Fix display of non-numeric DataFrame index columns

news/2 Fixes/5253.md

@@ -0,0 +1 @@
+Fix data viewer display of non-numeric index columns in DataFrames.

pythonFiles/vscode_datascience_helpers/dataframes/vscodeDataFrame.py

@@ -170,6 +170,17 @@ def _VSCODE_getDataFrameInfo(df):
 
     columnTypes = _VSCODE_builtins.list(df.dtypes)
 
+    # Compute the index column. It may have been renamed
+    try:
+        indexColumn = df.index.name if df.index.name else "index"
+    except AttributeError:
+        indexColumn = "index"
+
+    # Make sure the index column exists
+    if indexColumn not in columnNames:
+        columnNames.insert(0, indexColumn)
+        columnTypes.insert(0, "int64")
+
     # Then loop and generate our output json
     columns = []
     for n in _VSCODE_builtins.range(0, _VSCODE_builtins.len(columnNames)):
@@ -184,6 +195,7 @@ def _VSCODE_getDataFrameInfo(df):
     # Save this in our target
     target = {}
     target["columns"] = columns
+    target["indexColumn"] = indexColumn
     target["rowCount"] = rowCount
 
     # return our json object as a string

src/datascience-ui/data-explorer/mainPanel.tsx

@@ -32,11 +32,13 @@ import '../react-common/codicon/codicon.css';
 import '../react-common/seti/seti.less';
 import { SliceControl } from './sliceControl';
 import { debounce } from 'lodash';
+import * as uuid from 'uuid/v4';
 
 import { initializeIcons } from '@fluentui/react';
 initializeIcons(); // Register all FluentUI icons being used to prevent developer console errors
 
 const SliceableTypes: Set<string> = new Set<string>(['ndarray', 'Tensor', 'EagerTensor']);
+const RowNumberColumnName = uuid(); // Unique key for our column containing row numbers
 
 // Our css has to come after in order to override body styles
 export interface IMainPanelProps {
@@ -257,7 +259,7 @@ export class MainPanel extends React.Component<IMainPanelProps, IMainPanelState>
             <ReactSlickGrid
                 ref={this.grid}
                 columns={this.state.gridColumns}
-                idProperty={this.state.indexColumn}
+                idProperty={RowNumberColumnName}
                 rowsAdded={this.gridAddEvent}
                 resetGridEvent={this.resetGridEvent}
                 columnsUpdated={this.gridColumnUpdateEvent}
@@ -368,7 +370,7 @@ export class MainPanel extends React.Component<IMainPanelProps, IMainPanelState>
             gridRows: newActual
         });
 
-        // Tell our grid about the new ros
+        // Tell our grid about the new rows
         this.updateRows(normalized);
 
         // Get the next chunk
@@ -384,23 +386,34 @@ export class MainPanel extends React.Component<IMainPanelProps, IMainPanelState>
     }
 
     private generateColumns(variable: IDataFrameInfo): Slick.Column<Slick.SlickData>[] {
-        // Generate an index column
-        const indexColumn = {
-            key: this.state.indexColumn,
-            type: ColumnType.Number
-        };
         if (variable.columns) {
-            const columns = [indexColumn].concat(variable.columns);
-            return columns.map((c: { key: string; type: ColumnType }, i: number) => {
-                return {
-                    type: c.type,
-                    field: c.key.toString(),
-                    id: `${i}`,
-                    name: c.key.toString(),
-                    sortable: true,
-                    formatter: cellFormatterFunc
-                };
-            });
+            // Generate a column for row numbers
+            const rowNumberColumn = {
+                key: RowNumberColumnName,
+                type: ColumnType.Number
+            };
+            const columns = [rowNumberColumn].concat(variable.columns);
+            return columns.reduce(
+                (accum: Slick.Column<Slick.SlickData>[], c: { key: string; type: ColumnType }, i: number) => {
+                    // Only show index column for pandas DataFrame and Series
+                    if (
+                        variable?.type === 'DataFrame' ||
+                        variable?.type === 'Series' ||
+                        c.key !== this.state.indexColumn
+                    ) {
+                        accum.push({
+                            type: c.type,
+                            field: c.key.toString(),
+                            id: `${i}`,
+                            name: c.key.toString(),
+                            sortable: true,
+                            formatter: cellFormatterFunc
+                        } as Slick.Column<Slick.SlickData>);
+                    }
+                    return accum;
+                },
+                []
+            );
         }
         return [];
     }
@@ -416,7 +429,7 @@ export class MainPanel extends React.Component<IMainPanelProps, IMainPanelState>
             if (!r) {
                 r = {};
             }
-            r[this.state.indexColumn] = this.state.fetchedRowCount + idx;
+            r[RowNumberColumnName] = this.state.fetchedRowCount + idx;
             for (let [key, value] of Object.entries(r)) {
                 switch (value) {
                     case 'nan':

src/datascience-ui/data-explorer/reactSlickGrid.tsx

@@ -473,7 +473,7 @@ export class ReactSlickGrid extends React.Component<ISlickGridProps, ISlickGridS
             const placeholder = '99999999999';
             const maxFieldWidth = measureText(placeholder, fontString);
             columns.forEach((c) => {
-                if (c.id !== '0') {
+                if (c.field !== this.props.idProperty) {
                     c.width = maxFieldWidth;
                 } else {
                     c.width = maxFieldWidth / 2;
@@ -520,7 +520,6 @@ export class ReactSlickGrid extends React.Component<ISlickGridProps, ISlickGridS
         this.dataView.setItems([]);
         const styledColumns = this.styleColumns(data.columns);
         this.setColumns(styledColumns);
-        this.autoResizeColumns();
     };
 
     private updateColumns = (_e: Slick.EventData, newColumns: Slick.Column<Slick.SlickData>[]) => {
@@ -535,6 +534,7 @@ export class ReactSlickGrid extends React.Component<ISlickGridProps, ISlickGridS
         // The solution is to force the header row to become visible just before sending our slice request.
         this.state.grid?.setHeaderRowVisibility(true);
         this.state.grid?.setColumns(newColumns);
+        this.autoResizeColumns();
     };
 
     private addedRows = (_e: Slick.EventData, data: ISlickGridAdd) => {
@@ -573,7 +573,7 @@ export class ReactSlickGrid extends React.Component<ISlickGridProps, ISlickGridS
     }
 
     private renderFilterCell = (_e: Slick.EventData, args: Slick.OnHeaderRowCellRenderedEventArgs<Slick.SlickData>) => {
-        if (args.column.id === '0') {
+        if (args.column.field === this.props.idProperty) {
             const tooltipText = getLocString('DataScience.clearFilters', 'Clear all filters');
             ReactDOM.render(
                 <div

src/test/datascience/dataviewer.functional.test.tsx

@@ -284,7 +284,19 @@ suite('DataScience DataViewer tests', () => {
         assert.ok(dv, 'DataViewer not created');
         await gotAllRows;
 
-        verifyRows(wrapper.wrapper, [0, 0, 1, 1, 2, 2, 3, 3]);
+        verifyRows(wrapper.wrapper, [0, 0, 0, 1, 1, 1, 2, 2, 2, 3, 3, 3]);
+    });
+
+    runMountedTest('Transposed Data Frame', async (wrapper) => {
+        await injectCode(
+            'import pandas as pd\r\ndata = [["tom", 10], ["nick", 15], ["juli", 14]]\r\ndf = pd.DataFrame(data, columns=["Name", "Age"])\r\ndf = df.transpose()'
+        );
+        const gotAllRows = getCompletedPromise(wrapper);
+        const dv = await createJupyterVariableDataViewer('df', 'DataFrame');
+        assert.ok(dv, 'DataViewer not created');
+        await gotAllRows;
+
+        verifyRows(wrapper.wrapper, [0, 'Name', 'tom', 'nick', 'juli', 1, 'Age', '10', '15', '14']);
     });
 
     runMountedTest('List', async (wrapper) => {
@@ -304,7 +316,7 @@ suite('DataScience DataViewer tests', () => {
         assert.ok(dv, 'DataViewer not created');
         await gotAllRows;
 
-        verifyRows(wrapper.wrapper, [0, 0, 1, 1, 2, 2, 3, 3]);
+        verifyRows(wrapper.wrapper, [0, 0, 0, 1, 1, 1, 2, 2, 2, 3, 3, 3]);
     });
 
     runMountedTest('np.array', async (wrapper) => {