commit 22f903ad46688f98820a112bc83678b6beb292f9
Author: Tanmai Khanna <khanna.tanmai@gmail.com>
Date:   Mon Jul 20 13:42:30 2020 +0530

    Output wordbound blank automatically if there's only one LU in the matching pattern (#94)
    
    Output wordbound blank when no rule matches in single stage transfer; If matched pattern has one LU, blank goes on all output LUs automatically; If input chunk in postchunk rule has one LU, blank outputs on all LUs in rule output; Add tests

diff --git a/apertium/interchunk_word.cc b/apertium/interchunk_word.cc
index fc7b000..d5c84ff 100644
--- a/apertium/interchunk_word.cc
+++ b/apertium/interchunk_word.cc
@@ -89,7 +89,7 @@ InterchunkWord::init(string const &chunk)
   
   if(b_end > 0)
   {
-    this->blank = chunk.substr(0, b_end);
+    this->wblank = chunk.substr(0, b_end);
     this->chunk = chunk.substr(b_end);
   }
   else
@@ -126,9 +126,9 @@ InterchunkWord::chunkPart(ApertiumRE const &part)
 }
 
 string
-InterchunkWord::getBlank()
+InterchunkWord::getWblank()
 {
-  return blank;
+  return wblank;
 }
 
 bool
diff --git a/apertium/interchunk_word.h b/apertium/interchunk_word.h
index 8608b7d..db0d56b 100644
--- a/apertium/interchunk_word.h
+++ b/apertium/interchunk_word.h
@@ -43,7 +43,7 @@ private:
   /**
    * Wordbound blank (for postchunk)
    */
-  string blank;
+  string wblank;
 
   /**
    * Copy method
@@ -100,9 +100,9 @@ public:
   
   /**
    * Reference the wordbound blank (for postchunk)
-   * @returns reference to the part of string matched
+   * @returns reference to the wblank string
    */
-  string getBlank();
+  string getWblank();
 
   /**
    * Sets a value for a chunk part
diff --git a/apertium/postchunk.cc b/apertium/postchunk.cc
index 5a27a4a..1ac8110 100644
--- a/apertium/postchunk.cc
+++ b/apertium/postchunk.cc
@@ -315,15 +315,15 @@ Postchunk::evalString(xmlNode *element)
       case ti_clip_tl:
         if(checkIndex(element, ti.getPos(), lword))
         {
-          if(gettingLemmaFromWord(ti.getContent()))
+          if(gettingLemmaFromWord(ti.getContent()) && lword > 1)
           {
             if(in_lu)
             {
-              out_wblank = combineWblanks(out_wblank, word[ti.getPos()]->getBlank());
+              out_wblank = combineWblanks(out_wblank, word[ti.getPos()]->getWblank());
             }
             else if(in_let_var)
             {
-              var_out_wblank[var_val] = combineWblanks(var_out_wblank[var_val], word[ti.getPos()]->getBlank());
+              var_out_wblank[var_val] = combineWblanks(var_out_wblank[var_val], word[ti.getPos()]->getWblank());
             }
           }
           
@@ -335,7 +335,10 @@ Postchunk::evalString(xmlNode *element)
         return StringUtils::itoa_string(tmpword.size());
 
       case ti_var:
+        if(lword > 1)
+        {
         out_wblank = combineWblanks(out_wblank, var_out_wblank[ti.getContent()]);
+        }
         
         return variables[ti.getContent()];
 
@@ -483,6 +486,11 @@ Postchunk::evalString(xmlNode *element)
     }
     
     in_lu = false;
+    
+    if(lword == 1)
+    {
+      out_wblank = word[1]->getWblank();
+    }
 
     if(myword != "")
     {
@@ -536,6 +544,11 @@ Postchunk::evalString(xmlNode *element)
 	value.append(myword);
       }
     }
+    
+    if(lword == 1)
+    {
+      out_wblank = word[1]->getWblank();
+    }
 
     if(value != "")
     {
@@ -579,6 +592,11 @@ Postchunk::processOut(xmlNode *localroot)
         
         in_lu = false;
         
+        if(lword == 1)
+        {
+          out_wblank = word[1]->getWblank();
+        }
+        
         if(myword != "")
         {
           fputws_unlocked(UtfConverter::fromUtf8(out_wblank).c_str(), output);
@@ -628,6 +646,11 @@ Postchunk::processOut(xmlNode *localroot)
             myword.append(mylocalword);
           }
         }
+        
+        if(lword == 1)
+        {
+          out_wblank = word[1]->getWblank();
+        }
 
         fputws_unlocked(UtfConverter::fromUtf8(out_wblank).c_str(), output);
         fputwc_unlocked('^', output);
diff --git a/apertium/transfer.cc b/apertium/transfer.cc
index 0bc0433..49a14ad 100644
--- a/apertium/transfer.cc
+++ b/apertium/transfer.cc
@@ -54,6 +54,7 @@ word(0),
 blank(0),
 lword(0),
 lblank(0),
+last_lword(0),
 output(0),
 any_char(0),
 any_tag(0),
@@ -375,15 +376,15 @@ Transfer::evalString(xmlNode *element)
       case ti_clip_sl:
         if(checkIndex(element, ti.getPos(), lword))
         {
-          if(gettingLemmaFromWord(ti.getContent()))
+          if(gettingLemmaFromWord(ti.getContent()) && last_lword > 1)
           {
             if(in_lu)
             {
-              out_wblank = combineWblanks(out_wblank, word[ti.getPos()]->blank());
+              out_wblank = combineWblanks(out_wblank, word[ti.getPos()]->getWblank());
             }
             else if(in_let_var)
             {
-              var_out_wblank[var_val] = combineWblanks(var_out_wblank[var_val], word[ti.getPos()]->blank());
+              var_out_wblank[var_val] = combineWblanks(var_out_wblank[var_val], word[ti.getPos()]->getWblank());
             }
           }
           
@@ -394,15 +395,15 @@ Transfer::evalString(xmlNode *element)
       case ti_clip_tl:
         if(checkIndex(element, ti.getPos(), lword))
         {
-          if(gettingLemmaFromWord(ti.getContent()))
+          if(gettingLemmaFromWord(ti.getContent()) && last_lword > 1)
           {
             if(in_lu)
             {
-              out_wblank = combineWblanks(out_wblank, word[ti.getPos()]->blank());
+              out_wblank = combineWblanks(out_wblank, word[ti.getPos()]->getWblank());
             }
             else if(in_let_var)
             {
-              var_out_wblank[var_val] = combineWblanks(var_out_wblank[var_val], word[ti.getPos()]->blank());
+              var_out_wblank[var_val] = combineWblanks(var_out_wblank[var_val], word[ti.getPos()]->getWblank());
             }
           }
             
@@ -460,7 +461,10 @@ Transfer::evalString(xmlNode *element)
         break;
 
       case ti_var:
-        out_wblank = combineWblanks(out_wblank, var_out_wblank[ti.getContent()]);
+        if(last_lword > 1)
+        {
+          out_wblank = combineWblanks(out_wblank, var_out_wblank[ti.getContent()]);
+        }
         return variables[ti.getContent()];
 
       case ti_lit_tag:
@@ -674,6 +678,11 @@ Transfer::evalString(xmlNode *element)
     }
     
     in_lu = false;
+    
+    if(last_lword == 1)
+    {
+      out_wblank = word[0]->getWblank();
+    }
       
     if(myword != "")
     {
@@ -728,6 +737,11 @@ Transfer::evalString(xmlNode *element)
       }
     }
 
+    if(last_lword == 1)
+    {
+      out_wblank = word[0]->getWblank();
+    }
+    
     if(value != "")
     {
       return out_wblank+"^"+value+"$";
@@ -774,6 +788,11 @@ Transfer::processOut(xmlNode *localroot)
           }
             
           in_lu = false;
+          
+          if(last_lword == 1)
+          {
+            out_wblank = word[0]->getWblank();
+          }
 
           if(myword != "")
           {
@@ -825,6 +844,11 @@ Transfer::processOut(xmlNode *localroot)
             }
           }
           
+          if(last_lword == 1)
+          {
+            out_wblank = word[0]->getWblank();
+          }
+          
           if(myword != "")
           {
             fputws_unlocked(UtfConverter::fromUtf8(out_wblank).c_str(), output);
@@ -933,7 +957,13 @@ Transfer::processChunk(xmlNode *localroot)
             myword.append(evalString(j));
           }
         }
-          in_lu = false;
+        
+        in_lu = false;
+        
+        if(last_lword == 1)
+        {
+          out_wblank = word[0]->getWblank();
+        }
           
         if(myword != "")
         {
@@ -981,6 +1011,12 @@ Transfer::processChunk(xmlNode *localroot)
           }
           myword.append(mylocalword);
         }
+        
+        if(last_lword == 1)
+        {
+          out_wblank = word[0]->getWblank();
+        }
+        
         if(myword != "")
         {
           result.append(out_wblank);
@@ -2440,6 +2476,7 @@ Transfer::transfer(FILE *in, FILE *out)
 	  {
 	    if(defaultAttrs == lu)
 	    {
+        fputws_unlocked(tr_wblank.c_str(), output);
         if(tr.first[0] != L'[' || tr.first[1] != L'[')
         {
           fputwc_unlocked(L'^', output);
@@ -2494,6 +2531,7 @@ Transfer::transfer(FILE *in, FILE *out)
       lastrule = rule_map[val-1];
       lastrule_id = val;
       last = input_buffer.getPos();
+      last_lword = tmpword.size();
 
       if(trace)
       {
diff --git a/apertium/transfer.h b/apertium/transfer.h
index ca3bfad..15b7400 100644
--- a/apertium/transfer.h
+++ b/apertium/transfer.h
@@ -57,6 +57,7 @@ private:
   TransferWord **word;
   string **blank;
   int lword, lblank;
+  int last_lword;
   string noblank = "";
   Buffer<TransferToken> input_buffer;
   vector<wstring *> tmpword;
diff --git a/apertium/transfer_word.cc b/apertium/transfer_word.cc
index 86862fe..51f15eb 100644
--- a/apertium/transfer_word.cc
+++ b/apertium/transfer_word.cc
@@ -26,7 +26,7 @@ TransferWord::copy(TransferWord const &o)
   s_str = o.s_str;
   t_str = o.t_str;
   r_str = o.r_str;
-  b_str = o.b_str;
+  wb_str = o.wb_str;
   queue_length = o.queue_length;
 }
 
@@ -40,9 +40,9 @@ queue_length(0)
 {
 }
 
-TransferWord::TransferWord(string const &src, string const &tgt, string const &ref, string const &blank, int queue)
+TransferWord::TransferWord(string const &src, string const &tgt, string const &ref, string const &wblank, int queue)
 {
-  init(src, tgt, ref, blank);
+  init(src, tgt, ref, wblank);
   queue_length = queue;
 }
 
@@ -68,12 +68,12 @@ TransferWord::operator =(TransferWord const &o)
 }
 
 void
-TransferWord::init(string const &src, string const &tgt, string const &ref, string const &blank)
+TransferWord::init(string const &src, string const &tgt, string const &ref, string const &wblank)
 {
   s_str = src;
   t_str = tgt;
   r_str = ref;
-  b_str = blank;
+  wb_str = wblank;
 }
 
 string
@@ -116,9 +116,9 @@ TransferWord::reference(ApertiumRE const &part, bool with_queue)
 }
 
 string
-TransferWord::blank()
+TransferWord::getWblank()
 {
-  return b_str;
+  return wb_str;
 }
 
 bool
diff --git a/apertium/transfer_word.h b/apertium/transfer_word.h
index 3ff35d8..03802d8 100644
--- a/apertium/transfer_word.h
+++ b/apertium/transfer_word.h
@@ -49,7 +49,7 @@ private:
   /**
    * Wordbound blank
    */
-  string b_str;
+  string wb_str;
 
   /**
    * Queue length
@@ -105,10 +105,10 @@ public:
    * @param src source word
    * @param tgt target word
    * @param ref reference word
-   * @param blank wordbound blank
+   * @param wblank wordbound blank
    * @param queue queue lenght
    */
-  TransferWord(string const &src, string const &tgt, string const &ref, string const &blank, int queue = 0);
+  TransferWord(string const &src, string const &tgt, string const &ref, string const &wblank, int queue = 0);
 
   /**
    * Assignment operator
@@ -123,9 +123,9 @@ public:
    * @param src source word
    * @param tgt target word
    * @param ref reference word
-   * @param blank wordbound blank
+   * @param wblank wordbound blank
    */
-  void init(string const &src, string const &tgt, string const &ref, string const &blank);
+  void init(string const &src, string const &tgt, string const &ref, string const &wblank);
 
   /**
    * Reference a source language word part
@@ -155,7 +155,7 @@ public:
    * Reference the wordbound blank part
    * @returns reference to the wordbound blank
    */
-  string blank();
+  string getWblank();
 
   /**
    * Sets a value for a source language word part
diff --git a/tests/data/apertium-nno-nob.nno-nob.t3x b/tests/data/apertium-nno-nob.nno-nob.t3x
index a52713d..df91750 100644
--- a/tests/data/apertium-nno-nob.nno-nob.t3x
+++ b/tests/data/apertium-nno-nob.nno-nob.t3x
@@ -17,6 +17,9 @@
     <def-cat n="n_n">
       <cat-item name="n_n"/>
     </def-cat>
+    <def-cat n="thing_wb">
+      <cat-item name="thing_wb"/>
+    </def-cat>
   </section-def-cats>
 
 
@@ -155,5 +158,24 @@
       </action>
     </rule>
 
+    <rule comment="CHUNK: thing_wblank">
+      <pattern>
+        <pattern-item n="thing_wb"/>
+      </pattern>
+      <action>
+        <call-macro n="default_to_chunk_number"><with-param pos="1"/><with-param pos="0"/></call-macro>
+        <out>
+          <lu><concat><lit v="newthing"/><lit-tag v="adj.sg"/></concat></lu>
+          <b/>
+          <lu><clip pos="1" part="whole"/></lu>
+          <b/>
+          <mlu>
+            <lu><clip pos="1" part="whole"/></lu>
+            <lu><concat><lit v="newpr"/><lit-tag v="pr"/></concat></lu>
+          </mlu>
+        </out>
+      </action>
+    </rule>
+
   </section-rules>
 </postchunk>
diff --git a/tests/data/apertium-nno-nob.nob-nno.t1x b/tests/data/apertium-nno-nob.nob-nno.t1x
index ba6ab3d..1de6011 100644
--- a/tests/data/apertium-nno-nob.nob-nno.t1x
+++ b/tests/data/apertium-nno-nob.nob-nno.t1x
@@ -110,7 +110,6 @@
             <tags><tag><clip pos="1" side="tl" part="tags"/></tag></tags>
             <mlu>
               <lu><clip pos="1" side="tl" part="whole"/></lu>
-              <b/>
               <lu><clip pos="3" side="tl" part="whole"/></lu>
             </mlu>
           </chunk>
@@ -118,5 +117,32 @@
         </out>
       </action>
     </rule>
+
+    <rule comment="blankrule_single">
+      <pattern>
+        <pattern-item n="nacr"/>
+      </pattern>
+      <action>
+        <out>
+          <chunk name="nacr" case="caseFirstWord">
+            <tags><tag><clip pos="1" side="tl" part="tags"/></tag></tags>
+            <lu><concat><lit v="test"/><lit-tag v="n.sg"/></concat></lu>
+            <b/>
+            <lu><clip pos="1" side="tl" part="whole"/></lu>
+          </chunk>
+          <b/>
+          <chunk name="nacr2" case="caseFirstWord">
+            <tags><tag><clip pos="1" side="tl" part="tags"/></tag></tags>
+            <lu><concat><lit v="testlem"/><lit-tag v="n.pl"/></concat></lu>
+            <b/>
+            <mlu>
+              <lu><clip pos="1" side="tl" part="whole"/></lu>
+              <lu><concat><lit v="postp"/><lit-tag v="adj.sg"/></concat></lu>
+            </mlu>
+          </chunk>
+        </out>
+      </action>
+    </rule>
+
   </section-rules>
 </transfer>
diff --git a/tests/postchunk/__init__.py b/tests/postchunk/__init__.py
index 52c9ef3..e83de7f 100644
--- a/tests/postchunk/__init__.py
+++ b/tests/postchunk/__init__.py
@@ -111,6 +111,9 @@ class WordboundBlankTest(PostchunkTest):
     inputs =          ["^n_n<SN><sg>{[[t:b:123456]]^worda<n><ND><m>$ ;[testblank] [[t:s:xyzab12]]^wordb# xyz<n><ND><f>$}$"]
     expectedOutputs = ["[[t:s:xyzab12]]^wordb# xyz<n><ND><f>$ ;[testblank] [[t:b:123456]]^worda<n><ND><m>$ [[t:b:123456; t:s:xyzab12]]^worda+wordb# xyz$"]
 
+class SingleLUWordboundBlankTest(PostchunkTest):
+    inputs =          ["^thing_wb<thing><sg>{^[[t:i:xyzabc]]thing<thing><ND>$}$ ^n_n<SN><sg>{[[t:b:123456]]^worda<n><ND><m>$ ;[testblank] [[t:s:xyzab12]]^wordb# xyz<n><ND><f>$}$ [blanks] ^thing_wb<thing><sg>{^[[t:i:xyzabc]]thing<thing><ND>$}$ [blankx] ^vblex<vblex><imp>{[[t:b:123zbc]]^gå<vblex><imp>$}$^default<default>{^.<sent><clb>$}$ [blanks3] ^thing<thing><sg>{^[[t:i:xyzabc]]thing<thing><ND>$}$"]
+    expectedOutputs = ["[[t:i:xyzabc]]^newthing<adj><sg>$ [[t:i:xyzabc]]^thing<thing><sg>$ [[t:i:xyzabc]]^thing<thing><sg>+newpr<pr>$ [[t:s:xyzab12]]^wordb# xyz<n><ND><f>$ ;[testblank] [[t:b:123456]]^worda<n><ND><m>$ [[t:b:123456; t:s:xyzab12]]^worda+wordb# xyz$ [blanks] [[t:i:xyzabc]]^newthing<adj><sg>$ [[t:i:xyzabc]]^thing<thing><sg>$ [[t:i:xyzabc]]^thing<thing><sg>+newpr<pr>$ [blankx] [[t:b:123zbc]]^gå<vblex><imp>$^.<sent><clb>$ [blanks3] [[t:i:xyzabc]]^thing<thing><sg>$"]
 class BincompatTest(SimplePostchunkTest):
     bindata = "data/bincompat.t3x.bin"
 
diff --git a/tests/transfer/__init__.py b/tests/transfer/__init__.py
index 4eda9c1..805b0f1 100644
--- a/tests/transfer/__init__.py
+++ b/tests/transfer/__init__.py
@@ -99,6 +99,11 @@ class WordboundBlankTest(TransferTest):
     inputs =          ["[blank1] [[t:s:123456]]^worda<n><acr>/wordta<n><acr>$ ;[blank2]; [[t:b:xyz123; t:l:xyz347]]^wordb<n><acr>/wordtb<n><acr>$ [blank3];  [[t:i:abc123; t:s:abc123]]^hun<prn><f>/ho<prn><f>$"]
     expectedOutputs = ["[blank1] ^prn<prn><f>{[[t:i:abc123; t:s:abc123]]^ho<prn><f>$ [[t:b:xyz123; t:l:xyz347]]^wordtb<nacr><sg><m>$}$ ;[blank2]; ^det<n><acr>{[[t:s:123456; t:i:abc123; t:s:abc123]]^wordta<n><acr>+ho<prn><f>$}$ [blank3];  "]
 
+class SingleLUWordboundBlankTest(TransferTest):
+    inputs =          ["[blank1] [[t:s:123456]]^worda<n><acr>/wordta<n><acr>$ ;[blank2]; [[t:b:xyz123; t:l:xyz347]]^wordb<n><acr>/wordtb<n><acr>$ [blank3]; "]
+    expectedOutputs = ["[blank1] ^nacr<n><acr>{[[t:s:123456]]^test<n><sg>$ [[t:s:123456]]^wordta<n><acr>$}$ ^nacr2<n><acr>{[[t:s:123456]]^testlem<n><pl>$ [[t:s:123456]]^wordta<n><acr>+postp<adj><sg>$}$ ;[blank2]; ^nacr<n><acr>{[[t:b:xyz123; t:l:xyz347]]^test<n><sg>$ [[t:b:xyz123; t:l:xyz347]]^wordtb<n><acr>$}$ ^nacr2<n><acr>{[[t:b:xyz123; t:l:xyz347]]^testlem<n><pl>$ [[t:b:xyz123; t:l:xyz347]]^wordtb<n><acr>+postp<adj><sg>$}$ [blank3]; "]
+
+
 class BincompatTest(BasicTransferTest):
     bindata = "data/bincompat.t1x.bin"