNextWord();
}
}
if (countWords == Defs.MAX_SENTENCE_LENGTH)
{
//shouldn't get here, but if we do, put any random sentence end
sb.Append(SentenceEnd.FirstOrDefault());
}
return(sb.ToString());
}
/// <summary>
/// Initialize pool of random keys
/// The probability of picking a key is proportional to the number of values in the ChainMap
/// </summary>
public void Initialize()
{
_alwaysTrueCondition = new ChainCondition(p => true, p => true);
_keys = CreateKeysArray(_chains, _alwaysTrueCondition.KeyCondition);
_currentOnEndCondition = _alwaysTrueCondition;
}
/// <summary>
/// Reset word generator with same condition for initial chain as final chain
/// </summary>
/// <param name="initialCondition"></param>
public void ResetSubchains(ChainCondition initialCondition)
{
Reset(initialCondition, null);
}
/// <summary>
/// Reset CurrentWord based on conditions
/// </summary>
/// <param name="keyCondition"> Condition for key of first word </param>
/// <param name="onEndCondition"> Condition for new word after there are no more chains
/// By default this is the same as @keyCondition </param>
private void Reset(ChainCondition initialCondition, ChainCondition onEndCondition)
{
ChainCondition realInitialConditions = initialCondition;
//Set null conditions to always true conditions to simplify our code
if (realInitialConditions == null)
{
realInitialConditions = _alwaysTrueCondition;
}
else
{
if (realInitialConditions.KeyCondition == null)
{
realInitialConditions.KeyCondition = _alwaysTrueCondition.KeyCondition;
}
if (realInitialConditions.WordCondition == null)
{
realInitialConditions.WordCondition = _alwaysTrueCondition.WordCondition;
}
}
ChainCondition realOnEndCondition = onEndCondition;
//Set on end condition to start condition if it is null
if (realOnEndCondition == null)
{
realOnEndCondition = realInitialConditions;
}
else
{
if (realOnEndCondition.KeyCondition == null)
{
realOnEndCondition.KeyCondition = _alwaysTrueCondition.KeyCondition;
}
if (realOnEndCondition.WordCondition == null)
{
realOnEndCondition.WordCondition = _alwaysTrueCondition.WordCondition;
}
}
ChainMap initialConditionMap;
//Only generate subchains if we need to
if (realInitialConditions != _alwaysTrueCondition)
{
initialConditionMap = GetSubChain(realInitialConditions);
}
else
{
initialConditionMap = Chains;
}
ChainKey[] initialConditionSubKeys = CreateKeysArray(initialConditionMap,
_alwaysTrueCondition.KeyCondition);
//All chains and words should match condition so we can just get random
ChainKey currKey = GetRandomKey(initialConditionSubKeys);
string currWord;
GetCandidate(initialConditionMap.GetValues(currKey), null, out currWord);
if (realOnEndCondition != _alwaysTrueCondition)
{
//Verify on end map has valid elements. If no exception is thrown we are good to go
_onEndSubChains = GetSubChain(realOnEndCondition);
}
else
{
_onEndSubChains = initialConditionMap;
}
CurrentWord = currWord;
_currKey = currKey;
_initialConditionSubChains = initialConditionMap;
_initialConditionSubKeys = initialConditionSubKeys;
_currentOnEndCondition = realOnEndCondition;
SubchainsInitialized = true;
}
/// <summary>
/// Generate a new sentece that contains a minimum number of non punctuation words
/// This minimum number of words is not guaranteed, but it is a best try
/// </summary>
/// <param name="minLength">Minimum length of sentence. Use 0 to indicate
/// no minimum</param>
/// <returns></returns>
public string GenerateSentence(int minLength)
{
if (minLength < 0)
{
throw new Exceptions.InvalidArguments("Number of words generated must be positive");
}
if (minLength > Defs.MAX_SENTENCE_LENGTH)
{
throw new Exceptions.InvalidArguments("Minimum sentence length greater than default maximum of " +
Defs.MAX_SENTENCE_LENGTH);
}
StringBuilder sb = new StringBuilder();
//Word generator should reset to beginning of a sentence if needed
ChainCondition sentenceStartCondition = new ChainCondition(
ck => IsSentenceEnd(ck.Words[ck.Words.Length - 1]) ||
ck.Words[ck.Words.Length-1] == null, w => !IsPunctuation(w)) ;
//redundant for clarity
if(!_sentenceGenerator.SubchainsInitialized)
{
_sentenceGenerator.ResetSubchains(sentenceStartCondition, null);
}
else if(IsSentenceEnd(_sentenceGenerator.CurrentWord))
{
//We are already at the beginning of a new sentence
//This is to try to optimize perfomance so that we don't do a reset more than necessary
//A reset is an expensive operation as all chains need to be parsed
_sentenceGenerator.GetNextWord(sentenceStartCondition.WordCondition, Defs.MAX_ITERATIONS) ;
}
if(!(sentenceStartCondition.KeyCondition(_sentenceGenerator.CurrKey) &&
(sentenceStartCondition.WordCondition(_sentenceGenerator.CurrentWord))))
{
_sentenceGenerator.ResetReadOnly();
}
int countWords ;
for(countWords = 0 ; countWords < Defs.MAX_SENTENCE_LENGTH ; countWords++)
{
string nextWord = _sentenceGenerator.CurrentWord;
if(!(countWords == 0 || IsPunctuation(nextWord)))
{
//prepend space if not first word or punctuation
sb.Append(" ") ;
sb.Append(nextWord) ;
}
else if (countWords == 0)
{
//capitalize first word
if(!string.IsNullOrEmpty(nextWord))
{
sb.Append(nextWord.Capitalize()) ;
}
}
else
{
sb.Append(nextWord) ;
}
//If we get to sentence end, we are done
if(IsSentenceEnd(nextWord))
{
break ;
}
if(countWords < minLength)
{
_sentenceGenerator.GetNextWord(w => !IsSentenceEnd(w), Defs.MAX_ITERATIONS) ;
}
else
{
_sentenceGenerator.GetNextWord() ;
}
}
if(countWords == Defs.MAX_SENTENCE_LENGTH)
{
//shouldn't get here, but if we do, put any random sentence end
sb.Append(SentenceEnd.FirstOrDefault()) ;
}
return sb.ToString();
}
/// <summary>
/// Reset using new initial and onEndCondition. This requires quite a bit of CPU
/// as all the chains are processed to get a valid subset for the initial and on end
/// conditions
/// </summary>
public void ResetSubchains(ChainCondition initialCondition, ChainCondition onEndCondition)
{
Reset(initialCondition, onEndCondition);
}
/// <summary>
/// Reset CurrentWord based on conditions
/// </summary>
/// <param name="keyCondition"> Condition for key of first word </param>
/// <param name="onEndCondition"> Condition for new word after there are no more chains
/// By default this is the same as @keyCondition </param>
private void Reset(ChainCondition initialCondition, ChainCondition onEndCondition)
{
ChainCondition realInitialConditions = initialCondition;
//Set null conditions to always true conditions to simplify our code
if(realInitialConditions == null)
{
realInitialConditions = _alwaysTrueCondition;
}
else
{
if (realInitialConditions.KeyCondition == null)
{
realInitialConditions.KeyCondition = _alwaysTrueCondition.KeyCondition;
}
if (realInitialConditions.WordCondition == null)
{
realInitialConditions.WordCondition = _alwaysTrueCondition.WordCondition;
}
}
ChainCondition realOnEndCondition = onEndCondition ;
//Set on end condition to start condition if it is null
if(realOnEndCondition == null)
{
realOnEndCondition = realInitialConditions ;
}
else
{
if (realOnEndCondition.KeyCondition == null)
{
realOnEndCondition.KeyCondition = _alwaysTrueCondition.KeyCondition;
}
if (realOnEndCondition.WordCondition == null)
{
realOnEndCondition.WordCondition = _alwaysTrueCondition.WordCondition;
}
}
ChainMap initialConditionMap ;
//Only generate subchains if we need to
if(realInitialConditions != _alwaysTrueCondition)
{
initialConditionMap = GetSubChain(realInitialConditions) ;
}
else
{
initialConditionMap = Chains;
}
ChainKey[] initialConditionSubKeys = CreateKeysArray(initialConditionMap,
_alwaysTrueCondition.KeyCondition) ;
//All chains and words should match condition so we can just get random
ChainKey currKey = GetRandomKey(initialConditionSubKeys) ;
string currWord ;
GetCandidate(initialConditionMap.GetValues(currKey), null, out currWord) ;
if(realOnEndCondition != _alwaysTrueCondition)
{
//Verify on end map has valid elements. If no exception is thrown we are good to go
_onEndSubChains = GetSubChain(realOnEndCondition) ;
}
else
{
_onEndSubChains = initialConditionMap;
}
CurrentWord = currWord;
_currKey = currKey;
_initialConditionSubChains = initialConditionMap ;
_initialConditionSubKeys = initialConditionSubKeys;
_currentOnEndCondition = realOnEndCondition;
SubchainsInitialized = true ;
}
/// <summary>
/// Get subchain of chainkeys and words that satisfies the passed in condition
/// </summary>
/// <param name="condition"></param>
/// <returns></returns>
internal ChainMap GetSubChain(ChainCondition condition)
{
//If conditions won't be applied return original chains
if(condition == null || condition == _alwaysTrueCondition ||
(condition.WordCondition == null && condition.KeyCondition == null))
{
return _chains ;
}
ChainMap newMap = new ChainMap() ;
bool valuesFound = false ;
foreach(ChainKey key in _chains.GetAllKeys())
{
if(condition.KeyCondition != null)
{
if(condition.KeyCondition.Invoke(key))
{
foreach(string word in _chains.GetValues(key))
{
if(condition.WordCondition != null)
{
if(condition.WordCondition.Invoke(word))
{
newMap.AddToChain(key, word) ;
valuesFound = true ;
}
}
}
}
}
}
if(!valuesFound)
{
throw new Exceptions.NoPossibleElements("Chain map filter filtered everything out") ;
}
return newMap ;
}
/// <summary>
/// Reset word generator with same condition for initial chain as final chain
/// </summary>
/// <param name="initialCondition"></param>
public void ResetSubchains(ChainCondition initialCondition)
{
Reset(initialCondition, null) ;
}
/// <summary>
/// Reset using new initial and onEndCondition. This requires quite a bit of CPU
/// as all the chains are processed to get a valid subset for the initial and on end
/// conditions
/// </summary>
public void ResetSubchains(ChainCondition initialCondition, ChainCondition onEndCondition)
{
Reset(initialCondition, onEndCondition) ;
}
/// <summary>
/// Initialize pool of random keys
/// The probability of picking a key is proportional to the number of values in the ChainMap
/// </summary>
public void Initialize()
{
_alwaysTrueCondition = new ChainCondition(p => true, p => true);
_keys = CreateKeysArray(_chains, _alwaysTrueCondition.KeyCondition) ;
_currentOnEndCondition = _alwaysTrueCondition ;
}