/// <summary>
///
/// </summary>
/// <param name="lsett"></param>
/// <param name="elExamples"></param>
/// <param name="iStart">Index of the first word of the current group</param>
/// <param name="iEnd">Index of the last word of the current group</param>
/// <param name="ltnParentNode"></param>
private LemmaTreeNode(LemmatizerSettings lsett, ExampleList elExamples, int iStart, int iEnd,
LemmaTreeNode ltnParentNode) : this(lsett)
{
this.ltnParentNode = ltnParentNode;
dictSubNodes = null;
this.iStart = iStart;
this.iEnd = iEnd;
this.elExamples = elExamples;
if (iStart >= elExamples.Count || iEnd >= elExamples.Count || iStart > iEnd)
{
lrBestRule = elExamples.Rules.DefaultRule;
aBestRules = new RuleWeighted[1];
aBestRules[0] = new RuleWeighted(lrBestRule, 0);
dWeight = 0;
return;
}
int iConditionLength = Math.Min(ltnParentNode == null ? 0 : ltnParentNode.iSimilarity + 1,
elExamples[iStart].Word.Length);
sCondition = elExamples[iStart].Word.Substring(elExamples[iStart].Word.Length - iConditionLength);
iSimilarity = elExamples[iStart].Similarity(elExamples[iEnd]);
bWholeWord = ltnParentNode == null ? false : elExamples[iEnd].Word.Length == ltnParentNode.iSimilarity;
FindBestRules();
AddSubAll();
//TODO check this heuristics, can be problematic when there are more applicable rules
if (dictSubNodes != null)
{
List <KeyValuePair <char, LemmaTreeNode> > lReplaceNodes = new List <KeyValuePair <char, LemmaTreeNode> >();
foreach (KeyValuePair <char, LemmaTreeNode> kvpChild in dictSubNodes)
{
if (kvpChild.Value.dictSubNodes != null && kvpChild.Value.dictSubNodes.Count == 1)
{
IEnumerator <LemmaTreeNode> enumChildChild = kvpChild.Value.dictSubNodes.Values.GetEnumerator();
enumChildChild.MoveNext();
LemmaTreeNode ltrChildChild = enumChildChild.Current;
if (kvpChild.Value.lrBestRule == lrBestRule)
{
lReplaceNodes.Add(new KeyValuePair <char, LemmaTreeNode>(kvpChild.Key, ltrChildChild));
}
}
}
foreach (KeyValuePair <char, LemmaTreeNode> kvpChild in lReplaceNodes)
{
dictSubNodes[kvpChild.Key] = kvpChild.Value;
kvpChild.Value.ltnParentNode = this;
}
}
}
public void Deserialize(BinaryReader binRead, LemmatizerSettings lsett, ExampleList elExamples, LemmaTreeNode ltnParentNode)
{
this.lsett = lsett;
// read is not null?
if (binRead.ReadBoolean())
{
// read all dictionary (key + value)
dictSubNodes = new ConcurrentDictionary <char, LemmaTreeNode>();
int iCount = binRead.ReadInt32();
for (int i = 0; i < iCount; i++)
{
char cKey = binRead.ReadChar();
var ltrSub = new LemmaTreeNode(binRead, this.lsett, elExamples, this);
dictSubNodes.TryAdd(cKey, ltrSub);
}
}
else
{
dictSubNodes = null;
}
this.ltnParentNode = ltnParentNode;
// read similarity, condition and wholeword?
iSimilarity = binRead.ReadInt32();
sCondition = binRead.ReadString();
bWholeWord = binRead.ReadBoolean();
// best rule signature
lrBestRule = elExamples.Rules[binRead.ReadString()];
// best rules
int iCountBest = binRead.ReadInt32();
aBestRules = new RuleWeighted[iCountBest];
for (int i = 0; i < iCountBest; i++)
{
aBestRules[i] = new RuleWeighted(elExamples.Rules[binRead.ReadString()], binRead.ReadDouble());
}
// weight, start, end
dWeight = binRead.ReadDouble();
iStart = binRead.ReadInt32();
iEnd = binRead.ReadInt32();
this.elExamples = elExamples;
}
public void Deserialize(BinaryReader binRead, LemmatizerSettings lsett, ExampleList elExamples,
LemmaTreeNode ltnParentNode)
{
this.lsett = lsett;
if (binRead.ReadBoolean())
{
dictSubNodes = new Dictionary <char, LemmaTreeNode>();
int iCount = binRead.ReadInt32();
for (int i = 0; i < iCount; i++)
{
char cKey = binRead.ReadChar();
LemmaTreeNode ltrSub = new LemmaTreeNode(binRead, this.lsett, elExamples, this);
dictSubNodes.Add(cKey, ltrSub);
}
}
else
{
dictSubNodes = null;
}
this.ltnParentNode = ltnParentNode;
iSimilarity = binRead.ReadInt32();
sCondition = binRead.ReadString();
bWholeWord = binRead.ReadBoolean();
lrBestRule = elExamples.Rules[binRead.ReadString()];
int iCountBest = binRead.ReadInt32();
aBestRules = new RuleWeighted[iCountBest];
for (int i = 0; i < iCountBest; i++)
{
aBestRules[i] = new RuleWeighted(elExamples.Rules[binRead.ReadString()], binRead.ReadDouble());
}
dWeight = binRead.ReadDouble();
iStart = binRead.ReadInt32();
iEnd = binRead.ReadInt32();
this.elExamples = elExamples;
}
public void Deserialize(BinaryReader binRead, LemmatizerSettings lsett, ExampleList elExamples, LemmaTreeNode ltnParentNode)
{
this.lsett = lsett;
if (binRead.ReadBoolean())
{
dictSubNodes = new Dictionary <char, LemmaTreeNode>();
int iCount = binRead.ReadInt32();
for (int i = 0; i < iCount; i++)
{
char cKey = binRead.ReadChar();
LemmaTreeNode ltrSub = new LemmaTreeNode(binRead, this.lsett, elExamples, this);
dictSubNodes.Add(cKey, ltrSub);
}
}
else
{
dictSubNodes = null;
}
this.ltnParentNode = ltnParentNode;
iSimilarity = binRead.ReadInt32();
sCondition = binRead.ReadString();
bWholeWord = binRead.ReadBoolean();
lrBestRule = elExamples.Rules[binRead.ReadString()];
int iCountBest = binRead.ReadInt32();
aBestRules = new RuleWeighted[iCountBest];
for (int i = 0; i < iCountBest; i++)
{
aBestRules[i] = new RuleWeighted(elExamples.Rules[binRead.ReadString()], binRead.ReadDouble());
}
dWeight = binRead.ReadDouble();
//deserialize dictMsdBestRules dictionary
int dictMsdBestRulesCount = binRead.ReadInt32();
if (dictMsdBestRulesCount == -1)
{
dictMsdBestRules = null;
}
else
{
dictMsdBestRules = new Dictionary <string, RuleWeighted[]>();
for (int msdId = 0; msdId < dictMsdBestRulesCount; msdId++)
{
string sMsd = binRead.ReadString();
RuleWeighted[] lRuleWeighted;
int ruleWeightedCount = binRead.ReadInt32();
if (ruleWeightedCount == -1)
{
lRuleWeighted = null;
}
else
{
lRuleWeighted = new RuleWeighted[ruleWeightedCount];
for (int ruleId = 0; ruleId < ruleWeightedCount; ruleId++)
{
string ruleSignature = binRead.ReadString();
double ruleWeight = binRead.ReadDouble();
LemmaRule rule = elExamples.Rules[ruleSignature];
lRuleWeighted[ruleId] = new RuleWeighted(rule, ruleWeight);
}
}
dictMsdBestRules.Add(sMsd, lRuleWeighted);
}
}
//deserialize dictMsdWeights dictionary
int dictMsdWeightsCount = binRead.ReadInt32();
if (dictMsdWeightsCount == -1)
{
dictMsdWeights = null;
}
else
{
dictMsdWeights = new Dictionary <string, double>();
for (int msdId = 0; msdId < dictMsdWeightsCount; msdId++)
{
string sMsd = binRead.ReadString();
double dMsdWeight = binRead.ReadDouble();
dictMsdWeights.Add(sMsd, dMsdWeight);
}
}
iStart = binRead.ReadInt32();
iEnd = binRead.ReadInt32();
this.elExamples = elExamples;
}