Merge pull request #1290 from claussni/cross-func-split

Extend cross() function to support multiple-value-cell-input

main/src/com/google/refine/InterProjectModel.java

@@ -68,22 +68,38 @@ public class InterProjectModel {
             this.toProjectID = toProjectID;
             this.toProjectColumnName = toProjectColumnName;
         }
-        
+
-        public HasFieldsListImpl getRows(Object value) {
+        public HasFieldsListImpl getRows(final Object rowKey, String separatorRegexp) {
-            if (ExpressionUtils.isNonBlankData(value) && valueToRowIndices.containsKey(value)) {
+            Project toProject = ProjectManager.singleton.getProject(toProjectID);
-                Project toProject = ProjectManager.singleton.getProject(toProjectID);
+            if (toProject == null) {
-                if (toProject != null) {
+                return null;
-                    HasFieldsListImpl rows = new HasFieldsListImpl();
+            }
-                    for (Integer r : valueToRowIndices.get(value)) {
+
-                        Row row = toProject.rows.get(r);
+            HasFieldsListImpl resultFieldList = null;
-                        rows.add(new WrappedRow(toProject, r, row));
+
+            if (ExpressionUtils.isNonBlankData(rowKey)) {
+                Object[] rowKeys;
+                if (separatorRegexp != null && !separatorRegexp.isEmpty() && rowKey instanceof String) {
+                    rowKeys = ((String) rowKey).split(separatorRegexp);
+                } else {
+                    rowKeys = new Object[]{rowKey};
+                }
+
+                resultFieldList = new HasFieldsListImpl();
+                for (Object k : rowKeys) {
+                    if (valueToRowIndices.containsKey(k)) {
+                        for (Integer rowIndex : valueToRowIndices.get(k)) {
+                            Row row = toProject.rows.get(rowIndex);
+                            resultFieldList.add(new WrappedRow(toProject, rowIndex, row));
+                        }
                     }
-                    
-                    return rows;
                 }
             }
-            return null;
+
+            // Returning null instead of an empty list is expected
+            return resultFieldList.isEmpty() ? null : resultFieldList;
         }
+
     }
     
     protected Map<String, ProjectJoin> _joins = new HashMap<String, ProjectJoin>();
@@ -95,9 +111,10 @@ public class InterProjectModel {
      * @param fromColumn
      * @param toProject
      * @param toColumn
+     * @param separatorRegexp
      * @return
      */
-    public ProjectJoin getJoin(String fromProject, String fromColumn, String toProject, String toColumn) {
+    public ProjectJoin getJoin(String fromProject, String fromColumn, String toProject, String toColumn, String separatorRegexp) {
         String key = fromProject + ";" + fromColumn + ";" + toProject + ";" + toColumn;
         if (!_joins.containsKey(key)) {
             ProjectJoin join = new ProjectJoin(
@@ -106,8 +123,8 @@ public class InterProjectModel {
                 ProjectManager.singleton.getProjectID(toProject), 
                 toColumn
             );
-            
+
-            computeJoin(join);
+            computeJoin(join, separatorRegexp);
             
             synchronized (_joins) {
                 _joins.put(key, join);
@@ -142,7 +159,7 @@ public class InterProjectModel {
         }
     }
 
-    protected void computeJoin(ProjectJoin join) {
+    protected void computeJoin(ProjectJoin join, String separatorRegexp) {
         if (join.fromProjectID < 0 || join.toProjectID < 0) {
             return;
         }
@@ -158,11 +175,21 @@ public class InterProjectModel {
         if (fromColumn == null || toColumn == null) {
             return;
         }
-        
+
         for (Row fromRow : fromProject.rows) {
-            Object value = fromRow.getCellValue(fromColumn.getCellIndex());
+            Object fromRowKey = fromRow.getCellValue(fromColumn.getCellIndex());
-            if (ExpressionUtils.isNonBlankData(value) && !join.valueToRowIndices.containsKey(value)) {
+            if (ExpressionUtils.isNonBlankData(fromRowKey)) {
-                join.valueToRowIndices.put(value, new ArrayList<Integer>());
+                Object[] fromRowKeys;
+                if (separatorRegexp != null && !separatorRegexp.isEmpty() && fromRowKey instanceof String) {
+                    fromRowKeys = ((String) fromRowKey).split(separatorRegexp);
+                } else {
+                    fromRowKeys = new Object[]{fromRowKey};
+                }
+                for (Object k : fromRowKeys) {
+                    if (!join.valueToRowIndices.containsKey(k)) {
+                        join.valueToRowIndices.put(k, new ArrayList<Integer>());
+                    }
+                }
             }
         }
         
@@ -176,4 +203,5 @@ public class InterProjectModel {
             }
         }
     }
+
 }

main/src/com/google/refine/expr/functions/Cross.java

@@ -47,33 +47,39 @@ import com.google.refine.grel.Function;
 import com.google.refine.model.Project;
 
 public class Cross implements Function {
-    
+
+    public static final String EVAL_ERROR_MESSAGE =
+            " expects a string or cell, a project name to join with, and a column name in that project. " +
+                    "Optional accepts a regular expression separator for source values.";
+
     @Override
     public Object call(Properties bindings, Object[] args) {
-        if (args.length == 3) {
+        if (args.length >= 3) {
             // 1st argument can take either value or cell(for backward compatibility)
             Object v = args[0];
             Object toProjectName = args[1];
             Object toColumnName = args[2];
-            
+            String separatorRegexp = (args.length > 3) ? String.valueOf(args[3]) : null;
-            if (v != null && 
+
+            if (v != null &&
                 ( v instanceof String || v instanceof WrappedCell ) &&
                 toProjectName != null && toProjectName instanceof String &&
                 toColumnName != null && toColumnName instanceof String) {
-                
+
                 ProjectJoin join = ProjectManager.singleton.getInterProjectModel().getJoin(
                         ProjectManager.singleton.getProjectMetadata(((Project) bindings.get("project")).id).getName(),
                         (String) bindings.get("columnName"), 
-                        (String) toProjectName, 
+                        (String) toProjectName,
-                        (String) toColumnName
+                        (String) toColumnName,
-                        );
+                        separatorRegexp
+                );
                 
                 String srcValue = v instanceof String ? (String)v : (String)((WrappedCell) v).cell.value;
-                        
+
-                return join.getRows(srcValue);
+                return join.getRows(srcValue, separatorRegexp);
             }
         }
-        return new EvalError(ControlFunctionRegistry.getFunctionName(this) + " expects a string or cell, a project name to join with, and a column name in that project");
+        return new EvalError(ControlFunctionRegistry.getFunctionName(this) + EVAL_ERROR_MESSAGE);
     }
     
     @Override
@@ -82,7 +88,7 @@ public class Cross implements Function {
     
         writer.object();
         writer.key("description"); writer.value("join with another project by column");
-        writer.key("params"); writer.value("cell c or string value, string projectName, string columnName");
+        writer.key("params"); writer.value("cell c or string value, string projectName, string columnName(, string separatorRegexp)");
         writer.key("returns"); writer.value("array");
         writer.endObject();
     }

main/tests/server/src/com/google/refine/tests/expr/functions/CrossFunctionTests.java

@@ -9,6 +9,7 @@ import java.util.Calendar;
 import java.util.List;
 import java.util.Properties;
 
+import com.google.refine.expr.functions.Cross;
 import org.json.JSONObject;
 import org.slf4j.LoggerFactory;
 import org.testng.Assert;
@@ -163,17 +164,34 @@ public class CrossFunctionTests extends RefineTest {
      *  rest of cells shows "Error: cross expects a string or cell, a project name to join with, and a column name in that project"
      */
     @Test
-    public void crossFunctionNonLiteralValue() throws Exception {
+    public void crossFunctionIntegerValue() throws Exception {
-        Assert.assertEquals(((EvalError) invoke("cross", 1, "My Address Book", "friend")).message, 
+        String message = ((EvalError) invoke("cross", 1, "My Address Book", "friend")).message;
-                "cross expects a string or cell, a project name to join with, and a column name in that project");
+        Assert.assertTrue(message.contains(Cross.EVAL_ERROR_MESSAGE),
-        
+                String.format("Message should contain `%s` but is `%s`", Cross.EVAL_ERROR_MESSAGE, message));
-        Assert.assertEquals(((EvalError) invoke("cross", null, "My Address Book", "friend")).message, 
-                "cross expects a string or cell, a project name to join with, and a column name in that project");
-        
-        Assert.assertEquals(((EvalError) invoke("cross", Calendar.getInstance(), "My Address Book", "friend")).message, 
-                "cross expects a string or cell, a project name to join with, and a column name in that project");
     }
-    
+
+    /**
+     * rest of cells shows "Error: cross expects a string or cell, a project name to join with, and a column name in
+     * that project"
+     */
+    @Test
+    public void crossFunctionNull() throws Exception {
+        String message = ((EvalError) invoke("cross", null, "My Address Book", "friend")).message;
+        Assert.assertTrue(message.contains(Cross.EVAL_ERROR_MESSAGE),
+                String.format("Message should contain `%s` but is `%s`", Cross.EVAL_ERROR_MESSAGE, message));
+    }
+
+    /**
+     * rest of cells shows "Error: cross expects a string or cell, a project name to join with, and a column name in
+     * that project"
+     */
+    @Test
+    public void crossFunctionCalendarInstance() throws Exception {
+        String message = ((EvalError) invoke("cross", Calendar.getInstance(), "My Address Book", "friend")).message;
+        Assert.assertTrue(message.contains(Cross.EVAL_ERROR_MESSAGE),
+                String.format("Message should contain `%s` but is `%s`", Cross.EVAL_ERROR_MESSAGE, message));
+    }
+
     /**
      * Lookup a control function by name and invoke it with a variable number of args
      */