MCOL-1559 experimental

dbcon/execplan/predicateoperator.h

@@ -114,6 +114,8 @@ private:
     template <typename result_t>
     inline bool numericCompare(result_t op1, result_t op2);
     inline bool strCompare(const std::string& op1, const std::string& op2);
+    // MCOL-1559
+    inline bool strTrimCompare(const std::string& op1, const std::string& op2);
 };
 
 inline bool PredicateOperator::getBoolVal(rowgroup::Row& row, bool& isNull, ReturnedColumn* lop, ReturnedColumn* rop)
@@ -457,19 +459,15 @@ inline bool PredicateOperator::getBoolVal(rowgroup::Row& row, bool& isNull, Retu
                 return !ret;
             }
 
-            // MCOL-1559
-	    std::string val1 = lop->getStrVal(row, isNull);
             if (isNull)
                 return false;
 
-            std::string val2 = rop->getStrVal(row, isNull);
+            const std::string& val1 = lop->getStrVal(row, isNull);
             if (isNull)
                 return false;
 
-            boost::trim_right_if(val1, boost::is_any_of(" "));
+            return strTrimCompare(val1, rop->getStrVal(row, isNull)) && !isNull;
-            boost::trim_right_if(val2, boost::is_any_of(" "));
 
-            return strCompare(val1, val2);
         }
 
         //FIXME: ???
@@ -553,6 +551,37 @@ inline bool PredicateOperator::strCompare(const std::string& op1, const std::str
     }
 }
 
+inline bool PredicateOperator::strTrimCompare(const std::string& op1, const std::string& op2)
+{
+    switch (fOp)
+    {
+        case OP_EQ:
+            return funcexp::utf8::idb_strtrimcoll(op1, op2) == 0;
+
+        case OP_NE:
+            return funcexp::utf8::idb_strtrimcoll(op1, op2) != 0;
+
+        case OP_GT:
+            return funcexp::utf8::idb_strtrimcoll(op1, op2) > 0;
+
+        case OP_GE:
+            return funcexp::utf8::idb_strtrimcoll(op1, op2) >= 0;
+
+        case OP_LT:
+            return funcexp::utf8::idb_strtrimcoll(op1, op2) < 0;
+
+        case OP_LE:
+            return funcexp::utf8::idb_strtrimcoll(op1, op2) <= 0;
+
+        default:
+        {
+            std::ostringstream oss;
+            oss << "Non support predicate operation: " << fOp;
+            throw logging::InvalidOperationExcept(oss.str());
+        }
+    }
+}
+
 std::ostream& operator<<(std::ostream& os, const PredicateOperator& rhs);
 }
 

utils/funcexp/utils_utf8.h

@@ -130,6 +130,81 @@ int idb_strcoll(const char* str1, const char* str2)
         return strcoll(str1, str2);
 }
 
+// MCOL-1559 Add a trimmed version of strcoll
+// We want to compare str1 and str2 ignoring any trailing whitespace
+// without making a copy of the strings (performance hit).
+// I can't find any library that does this while paying attention to
+// Locale. string::compare can be used to compare substrings, but
+// it's a byte by byte compare.
+// We find the last real character, and if a space is following, we
+// temporarily replace it with a NULL, do the compare, and restore the
+// original value to that spot. 
+// WARNING: This is not thread safe. It temporarily modifies the
+// strings and assumes it is free to do so.
+inline
+int idb_strtrimcoll(const std::string& str1, const std::string& str2)
+{
+    const std::string whitespaces (" \t\f\v\n\r");
+    int rtn = 0;
+    char orig1;
+    char orig2;
+    char* s1 = NULL;
+    char* s2 = NULL;
+
+    // Set found1 to the first whitespace char in str1
+    std::size_t found1 = str1.find_last_not_of(whitespaces);
+    if (found1 == std::string::npos) // Either the string is empty or all whitespace.
+    {
+        if (strlen(str1) > 0) // Is all whitespace
+            found1 = 0;       // First whitespace position
+    }
+    else
+    {
+        if (strlen(str1) > found1+1)
+            ++found1; // move to the first whitespace position
+        else
+            found1 = std::string::npos; // No trailing whitespace
+    }
+    // Save the value at found1 and set to NULL
+    if (found1 != std::string::npos)
+    {
+        s1 = &const_cast<std::string&>(str1)[found1];
+        orig1 = *s1;
+        *s1 = 0;
+    }
+
+    // Set found2 to the first whitespace char in str2
+    std::size_t found2 = str2.find_last_not_of(whitespaces);
+    if (found2 == std::string::npos) // Either the string is empty or all whitespace.
+    {
+        if (strlen(str2) > 0) // Is all whitespace
+            found2 = 0;       // First whitespace position
+    }
+    else
+    {
+        if (strlen(str2) > found2+1)
+            ++found2; // move to the first whitespace position
+        else
+            found2 = std::string::npos; // No trailing whitespace
+    }
+    // Save the value at found2 and set to NULL
+    if (found2 != std::string::npos)
+    {
+        s2 = &const_cast<std::string&>(str2)[found2];
+        orig2 = *s2;
+        *s2 = 0;
+    }
+        
+    // Compare the trimmed strings
+    rtn = idb_strcoll(str1.c_str(), str2.c_str());
+
+    // Restore the whitespace
+    if (s1)
+        *s1 = orig1;
+    if (s2)
+        *s2 = orig2;
+    return rtn;
+}
 
 // BUG 5241
 // Infinidb specific mbstowcs(). This will handle both windows and unix platforms