svn commit: r1488584 - in /pig/trunk: CHANGES.txt src/org/apache/pig/builtin/IN.java src/org/apache/pig/parser/LogicalPlanGenerator.g src/org/apache/pig/parser/QueryParser.g test/org/apache/pig/builtin/TestInUdf.java

[ch...@apache.org]

Author: cheolsoo
Date: Sat Jun  1 20:03:58 2013
New Revision: 1488584

URL: http://svn.apache.org/r1488584
Log:
PIG-3336: Change IN operator to use or-expressions instead of EvalFunc (cheolsoo)

Removed:
    pig/trunk/src/org/apache/pig/builtin/IN.java
    pig/trunk/test/org/apache/pig/builtin/TestInUdf.java
Modified:
    pig/trunk/CHANGES.txt
    pig/trunk/src/org/apache/pig/parser/LogicalPlanGenerator.g
    pig/trunk/src/org/apache/pig/parser/QueryParser.g

Modified: pig/trunk/CHANGES.txt
URL: http://svn.apache.org/viewvc/pig/trunk/CHANGES.txt?rev=1488584&r1=1488583&r2=1488584&view=diff
==============================================================================
--- pig/trunk/CHANGES.txt (original)
+++ pig/trunk/CHANGES.txt Sat Jun  1 20:03:58 2013
@@ -28,9 +28,7 @@ PIG-3174:  Remove rpm and deb artifacts 
 
 IMPROVEMENTS
 
-PIG-3335: TestErrorHandling.tesNegative7 fails on MR2 (xuefuz)
-
-PIG-3316: Pig failed to interpret DateTime values in some special cases (xuefuz)
+PIG-3336: Change IN operator to use or-expressions instead of EvalFunc (cheolsoo)
 
 PIG-3339: Move pattern compilation in ToDate as a static variable (rohini)
 
@@ -192,6 +190,10 @@ PIG-3013: BinInterSedes improve chararra
 
 BUG FIXES
 
+PIG-3335: TestErrorHandling.tesNegative7 fails on MR2 (xuefuz)
+
+PIG-3316: Pig failed to interpret DateTime values in some special cases (xuefuz)
+
 PIG-2956: Invalid cache specification for some streaming statement (daijy)
 
 PIG-3310: ImplicitSplitInserter does not generate new uids for nested schema fields, leading to miscomputations (cstenac via daijy)

Modified: pig/trunk/src/org/apache/pig/parser/LogicalPlanGenerator.g
URL: http://svn.apache.org/viewvc/pig/trunk/src/org/apache/pig/parser/LogicalPlanGenerator.g?rev=1488584&r1=1488583&r2=1488584&view=diff
==============================================================================
--- pig/trunk/src/org/apache/pig/parser/LogicalPlanGenerator.g (original)
+++ pig/trunk/src/org/apache/pig/parser/LogicalPlanGenerator.g Sat Jun  1 20:03:58 2013
@@ -783,13 +783,27 @@ cond[LogicalExpressionPlan exprPlan] ret
 
 in_eval[LogicalExpressionPlan plan] returns[LogicalExpression expr]
 @init {
-    List<LogicalExpression> args = new ArrayList<LogicalExpression>();
+    List<LogicalExpression> exprs = new ArrayList<LogicalExpression>();
 }
- : ^( IN exp1 = expr[$plan] { args.add( $exp1.expr ); } ( exp2 = expr[$plan] { args.add( $exp2.expr ); } )+ )
-   {
-       SourceLocation loc = new SourceLocation( (PigParserNode)$IN );
-       $expr = builder.buildUDF( loc, $plan, "IN", args );
-   }
+ : ^( IN ( expr[$plan] { exprs.add($expr.expr); } )+ )
+    {
+        // Convert IN tree to nested or expressions. Please also see
+        // QueryParser.g for how IN tree is constructed from IN expression.
+        EqualExpression firstBoolExpr = new EqualExpression(plan, exprs.get(0), exprs.get(1));
+        if (exprs.size() == 2) {
+            $expr = firstBoolExpr;
+        } else {
+            OrExpression currOrExpr = null;
+            OrExpression prevOrExpr = null;
+            for (int i = 2; i < exprs.size(); i = i + 2) {
+                EqualExpression boolExpr = new EqualExpression(plan, exprs.get(i), exprs.get(i+1));
+                currOrExpr = new OrExpression( $plan, prevOrExpr == null ? firstBoolExpr : prevOrExpr, boolExpr );
+                prevOrExpr = currOrExpr;
+            }
+            $expr = currOrExpr;
+        }
+        $expr.setLocation( new SourceLocation( (PigParserNode)$in_eval.start ) );
+    }
 ;
 
 func_eval[LogicalExpressionPlan plan] returns[LogicalExpression expr]

Modified: pig/trunk/src/org/apache/pig/parser/QueryParser.g
URL: http://svn.apache.org/viewvc/pig/trunk/src/org/apache/pig/parser/QueryParser.g?rev=1488584&r1=1488583&r2=1488584&view=diff
==============================================================================
--- pig/trunk/src/org/apache/pig/parser/QueryParser.g (original)
+++ pig/trunk/src/org/apache/pig/parser/QueryParser.g Sat Jun  1 20:03:58 2013
@@ -625,17 +625,36 @@ unary_cond
         // brackets, and otherwise we'll assume its an "expr" (and so
         // we'll have to strip off the BOOL_COND token the "cast_expr"
         // rule added)
-        Tree tree = (Tree)retval.getTree();
+        BaseTree tree = (BaseTree) retval.getTree();
         if(tree.getType() == BOOL_COND
         && tree.getChild(0).getType() == EXPR_IN_PAREN
         && BOOLEAN_TOKENS.contains(tree.getChild(0).getChild(0).getType())) {
             retval.tree = tree.getChild(0).getChild(0);
             adaptor.setTokenBoundaries(retval.tree, retval.start, retval.stop);
         }
+
+        // For IN expression, we clone the lhs expression (1st child of the
+        // returned tree) and insert it before every rhs expression. For example,
+        //
+        //   lhs IN (rhs1, rhs2, rhs3)
+        // =>
+        //   ^( IN lhs, rhs1, lhs, rhs2, lhs, rhs3 )
+        //
+        // Note that lhs appears three times at index 0, 2 and 4.
+        //
+        // This is needed because in LogicalPlanGenerator.g, we translate this
+        // tree to nested or expressions, and we need to construct a new
+        // LogicalExpression object per rhs expression.
+        if(tree.getType() == IN) {
+            Tree lhs = tree.getChild(0);
+            for(int i = 2; i < tree.getChildCount(); i = i + 2) {
+                tree.insertChild(i, deepCopy(lhs));
+            }
+        }
     }
     : exp1 = expr
         ( ( IS NOT? NULL -> ^( NULL $exp1 NOT? ) )
-        | ( IN LEFT_PAREN ( expr ( COMMA expr )* ) RIGHT_PAREN -> ^( IN $exp1 expr+ ) )
+        | ( IN LEFT_PAREN ( expr ( COMMA expr )* ) RIGHT_PAREN -> ^( IN expr+ ) )
         | ( rel_op exp2 = expr -> ^( rel_op $exp1 $exp2 ) )
         | ( -> ^(BOOL_COND expr) ) )
 ;