positional column denomiation foreach

elisehuard · Dec 17, 2013 · bef2360 · bef2360
1 parent fa88129
commit bef2360
Show file tree

Hide file tree

Showing 3 changed files with 16 additions and 1 deletion.
diff --git a/src/Language/Pig/Parser/AST.hs b/src/Language/Pig/Parser/AST.hs
@@ -44,8 +44,11 @@ data Transform = Flatten String Tuple -- foreach flatten transform
                | FunctionTransform Function Alias
                | EnvTransform Scalar Alias
                | IdentityTransform Alias
+               | PositionalTypeTransform SimpleType Pos Alias
                DERIVE
 
+type Pos = Integer
+
 data Join = Join String String
             DERIVE
 

diff --git a/src/Language/Pig/Parser/Parser.hs b/src/Language/Pig/Parser/Parser.hs
@@ -49,8 +49,8 @@ lexer = Token.makeTokenParser pigLanguageDef
 identifier = Token.identifier lexer
 reserved = Token.reserved lexer
 reservedOp = Token.reservedOp lexer
-integer = Token.integer lexer
 naturalOrFloat = Token.naturalOrFloat lexer
+integer = Token.integer lexer
 semi = Token.semi lexer
 comma = Token.comma lexer
 whiteSpace = Token.whiteSpace lexer
@@ -256,6 +256,7 @@ transform :: Parser Transform
 transform = try(aliasTransform)
          <|> flattenTransform
          <|> tupleFieldGlob
+         <|> try(positionalTypeTransform)
          <|> expressionTransform
          <|> try(functionTransform)
          <|> identityTransform
@@ -297,6 +298,14 @@ identityTransform :: Parser Transform
 identityTransform = IdentityTransform <$>
                       (Identifier <$> pigIdentifier)
 
+positionalTypeTransform :: Parser Transform
+positionalTypeTransform = PositionalTypeTransform <$>
+                             (parens pigType) <*>
+                             (char '$' *>
+                                integer) <*>
+                             (reserved "AS" *>
+                                (Identifier <$> identifier))
+
 -- general expression:
 -- fieldExpression or literal or function or binary operation (+-*/%) or bincond (?:)
 -- bincond: boolean expression (==, !=, >, <, >=, <=) (and, or, not)

diff --git a/test/Language/Pig/Parser/Test.hs b/test/Language/Pig/Parser/Test.hs
@@ -44,6 +44,9 @@ desktop_client_dates = FOREACH desktop_client GENERATE server_date AS server_dat
    , testCase "foreach stmt with quoted string" (testStmt "report = FOREACH report GENERATE '$date' AS date, *;"
                                                           "Seq [Assignment (Identifier \"report\") (ForeachClause (Identifier \"report\") (GenBlock [EnvTransform (String \"$date\") (Identifier \"date\"),TupleFieldGlob]))]")
 
+   , testCase "foreach stmt with field positions" (testStmt "users = FOREACH users GENERATE (long)$0 AS user_id, (chararray)$7 AS registration_date;"
+                                                            "Seq [Assignment (Identifier \"users\") (ForeachClause (Identifier \"users\") (GenBlock [PositionalTypeTransform Long 0 (Identifier \"user_id\"),PositionalTypeTransform CharArray 7 (Identifier \"registration_date\")]))]")
+
    , testCase "join stmt" (testStmt "active_users = JOIN users BY user_id, active_users BY user_id;"
                                     "Seq [Assignment (Identifier \"active_users\") (InnerJoinClause [Join \"users\" \"user_id\",Join \"active_users\" \"user_id\"])]")