public override DetokenizationOperation[] GetDetokenizationOperations(string[] tokens)
{
var operations = new DetokenizationOperation[tokens.Length];
var matchingTokens = new HashSet <string>();
for (int i = 0; i < tokens.Length; i++)
{
if (!_tokenToDetokenizationOperation.ContainsKey(tokens[i]))
{
operations[i] = DetokenizationOperation.NO_OPERATION;
continue;
}
DetokenizationOperation dictOperation = _tokenToDetokenizationOperation[tokens[i]];
if (dictOperation == DetokenizationOperation.MERGE_TO_LEFT ||
dictOperation == DetokenizationOperation.MERGE_TO_RIGHT ||
dictOperation == DetokenizationOperation.MERGE_BOTH)
{
operations[i] = dictOperation;
}
else if (dictOperation == DetokenizationOperation.MERGE_BOTH_IF_SURROUNDED_BY_WORDS)
{
if (0 < i && i < tokens.Length - 1 && WordRegex.IsMatch(tokens[i - 1]) && WordRegex.IsMatch(tokens[i + 1]))
{
operations[i] = DetokenizationOperation.MERGE_BOTH;
}
else
{
operations[i] = DetokenizationOperation.NO_OPERATION;
}
}
else if (dictOperation == DetokenizationOperation.RIGHT_LEFT_MATCHING)
{
if (matchingTokens.Contains(tokens[i]))
{
// The token already occurred once, move it to the left and clear the occurrence flag
operations[i] = DetokenizationOperation.MERGE_TO_LEFT;
matchingTokens.Remove(tokens[i]);
}
else
{
// First time this token is seen, move it to the right and remember it
operations[i] = DetokenizationOperation.MERGE_TO_RIGHT;
matchingTokens.Add(tokens[i]);
}
}
else
{
#if DNF
throw new InvalidEnumArgumentException("Unknown operation: " + dictOperation);
#else
throw new ArgumentException("Unknown operation: " + dictOperation);
#endif
}
}
return(operations);
}
public override DetokenizationOperation[] GetDetokenizationOperations(string[] tokens)
{
var operations = new DetokenizationOperation[tokens.Length];
var matchingTokens = new HashSet<string>();
for (int i = 0; i < tokens.Length; i++)
{
if (!_tokenToDetokenizationOperation.ContainsKey(tokens[i]))
{
operations[i] = DetokenizationOperation.NO_OPERATION;
continue;
}
DetokenizationOperation dictOperation = _tokenToDetokenizationOperation[tokens[i]];
if (dictOperation == DetokenizationOperation.MERGE_TO_LEFT
|| dictOperation == DetokenizationOperation.MERGE_TO_RIGHT
|| dictOperation == DetokenizationOperation.MERGE_BOTH)
{
operations[i] = dictOperation;
}
else if (dictOperation == DetokenizationOperation.MERGE_BOTH_IF_SURROUNDED_BY_WORDS)
{
if(0 < i && i < tokens.Length - 1 && WordRegex.IsMatch(tokens[i-1]) && WordRegex.IsMatch(tokens[i+1])){
operations[i] = DetokenizationOperation.MERGE_BOTH;
}
else
{
operations[i] = DetokenizationOperation.NO_OPERATION;
}
}
else if (dictOperation == DetokenizationOperation.RIGHT_LEFT_MATCHING)
{
if (matchingTokens.Contains(tokens[i]))
{
// The token already occurred once, move it to the left and clear the occurrence flag
operations[i] = DetokenizationOperation.MERGE_TO_LEFT;
matchingTokens.Remove(tokens[i]);
}
else
{
// First time this token is seen, move it to the right and remember it
operations[i] = DetokenizationOperation.MERGE_TO_RIGHT;
matchingTokens.Add(tokens[i]);
}
}
else
{
throw new InvalidEnumArgumentException("Unknown operation: " + dictOperation);
}
}
return operations;
}
/// <summary>Detokenizes the specified tokens.</summary>
/// <param name="tokens">The tokens to detokenize.</param>
/// <returns>The merge operations to detokenize the input tokens.</returns>
public DetokenizationOperation[] Detokenize(string[] tokens)
{
var operations = new DetokenizationOperation[tokens.Length];
var matchingTokens = new HashSet <string>();
for (var i = 0; i < tokens.Length; i++)
{
var dictOperation = dictionary[tokens[i]];
switch (dictOperation)
{
case DetokenizationDictionary.Operation.MoveRight:
operations[i] = DetokenizationOperation.MergeToRight;
break;
case DetokenizationDictionary.Operation.MoveLeft:
operations[i] = DetokenizationOperation.MergeToLeft;
break;
case DetokenizationDictionary.Operation.MoveBoth:
operations[i] = DetokenizationOperation.MergeBoth;
break;
case DetokenizationDictionary.Operation.RightLeftMatching:
if (matchingTokens.Contains(tokens[i]))
{
// The token already occurred once, move it to the left
// and clear the occurrence flag
operations[i] = DetokenizationOperation.MergeToLeft;
matchingTokens.Remove(tokens[i]);
}
else
{
// First time this token is seen, move it to the right
// and remember it
operations[i] = DetokenizationOperation.MergeToRight;
matchingTokens.Add(tokens[i]);
}
break;
case null:
operations[i] = DetokenizationOperation.NoOperation;
break;
default:
throw new ArgumentOutOfRangeException("Unknown operation: " + dictionary[tokens[i]]);
}
}
return(operations);
}
/// <summary>Detokenizes the specified tokens.</summary>
/// <param name="tokens">The tokens to detokenize.</param>
/// <returns>The merge operations to detokenize the input tokens.</returns>
public DetokenizationOperation[] Detokenize(string[] tokens) {
var operations = new DetokenizationOperation[tokens.Length];
var matchingTokens = new HashSet<string>();
for (var i = 0; i < tokens.Length; i++) {
var dictOperation = dictionary[tokens[i]];
switch (dictOperation) {
case DetokenizationDictionary.Operation.MoveRight:
operations[i] = DetokenizationOperation.MergeToRight;
break;
case DetokenizationDictionary.Operation.MoveLeft:
operations[i] = DetokenizationOperation.MergeToLeft;
break;
case DetokenizationDictionary.Operation.MoveBoth:
operations[i] = DetokenizationOperation.MergeBoth;
break;
case DetokenizationDictionary.Operation.RightLeftMatching:
if (matchingTokens.Contains(tokens[i])) {
// The token already occurred once, move it to the left
// and clear the occurrence flag
operations[i] = DetokenizationOperation.MergeToLeft;
matchingTokens.Remove(tokens[i]);
} else {
// First time this token is seen, move it to the right
// and remember it
operations[i] = DetokenizationOperation.MergeToRight;
matchingTokens.Add(tokens[i]);
}
break;
case null:
operations[i] = DetokenizationOperation.NoOperation;
break;
default:
throw new ArgumentOutOfRangeException("Unknown operation: " + dictionary[tokens[i]]);
}
}
return operations;
}
internal static bool IsMergeToLeft(DetokenizationOperation operation)
{
return(operation == DetokenizationOperation.MergeToLeft ||
operation == DetokenizationOperation.MergeBoth);
}
internal static bool IsMergeToLeft(DetokenizationOperation operation) {
return operation == DetokenizationOperation.MergeToLeft ||
operation == DetokenizationOperation.MergeBoth;
}