starting placeholder

1 files changed, 16 insertions, 3 deletions

diff --git a/moses/XmlOption.cpp b/moses/XmlOption.cpp
index 4b703b247..069166e1f 100644
--- a/moses/XmlOption.cpp
+++ b/moses/XmlOption.cpp
@@ -150,10 +150,13 @@ vector<string> TokenizeXml(const string& str, const std::string& lbrackStr, cons
  * \param rbrackStr xml tag's right bracket string, typically ">"
  */
 bool ProcessAndStripXMLTags(string &line, vector<XmlOption*> &res, ReorderingConstraint &reorderingConstraint, vector< size_t > &walls,
+							std::vector< std::pair<size_t, std::string> > &placeholders,
                             const std::string& lbrackStr, const std::string& rbrackStr)
 {
   //parse XML markup in translation line
 
+  const StaticData &staticData = StaticData::Instance();
+
   // no xml tag? we're done.
 //if (line.find_first_of('<') == string::npos) {
   if (line.find(lbrackStr) == string::npos) {
@@ -172,8 +175,8 @@ bool ProcessAndStripXMLTags(string &line, vector<XmlOption*> &res, ReorderingCon
   string cleanLine; // return string (text without xml)
   size_t wordPos = 0; // position in sentence (in terms of number of words)
 
-  const vector<FactorType> &outputFactorOrder = StaticData::Instance().GetOutputFactorOrder();
-  const string &factorDelimiter = StaticData::Instance().GetFactorDelimiter();
+  const vector<FactorType> &outputFactorOrder = staticData.GetOutputFactorOrder();
+  const string &factorDelimiter = staticData.GetFactorDelimiter();
 
   // loop through the tokens
   for (size_t xmlTokenPos = 0 ; xmlTokenPos < xmlTokens.size() ; xmlTokenPos++) {
@@ -290,6 +293,16 @@ bool ProcessAndStripXMLTags(string &line, vector<XmlOption*> &res, ReorderingCon
           reorderingConstraint.SetZone( startPos, endPos-1 );
         }
 
+        // name-entity placeholder
+        else if (tagName == "ne") {
+        	if (startPos != (endPos - 1)) {
+        	            TRACE_ERR("ERROR: Placeholder must only span 1 word: " << line << endl);
+        	            return false;
+        	}
+        	string entity = ParseXmlTagAttribute(tagContent,"entity");
+        	placeholders.push_back(std::pair<size_t, std::string>(startPos, entity));
+        }
+
         // default: opening tag that specifies translation options
         else {
           if (startPos >= endPos) {
@@ -329,7 +342,7 @@ bool ProcessAndStripXMLTags(string &line, vector<XmlOption*> &res, ReorderingCon
           }
 
           // store translation options into members
-          if (StaticData::Instance().GetXmlInputType() != XmlIgnore) {
+          if (staticData.GetXmlInputType() != XmlIgnore) {
             // only store options if we aren't ignoring them
             for (size_t i=0; i<altTexts.size(); ++i) {
               Phrase sourcePhrase; // TODO don't know what the source phrase is