commit 8f4dd5dd16e5f8df97a1d68e0a440914cb54181e
Author: Tanmai Khanna <khanna.tanmai@gmail.com>
Date:   Mon May 11 17:07:18 2020 +0530

    getting secondary tags in output from the word which provides the lemma | modifying regex in transfer_data for tags and stags

diff --git a/apertium/transfer.cc b/apertium/transfer.cc
index eff15a4..50a397d 100644
--- a/apertium/transfer.cc
+++ b/apertium/transfer.cc
@@ -304,6 +304,12 @@ Transfer::evalString(xmlNode *element)
       case ti_clip_sl:
         if(checkIndex(element, ti.getPos(), lword))
         {
+          if(ti.getContent().compare("lem") == 0 || ti.getContent().compare("lemh") == 0) //only get secondary tags if we're getting the lemma from the word
+          {
+              stags = word[ti.getPos()]->source(attr_items["stags"], ti.getCondition());
+              //cout << "\n##SLSTAGS::" <<stags << "##SLSTAGS##\n";
+          }
+            
           return word[ti.getPos()]->source(attr_items[ti.getContent()], ti.getCondition());
         }
         break;
@@ -311,6 +317,12 @@ Transfer::evalString(xmlNode *element)
       case ti_clip_tl:
         if(checkIndex(element, ti.getPos(), lword))
         {
+          if(ti.getContent().compare("lem") == 0 || ti.getContent().compare("lemh") == 0)
+          {
+              stags = word[ti.getPos()]->target(attr_items["stags"], ti.getCondition());
+              //cout << "\n##TLSTAGS::" <<stags << "##TLSTAGS##\n";
+          }
+            
           return word[ti.getPos()]->target(attr_items[ti.getContent()], ti.getCondition());
         }
         break;
@@ -651,7 +663,7 @@ Transfer::processOut(xmlNode *localroot)
   {
     if(i->type == XML_ELEMENT_NODE)
     {
-      if(defaultAttrs == lu)
+      if(defaultAttrs == lu) //stream_modification TODO
       {
         if(!xmlStrcmp(i->name, (const xmlChar *) "lu"))
         {
@@ -793,8 +805,11 @@ Transfer::processChunk(xmlNode *localroot)
         result.append(processTags(i));
         result.append("{");
       }
-      else if(!xmlStrcmp(i->name, (const xmlChar *) "lu"))
+      else if(!xmlStrcmp(i->name, (const xmlChar *) "lu")) //Need to add secondary tags here!
       {
+          //cout << "\n%%OUTNOW%%\n";
+          stags.clear();
+          
         string myword;
         for(xmlNode *j = i->children; j != NULL; j = j->next)
         {
@@ -803,6 +818,9 @@ Transfer::processChunk(xmlNode *localroot)
             myword.append(evalString(j));
           }
         }
+          //cout << "\n###MYWORD###\n" << myword << "\n###MYWORD###\n";
+          myword.append(stags); //from the LU that the lem or lemh has come from
+          //cout << "\n###WITHSTAGS###\n" << myword << "\n###WITHSTAGS###\n";
         if(myword != "")
         {
           result.append("^");
@@ -810,7 +828,7 @@ Transfer::processChunk(xmlNode *localroot)
           result.append("$");
         }
       }
-      else if(!xmlStrcmp(i->name, (const xmlChar *) "mlu"))
+      else if(!xmlStrcmp(i->name, (const xmlChar *) "mlu")) //stream modification TODO
       {
         bool first_time = true;
         string myword;
diff --git a/apertium/transfer.h b/apertium/transfer.h
index 2a1135f..7ee5722 100644
--- a/apertium/transfer.h
+++ b/apertium/transfer.h
@@ -61,7 +61,9 @@ private:
   Buffer<TransferToken> input_buffer;
   vector<wstring *> tmpword;
   vector<wstring *> tmpblank;
-
+  
+  string stags; //stores secondary tags of the LU that is being output
+    
   FSTProcessor fstp;
   FSTProcessor extended;
   bool isExtended;
diff --git a/apertium/transfer_data.cc b/apertium/transfer_data.cc
index 78015f3..adf8b8b 100644
--- a/apertium/transfer_data.cc
+++ b/apertium/transfer_data.cc
@@ -50,7 +50,8 @@ TransferData::TransferData()
   attr_items[L"lemq"] = L"\\#[- _][^<]+";
   attr_items[L"lemh"] = L"^(([^<#]|\"\\<\"|\"\\#\")+)";
   attr_items[L"whole"] = L"(.+)";
-  attr_items[L"tags"] = L"((<[^>]+>)+)";
+  attr_items[L"tags"] = L"((<[^:>]+>)+)"; //match all tags excluding secondary tags
+  attr_items[L"stags"] = L"((<[^>]+:[^>]+>)+)"; //match all secondary tags
   attr_items[L"chname"] = L"({([^/]+)\\/)"; // includes delimiters { and / !!!
   attr_items[L"chcontent"] = L"(\\{.+)";
   attr_items[L"content"] = L"(\\{.+)";