//JAVA TO C# CONVERTER WARNING: Method 'throws' clauses are not available in .NET:
//ORIGINAL LINE: private void bracketing(org.maltparser.core.syntaxgraph.PhraseStructure phraseStructure, int start, int end, org.maltparser.core.syntaxgraph.node.PhraseStructureNode parent) throws org.maltparser.core.exception.MaltChainedException
private void bracketing(PhraseStructure phraseStructure, int start, int end, PhraseStructureNode parent)
{
int bracketsdepth = 0;
int startpos = start - 1;
for (int i = start, n = end; i < n; i++)
{
if (input[i] == STARTING_BRACKET && (i == 0 || input[i - 1] != '\\'))
{
if (bracketsdepth == 0)
{
startpos = i;
}
bracketsdepth++;
}
else if (input[i] == CLOSING_BRACKET && (i == 0 || input[i - 1] != '\\'))
{
bracketsdepth--;
if (bracketsdepth == 0)
{
extract(phraseStructure, startpos + 1, i, parent);
}
}
}
}
//JAVA TO C# CONVERTER WARNING: Method 'throws' clauses are not available in .NET:
//ORIGINAL LINE: public void writeSentence(org.maltparser.core.syntaxgraph.TokenStructure syntaxGraph) throws org.maltparser.core.exception.MaltChainedException
public virtual void writeSentence(ITokenStructure syntaxGraph)
{
if (syntaxGraph == null || dataFormatInstance == null)
{
return;
}
if (syntaxGraph.HasTokens())
{
sentenceCount++;
//JAVA TO C# CONVERTER WARNING: The original Java variable was marked 'final':
//ORIGINAL LINE: final org.maltparser.core.syntaxgraph.PhraseStructure phraseStructure = (org.maltparser.core.syntaxgraph.PhraseStructure)syntaxGraph;
PhraseStructure phraseStructure = (PhraseStructure)syntaxGraph;
try
{
sentenceID.Length = 0;
sentenceID.Append(sentencePrefix);
if (phraseStructure.SentenceID != 0)
{
sentenceID.Append(Convert.ToString(phraseStructure.SentenceID));
}
else
{
sentenceID.Append(Convert.ToString(sentenceCount));
}
writer.Write(" <s id=\"");
writer.Write(sentenceID.ToString());
writer.Write("\">\n");
RootID = phraseStructure;
writer.Write(" <graph root=\"");
writer.Write(rootID.ToString());
writer.Write("\" ");
writer.Write("discontinuous=\"");
writer.Write(Convert.ToString(!phraseStructure.Continuous));
writer.Write("\">\n");
writeTerminals(phraseStructure);
if (phraseStructure.NTokenNode() != 1 || rootHandling.Equals(RootHandling.TALBANKEN))
{
writeNonTerminals(phraseStructure);
}
else
{
writer.Write(" <nonterminals/>\n");
}
writer.Write(" </graph>\n");
writer.Write(" </s>\n");
}
catch (IOException e)
{
throw new DataFormatException("The TigerXML writer could not write to file. ", e);
}
}
}
//JAVA TO C# CONVERTER WARNING: Method 'throws' clauses are not available in .NET:
//ORIGINAL LINE: private void calculateIndices(org.maltparser.core.syntaxgraph.PhraseStructure phraseStructure) throws org.maltparser.core.exception.MaltChainedException
private void calculateIndices(PhraseStructure phraseStructure)
{
//JAVA TO C# CONVERTER WARNING: The original Java variable was marked 'final':
//ORIGINAL LINE: final java.util.SortedMap<int,int> heights = new java.util.TreeMap<int,int>();
SortedDictionary <int, int> heights = new SortedDictionary <int, int>();
foreach (int index in phraseStructure.NonTerminalIndices)
{
heights[index] = ((NonTerminalNode)phraseStructure.getNonTerminalNode(index)).Height;
}
bool done = false;
int h = 1;
int ntid = START_ID_OF_NONTERMINALS;
nonTerminalIndexMap.clear();
while (!done)
{
done = true;
foreach (int index in phraseStructure.NonTerminalIndices)
{
if (heights[index] == h)
{
NonTerminalNode nt = (NonTerminalNode)phraseStructure.getNonTerminalNode(index);
nonTerminalIndexMap.put(nt.Index, ntid++);
// nonTerminalIndexMap.put(nt.getIndex(), nt.getIndex()+START_ID_OF_NONTERMINALS-1);
done = false;
}
}
h++;
}
// boolean done = false;
// int h = 1;
//// int ntid = START_ID_OF_NONTERMINALS;
//// nonTerminalIndexMap.clear();
// while (!done) {
// done = true;
// for (int index : phraseStructure.getNonTerminalIndices()) {
// if (heights.get(index) == h) {
// NonTerminalNode nt = (NonTerminalNode)phraseStructure.getNonTerminalNode(index);
//// nonTerminalIndexMap.put(nt.getIndex(), ntid++);
// nonTerminalIndexMap.put(nt.getIndex(), nt.getIndex()+START_ID_OF_NONTERMINALS-1);
// done = false;
// }
// }
// h++;
// }
}
//JAVA TO C# CONVERTER WARNING: Method 'throws' clauses are not available in .NET:
//ORIGINAL LINE: public void writeSentence(org.maltparser.core.syntaxgraph.TokenStructure syntaxGraph) throws org.maltparser.core.exception.MaltChainedException
public virtual void writeSentence(ITokenStructure syntaxGraph)
{
if (syntaxGraph == null || dataFormatInstance == null || !(syntaxGraph is PhraseStructure) || !syntaxGraph.HasTokens())
{
return;
}
PhraseStructure phraseStructure = (PhraseStructure)syntaxGraph;
sentenceCount++;
try
{
writer.Write("#BOS ");
if (phraseStructure.SentenceID != 0)
{
writer.Write(Convert.ToString(phraseStructure.SentenceID));
}
else
{
writer.Write(Convert.ToString(sentenceCount));
}
writer.BaseStream.WriteByte('\n');
if (phraseStructure.hasNonTerminals())
{
calculateIndices(phraseStructure);
writeTerminals(phraseStructure);
writeNonTerminals(phraseStructure);
}
else
{
writeTerminals(phraseStructure);
}
writer.Write("#EOS ");
if (phraseStructure.SentenceID != 0)
{
writer.Write(Convert.ToString(phraseStructure.SentenceID));
}
else
{
writer.Write(Convert.ToString(sentenceCount));
}
writer.BaseStream.WriteByte('\n');
}
catch (IOException e)
{
throw new DataFormatException("Could not write to the output file. ", e);
}
}
//JAVA TO C# CONVERTER WARNING: Method 'throws' clauses are not available in .NET:
//ORIGINAL LINE: public void writeNonTerminals(org.maltparser.core.syntaxgraph.PhraseStructure phraseStructure) throws org.maltparser.core.exception.MaltChainedException
public virtual void writeNonTerminals(PhraseStructure phraseStructure)
{
try
{
SortedDictionary <int, int> heights = new SortedDictionary <int, int>();
foreach (int index in phraseStructure.NonTerminalIndices)
{
heights[index] = ((NonTerminalNode)phraseStructure.getNonTerminalNode(index)).Height;
}
writer.Write(" <nonterminals>\n");
bool done = false;
int h = 1;
while (!done)
{
done = true;
foreach (int index in phraseStructure.NonTerminalIndices)
{
if (heights[index] == h)
{
NonTerminalNode nt = (NonTerminalNode)phraseStructure.getNonTerminalNode(index);
tmpID.Length = 0;
tmpID.Append(sentenceID);
tmpID.Append('_');
tmpID.Append(Convert.ToString(nt.Index + START_ID_OF_NONTERMINALS - 1));
writeNonTerminal(phraseStructure.SymbolTables, nt, tmpID.ToString());
done = false;
}
}
h++;
}
writeNonTerminal(phraseStructure.SymbolTables, (NonTerminalNode)phraseStructure.PhraseStructureRoot, rootID.ToString());
writer.Write(" </nonterminals>\n");
}
catch (IOException e)
{
throw new DataFormatException("The TigerXML writer is not able to write. ", e);
}
}
//JAVA TO C# CONVERTER WARNING: Method 'throws' clauses are not available in .NET:
//ORIGINAL LINE: private void writeTerminals(org.maltparser.core.syntaxgraph.PhraseStructure phraseStructure) throws org.maltparser.core.exception.MaltChainedException
private void writeTerminals(PhraseStructure phraseStructure)
{
try
{
writer.Write(" <terminals>\n");
foreach (int index in phraseStructure.TokenIndices)
{
//JAVA TO C# CONVERTER WARNING: The original Java variable was marked 'final':
//ORIGINAL LINE: final org.maltparser.core.syntaxgraph.node.PhraseStructureNode t = phraseStructure.getTokenNode(index);
PhraseStructureNode t = phraseStructure.GetTokenNode(index);
writer.Write(" <t ");
if (!labeledTerminalID)
{
tmpID.Length = 0;
tmpID.Append(sentenceID);
tmpID.Append('_');
tmpID.Append(Convert.ToString(t.Index));
writer.Write("id=\"");
writer.Write(tmpID.ToString());
writer.Write("\" ");
}
foreach (ColumnDescription column in dataFormatInstance.InputColumnDescriptionSet)
{
writer.Write(column.Name.ToLower());
writer.Write("=\"");
writer.Write(Util.xmlEscape(t.getLabelSymbol(phraseStructure.SymbolTables.getSymbolTable(column.Name))));
writer.Write("\" ");
}
writer.Write("/>\n");
}
writer.Write(" </terminals>\n");
}
catch (IOException e)
{
throw new DataFormatException("The TigerXML writer is not able to write. ", e);
}
}
//JAVA TO C# CONVERTER WARNING: Method 'throws' clauses are not available in .NET:
//ORIGINAL LINE: private void extract(org.maltparser.core.syntaxgraph.PhraseStructure phraseStructure, int begin, int end, org.maltparser.core.syntaxgraph.node.PhraseStructureNode parent) throws org.maltparser.core.exception.MaltChainedException
private void extract(PhraseStructure phraseStructure, int begin, int end, PhraseStructureNode parent)
{
SymbolTableHandler symbolTables = phraseStructure.SymbolTables;
int index = -1;
for (int i = begin; i < end; i++)
{
if (input[i] == STARTING_BRACKET && (i == begin || input[i - 1] != '\\'))
{
index = i;
break;
}
}
if (index == -1)
{
TokenNode t = phraseStructure.AddTokenNode(terminalCounter);
if (t == null)
{
close();
throw new MaltChainedException("Bracket Reader error: could not create a terminal node. ");
}
terminalCounter++;
Edge.Edge e = null;
if (parent != null)
{
e = phraseStructure.addPhraseStructureEdge(parent, (PhraseStructureNode)t);
}
else
{
close();
throw new MaltChainedException("Bracket Reader error: could not find the parent node. ");
}
int start = begin;
IEnumerator <string> inputColumnsIterator = inputColumns.Keys.GetEnumerator();
IEnumerator <string> edgeLabelsColumnsIterator = edgeLabelColumns.Keys.GetEnumerator();
bool noneNode = false;
bool edgeLabels = false;
for (int i = begin; i < end; i++)
{
if (input[i] == EDGELABEL_SEPARATOR || (input[i] == INPUT_SEPARATOR && (i == begin || input[i - 1] != '\\')) || i == end - 1)
{
if (i == begin && input[i] == EDGELABEL_SEPARATOR)
{
noneNode = true;
}
else if (start == begin)
{
if ((noneNode && input[i] != EDGELABEL_SEPARATOR) || !noneNode)
{
//JAVA TO C# CONVERTER TODO TASK: Java iterators are only converted within the context of 'while' and 'for' loops:
if (inputColumnsIterator.hasNext())
{
//JAVA TO C# CONVERTER TODO TASK: Java iterators are only converted within the context of 'while' and 'for' loops:
t.addLabel(symbolTables.getSymbolTable(inputColumns[inputColumnsIterator.next()].Name), decodeString((i == end - 1)?input.substring(start, end - start):input.substring(start, i - start)));
}
start = i + 1;
if (input[i] == EDGELABEL_SEPARATOR)
{
edgeLabels = true;
}
}
}
else if (edgeLabels && e != null)
{
//JAVA TO C# CONVERTER TODO TASK: Java iterators are only converted within the context of 'while' and 'for' loops:
if (edgeLabelsColumnsIterator.hasNext())
{
//JAVA TO C# CONVERTER TODO TASK: Java iterators are only converted within the context of 'while' and 'for' loops:
e.addLabel(symbolTables.getSymbolTable(edgeLabelColumns[edgeLabelsColumnsIterator.next()].Name), (i == end - 1)?input.substring(start, end - start):input.substring(start, i - start));
}
start = i + 1;
if (input[i] == INPUT_SEPARATOR && (i == begin || input[i - 1] != '\\'))
{
edgeLabels = false;
}
}
else if (input[i] == EDGELABEL_SEPARATOR && i != end - 1 && (input[i + 1] != INPUT_SEPARATOR && (i == begin || input[i - 1] != '\\')))
{
}
else
{
//JAVA TO C# CONVERTER TODO TASK: Java iterators are only converted within the context of 'while' and 'for' loops:
if (inputColumnsIterator.hasNext())
{
//JAVA TO C# CONVERTER TODO TASK: Java iterators are only converted within the context of 'while' and 'for' loops:
t.addLabel(symbolTables.getSymbolTable(inputColumns[inputColumnsIterator.next()].Name), (i == end - 1)?input.substring(start, end - start):input.substring(start, i - start));
}
start = i + 1;
}
}
}
}
else
{
PhraseStructureNode nt;
Edge.Edge e = null;
if (parent == null)
{
nt = phraseStructure.PhraseStructureRoot;
}
else
{
nt = phraseStructure.addNonTerminalNode(nonTerminalCounter);
if (nt == null)
{
close();
throw new MaltChainedException("Bracket Reader error: could not create a nonterminal node. ");
}
nonTerminalCounter++;
e = phraseStructure.addPhraseStructureEdge(parent, nt);
}
IEnumerator <string> phraseLabelColumnsIterator = phraseLabelColumns.Keys.GetEnumerator();
IEnumerator <string> edgeLabelsColumnsIterator = edgeLabelColumns.Keys.GetEnumerator();
int newbegin = begin;
int start = begin;
for (int i = begin; i < index; i++)
{
if (input[i] == EDGELABEL_SEPARATOR || i == index - 1)
{
if (start == newbegin)
{
//JAVA TO C# CONVERTER TODO TASK: Java iterators are only converted within the context of 'while' and 'for' loops:
if (phraseLabelColumnsIterator.hasNext())
{
//JAVA TO C# CONVERTER TODO TASK: Java iterators are only converted within the context of 'while' and 'for' loops:
nt.addLabel(symbolTables.getSymbolTable(phraseLabelColumns[phraseLabelColumnsIterator.next()].Name), (i == index - 1)?input.substring(start, index - start):input.substring(start, i - start));
}
start = i + 1;
}
else if (e != null)
{
//JAVA TO C# CONVERTER TODO TASK: Java iterators are only converted within the context of 'while' and 'for' loops:
if (edgeLabelsColumnsIterator.hasNext())
{
//JAVA TO C# CONVERTER TODO TASK: Java iterators are only converted within the context of 'while' and 'for' loops:
e.addLabel(symbolTables.getSymbolTable(edgeLabelColumns[edgeLabelsColumnsIterator.next()].Name), (i == index - 1)?input.substring(start, index - start):input.substring(start, i - start));
}
start = i + 1;
}
}
else if (input[i] == BLANK)
{
start++;
newbegin++;
}
}
bracketing(phraseStructure, index, end, nt);
}
}
//JAVA TO C# CONVERTER WARNING: Method 'throws' clauses are not available in .NET:
//ORIGINAL LINE: public boolean readSentence(org.maltparser.core.syntaxgraph.TokenStructure syntaxGraph) throws org.maltparser.core.exception.MaltChainedException
public virtual bool readSentence(ITokenStructure syntaxGraph)
{
if (syntaxGraph == null || !(syntaxGraph is PhraseStructure))
{
return(false);
}
syntaxGraph.Clear();
//JAVA TO C# CONVERTER WARNING: The original Java variable was marked 'final':
//ORIGINAL LINE: final org.maltparser.core.syntaxgraph.PhraseStructure phraseStructure = (org.maltparser.core.syntaxgraph.PhraseStructure)syntaxGraph;
PhraseStructure phraseStructure = (PhraseStructure)syntaxGraph;
//JAVA TO C# CONVERTER WARNING: The original Java variable was marked 'final':
//ORIGINAL LINE: final org.maltparser.core.symbol.SymbolTableHandler symbolTables = phraseStructure.getSymbolTables();
SymbolTableHandler symbolTables = phraseStructure.SymbolTables;
PhraseStructureNode parent = null;
PhraseStructureNode child = null;
currentHeaderTable = NegraTables.UNDEF;
string line = null;
syntaxGraph.Clear();
nonterminals.Clear();
try
{
while (true)
{
line = reader.ReadLine();
if (ReferenceEquals(line, null))
{
if (syntaxGraph.HasTokens())
{
sentenceCount++;
if (syntaxGraph is MappablePhraseStructureGraph)
{
((MappablePhraseStructureGraph)syntaxGraph).Mapping.updateDependenyGraph(((MappablePhraseStructureGraph)syntaxGraph), ((PhraseStructure)syntaxGraph).PhraseStructureRoot);
}
}
if (cIterations < nIterations)
{
cIterations++;
reopen();
return(true);
}
return(false);
}
else if (line.StartsWith("#EOS", StringComparison.Ordinal))
{
currentTerminalSize = 0;
currentNonTerminalSize = 0;
currentHeaderTable = NegraTables.UNDEF;
if (syntaxGraph is MappablePhraseStructureGraph)
{
((MappablePhraseStructureGraph)syntaxGraph).Mapping.updateDependenyGraph(((MappablePhraseStructureGraph)syntaxGraph), ((PhraseStructure)syntaxGraph).PhraseStructureRoot);
}
return(true);
}
else if (line.StartsWith("#BOS", StringComparison.Ordinal))
{
currentHeaderTable = NegraTables.SENTENCE;
int s = -1, e = -1;
for (int i = 5, n = line.Length; i < n; i++)
{
if (char.IsDigit(line[i]) && s == -1)
{
s = i;
}
if (line[i] == ' ')
{
e = i;
break;
}
}
if (s != e && s != -1 && e != -1)
{
phraseStructure.SentenceID = int.Parse(line.Substring(s, e - s));
}
sentenceCount++;
}
else if (currentHeaderTable == NegraTables.SENTENCE)
{
if (line.Length >= 2 && line[0] == '#' && char.IsDigit(line[1]))
{ // Non-terminal
IEnumerator <ColumnDescription> columns = dataFormatInstance.GetEnumerator();
ColumnDescription column = null;
currentNonTerminalSize++;
char[] lineChars = line.ToCharArray();
int start = 0;
int secedgecounter = 0;
for (int i = 0, n = lineChars.Length; i < n; i++)
{
if (lineChars[i] == '\t' && start == i)
{
start++;
}
else if (lineChars[i] == '\t' || i == n - 1)
{
//JAVA TO C# CONVERTER TODO TASK: Java iterators are only converted within the context of 'while' and 'for' loops:
if (columns.hasNext())
{
//JAVA TO C# CONVERTER TODO TASK: Java iterators are only converted within the context of 'while' and 'for' loops:
column = columns.next();
}
if (column.Position == 0)
{
int index = int.Parse((i == n - 1)?line.Substring(start + 1):line.Substring(start + 1, i - (start + 1)));
child = nonterminals[index];
if (child == null)
{
if (index != 0)
{
child = ((PhraseStructure)syntaxGraph).addNonTerminalNode(index - START_ID_OF_NONTERMINALS + 1);
}
nonterminals[index] = child;
}
}
else if (column.Position == 2 && child != null)
{
syntaxGraph.AddLabel(child, "CAT", (i == n - 1)?line.Substring(start):line.Substring(start, i - start));
}
else if (column.Category == ColumnDescription.PHRASE_STRUCTURE_EDGE_LABEL)
{
edgelabelSymbol.Length = 0;
edgelabelSymbol.Append((i == n - 1)?line.Substring(start):line.Substring(start, i - start));
edgelabelTableName.Length = 0;
edgelabelTableName.Append(column.Name);
}
else if (column.Category == ColumnDescription.PHRASE_STRUCTURE_NODE_LABEL && child != null)
{
int index = int.Parse((i == n - 1)?line.Substring(start):line.Substring(start, i - start));
parent = nonterminals[index];
if (parent == null)
{
if (index == 0)
{
parent = phraseStructure.PhraseStructureRoot;
}
else
{
parent = phraseStructure.addNonTerminalNode(index - START_ID_OF_NONTERMINALS + 1);
}
nonterminals[index] = parent;
}
Edge.Edge e = phraseStructure.addPhraseStructureEdge(parent, child);
syntaxGraph.AddLabel(e, edgelabelTableName.ToString(), edgelabelSymbol.ToString());
}
else if (column.Category == ColumnDescription.SECONDARY_EDGE_LABEL && child != null)
{
if (secedgecounter % 2 == 0)
{
edgelabelSymbol.Length = 0;
edgelabelSymbol.Append((i == n - 1)?line.Substring(start):line.Substring(start, i - start));
secedgecounter++;
}
else
{
int index = int.Parse((i == n - 1)?line.Substring(start):line.Substring(start, i - start));
if (index == 0)
{
parent = phraseStructure.PhraseStructureRoot;
}
else if (index < START_ID_OF_NONTERMINALS)
{
parent = phraseStructure.GetTokenNode(index);
}
else
{
parent = nonterminals[index];
if (parent == null)
{
parent = phraseStructure.addNonTerminalNode(index - START_ID_OF_NONTERMINALS + 1);
nonterminals[index] = parent;
}
}
Edge.Edge e = phraseStructure.addSecondaryEdge(parent, child);
e.addLabel(symbolTables.getSymbolTable(column.Name), edgelabelSymbol.ToString());
secedgecounter++;
}
}
start = i + 1;
}
}
}
else
{ // Terminal
IEnumerator <ColumnDescription> columns = dataFormatInstance.GetEnumerator();
ColumnDescription column = null;
currentTerminalSize++;
child = syntaxGraph.AddTokenNode(currentTerminalSize);
char[] lineChars = line.ToCharArray();
int start = 0;
int secedgecounter = 0;
for (int i = 0, n = lineChars.Length; i < n; i++)
{
if (lineChars[i] == '\t' && start == i)
{
start++;
}
else if (lineChars[i] == '\t' || i == n - 1)
{
//JAVA TO C# CONVERTER TODO TASK: Java iterators are only converted within the context of 'while' and 'for' loops:
if (columns.hasNext())
{
//JAVA TO C# CONVERTER TODO TASK: Java iterators are only converted within the context of 'while' and 'for' loops:
column = columns.next();
}
if (column.Category == ColumnDescription.INPUT && child != null)
{
syntaxGraph.AddLabel(child, column.Name, (i == n - 1)?line.Substring(start):line.Substring(start, i - start));
}
else if (column.Category == ColumnDescription.PHRASE_STRUCTURE_EDGE_LABEL && child != null)
{ // && column.getName().equals("EDGELABEL")) {
edgelabelSymbol.Length = 0;
edgelabelSymbol.Append((i == n - 1)?line.Substring(start):line.Substring(start, i - start));
edgelabelTableName.Length = 0;
edgelabelTableName.Append(column.Name);
}
else if (column.Category == ColumnDescription.PHRASE_STRUCTURE_NODE_LABEL && child != null)
{
int index = int.Parse((i == n - 1)?line.Substring(start):line.Substring(start, i - start));
parent = nonterminals[index];
if (parent == null)
{
if (index == 0)
{
parent = phraseStructure.PhraseStructureRoot;
}
else
{
parent = phraseStructure.addNonTerminalNode(index - START_ID_OF_NONTERMINALS + 1);
}
nonterminals[index] = parent;
}
Edge.Edge e = phraseStructure.addPhraseStructureEdge(parent, child);
syntaxGraph.AddLabel(e, edgelabelTableName.ToString(), edgelabelSymbol.ToString());
}
else if (column.Category == ColumnDescription.SECONDARY_EDGE_LABEL && child != null)
{
if (secedgecounter % 2 == 0)
{
edgelabelSymbol.Length = 0;
edgelabelSymbol.Append((i == n - 1)?line.Substring(start):line.Substring(start, i - start));
secedgecounter++;
}
else
{
int index = int.Parse((i == n - 1)?line.Substring(start):line.Substring(start, i - start));
if (index == 0)
{
parent = phraseStructure.PhraseStructureRoot;
}
else if (index < START_ID_OF_NONTERMINALS)
{
parent = phraseStructure.GetTokenNode(index);
}
else
{
parent = nonterminals[index];
if (parent == null)
{
parent = phraseStructure.addNonTerminalNode(index - START_ID_OF_NONTERMINALS + 1);
nonterminals[index] = parent;
}
}
Edge.Edge e = phraseStructure.addSecondaryEdge(parent, child);
e.addLabel(symbolTables.getSymbolTable(column.Name), edgelabelSymbol.ToString());
secedgecounter++;
}
}
start = i + 1;
}
}
}
}
else if (line.StartsWith("%%", StringComparison.Ordinal))
{ // comment skip
}
else if (line.StartsWith("#FORMAT", StringComparison.Ordinal))
{
// int index = line.indexOf(' ');
// if (index > -1) {
// try {
// formatVersion = Integer.parseInt(line.substring(index+1));
// } catch (NumberFormatException e) {
//
// }
// }
}
else if (line.StartsWith("#BOT", StringComparison.Ordinal))
{
// int index = line.indexOf(' ');
// if (index > -1) {
// if (line.substring(index+1).equals("ORIGIN")) {
// currentHeaderTable = NegraTables.ORIGIN;
// } else if (line.substring(index+1).equals("EDITOR")) {
// currentHeaderTable = NegraTables.EDITOR;
// } else if (line.substring(index+1).equals("WORDTAG")) {
// currentHeaderTable = NegraTables.WORDTAG;
// } else if (line.substring(index+1).equals("MORPHTAG")) {
// currentHeaderTable = NegraTables.MORPHTAG;
// } else if (line.substring(index+1).equals("NODETAG")) {
// currentHeaderTable = NegraTables.NODETAG;
// } else if (line.substring(index+1).equals("EDGETAG")) {
// currentHeaderTable = NegraTables.EDGETAG;
// } else if (line.substring(index+1).equals("SECEDGETAG")) {
// currentHeaderTable = NegraTables.SECEDGETAG;
// } else {
// currentHeaderTable = NegraTables.UNDEF;
// }
// }
}
else if (line.StartsWith("#EOT", StringComparison.Ordinal))
{
currentHeaderTable = NegraTables.UNDEF;
}
}
}
catch (IOException e)
{
throw new DataFormatException("Error when reading from the input file. ", e);
}
}
//JAVA TO C# CONVERTER WARNING: Method 'throws' clauses are not available in .NET:
//ORIGINAL LINE: public boolean readSentence(org.maltparser.core.syntaxgraph.TokenStructure syntaxGraph) throws org.maltparser.core.exception.MaltChainedException
public virtual bool readSentence(ITokenStructure syntaxGraph)
{
if (syntaxGraph == null || !(syntaxGraph is PhraseStructure))
{
return(false);
}
syntaxGraph.Clear();
//JAVA TO C# CONVERTER WARNING: The original Java variable was marked 'final':
//ORIGINAL LINE: final org.maltparser.core.syntaxgraph.PhraseStructure phraseStructure = (org.maltparser.core.syntaxgraph.PhraseStructure)syntaxGraph;
PhraseStructure phraseStructure = (PhraseStructure)syntaxGraph;
PhraseStructureNode parent = null;
PhraseStructureNode child = null;
// if (header == null) {
// header = new TigerXMLHeader(syntaxGraph.getSymbolTables());
// }
try
{
while (true)
{
int @event = reader.next();
if (@event == XMLStreamConstants.START_ELEMENT)
{
if (reader.LocalName.length() == 0)
{
continue;
}
if (reader.LocalName.charAt(0) == 'e')
{
// e -> edge, edgelabel
if (reader.LocalName.length() == 4)
{ //edge
int childid = -1;
int indexSep = reader.getAttributeValue(null, "idref").IndexOf('_');
try
{
if (indexSep != -1)
{
childid = int.Parse(reader.getAttributeValue(null, "idref").substring(indexSep + 1));
}
else
{
childid = int.Parse(reader.getAttributeValue(null, "idref"));
}
if (childid == -1)
{
throw new SyntaxGraphException("The tiger reader couldn't recognize the idref attribute '" + reader.getAttributeValue(null, "idref") + "' of the edge element. ");
}
}
catch (System.FormatException)
{
throw new SyntaxGraphException("The tiger reader couldn't recognize the idref attribute '" + reader.getAttributeValue(null, "idref") + "' of the edge element. ");
}
if (childid < START_ID_OF_NONTERMINALS)
{
child = phraseStructure.GetTokenNode(childid);
}
else
{
child = phraseStructure.getNonTerminalNode(childid - START_ID_OF_NONTERMINALS + 1);
}
Edge.Edge e = phraseStructure.addPhraseStructureEdge(parent, child);
SortedDictionary <string, SymbolTable> inputTables = dataFormatInstance.getPhraseStructureEdgeLabelSymbolTables(phraseStructure.SymbolTables);
foreach (string name in inputTables.Keys)
{
e.addLabel(inputTables[name], reader.getAttributeValue(null, name.ToLower()));
}
}
else if (reader.LocalName.Equals("edgelabel"))
{ // edgelabel
// domain = Domain.EL;
}
}
else if (reader.LocalName.charAt(0) == 'n')
{
// n -> nt, nonterminals, name
if (reader.LocalName.length() == 2)
{ // nt
//JAVA TO C# CONVERTER WARNING: The original Java variable was marked 'final':
//ORIGINAL LINE: final String id = reader.getAttributeValue(null, "id");
string id = reader.getAttributeValue(null, "id");
if (graphRootID.Length == id.Length && graphRootID.ToString().Equals(id))
{
parent = phraseStructure.PhraseStructureRoot;
}
else
{
int index = id.IndexOf('_');
if (index != -1)
{
parent = phraseStructure.addNonTerminalNode(int.Parse(id.Substring(index + 1)) - START_ID_OF_NONTERMINALS + 1);
}
}
SortedDictionary <string, SymbolTable> inputTables = dataFormatInstance.getPhraseStructureNodeLabelSymbolTables(phraseStructure.SymbolTables);
foreach (string name in inputTables.Keys)
{
parent.addLabel(inputTables[name], reader.getAttributeValue(null, name.ToLower()));
}
}
else if (reader.LocalName.Equals("name"))
{ // name
// elementContent.setLength(0);
// collectChar = true;
}
}
else if (reader.LocalName.charAt(0) == 't')
{
// t -> t, terminals
if (reader.LocalName.length() == 1)
{ // t
SortedDictionary <string, SymbolTable> inputTables = dataFormatInstance.getInputSymbolTables(phraseStructure.SymbolTables);
child = syntaxGraph.AddTokenNode();
foreach (string name in inputTables.Keys)
{
child.addLabel(inputTables[name], reader.getAttributeValue(null, name.ToLower()));
}
}
}
else if (reader.LocalName.charAt(0) == 's')
{
// s -> subcorpus, secedge, s, secedgelabel
if (reader.LocalName.length() == 1)
{ // s
string id = reader.getAttributeValue(null, "id");
bool indexable = false;
int index = -1;
if (!ReferenceEquals(id, null) && id.Length > 0)
{
for (int i = 0, n = id.Length; i < n; i++)
{
if (char.IsDigit(id[i]))
{
if (index == -1)
{
index = i;
}
indexable = true;
}
}
}
if (indexable)
{
phraseStructure.SentenceID = int.Parse(id.Substring(index));
}
else
{
phraseStructure.SentenceID = sentenceCount + 1;
}
}
}
else if (reader.LocalName.charAt(0) == 'v')
{
// v -> variable, value
// if (reader.getLocalName().equals("value")) {
// valueName.setLength(0);
// valueName.append(reader.getAttributeValue(null, "name"));
// elementContent.setLength(0);
// collectChar = true;
// }
}
else
{
// a -> annotation, author
// b -> body
// c -> corpus
// d -> date, description,
// f -> feature, format
// g -> graph
// h -> head, history
// m -> matches, match
if (reader.LocalName.Equals("graph"))
{
graphRootID.Length = 0;
graphRootID.Append(reader.getAttributeValue(null, "root"));
}
else if (reader.LocalName.Equals("corpus"))
{
// header.setCorpusID(reader.getAttributeValue(null, "id"));
// header.setCorpusID(reader.getAttributeValue(null, "version"));
}
else if (reader.LocalName.Equals("feature"))
{
// if (header != null) {
// currentFeatureName.setLength(0);
// currentFeatureName.append(reader.getAttributeValue(null, "name"));
// header.addFeature(reader.getAttributeValue(null, "name"), reader.getAttributeValue(null, "domain"));
// }
// domain = Domain.valueOf(reader.getAttributeValue(null, "domain"));
}
else if (reader.LocalName.Equals("secedgelabel"))
{
// domain = Domain.SEL;
}
else if (reader.LocalName.Equals("author"))
{
// elementContent.setLength(0);
// collectChar = true;
}
else if (reader.LocalName.Equals("date"))
{
// elementContent.setLength(0);
// collectChar = true;
}
else if (reader.LocalName.Equals("description"))
{
// elementContent.setLength(0);
// collectChar = true;
}
else if (reader.LocalName.Equals("format"))
{
// elementContent.setLength(0);
// collectChar = true;
}
else if (reader.LocalName.Equals("history"))
{
// elementContent.setLength(0);
// collectChar = true;
}
}
}
else if (@event == XMLStreamConstants.END_ELEMENT)
{
if (reader.LocalName.length() == 0)
{
continue;
}
if (reader.LocalName.charAt(0) == 'e')
{
// e -> edge, edgelabel
}
else if (reader.LocalName.charAt(0) == 'n')
{
// n -> nt, nonterminals, name
if (reader.LocalName.Equals("nt"))
{
ntid.Length = 0;
}
else if (reader.LocalName.Equals("nonterminals"))
{
if (phraseStructure.NTokenNode() == 1 && phraseStructure.nNonTerminals() == 0 && ((NonTerminalNode)phraseStructure.PhraseStructureRoot).nChildren() == 0)
{
Edge.Edge e = phraseStructure.addPhraseStructureEdge(phraseStructure.PhraseStructureRoot, phraseStructure.GetTokenNode(1));
SortedDictionary <string, SymbolTable> inputTables = dataFormatInstance.getPhraseStructureEdgeLabelSymbolTables(phraseStructure.SymbolTables);
foreach (string name in inputTables.Keys)
{
e.addLabel(inputTables[name], "--");
}
}
}
// else if (reader.getLocalName().equals("name")) {
// if (header != null) {
// header.setMetaName(elementContent.toString());
// }
// collectChar = false;
// }
}
else if (reader.LocalName.charAt(0) == 't')
{
// t -> t, terminals
}
else if (reader.LocalName.charAt(0) == 's')
{
// s -> subcorpus, secedge, s, secedgelabel
if (reader.LocalName.Equals("s"))
{
if (syntaxGraph.HasTokens())
{
sentenceCount++;
}
if (syntaxGraph is MappablePhraseStructureGraph)
{
((MappablePhraseStructureGraph)syntaxGraph).Mapping.updateDependenyGraph(((MappablePhraseStructureGraph)syntaxGraph), ((PhraseStructure)syntaxGraph).PhraseStructureRoot);
}
return(true);
}
}
else if (reader.LocalName.charAt(0) == 'v')
{
// v -> variable, value
// if (reader.getLocalName().equals("value")) {
// if (header != null) {
// if (domain == Domain.T || domain == Domain.NT || domain == Domain.FREC) {
// header.addFeatureValue(currentFeatureName.toString(), valueName.toString(), elementContent.toString());
// } else if (domain == Domain.EL) {
// header.addEdgeLabelValue(valueName.toString(), elementContent.toString());
// } else if (domain == Domain.SEL) {
// header.addSecEdgeLabelValue(valueName.toString(), elementContent.toString());
// }
// }
// collectChar = false;
// }
}
else
{
// a -> annotation, author
// b -> body
// c -> corpus
// d -> date, description,
// f -> feature, format
// g -> graph
// h -> head, history
// m -> matches, match
if (reader.LocalName.Equals("body"))
{
//sentence = dataStructures.getSentence();
//phraseTree = dataStructures.getInPhraseTree();
//sentence.clear();
//phraseTree.clear();
//dataStructures.setLastProcessObject(true);
}
else if (reader.LocalName.Equals("author"))
{
// if (header != null) {
// header.setMetaAuthor(elementContent.toString());
// }
// collectChar = false;
}
else if (reader.LocalName.Equals("date"))
{
// if (header != null) {
// header.setMetaInDate(elementContent.toString());
// }
// collectChar = false;
}
else if (reader.LocalName.Equals("description"))
{
// if (header != null) {
// header.setMetaDescription(elementContent.toString());
// }
// collectChar = false;
}
else if (reader.LocalName.Equals("format"))
{
// if (header != null) {
// header.setMetaFormat(elementContent.toString());
// }
// collectChar = false;
}
else if (reader.LocalName.Equals("history"))
{
// if (header != null) {
// header.setMetaHistory(elementContent.toString());
// }
// collectChar = false;
} /* else if (reader.getLocalName().equals("annotation")) {
* if (header != null) {
* System.out.println(header.toTigerXML());
* }
* collectChar = false;
* } */
}
}
else if (@event == XMLStreamConstants.END_DOCUMENT)
{
if (syntaxGraph.HasTokens())
{
sentenceCount++;
}
if (cIterations < nIterations)
{
cIterations++;
reopen();
return(true);
}
return(false);
}
else if (@event == XMLStreamConstants.CHARACTERS)
{
// if (collectChar) {
// char[] ch = reader.getTextCharacters();
// final int size = reader.getTextStart()+reader.getTextLength();
// for (int i = reader.getTextStart(); i < size; i++) {
// elementContent.append(ch[i]);
// }
// }
}
}
}
catch (XMLStreamException e)
{
throw new DataFormatException("", e);
}
}
//JAVA TO C# CONVERTER WARNING: Method 'throws' clauses are not available in .NET:
//ORIGINAL LINE: private void writeNonTerminals(org.maltparser.core.syntaxgraph.PhraseStructure phraseStructure) throws org.maltparser.core.exception.MaltChainedException
private void writeNonTerminals(PhraseStructure phraseStructure)
{
//JAVA TO C# CONVERTER WARNING: The original Java variable was marked 'final':
//ORIGINAL LINE: final org.maltparser.core.symbol.SymbolTableHandler symbolTables = phraseStructure.getSymbolTables();
SymbolTableHandler symbolTables = phraseStructure.SymbolTables;
foreach (int index in nonTerminalIndexMap.Keys)
{
// for (int index : phraseStructure.getNonTerminalIndices()) {
NonTerminalNode nonTerminal = (NonTerminalNode)phraseStructure.getNonTerminalNode(index);
if (nonTerminal == null || nonTerminal.Root)
{
return;
}
try
{
writer.BaseStream.WriteByte('#');
// writer.write(Integer.toString(index+START_ID_OF_NONTERMINALS-1));
writer.Write(Convert.ToString(nonTerminalIndexMap.get(index)));
writer.Write("\t\t\t--\t\t\t");
if (nonTerminal.hasLabel(symbolTables.getSymbolTable("CAT")))
{
writer.BaseStream.WriteByte(nonTerminal.getLabelSymbol(symbolTables.getSymbolTable("CAT")));
}
else
{
writer.Write("--");
}
writer.Write("\t--\t\t");
if (nonTerminal.hasParentEdgeLabel(symbolTables.getSymbolTable("LABEL")))
{
writer.Write(nonTerminal.getParentEdgeLabelSymbol(symbolTables.getSymbolTable("LABEL")));
}
else
{
writer.Write("--");
}
writer.BaseStream.WriteByte('\t');
if (nonTerminal.Parent == null || nonTerminal.Parent.Root)
{
writer.BaseStream.WriteByte('0');
}
else
{
// writer.write(Integer.toString(nonTerminal.getParent().getIndex()+START_ID_OF_NONTERMINALS-1));
writer.Write(Convert.ToString(nonTerminalIndexMap.get(nonTerminal.Parent.Index)));
}
foreach (Edge.Edge e in nonTerminal.IncomingSecondaryEdges)
{
if (e.hasLabel(symbolTables.getSymbolTable("SECEDGELABEL")))
{
writer.BaseStream.WriteByte('\t');
writer.Write(e.getLabelSymbol(symbolTables.getSymbolTable("SECEDGELABEL")));
writer.BaseStream.WriteByte('\t');
if (e.Source is NonTerminalNode)
{
// writer.write(Integer.toString(e.getSource().getIndex()+START_ID_OF_NONTERMINALS-1));
writer.Write(Convert.ToString(nonTerminalIndexMap.get(e.Source.Index)));
}
else
{
writer.Write(Convert.ToString(e.Source.Index));
}
}
}
writer.Write("\n");
}
catch (IOException e)
{
throw new DataFormatException("The Negra writer is not able to write the non-terminals. ", e);
}
}
}
//JAVA TO C# CONVERTER WARNING: Method 'throws' clauses are not available in .NET:
//ORIGINAL LINE: private void writeTerminals(org.maltparser.core.syntaxgraph.PhraseStructure phraseStructure) throws org.maltparser.core.exception.MaltChainedException
private void writeTerminals(PhraseStructure phraseStructure)
{
try
{
//JAVA TO C# CONVERTER WARNING: The original Java variable was marked 'final':
//ORIGINAL LINE: final org.maltparser.core.symbol.SymbolTableHandler symbolTables = phraseStructure.getSymbolTables();
SymbolTableHandler symbolTables = phraseStructure.SymbolTables;
foreach (int index in phraseStructure.TokenIndices)
{
//JAVA TO C# CONVERTER WARNING: The original Java variable was marked 'final':
//ORIGINAL LINE: final org.maltparser.core.syntaxgraph.node.PhraseStructureNode terminal = phraseStructure.getTokenNode(index);
PhraseStructureNode terminal = phraseStructure.GetTokenNode(index);
//JAVA TO C# CONVERTER WARNING: The original Java variable was marked 'final':
//ORIGINAL LINE: final java.util.Iterator<org.maltparser.core.io.dataformat.ColumnDescription> columns = dataFormatInstance.iterator();
IEnumerator <ColumnDescription> columns = dataFormatInstance.GetEnumerator();
ColumnDescription column = null;
int ti = 1;
while (columns.MoveNext())
{
column = columns.Current;
if (column.Category == ColumnDescription.INPUT)
{
SymbolTable table = symbolTables.getSymbolTable(column.Name);
writer.Write(terminal.getLabelSymbol(table));
int nTabs = 1;
if (ti == 1 || ti == 2)
{
nTabs = 3 - (terminal.getLabelSymbol(table).Length / 8);
}
else if (ti == 3)
{
nTabs = 1;
}
else if (ti == 4)
{
nTabs = 2 - (terminal.getLabelSymbol(table).Length / 8);
}
if (nTabs < 1)
{
nTabs = 1;
}
for (int j = 0; j < nTabs; j++)
{
writer.BaseStream.WriteByte('\t');
}
ti++;
}
else if (column.Category == ColumnDescription.PHRASE_STRUCTURE_EDGE_LABEL)
{
SymbolTable table = symbolTables.getSymbolTable(column.Name);
if (terminal.Parent != null && terminal.hasParentEdgeLabel(table))
{
writer.Write(terminal.getParentEdgeLabelSymbol(table));
writer.BaseStream.WriteByte('\t');
}
else
{
writer.Write("--\t");
}
}
else if (column.Category == ColumnDescription.PHRASE_STRUCTURE_NODE_LABEL)
{
if (terminal.Parent == null || terminal.Parent == phraseStructure.PhraseStructureRoot)
{
writer.BaseStream.WriteByte('0');
}
else
{
writer.Write(Convert.ToString(nonTerminalIndexMap.get(terminal.Parent.Index)));
// writer.write(Integer.toString(terminal.getParent().getIndex()+START_ID_OF_NONTERMINALS-1));
}
}
}
SymbolTable table = symbolTables.getSymbolTable(column.Name);
foreach (Edge.Edge e in terminal.IncomingSecondaryEdges)
{
if (e.hasLabel(table))
{
writer.BaseStream.WriteByte('\t');
writer.Write(e.getLabelSymbol(table));
writer.BaseStream.WriteByte('\t');
if (e.Source is NonTerminalNode)
{
writer.Write(Convert.ToString(nonTerminalIndexMap.get(e.Source.Index)));
// writer.write(Integer.toString(e.getSource().getIndex()+START_ID_OF_NONTERMINALS-1));
}
else
{
writer.Write(Convert.ToString(e.Source.Index));
}
}
}
writer.Write("\n");
}
}
catch (IOException e)
{
throw new DataFormatException("The Negra writer is not able to write. ", e);
}
}
