commit f5697bcd8704afc6ec571590fda82e5a76a1c909
Author: Tanmai Khanna <khanna.tanmai@gmail.com>
Date:   Wed Jul 15 17:30:00 2020 +0530

    Fix wordbound blank merging | Add tests for wordbound blanks

diff --git a/src/lsx_processor.cc b/src/lsx_processor.cc
index 0163aa4..e50edba 100644
--- a/src/lsx_processor.cc
+++ b/src/lsx_processor.cc
@@ -293,27 +293,28 @@ LSXProcessor::processWord(FILE* input, FILE* output)
       
       if(wblank.empty())
       {
-        for(size_t j = 0; j < out_lus.size(); j++)
+        bool first_bound_blank = true;
+        for(size_t j = 0; j < bound_blank_queue.size()-1; j++) //ignore last bound_blank as it belongs to word after matched pattern
         {
           if(bound_blank_queue[j].size() > 0)
           {
-            if(j == 0)
+            if(first_bound_blank)
             {
               wblank += L"[[";
+              first_bound_blank = false;
             }
-            else if(j > 0)
+            else
             {
               wblank += L"; ";
             }
             
             wblank += bound_blank_queue[j].c_str();
-            
-            if(j == out_lus.size() - 1)
-            {
-              wblank += L"]]";
-            }
           }
         }
+        if(!first_bound_blank) //if there were any wordbound blanks
+        {
+          wblank += L"]]";
+        }
         
         fputws_unlocked(wblank.c_str(), output);
       }
diff --git a/tests/lsx_proc/__init__.py b/tests/lsx_proc/__init__.py
index 65fe3b2..d3acaf5 100644
--- a/tests/lsx_proc/__init__.py
+++ b/tests/lsx_proc/__init__.py
@@ -8,8 +8,24 @@ from proctest import ProcTest
 class nullFlushTest(unittest.TestCase, ProcTest):
     procdix = "data/short-example.dix"
     procflags = ["-z"]
-    inputs = ["^take<vblex><pres>$ ^it<prn><obj>$ ^out<adv>$","^take<vblex><pres>$ ^me<prn><obj>$ ^out<adv>$"]
-    expectedOutputs = ["^take# out<vblex><pres>$ ^it<prn><obj>$","^take# out<vblex><pres>$ ^me<prn><obj>$"]
-
+    inputs = ["^take<vblex><pres>$ ^it<prn><obj>$ ^out<adv>$",
+    "^take<vblex><pres>$ ^me<prn><obj>$ ^out<adv>$",
+     "^the<det>$ !!^Aragonese<n><sg>$ ;^take<vblex><pres>$ ;.^it<prn><obj>$   !;^out<adv>$ ^a<det><sg>$", #out of LU characters test
+     "^the<det>$ [<x>]^Aragonese<n><sg>$[</x>] [<y>]^take<vblex><pres>$ [<z>]^it<prn><obj>$ [</y></z>]^out<adv>$ [<a>]^a<det><sg>$[</a>]", #normal blanks test
+     "^the<det>$ [[t:b:123456]]^Aragonese<n><sg>$ [[t:s:abc123]]^take<vblex><pres>$ [[t:b:xyz567]]^it<prn><obj>$ [[t:p:yui124]]^out<adv>$ [[t:b:uvw674]]^a<det><sg>$", #wordbound blank tests
+     "^the<det>$ [[t:b:123456]]^Aragonese<n><sg>$ [[t:s:abc123; t:p:hgb650]]^take<vblex><pres>$ ^it<prn><obj>$ [[t:p:yui124]]^out<adv>$ [[t:b:uvw674]]^a<det><sg>$",
+     "^the<det>$ [[t:b:123456]]^Aragonese<n><sg>$ ^take<vblex><pres>$ [[t:b:xyz567]]^it<prn><obj>$ ^out<adv>$ [[t:b:uvw674]]^a<det><sg>$",
+     "^the<det>$ [[t:b:123456]]^Aragonese<n><sg>$ ^take<vblex><pres>$ ^it<prn><obj>$ [[t:p:yui124]]^out<adv>$ [[t:b:uvw674]]^a<det><sg>$",
+     "^the<det>$ [[t:b:123456]]^Aragonese<n><sg>$ [[t:s:abc123; t:p:hgb650]]^take<vblex><pres>$ ^it<prn><obj>$ [[t:p:yui124; t:x:puhbj23]]^out<adv>$ [[t:b:uvw674]]^a<det><sg>$"]
+     
+    expectedOutputs = ["^take# out<vblex><pres>$ ^it<prn><obj>$",
+    "^take# out<vblex><pres>$ ^me<prn><obj>$",
+     "^the<det>$ !!^Aragonese<n><sg>$ ;^take# out<vblex><pres>$ ;.^it<prn><obj>$   !; ^a<det><sg>$",
+     "^the<det>$ [<x>]^Aragonese<n><sg>$[</x>] [<y>]^take# out<vblex><pres>$ [<z>]^it<prn><obj>$ [</y></z>] [<a>]^a<det><sg>$[</a>]",
+     "^the<det>$ [[t:b:123456]]^Aragonese<n><sg>$ [[t:s:abc123; t:b:xyz567; t:p:yui124]]^take# out<vblex><pres>$ [[t:s:abc123; t:b:xyz567; t:p:yui124]]^it<prn><obj>$ [[t:b:uvw674]]^a<det><sg>$",
+     "^the<det>$ [[t:b:123456]]^Aragonese<n><sg>$ [[t:s:abc123; t:p:hgb650; t:p:yui124]]^take# out<vblex><pres>$ [[t:s:abc123; t:p:hgb650; t:p:yui124]]^it<prn><obj>$ [[t:b:uvw674]]^a<det><sg>$",
+     "^the<det>$ [[t:b:123456]]^Aragonese<n><sg>$ [[t:b:xyz567]]^take# out<vblex><pres>$ [[t:b:xyz567]]^it<prn><obj>$ [[t:b:uvw674]]^a<det><sg>$",
+     "^the<det>$ [[t:b:123456]]^Aragonese<n><sg>$ [[t:p:yui124]]^take# out<vblex><pres>$ [[t:p:yui124]]^it<prn><obj>$ [[t:b:uvw674]]^a<det><sg>$",
+     "^the<det>$ [[t:b:123456]]^Aragonese<n><sg>$ [[t:s:abc123; t:p:hgb650; t:p:yui124; t:x:puhbj23]]^take# out<vblex><pres>$ [[t:s:abc123; t:p:hgb650; t:p:yui124; t:x:puhbj23]]^it<prn><obj>$ [[t:b:uvw674]]^a<det><sg>$"]
 # These fail on some systems:
 #from null_flush_invalid_stream_format import *