Simplification in lngmisc with accompanying tests

3 files changed, 47 insertions, 33 deletions

diff --git a/svl/inc/svl/lngmisc.hxx b/svl/inc/svl/lngmisc.hxx
index a473ab69f910..6c71b45e2c0b 100644
--- a/svl/inc/svl/lngmisc.hxx
+++ b/svl/inc/svl/lngmisc.hxx
@@ -54,6 +54,7 @@ inline bool IsHyphen( sal_Unicode cChar )
 
 inline bool IsControlChar( sal_Unicode cChar )
 {
+    // TODO: why doesn't this include 0x0F DEL?
     return cChar < static_cast<sal_Unicode>(' ');
 }
 
diff --git a/svl/qa/unit/test_lngmisc.cxx b/svl/qa/unit/test_lngmisc.cxx
index 47f671a9dfe6..6411e208cd5e 100644
--- a/svl/qa/unit/test_lngmisc.cxx
+++ b/svl/qa/unit/test_lngmisc.cxx
@@ -16,14 +16,14 @@ namespace
   {
   private:
     void testRemoveHyphens();
-    //    void testRemoveControlChars();
+    void testRemoveControlChars();
     //    void testReplaceControlChars();
     //    void testGetThesaurusReplaceText();
 
     CPPUNIT_TEST_SUITE(LngMiscTest);
 
     CPPUNIT_TEST(testRemoveHyphens);
-    //    CPPUNIT_TEST(testRemoveControlChars);
+    CPPUNIT_TEST(testRemoveControlChars);
     //    CPPUNIT_TEST(testReplaceControlChars);
     //    CPPUNIT_TEST(testGetThesaurusReplaceText);
 
@@ -61,12 +61,38 @@ namespace
     CPPUNIT_ASSERT(str4.equalsAsciiL(RTL_CONSTASCII_STRINGPARAM("asdf")));
   }
 
-  /*
   void LngMiscTest::testRemoveControlChars()
   {
-    CPPUNIT_ASSERT(true);
+    ::rtl::OUString str1(RTL_CONSTASCII_USTRINGPARAM(""));
+    ::rtl::OUString str2(RTL_CONSTASCII_USTRINGPARAM("asdf"));
+    ::rtl::OUString str3(RTL_CONSTASCII_USTRINGPARAM("asdf\nasdf"));
+
+    ::rtl::OUStringBuffer str4Buf(33);
+    str4Buf.setLength(33);
+    for(int i = 0; i < 33; i++)
+      str4Buf[i] = static_cast<sal_Unicode>(i);
+    //    TODO: is this a bug? shouldn't RemoveControlChars remove this?
+    //    str4Buf[33] = static_cast<sal_Unicode>(0x7F);
+    ::rtl::OUString str4(str4Buf.makeStringAndClear());
+
+    bool bModified = linguistic::RemoveControlChars(str1);
+    CPPUNIT_ASSERT(!bModified);
+    CPPUNIT_ASSERT(str1.isEmpty());
+
+    bModified = linguistic::RemoveControlChars(str2);
+    CPPUNIT_ASSERT(!bModified);
+    CPPUNIT_ASSERT(str2.equalsAsciiL(RTL_CONSTASCII_STRINGPARAM("asdf")));
+
+    bModified = linguistic::RemoveControlChars(str3);
+    CPPUNIT_ASSERT(bModified);
+    CPPUNIT_ASSERT(str3.equalsAsciiL(RTL_CONSTASCII_STRINGPARAM("asdfasdf")));
+
+    bModified = linguistic::RemoveControlChars(str4);
+    CPPUNIT_ASSERT(bModified);
+    CPPUNIT_ASSERT(str4.equalsAsciiL(RTL_CONSTASCII_STRINGPARAM(" ")));
   }
 
+  /*
   void LngMiscTest::testReplaceControlChars()
   {
     CPPUNIT_ASSERT(true);
diff --git a/svl/source/misc/lngmisc.cxx b/svl/source/misc/lngmisc.cxx
index 3c32099db5ee..56c10661887c 100644
--- a/svl/source/misc/lngmisc.cxx
+++ b/svl/source/misc/lngmisc.cxx
@@ -47,12 +47,9 @@ namespace linguistic
 sal_Int32 GetNumControlChars( const OUString &rTxt )
 {
     sal_Int32 nCnt = 0;
-    sal_Int32 nLen = rTxt.getLength();
-    for (sal_Int32 i = 0;  i < nLen;  ++i)
-    {
-        if (IsControlChar( rTxt[i] ))
+    for (sal_Int32 i = 0; i < rTxt.getLength(); ++i)
+        if (IsControlChar(rTxt[i]))
             ++nCnt;
-    }
     return nCnt;
 }
 
@@ -66,31 +63,21 @@ bool RemoveHyphens( OUString &rTxt )
 
 bool RemoveControlChars( OUString &rTxt )
 {
-    bool bModified = false;
-    sal_Int32 nCtrlChars = GetNumControlChars( rTxt );
-    if (nCtrlChars)
-    {
-        sal_Int32 nLen  = rTxt.getLength();
-        sal_Int32 nSize = nLen - nCtrlChars;
-        OUStringBuffer aBuf( nSize );
-        aBuf.setLength( nSize );
-        sal_Int32 nCnt = 0;
-        for (sal_Int32 i = 0;  i < nLen;  ++i)
-        {
-            sal_Unicode cChar = rTxt[i];
-            if (!IsControlChar( cChar ))
-            {
-                DBG_ASSERT( nCnt < nSize, "index out of range" );
-                aBuf.setCharAt( nCnt++, cChar );
-            }
-        }
-        DBG_ASSERT( nCnt == nSize, "wrong size" );
-        rTxt = aBuf.makeStringAndClear();
-        bModified = true;
-    }
-    return bModified;
-}
+    sal_Int32 nSize = rTxt.getLength() - GetNumControlChars(rTxt);
+    if(nSize == rTxt.getLength())
+        return false;
+
+    OUStringBuffer aBuf(nSize);
+    aBuf.setLength(nSize);
+    for (sal_Int32 i = 0, j = 0; i < rTxt.getLength() && j < nSize; ++i)
+        if (!IsControlChar(rTxt[i]))
+            aBuf[j++] = rTxt[i];
 
+    rTxt = aBuf.makeStringAndClear();
+    DBG_ASSERT(rTxt.getLength() == nSize, "GetNumControlChars returned a different number of control characters than were actually removed.");
+
+    return true;
+}
 
 // non breaking field character
 #define CH_TXTATR_INWORD    ((sal_Char) 0x02)