/// <summary>
/// Given a path to a file containing a list of SsurgeonPatterns, returns
/// TODO: deal with resources
/// </summary>
/// <exception cref="System.Exception"/>
public virtual IList <SsurgeonPattern> ReadFromFile(File file)
{
IList <SsurgeonPattern> retList = new List <SsurgeonPattern>();
IDocument doc = DocumentBuilderFactory.NewInstance().NewDocumentBuilder().Parse(file);
INodeList patternNodes = doc.GetElementsByTagName(SsurgeonPattern.SsurgeonElemTag);
for (int i = 0; i < patternNodes.GetLength(); i++)
{
INode node = patternNodes.Item(i);
if (node.GetNodeType() == NodeConstants.ElementNode)
{
IElement elt = (IElement)node;
SsurgeonPattern pattern = SsurgeonPatternFromXML(elt);
retList.Add(pattern);
}
}
INodeList resourceNodes = doc.GetElementsByTagName(SsurgeonPattern.ResourceTag);
for (int i_1 = 0; i_1 < resourceNodes.GetLength(); i_1++)
{
INode node = patternNodes.Item(i_1);
if (node.GetNodeType() == NodeConstants.ElementNode)
{
IElement resourceElt = (IElement)node;
SsurgeonWordlist wlRsrc = new SsurgeonWordlist(resourceElt);
AddResource(wlRsrc);
}
}
return(retList);
}
private static void GetMatchingNodes(INode node, string[] nodePath, int cur, IList <INode> res)
{
if (cur < 0 || cur >= nodePath.Length)
{
return;
}
bool last = (cur == nodePath.Length - 1);
string name = nodePath[cur];
if (node.HasChildNodes())
{
INodeList children = node.GetChildNodes();
for (int i = 0; i < children.GetLength(); i++)
{
INode c = children.Item(i);
if (name.Equals(c.GetNodeName()))
{
if (last)
{
res.Add(c);
}
else
{
GetMatchingNodes(c, nodePath, cur + 1, res);
}
}
}
}
}
/// <summary>
/// Given the root Element for a SemgrexPattern (SSURGEON_ELEM_TAG), converts
/// it into its corresponding SemgrexPattern object.
/// </summary>
/// <exception cref="System.Exception"/>
public static SsurgeonPattern SsurgeonPatternFromXML(IElement elt)
{
string uid = GetTagText(elt, SsurgeonPattern.UidElemTag);
string notes = GetTagText(elt, SsurgeonPattern.NotesElemTag);
string semgrexString = GetTagText(elt, SsurgeonPattern.SemgrexElemTag);
SemgrexPattern semgrexPattern = SemgrexPattern.Compile(semgrexString);
SsurgeonPattern retPattern = new SsurgeonPattern(uid, semgrexPattern);
retPattern.SetNotes(notes);
INodeList editNodes = elt.GetElementsByTagName(SsurgeonPattern.EditListElemTag);
for (int i = 0; i < editNodes.GetLength(); i++)
{
INode node = editNodes.Item(i);
if (node.GetNodeType() == NodeConstants.ElementNode)
{
IElement editElt = (IElement)node;
string editVal = GetEltText(editElt);
retPattern.AddEdit(Edu.Stanford.Nlp.Semgraph.Semgrex.Ssurgeon.Ssurgeon.ParseEditLine(editVal));
}
}
// If predicate available, parse
IElement predElt = GetFirstTag(elt, SsurgeonPattern.PredicateTag);
if (predElt != null)
{
ISsurgPred pred = AssemblePredFromXML(GetFirstChildElement(predElt));
retPattern.SetPredicate(pred);
}
return(retPattern);
}
private static void GetMatchingNodes(INode node, Pattern[] nodePath, int cur, IList <INode> res)
{
if (cur < 0 || cur >= nodePath.Length)
{
return;
}
bool last = (cur == nodePath.Length - 1);
Pattern pattern = nodePath[cur];
INodeList children = node.GetChildNodes();
for (int i = 0; i < children.GetLength(); i++)
{
INode c = children.Item(i);
if (pattern.Matcher(c.GetNodeName()).Matches())
{
if (last)
{
res.Add(c);
}
else
{
GetMatchingNodes(c, nodePath, cur + 1, res);
}
}
}
}
public static void RemoveChildren(INode e)
{
INodeList list = e.GetChildNodes();
for (int i = 0; i < list.GetLength(); i++)
{
INode n = list.Item(i);
e.RemoveChild(n);
}
}
/// <summary>Reconstructs the resource from the XML file</summary>
public SsurgeonWordlist(IElement rootElt)
{
id = rootElt.GetAttribute("id");
INodeList wordEltNL = rootElt.GetElementsByTagName(WordElt);
for (int i = 0; i < wordEltNL.GetLength(); i++)
{
INode node = wordEltNL.Item(i);
if (node.GetNodeType() == NodeConstants.ElementNode)
{
string word = Edu.Stanford.Nlp.Semgraph.Semgrex.Ssurgeon.Ssurgeon.GetEltText((IElement)node);
words.Add(word);
}
}
}
/// <summary>Searches for all immediate children with the given name</summary>
protected internal static IList <INode> GetChildrenByName(INode node, string name)
{
IList <INode> matches = new List <INode>();
INodeList children = node.GetChildNodes();
// search children
for (int i = 0; i < children.GetLength(); i++)
{
INode child = children.Item(i);
if (child.GetNodeName().Equals(name))
{
matches.Add(child);
}
}
return(matches);
}
public static string GetJustText(INode text)
{
StringBuilder sb = new StringBuilder();
INodeList textElems = text.GetChildNodes();
for (int i = 0; i < textElems.GetLength(); i++)
{
INode child = textElems.Item(i);
string str = child.GetTextContent();
//replace single occurrence of \n with " ", double occurrences with a single one.
str = str.ReplaceAll("\n(?!\n)", " ");
str = str.ReplaceAll("_", string.Empty);
//bug fix for sentence splitting
sb.Append(str + " ");
}
return(sb.ToString());
}
/// <summary>
/// For a given Element, treats the first child as a text element
/// and returns its value.
/// </summary>
public static string GetEltText(IElement element)
{
try
{
INodeList childNodeList = element.GetChildNodes();
if (childNodeList.GetLength() == 0)
{
return(string.Empty);
}
return(childNodeList.Item(0).GetNodeValue());
}
catch (Exception e)
{
log.Warning("Exception e=" + e.Message + " thrown calling getEltText on element=" + element);
}
return(string.Empty);
}
/// <summary>Returns the first child whose node type is Element under the given Element.</summary>
private static IElement GetFirstChildElement(IElement element)
{
try
{
INodeList nodeList = element.GetChildNodes();
for (int i = 0; i < nodeList.GetLength(); i++)
{
INode node = nodeList.Item(i);
if (node.GetNodeType() == NodeConstants.ElementNode)
{
return((IElement)node);
}
}
}
catch (Exception e)
{
log.Warning("Error getting first child Element for element=" + element + ", exception=" + e);
}
return(null);
}
//Silently ignore
public virtual Tree ReadTree()
{
Tree t = null;
while (t == null && sentences != null && sentIdx < sentences.GetLength())
{
INode sentRoot = sentences.Item(sentIdx++);
t = GetTreeFromXML(sentRoot);
if (t != null)
{
t = treeNormalizer.NormalizeWholeTree(t, treeFactory);
if (t.Label() is CoreLabel)
{
string ftbId = ((IElement)sentRoot).GetAttribute(AttrNumber);
((CoreLabel)t.Label()).Set(typeof(CoreAnnotations.SentenceIDAnnotation), ftbId);
}
}
}
return(t);
}
//Silently ignore
public virtual Tree ReadTree()
{
Tree t = null;
while (t == null && sentences != null && sentIdx < sentences.GetLength())
{
int thisSentenceId = sentIdx++;
INode sentRoot = sentences.Item(thisSentenceId);
t = GetTreeFromXML(sentRoot);
if (t != null)
{
t = treeNormalizer.NormalizeWholeTree(t, treeFactory);
if (t.Label() is CoreLabel)
{
((CoreLabel)t.Label()).Set(typeof(CoreAnnotations.SentenceIDAnnotation), int.ToString(thisSentenceId));
}
}
}
return(t);
}
/// <summary>Searches (recursively) for the first child that has the given name</summary>
protected internal static INode GetChildByName(INode node, string name)
{
INodeList children = node.GetChildNodes();
// this node matches
if (node.GetNodeName().Equals(name))
{
return(node);
}
// search children
for (int i = 0; i < children.GetLength(); i++)
{
INode found = GetChildByName(children.Item(i), name);
if (found != null)
{
return(found);
}
}
// failed
return(null);
}
public static IList <Person> ReadXMLCharacterList(IDocument doc)
{
IList <Person> personList = new List <Person>();
INodeList characters = doc.GetDocumentElement().GetElementsByTagName("characters").Item(0).GetChildNodes();
for (int i = 0; i < characters.GetLength(); i++)
{
INode child = characters.Item(i);
if (child.GetNodeName().Equals("character"))
{
string name = child.GetAttributes().GetNamedItem("name").GetNodeValue();
char[] cName = name.ToCharArray();
cName[0] = char.ToUpperCase(cName[0]);
name = new string(cName);
IList <string> aliases = Arrays.AsList(child.GetAttributes().GetNamedItem("aliases").GetNodeValue().Split(";"));
string gender = (child.GetAttributes().GetNamedItem("gender") == null) ? string.Empty : child.GetAttributes().GetNamedItem("gender").GetNodeValue();
personList.Add(new Person(child.GetAttributes().GetNamedItem("name").GetNodeValue(), gender, aliases));
}
}
return(personList);
}
/// <summary>Returns all of the Element typed children from the given element.</summary>
/// <remarks>
/// Returns all of the Element typed children from the given element. Note: disregards
/// other node types.
/// </remarks>
private static IList <IElement> GetChildElements(IElement element)
{
LinkedList <IElement> childElements = new LinkedList <IElement>();
try
{
INodeList nodeList = element.GetChildNodes();
for (int i = 0; i < nodeList.GetLength(); i++)
{
INode node = nodeList.Item(i);
if (node.GetNodeType() == NodeConstants.ElementNode)
{
childElements.Add((IElement)node);
}
}
}
catch (Exception e)
{
log.Warning("Exception thrown getting all children for element=" + element + ", e=" + e);
}
return(childElements);
}
/// <summary>Searches for children that have the given name and attribute</summary>
protected internal static INode GetChildByNameAndAttribute(INode node, string name, string attributeName, string attributeValue)
{
INodeList children = node.GetChildNodes();
INamedNodeMap attribs = node.GetAttributes();
INode attribute = null;
// this node matches
if (node.GetNodeName().Equals(name) && attribs != null && (attribute = attribs.GetNamedItem(attributeName)) != null && attribute.GetNodeValue().Equals(attributeValue))
{
return(node);
}
// search children
for (int i = 0; i < children.GetLength(); i++)
{
INode found = GetChildByAttribute(children.Item(i), attributeName, attributeValue);
if (found != null)
{
return(found);
}
}
// failed
return(null);
}
/// <summary>For the given element, finds the first child Element with the given tag.</summary>
private static IElement GetFirstTag(IElement element, string tag)
{
try
{
INodeList nodeList = element.GetElementsByTagName(tag);
if (nodeList.GetLength() == 0)
{
return(null);
}
for (int i = 0; i < nodeList.GetLength(); i++)
{
INode node = nodeList.Item(i);
if (node.GetNodeType() == NodeConstants.ElementNode)
{
return((IElement)node);
}
}
}
catch (Exception)
{
log.Warning("Error getting first tag " + tag + " under element=" + element);
}
return(null);
}
private bool Load(InputStream stream)
{
DocumentBuilder parser = XMLUtils.GetXmlParser();
if (parser == null)
{
return(false);
}
try
{
IDocument xmlDocument = parser.Parse(stream);
IElement root = xmlDocument.GetDocumentElement();
INodeList sentences = root.GetElementsByTagName(Sentence);
for (int i = 0; i < sentences.GetLength(); i++)
{
IElement sentence = (IElement)sentences.Item(i);
Lattice lattice = new Lattice();
//Create the node map
ISortedSet <int> nodes = new TreeSet <int>();
INodeList xmlNodes = sentence.GetElementsByTagName(Node);
for (int nodeIdx = 0; nodeIdx < xmlNodes.GetLength(); nodeIdx++)
{
IElement xmlNode = (IElement)xmlNodes.Item(nodeIdx);
int nodeName = System.Convert.ToInt32(xmlNode.GetAttribute(NodeId));
nodes.Add(nodeName);
}
IDictionary <int, int> nodeMap = Generics.NewHashMap();
int realNodeIdx = 0;
int lastBoundaryNode = -1;
foreach (int nodeName_1 in nodes)
{
if (lastBoundaryNode == -1)
{
System.Diagnostics.Debug.Assert(nodeName_1 % NodeOffset == 0);
lastBoundaryNode = realNodeIdx;
}
else
{
if (nodeName_1 % NodeOffset == 0)
{
ParserConstraint c = new ParserConstraint(lastBoundaryNode, realNodeIdx, ".*");
lattice.AddConstraint(c);
}
}
nodeMap[nodeName_1] = realNodeIdx;
realNodeIdx++;
}
//Read the edges
INodeList xmlEdges = sentence.GetElementsByTagName(Edge);
for (int edgeIdx = 0; edgeIdx < xmlEdges.GetLength(); edgeIdx++)
{
IElement xmlEdge = (IElement)xmlEdges.Item(edgeIdx);
string segment = xmlEdge.GetAttribute(Segment);
double weight = double.Parse(xmlEdge.GetAttribute(Weight));
//Input weights should be log scale
int from = System.Convert.ToInt32(xmlEdge.GetAttribute(FromNode));
int normFrom = nodeMap[from];
int to = System.Convert.ToInt32(xmlEdge.GetAttribute(ToNode));
int normTo = nodeMap[to];
LatticeEdge e = new LatticeEdge(segment, weight, normFrom, normTo);
// Set attributes below here
INodeList xmlAttrs = xmlEdge.GetElementsByTagName(EAttrNode);
for (int attrIdx = 0; attrIdx < xmlAttrs.GetLength(); attrIdx++)
{
IElement xmlAttr = (IElement)xmlAttrs.Item(attrIdx);
string key = xmlAttr.GetAttribute(EAttr);
string value = xmlAttr.GetAttribute(EAttrVal);
e.SetAttr(key, value);
}
lattice.AddEdge(e);
}
//Configure for parsing in ExhaustivePCFG parser
lattice.AddBoundary();
lattices.Add(lattice);
}
}
catch (IOException e)
{
System.Console.Error.Printf("%s: Error reading XML from input stream.%n", this.GetType().FullName);
Sharpen.Runtime.PrintStackTrace(e);
return(false);
}
catch (SAXException e)
{
Sharpen.Runtime.PrintStackTrace(e);
return(false);
}
return(true);
}
private static IList <ICoreMap> ToTimexCoreMaps(IElement docElem, ICoreMap originalDocument)
{
//--Collect Token Offsets
IDictionary <int, int> beginMap = Generics.NewHashMap();
IDictionary <int, int> endMap = Generics.NewHashMap();
bool haveTokenOffsets = true;
foreach (ICoreMap sent in originalDocument.Get(typeof(CoreAnnotations.SentencesAnnotation)))
{
foreach (CoreLabel token in sent.Get(typeof(CoreAnnotations.TokensAnnotation)))
{
int tokBegin = token.Get(typeof(CoreAnnotations.TokenBeginAnnotation));
int tokEnd = token.Get(typeof(CoreAnnotations.TokenEndAnnotation));
if (tokBegin == null || tokEnd == null)
{
haveTokenOffsets = false;
}
int charBegin = token.Get(typeof(CoreAnnotations.CharacterOffsetBeginAnnotation));
int charEnd = token.Get(typeof(CoreAnnotations.CharacterOffsetEndAnnotation));
beginMap[charBegin] = tokBegin;
endMap[charEnd] = tokEnd;
}
}
IList <ICoreMap> timexMaps = new List <ICoreMap>();
int offset = 0;
INodeList docNodes = docElem.GetChildNodes();
for (int i = 0; i < docNodes.GetLength(); i++)
{
INode content = docNodes.Item(i);
if (content is IText)
{
IText text = (IText)content;
offset += text.GetWholeText().Length;
}
else
{
if (content is IElement)
{
IElement child = (IElement)content;
if (child.GetNodeName().Equals("TIMEX3"))
{
Timex timex = new Timex(child);
if (child.GetChildNodes().GetLength() != 1)
{
throw new Exception("TIMEX3 should only contain text " + child);
}
string timexText = child.GetTextContent();
ICoreMap timexMap = new ArrayCoreMap();
timexMap.Set(typeof(TimeAnnotations.TimexAnnotation), timex);
timexMap.Set(typeof(CoreAnnotations.TextAnnotation), timexText);
int charBegin = offset;
timexMap.Set(typeof(CoreAnnotations.CharacterOffsetBeginAnnotation), offset);
offset += timexText.Length;
timexMap.Set(typeof(CoreAnnotations.CharacterOffsetEndAnnotation), offset);
int charEnd = offset;
//(tokens)
if (haveTokenOffsets)
{
int tokBegin = beginMap[charBegin];
int searchStep = 1;
//if no exact match, search around the character offset
while (tokBegin == null)
{
tokBegin = beginMap[charBegin - searchStep];
if (tokBegin == null)
{
tokBegin = beginMap[charBegin + searchStep];
}
searchStep += 1;
}
searchStep = 1;
int tokEnd = endMap[charEnd];
while (tokEnd == null)
{
tokEnd = endMap[charEnd - searchStep];
if (tokEnd == null)
{
tokEnd = endMap[charEnd + searchStep];
}
searchStep += 1;
}
timexMap.Set(typeof(CoreAnnotations.TokenBeginAnnotation), tokBegin);
timexMap.Set(typeof(CoreAnnotations.TokenEndAnnotation), tokEnd);
}
timexMaps.Add(timexMap);
}
else
{
throw new Exception("unexpected element " + child);
}
}
else
{
throw new Exception("unexpected content " + content);
}
}
}
return(timexMaps);
}
/// <summary>Parses one ACE specification</summary>
/// <returns>Simply displays the events to stdout</returns>
/// <exception cref="System.IO.IOException"/>
/// <exception cref="Org.Xml.Sax.SAXException"/>
/// <exception cref="Javax.Xml.Parsers.ParserConfigurationException"/>
public static AceDocument ParseDocument(File f)
{
// parse the Dom document
IDocument document = ReadDocument(f);
//
// create the ACE document object
//
INode docElement = document.GetElementsByTagName("document").Item(0);
AceDocument aceDoc = new AceDocument(GetAttributeValue(docElement, "DOCID"));
//
// read all entities
//
INodeList entities = document.GetElementsByTagName("entity");
int entityCount = 0;
for (int i = 0; i < entities.GetLength(); i++)
{
INode node = entities.Item(i);
//
// the entity type and subtype
//
string id = GetAttributeValue(node, "ID");
string type = GetAttributeValue(node, "TYPE");
string subtype = GetAttributeValue(node, "SUBTYPE");
string cls = GetAttributeValue(node, "CLASS");
// create the entity
AceEntity entity = new AceEntity(id, type, subtype, cls);
aceDoc.AddEntity(entity);
// fetch all mentions of this event
IList <INode> mentions = GetChildrenByName(node, "entity_mention");
// parse all its mentions
foreach (INode mention1 in mentions)
{
AceEntityMention mention = ParseEntityMention(mention1);
entity.AddMention(mention);
aceDoc.AddEntityMention(mention);
}
entityCount++;
}
//log.info("Parsed " + entityCount + " XML entities.");
//
// read all relations
//
INodeList relations = document.GetElementsByTagName("relation");
for (int i_1 = 0; i_1 < relations.GetLength(); i_1++)
{
INode node = relations.Item(i_1);
//
// the relation type, subtype, tense, and modality
//
string id = GetAttributeValue(node, "ID");
string type = GetAttributeValue(node, "TYPE");
string subtype = GetAttributeValue(node, "SUBTYPE");
string modality = GetAttributeValue(node, "MODALITY");
string tense = GetAttributeValue(node, "TENSE");
// create the relation
AceRelation relation = new AceRelation(id, type, subtype, modality, tense);
aceDoc.AddRelation(relation);
// XXX: fetch relation_arguments here!
// fetch all mentions of this relation
IList <INode> mentions = GetChildrenByName(node, "relation_mention");
// traverse all mentions
foreach (INode mention1 in mentions)
{
AceRelationMention mention = ParseRelationMention(mention1, aceDoc);
relation.AddMention(mention);
aceDoc.AddRelationMention(mention);
}
}
//
// read all events
//
INodeList events = document.GetElementsByTagName("event");
for (int i_2 = 0; i_2 < events.GetLength(); i_2++)
{
INode node = events.Item(i_2);
//
// the event type, subtype, tense, and modality
//
string id = GetAttributeValue(node, "ID");
string type = GetAttributeValue(node, "TYPE");
string subtype = GetAttributeValue(node, "SUBTYPE");
string modality = GetAttributeValue(node, "MODALITY");
string polarity = GetAttributeValue(node, "POLARITY");
string genericity = GetAttributeValue(node, "GENERICITY");
string tense = GetAttributeValue(node, "TENSE");
// create the event
AceEvent @event = new AceEvent(id, type, subtype, modality, polarity, genericity, tense);
aceDoc.AddEvent(@event);
// fetch all mentions of this relation
IList <INode> mentions = GetChildrenByName(node, "event_mention");
// traverse all mentions
foreach (INode mention1 in mentions)
{
AceEventMention mention = ParseEventMention(mention1, aceDoc);
@event.AddMention(mention);
aceDoc.AddEventMention(mention);
}
}
return(aceDoc);
}
/// <exception cref="System.Exception"/>
public static XMLToAnnotation.Data ReadXMLFormat(string fileName)
{
//Extract character list, gold quote speaker and mention information from the XML document.
IDocument doc = XMLUtils.ReadDocumentFromFile(fileName);
INode text = doc.GetDocumentElement().GetElementsByTagName("text").Item(0);
string docText = GetJustText(text);
Annotation document = GetAnnotatedFile(docText, fileName, GetProcessedCoreNLPProperties());
IList <ICoreMap> quotes = document.Get(typeof(CoreAnnotations.QuotationsAnnotation));
IList <CoreLabel> tokens = document.Get(typeof(CoreAnnotations.TokensAnnotation));
IList <XMLToAnnotation.GoldQuoteInfo> goldList = new List <XMLToAnnotation.GoldQuoteInfo>();
IDictionary <int, XMLToAnnotation.Mention> idToMention = new Dictionary <int, XMLToAnnotation.Mention>();
IList <Person> personList = ReadXMLCharacterList(doc);
IDictionary <string, IList <Person> > personMap = QuoteAttributionUtils.ReadPersonMap(personList);
IList <Pair <int, string> > mentionIdToSpeakerList = new List <Pair <int, string> >();
//there is at least 1 case in which the XML quote does not match up with the automatically-extracted quote. (Ex: quote by Mr. Collins that begins, "Hunsford, near Westerham, Kent, ...")
//as the dirty solution, we treat all quotes encapsulated within an XML quote as the same speaker (although this is not 100% accurate!)
int quoteIndex = 0;
INodeList textElems = text.GetChildNodes();
int tokenIndex = 0;
for (int i = 0; i < textElems.GetLength(); i++)
{
INode chapterNode = textElems.Item(i);
if (chapterNode.GetNodeName().Equals("chapter"))
{
INodeList chapElems = chapterNode.GetChildNodes();
for (int j = 0; j < chapElems.GetLength(); j++)
{
INode child = chapElems.Item(j);
if (child.GetNodeName().Equals("quote"))
{
//search for nested mentions
INodeList quoteChildren = child.GetChildNodes();
for (int k = 0; k < quoteChildren.GetLength(); k++)
{
INode quoteChild = quoteChildren.Item(k);
if (quoteChild.GetNodeName().Equals("mention"))
{
string mentionText = quoteChild.GetTextContent();
int id = System.Convert.ToInt32(Sharpen.Runtime.Substring(quoteChild.GetAttributes().GetNamedItem("id").GetTextContent(), 1));
IList <int> connections = ReadConnection(quoteChild.GetAttributes().GetNamedItem("connection").GetNodeValue());
int endIndex = GetEndIndex(tokenIndex, tokens, mentionText);
// mentions.put(id, new XMLMention(quoteChild.getTextContent(), tokenIndex, endIndex, id, connections));
idToMention[id] = new XMLToAnnotation.Mention(mentionText, tokenIndex, endIndex);
tokenIndex = endIndex + 1;
}
else
{
string quoteText = quoteChild.GetTextContent();
quoteText = quoteText.ReplaceAll("\n(?!\n)", " ");
//trim unnecessarily newlines
quoteText = quoteText.ReplaceAll("_", string.Empty);
tokenIndex = GetEndIndex(tokenIndex, tokens, quoteText) + 1;
}
}
string quoteText_1 = child.GetTextContent();
// tokenIndex = getEndIndex(tokenIndex, tokens, quoteText) + 1;
quoteText_1 = quoteText_1.ReplaceAll("\n(?!\n)", " ");
//trim unnecessarily newlines
quoteText_1 = quoteText_1.ReplaceAll("_", string.Empty);
int quotationOffset = 1;
if (quoteText_1.StartsWith("``"))
{
quotationOffset = 2;
}
IList <int> connections_1 = ReadConnection(child.GetAttributes().GetNamedItem("connection").GetTextContent());
int id_1 = System.Convert.ToInt32(Sharpen.Runtime.Substring(child.GetAttributes().GetNamedItem("id").GetTextContent(), 1));
int mention_id = null;
if (connections_1.Count > 0)
{
mention_id = connections_1[0];
}
else
{
System.Console.Out.WriteLine("quote w/ no mention. ID: " + id_1);
}
// Pair<Integer, Integer> mentionPair = idToMentionPair.get(mention_id);
mentionIdToSpeakerList.Add(new Pair <int, string>(mention_id, child.GetAttributes().GetNamedItem("speaker").GetTextContent()));
string annotatedQuoteText = quotes[quoteIndex].Get(typeof(CoreAnnotations.TextAnnotation));
while (!quoteText_1.EndsWith(annotatedQuoteText))
{
quoteIndex++;
annotatedQuoteText = quotes[quoteIndex].Get(typeof(CoreAnnotations.TextAnnotation));
mentionIdToSpeakerList.Add(new Pair <int, string>(mention_id, child.GetAttributes().GetNamedItem("speaker").GetTextContent()));
}
// idToMentionPair.put(id, new Pair<>(-1, -1));
// imention_id = connections.get(0);
// quotes.add(new XMLQuote(quoteText.substring(quotationOffset, quoteText.length() - quotationOffset), child.getAttributes().getNamedItem("speaker").getTextContent(), id, chapterIndex, mention_id));
quoteIndex++;
}
else
{
if (child.GetNodeName().Equals("mention"))
{
string mentionText = child.GetTextContent();
int id = System.Convert.ToInt32(Sharpen.Runtime.Substring(child.GetAttributes().GetNamedItem("id").GetTextContent(), 1));
IList <int> connections = ReadConnection(child.GetAttributes().GetNamedItem("connection").GetNodeValue());
int endIndex = GetEndIndex(tokenIndex, tokens, mentionText);
idToMention[id] = new XMLToAnnotation.Mention(mentionText, tokenIndex, endIndex);
// mentions.put(id, new XMLMention(child.getTextContent(), tokenIndex, endIndex, id, connections));
tokenIndex = endIndex + 1;
}
else
{
//#text
string nodeText = child.GetTextContent();
nodeText = nodeText.ReplaceAll("\n(?!\n)", " ");
nodeText = nodeText.ReplaceAll("_", string.Empty);
if (tokenIndex >= tokens.Count)
{
continue;
}
tokenIndex = GetEndIndex(tokenIndex, tokens, nodeText) + 1;
}
}
}
}
}
foreach (Pair <int, string> item in mentionIdToSpeakerList)
{
XMLToAnnotation.Mention mention = idToMention[item.first];
if (mention == null)
{
goldList.Add(new XMLToAnnotation.GoldQuoteInfo(-1, -1, item.second, null));
}
else
{
goldList.Add(new XMLToAnnotation.GoldQuoteInfo(mention.begin, mention.end, item.second, mention.text));
}
}
//verify
if (document.Get(typeof(CoreAnnotations.QuotationsAnnotation)).Count != goldList.Count)
{
throw new Exception("Quotes size and gold size don't match!");
}
return(new XMLToAnnotation.Data(goldList, personList, document));
}
