[SYSTEMDS-3857] set get names on dataframes

docs/site/dml-language-reference.md

@@ -2068,10 +2068,12 @@ The following example uses <code>transformapply()</code> with the input matrix a
 
 **Table F5**: Frame processing built-in functions
 
-Function | Description | Parameters | Example
--------- | ----------- | ---------- | -------
-map() | It will execute the given lambda expression on a frame (cell, row or column wise). | Input: (X &lt;frame&gt;, y &lt;String&gt;, \[margin &lt;int&gt;\]) <br/>Output: &lt;frame&gt;. <br/> X is a frame and <br/>y is a String containing the lambda expression to be executed on frame X. <br/> margin - how to apply the lambda expression (0 indicates each cell, 1 - rows, 2 - columns). Output matrix dimensions are always equal to the input. | [map](#map) 
-tokenize() | Transforms a frame to tokenized frame using specification. Tokenization is valid only for string columns. | Input:<br/> target = &lt;frame&gt; <br/> spec = &lt;json specification&gt; <br/> Outputs: &lt;matrix&gt;, &lt;frame&gt; | [tokenize](#tokenize)
+Function | Description                                                                                               | Parameters                                                                                                                                                                                                                                                                                                                                          | Example
+-------- |-----------------------------------------------------------------------------------------------------------|-----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------| -------
+map() | It will execute the given lambda expression on a frame (cell, row or column wise).                        | Input: (X &lt;frame&gt;, y &lt;String&gt;, \[margin &lt;int&gt;\]) <br/>Output: &lt;frame&gt;. <br/> X is a frame and <br/>y is a String containing the lambda expression to be executed on frame X. <br/> margin - how to apply the lambda expression (0 indicates each cell, 1 - rows, 2 - columns). Output matrix dimensions are always equal to the input. | [map](#map) 
+tokenize() | Transforms a frame to tokenized frame using specification. Tokenization is valid only for string columns. | Input:<br/> target = &lt;frame&gt; <br/> spec = &lt;json specification&gt; <br/> Outputs: &lt;matrix&gt;, &lt;frame&gt;                                                                                                                                                                                                                             | [tokenize](#tokenize)
+getNames() | Returns the column names of a frame as a single-row frame.                                                | Input: X &lt;frame&gt; <br/> Output: &lt;frame&gt;                                                                                                                                                                                                                                                                                                  | N = getNames(X)
+setNames() | Sets the column names of a frame from a single-row frame containing string values. | Input:<br/> X = &lt;frame&gt;<br/> N = &lt;frame&gt;<br/> Output:&lt;frame&gt;                                                                                                                                                                                                                                                                      | Y = setNames(X, N)
 
 #### map
 

src/main/java/org/apache/sysds/common/Builtins.java

@@ -154,6 +154,7 @@ public enum Builtins {
 	GARCH("garch", true),
 	GAUSSIAN_CLASSIFIER("gaussianClassifier", true),
 	GET_ACCURACY("getAccuracy", true),
+	GET_NAMES("getNames", false),
 	GET_CATEGORICAL_MASK("getCategoricalMask", false),
 	GLM("glm", true),
 	GLM_PREDICT("glmPredict", true),
@@ -310,6 +311,7 @@ public enum Builtins {
 	SELVARTHRESH("selectByVarThresh", true),
 	SEQ("seq", false),
 	SES("ses", true),
+	SET_NAMES("setNames", false),
 	SYMMETRICDIFFERENCE("symmetricDifference", true),
 	SHAPEXPLAINER("shapExplainer", true),
 	SHERLOCK("sherlock", true),

src/main/java/org/apache/sysds/common/Opcodes.java

@@ -352,6 +352,7 @@ public enum Opcodes {
 	MAPPM("map+*", InstructionType.Binary),
 	MAPMINUSMULT("map-*", InstructionType.Binary),
 	MAPDROPINVALIDLENGTH("mapdropInvalidLength", InstructionType.Binary),
+	SET_COLNAMES("set_colnames", InstructionType.Binary),
 
 	MAPGT("map>", InstructionType.Binary),
 	MAPGE("map>=", InstructionType.Binary),

src/main/java/org/apache/sysds/common/Types.java

@@ -641,7 +641,8 @@ public enum OpOp2 {
 		MINUS1_MULT(false), //1-X*Y
 		GET_CATEGORICAL_MASK(false), // get transformation mask
 		QUANTIZE_COMPRESS(false), //quantization-fused compression
-		UNION_DISTINCT(false);
+		UNION_DISTINCT(false),
+		SET_COLNAMES(false);
 
 		private final boolean _validOuter;
 

src/main/java/org/apache/sysds/parser/BuiltinFunctionExpression.java

@@ -1095,13 +1095,34 @@ else if( getAllExpr().length == 2 ) { //binary
 		case TYPEOF:
 		case DETECTSCHEMA:
 		case COLNAMES:
+		case GET_NAMES:
 			checkNumParameters(1);
 			checkMatrixFrameParam(getFirstExpr());
 			output.setDataType(DataType.FRAME);
 			output.setDimensions(1, id.getDim2());
 			output.setBlocksize (id.getBlocksize());
 			output.setValueType(ValueType.STRING);
 			break;
+		case SET_NAMES:
+			//check if we use 2 parameters (Frame on which nemas are set and vector for names)
+			checkNumParameters(2);
+
+			// check if first paramters is a frame
+			checkMatrixFrameParam(getFirstExpr());
+
+			// check if second paramters is a vector 1xn Frame
+			checkMatrixFrameParam(getSecondExpr());
+
+			//output should be a frame
+			output.setDataType(DataType.FRAME);
+
+
+			checkMatrixFrameParam(getFirstExpr());
+			output.setDataType(DataType.FRAME);
+			output.setDimensions(id.getDim1(), id.getDim2());
+			output.setBlocksize (id.getBlocksize());
+			output.setDataType(DataType.FRAME);
+			break;
 		case CAST_AS_FRAME:
 			// operation as.frame
 			// overloaded to take either one argument or 2 where second is column names

src/main/java/org/apache/sysds/parser/DMLTranslator.java

@@ -2762,6 +2762,34 @@ else if ( in.length == 2 )
 		case TYPEOF:
 		case DET:
 		case DETECTSCHEMA:
+			currBuiltinOp = new UnaryOp(
+					target.getName(),
+					target.getDataType(),
+					target.getValueType(),
+					OpOp1.valueOf(source.getOpCode().name()),
+					expr
+			);
+			break;
+
+		case SET_NAMES:
+			currBuiltinOp = new BinaryOp(
+					target.getName(),
+					target.getDataType(),
+					target.getValueType(),
+					OpOp2.SET_COLNAMES,
+					expr,
+					expr2
+			);
+			break;
+
+		case GET_NAMES:
+			currBuiltinOp = new UnaryOp(
+					target.getName(),
+					target.getDataType(),
+					target.getValueType(),
+					OpOp1.COLNAMES, expr
+			);
+			break;
 		case COLNAMES:
 			currBuiltinOp = new UnaryOp(target.getName(), target.getDataType(),
 				target.getValueType(), OpOp1.valueOf(source.getOpCode().name()), expr);

src/main/java/org/apache/sysds/runtime/instructions/InstructionUtils.java

@@ -686,6 +686,8 @@ else if( opcode.equalsIgnoreCase(Opcodes.VALUESWAP.toString()))
 			return new BinaryOperator(Builtin.getBuiltinFnObject("valueSwap"));
 		else if( opcode.equalsIgnoreCase(Opcodes.FREPLICATE.toString()))
 			return new BinaryOperator(Builtin.getBuiltinFnObject("freplicate"));
+		else if( opcode.equalsIgnoreCase(Opcodes.SET_COLNAMES.toString()))
+			return new BinaryOperator(Builtin.getBuiltinFnObject("set_colnames"));
 
 		throw new RuntimeException("Unknown binary opcode " + opcode);
 	}
@@ -923,6 +925,8 @@ else if ( opcode.equalsIgnoreCase(Opcodes.DROPINVALIDLENGTH.toString()) || opcod
 			return new BinaryOperator(Builtin.getBuiltinFnObject("dropInvalidLength"));
 		else if ( opcode.equalsIgnoreCase(Opcodes.VALUESWAP.toString()) || opcode.equalsIgnoreCase("mapValueSwap") )
 			return new BinaryOperator(Builtin.getBuiltinFnObject("valueSwap"));
+		else if (opcode.equalsIgnoreCase(Opcodes.SET_COLNAMES.toString()))
+			return new BinaryOperator(Builtin.getBuiltinFnObject("set_colnames"));
 
 		throw new DMLRuntimeException("Unknown binary opcode " + opcode);
 	}

src/main/java/org/apache/sysds/runtime/instructions/cp/BinaryFrameFrameCPInstruction.java

@@ -21,6 +21,7 @@
 
 import org.apache.sysds.common.Opcodes;
 import org.apache.sysds.runtime.controlprogram.context.ExecutionContext;
+import org.apache.sysds.runtime.DMLRuntimeException;
 import org.apache.sysds.runtime.frame.data.FrameBlock;
 import org.apache.sysds.runtime.frame.data.lib.FrameLibApplySchema;
 import org.apache.sysds.runtime.matrix.operators.BinaryOperator;
@@ -62,6 +63,40 @@ else if(getOpcode().equals(Opcodes.APPLYSCHEMA.toString())) {
 			final int k = ((MultiThreadedOperator)_optr).getNumThreads();
 			final FrameBlock out = FrameLibApplySchema.applySchema(inBlock1, inBlock2, k);
 			ec.setFrameOutput(output.getName(), out);
+		}
+		else if(getOpcode().equals(Opcodes.SET_COLNAMES.toString())) {
+
+			FrameBlock in = ec.getFrameInput(input1.getName());
+			FrameBlock names = ec.getFrameInput(input2.getName());
+
+			if (names == null)
+				throw new DMLRuntimeException("Column names cannot be null.");
+
+			if (names.getNumRows() != 1)
+				throw new DMLRuntimeException(
+						"Column names must be provided as a 1 x n frame.");
+
+			if (names.getNumColumns() != in.getNumColumns())
+				throw new DMLRuntimeException(
+						"Expected " + in.getNumColumns() +
+								" column names but got " + names.getNumColumns());
+
+			String[] colNames = new String[(int) names.getNumColumns()];
+			for(int i = 0; i < colNames.length; i++){
+				colNames[i] = names.get(0, i).toString();
+			}
+
+			FrameBlock out  = new FrameBlock(in);
+
+			out.setColumnNames(colNames);
+
+			ec.setFrameOutput(output.getName(), out);
+
+			ec.releaseFrameInput(input1.getName());
+
+			ec.releaseFrameInput(input2.getName());
+
+
 		}
 		else {
 			// Execute binary operations

src/main/java/org/apache/sysds/runtime/instructions/cp/UnaryFrameCPInstruction.java

@@ -52,6 +52,7 @@ else if(getOpcode().equals(Opcodes.COLNAMES.toString())) {
 			ec.releaseFrameInput(input1.getName());
 			ec.setFrameOutput(output.getName(), retBlock);
 		}
+
 		else
 			throw new DMLScriptException("Opcode '" + getOpcode() + "' is not a valid UnaryFrameCPInstruction");
 	}