public static void BuildAllLemmatizers(bool bIncludeExamples, bool bCompress, LemmatizerSettings lsett, string sNewFileMask)
{
DateTime dtStartAll = DateTime.Now;
foreach (LanguagePrebuilt lp in Enum.GetValues(typeof(LanguagePrebuilt))) {
DateTime dtStart = DateTime.Now;
string sDataFileName = GetResourceFileName(sFileMask, lp);
string sLemFileName = GetResourceFileName(sNewFileMask, lp);
Console.WriteLine("Building lemmatizer for {0} from {1}", lp, sDataFileName);
LemmatizerPrebuiltData lemPrebuild = new LemmatizerPrebuiltData(lp, lsett);
Console.WriteLine(" Building lemmatizer completed in: {0}", new TimeSpan(DateTime.Now.Ticks - dtStart.Ticks).ToString());
dtStart = DateTime.Now;
Console.WriteLine(" Serializing {0} lemmatizer to file {1}", (bIncludeExamples ? "full (examples included)" : "compact (examples excluded)"), sLemFileName);
Stream binStreamOut = File.Open(sLemFileName, FileMode.Create);
lemPrebuild.Serialize(binStreamOut, bIncludeExamples, bCompress);
binStreamOut.Close();
Console.WriteLine(" Successfully completed, time needed: {0}", new TimeSpan(DateTime.Now.Ticks - dtStart.Ticks).ToString());
}
Console.WriteLine("Time for building all lemmatizers: " + new TimeSpan(DateTime.Now.Ticks - dtStartAll.Ticks).ToString());
}
public ExampleList(LemmatizerSettings lsett)
: base()
{
this.lsett = lsett;
this.dictExamples = new Dictionary<string, LemmaExample>();
this.lstExamples = null;
this.rlRules = new RuleList(lsett);
}
public LemmatizerPrebuiltData(LanguagePrebuilt lang, LemmatizerSettings lsett)
: base(lang,lsett)
{
Stream stream = GetResourceStream(ResourceFileName);
StreamReader srIn = new StreamReader(stream, LanguageEncoding);
AddMultextFile(srIn, "WLM");
srIn.Close();
BuildModel();
}
public ExampleList(SerializationInfo info, StreamingContext context)
{
lsett = (LemmatizerSettings)info.GetValue("lsett", typeof(LemmatizerSettings));
this.dictExamples = new Dictionary<string, LemmaExample>();
this.lstExamples = null;
this.rlRules = new RuleList(lsett);
string[] aWords = (string[])info.GetValue("aWords", typeof(string[]));
string[] aLemmas = (string[])info.GetValue("aLemmas", typeof(string[]));
double[] aWeights = (double[])info.GetValue("aWeights", typeof(double[]));
string[] aMsds = (string[])info.GetValue("aMsds", typeof(string[]));
for (int iExm = 0; iExm < aWords.Length; iExm++)
AddExample(aWords[iExm], aLemmas[iExm], aWeights[iExm], aMsds[iExm]);
}
public LemmaRule(string sWord, string sLemma, int iId, LemmatizerSettings lsett)
{
this.lsett = lsett;
this.iId = iId;
int iSameStem = SameStem(sWord, sLemma);
sTo = sLemma.Substring(iSameStem);
iFrom = sWord.Length - iSameStem;
if (lsett.bUseFromInRules) {
sFrom = sWord.Substring(iSameStem);
sSignature = "[" + sFrom + "]==>[" + sTo + "]";
}
else {
sFrom = null;
sSignature = "[#" + iFrom + "]==>[" + sTo + "]";
}
}
/// <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;
this.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);
this.sCondition = elExamples[iStart].Word.Substring(elExamples[iStart].Word.Length - iConditionLength);
this.iSimilarity = elExamples[iStart].Similarity(elExamples[iEnd]);
this.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)
{
//load metadata
bool bThisTopObject = binRead.ReadBoolean();
//load value types --------------------------------------
iId = binRead.ReadInt32();
iFrom = binRead.ReadInt32();
if (binRead.ReadBoolean())
sFrom = binRead.ReadString();
else
sFrom = null;
sTo = binRead.ReadString();
sSignature = binRead.ReadString();
//load refernce types if needed -------------------------
if (bThisTopObject)
this.lsett = new LemmatizerSettings(binRead);
else
this.lsett = lsett;
}
public LemmaTreeNode(BinaryReader binRead, LemmatizerSettings lsett, ExampleList elExamples, LemmaTreeNode ltnParentNode)
{
Deserialize(binRead, lsett, elExamples, ltnParentNode);
}
public void Load(Latino.BinarySerializer binRead, LemmatizerSettings lsett, ExampleList elExamples, LemmaTreeNode ltnParentNode)
{
this.lsett = lsett;
if (binRead.ReadBool()) {
dictSubNodes = new Dictionary<char, LemmaTreeNode>();
int iCount = binRead.ReadInt();
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.ReadInt();
sCondition = binRead.ReadString();
bWholeWord = binRead.ReadBool();
lrBestRule = elExamples.Rules[binRead.ReadString()];
int iCountBest = binRead.ReadInt();
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.ReadInt();
iEnd = binRead.ReadInt();
this.elExamples = elExamples;
}
public LemmaTreeNode(Latino.BinarySerializer binRead, LemmatizerSettings lsett, ExampleList elExamples, LemmaTreeNode ltnParentNode)
{
Load(binRead, lsett, elExamples, ltnParentNode);
}
private string sCondition; //suffix that must match in order to lemmatize
#endregion Fields
#region Constructors
public LemmaTreeNode(LemmatizerSettings lsett, ExampleList elExamples)
: this(lsett, elExamples, 0, elExamples.Count-1, null)
{
}
public LemmatizerPrebuilt(LanguagePrebuilt lang, LemmatizerSettings lsett)
: base(lsett)
{
this.lang = lang;
}
public LemmaRule(System.IO.BinaryReader binRead, LemmatizerSettings lsett)
{
this.Deserialize(binRead, lsett);
}
public ExampleList(BinaryReader binRead, LemmatizerSettings lsett)
{
Deserialize(binRead, lsett);
}
public LemmaRule(Latino.BinarySerializer binRead, LemmatizerSettings lsett)
{
Load(binRead, lsett);
}
public ExampleList(StreamReader srIn, string sFormat, LemmatizerSettings lsett)
: this(lsett)
{
AddMultextFile(srIn, sFormat);
}
public void Load(Latino.BinarySerializer binRead)
{
lsett = new LemmatizerSettings(binRead);
elExamples = new ExampleList(binRead, lsett);
ltnRootNode = new LemmaTreeNode(binRead, lsett, elExamples, null);
}
public void Load(Latino.BinarySerializer binRead, LemmatizerSettings lsett)
{
//load metadata
bool bThisTopObject = binRead.ReadBool();
//load refernce types if needed -------------------------
if (bThisTopObject)
this.lsett = new LemmatizerSettings(binRead);
else
this.lsett = lsett;
rlRules = new RuleList(binRead, this.lsett);
bool bCreateLstExamples = binRead.ReadBool();
lstExamples = bCreateLstExamples ? new List<LemmaExample>() : null;
dictExamples = new Dictionary<string, LemmaExample>();
//load dictionary items
int iCount = binRead.ReadInt();
for (int iId = 0; iId < iCount; iId++) {
LemmaRule lrRule = rlRules[binRead.ReadString()];
LemmaExample le = new LemmaExample(binRead, this.lsett, lrRule);
dictExamples.Add(le.Signature, le);
if (bCreateLstExamples) lstExamples.Add(le);
}
}
public ExampleList(Latino.BinarySerializer binRead, LemmatizerSettings lsett)
{
Load(binRead, lsett);
}
public Lemmatizer(StreamReader srIn, string sFormat, LemmatizerSettings lsett) : this(lsett) {
AddMultextFile(srIn, sFormat);
}
private LemmaTreeNode(LemmatizerSettings lsett)
{
this.lsett = lsett;
}
public Lemmatizer(LemmatizerSettings lsett)
{
this.lsett = lsett;
this.elExamples = new ExampleList(lsett);
this.ltnRootNode = null;
}
public Lemmatizer(SerializationInfo info, StreamingContext context) : this()
{
lsett = (LemmatizerSettings)info.GetValue("lsett", typeof(LemmatizerSettings));
elExamples = (ExampleList)info.GetValue("elExamples", typeof(ExampleList));
this.BuildModel();
}
public LemmatizerPrebuilt(LanguagePrebuilt lang, LemmatizerSettings lsett)
: base(lsett)
{
this.lang = lang;
}