Compare commits

..
72 changed files with 512 additions and 2584 deletions
+1 -1
View File
@@ -48,7 +48,7 @@ jobs:
python-version: "3.11"
- name: Install uv
uses: astral-sh/setup-uv@ae62891fec2bb8e7d6c99fc78c9fec3a63790f8d # v10.0.0
uses: astral-sh/setup-uv@c771a70e6277c0a99b617c7a806ffedaca235ff9 # v9.0.0
with:
python-version: "3.11"
enable-cache: true
+2 -2
View File
@@ -67,7 +67,7 @@ jobs:
# Initializes the CodeQL tools for scanning.
- name: Initialize CodeQL
uses: github/codeql-action/init@ff2f1c621b7f889edc0d3c761ac2e6a3f8cdb0dd # v4.37.7
uses: github/codeql-action/init@5595ccaf912efad79be6eef63a5619ff05969be3 # v4.37.6
with:
languages: ${{ matrix.language }}
# If you wish to specify custom queries, you can do so here or in a config file.
@@ -78,6 +78,6 @@ jobs:
# queries: security-extended,security-and-quality
- name: Perform CodeQL Analysis
uses: github/codeql-action/analyze@ff2f1c621b7f889edc0d3c761ac2e6a3f8cdb0dd # v4.37.7
uses: github/codeql-action/analyze@5595ccaf912efad79be6eef63a5619ff05969be3 # v4.37.6
with:
category: "/language:${{matrix.language}}"
+4 -4
View File
@@ -64,8 +64,8 @@
"antd": "^6.6.0",
"baseline-browser-mapping": "^2.11.13",
"caniuse-lite": "^1.0.30001809",
"docusaurus-plugin-openapi-docs": "^5.2.0",
"docusaurus-theme-openapi-docs": "^5.2.0",
"docusaurus-plugin-openapi-docs": "^5.1.3",
"docusaurus-theme-openapi-docs": "^5.1.3",
"js-yaml": "^5.2.3",
"json-bigint": "^1.0.0",
"prism-react-renderer": "^2.4.1",
@@ -78,7 +78,7 @@
"remark-import-partial": "^0.0.2",
"reselect": "^5.2.0",
"storybook": "^10.5.7",
"swagger-ui-react": "^5.32.13",
"swagger-ui-react": "^5.32.12",
"swc-loader": "^0.2.7",
"tinycolor2": "^1.4.2",
"unist-util-visit": "^5.1.0"
@@ -93,7 +93,7 @@
"@typescript-eslint/parser": "^8.67.0",
"eslint": "^9.39.2",
"eslint-plugin-react": "^7.37.5",
"globals": "^17.11.0",
"globals": "^17.9.0",
"oxfmt": "^0.63.0",
"typescript": "~6.0.3",
"typescript-eslint": "^8.67.0",
+23 -23
View File
@@ -8014,10 +8014,10 @@ doctrine@^2.1.0:
dependencies:
esutils "^2.0.2"
docusaurus-plugin-openapi-docs@^5.2.0:
version "5.2.0"
resolved "https://registry.yarnpkg.com/docusaurus-plugin-openapi-docs/-/docusaurus-plugin-openapi-docs-5.2.0.tgz#8318ec90cd21fed023be57696211af7d72fd81db"
integrity sha512-MjrfRAMB64uvdxRVz6L9AXWe4QFjCdoBAzYs306yyI3nnXHsFj2lv2FnLA90JV9CAUZaGiYMvvkzBo2Nrkq/9w==
docusaurus-plugin-openapi-docs@^5.1.3:
version "5.1.3"
resolved "https://registry.yarnpkg.com/docusaurus-plugin-openapi-docs/-/docusaurus-plugin-openapi-docs-5.1.3.tgz#b8cd5f8451aaf881deb1a744a8295685f1681865"
integrity sha512-HnpblSBdXoR39VNTIW9zWERUsMJxXOpvdQoBKyaTkUBPwCM48Z76+ndo2yO2vADq+EhWjJlfxL1DUzCrgNjThQ==
dependencies:
"@apidevtools/json-schema-ref-parser" "^15.3.3"
"@redocly/openapi-core" "^2.25.2"
@@ -8035,10 +8035,10 @@ docusaurus-plugin-openapi-docs@^5.2.0:
swagger2openapi "^7.0.8"
xml-formatter "^3.6.6"
docusaurus-theme-openapi-docs@^5.2.0:
version "5.2.0"
resolved "https://registry.yarnpkg.com/docusaurus-theme-openapi-docs/-/docusaurus-theme-openapi-docs-5.2.0.tgz#6d93a74e2e3cf0ae77d24e1c4144bd2e74a52115"
integrity sha512-L0b80LzaMUfr76a9EQXRPCf8nxkEz8Xo6Aknnke1UeE2oXsgoiVki6U+RTE7GmJRjO8zSNKXyckGmGmqqWuHeA==
docusaurus-theme-openapi-docs@^5.1.3:
version "5.1.3"
resolved "https://registry.yarnpkg.com/docusaurus-theme-openapi-docs/-/docusaurus-theme-openapi-docs-5.1.3.tgz#e23644a63785352abbc76e42760c0dfdff3669e1"
integrity sha512-npbD1QahtjAEmrOet/86i5fTmcJX4/rPhVT+c0qKjm7StUNbyqjwchSVBQuU1rB69T51JOA9TpT/y6QcB9Xjvw==
dependencies:
"@hookform/error-message" "^2.0.1"
"@reduxjs/toolkit" "^2.8.2"
@@ -8123,7 +8123,7 @@ domhandler@^5.0.2, domhandler@^5.0.3:
dependencies:
domelementtype "^2.3.0"
dompurify@^3.3.3, dompurify@^3.4.13:
dompurify@^3.3.3, dompurify@^3.4.12:
version "3.4.13"
resolved "https://registry.yarnpkg.com/dompurify/-/dompurify-3.4.13.tgz#fc28949d59f92d62e28a3a764bcbeee35897a1be"
integrity sha512-2vmYIoqjze2d+kakP8S/nS5shfsl587kzwEjcGlTdiksUVgFHnFCsLYDVj/JNqJVOQZGSYBTmuycv0PodwmnMQ==
@@ -9174,10 +9174,10 @@ globals@^14.0.0:
resolved "https://registry.yarnpkg.com/globals/-/globals-14.0.0.tgz#898d7413c29babcf6bafe56fcadded858ada724e"
integrity sha512-oahGvuMGQlPw/ivIYBjVSrWAfWLBeku5tpPE2fOPLi+WHffIWbuh2tCjhyQhTBPMf5E9jDEH4FOmTYgYwbKwtQ==
globals@^17.11.0:
version "17.11.0"
resolved "https://registry.yarnpkg.com/globals/-/globals-17.11.0.tgz#d643485bb30220d7751e511cf4f68c73d3870d87"
integrity sha512-Z2I8hM+PbJDXQDq3Icgpzv+mPdwr68iZUU9d5WW4FuXfDUQfkZaZuvjMv42/5crNyw154+9+VWXbYrUgDXbxNw==
globals@^17.9.0:
version "17.9.0"
resolved "https://registry.yarnpkg.com/globals/-/globals-17.9.0.tgz#e43f252d6bbe71508da43902a1709c8895a59f70"
integrity sha512-m/MvAW61QVU5VDNF1Vj8axt016h8w7L5TU1e9zlab7XIttAT2YAlCwl75K1fOqvMM9apmD7lbCIRhpfkhmxhCg==
globalthis@^1.0.4:
version "1.0.4"
@@ -10284,10 +10284,10 @@ js-levenshtein@^1.1.6:
resolved "https://registry.yarnpkg.com/js-tokens/-/js-tokens-4.0.0.tgz#19203fb59991df98e3a287050d4647cdeaf32499"
integrity sha512-RdJUflcE3cUzKiMqQgsCu06FPu9UdIJO0beYbPhHN4k6apgJtifcoCtT9bcxOpYBtpD2kCM6Sbzg4CausW/PKQ==
js-yaml@4.1.0, js-yaml@=4.3.1, js-yaml@^4.1.0, js-yaml@^4.1.1, js-yaml@^4.2.0, js-yaml@^4.3.0:
version "4.3.1"
resolved "https://registry.yarnpkg.com/js-yaml/-/js-yaml-4.3.1.tgz#01216c001d67f48e2cd560d708c7af21090a3848"
integrity sha512-CY6crGq313MX8GkwvB7tzgp99vjQxY1++5y10/BKN/GUfHqWaOGQMNZkBvqSzsZKWk/ijwHlWzzkLulsGHhjWQ==
js-yaml@4.1.0, js-yaml@=4.3.0, js-yaml@^4.1.0, js-yaml@^4.1.1, js-yaml@^4.2.0, js-yaml@^4.3.0:
version "4.3.0"
resolved "https://registry.yarnpkg.com/js-yaml/-/js-yaml-4.3.0.tgz#d1900572a7f7cf0b5f540c83673e60bad3436592"
integrity sha512-1td788aAnnZ5qs7V2QIRl1owjtYpbKt749Y3xauqQgwIIGF/xXWz1wMTEBx5O3LK3lXLVuqXPdPxj2BoFHaW9Q==
dependencies:
argparse "^2.0.1"
@@ -15103,10 +15103,10 @@ swagger-client@^3.37.8:
"@swagger-api/apidom-parser-adapter-openapi-yaml-3-2" "^1.12.0"
"@swagger-api/apidom-parser-adapter-yaml-1-2" "^1.12.0"
swagger-ui-react@^5.32.13:
version "5.32.13"
resolved "https://registry.yarnpkg.com/swagger-ui-react/-/swagger-ui-react-5.32.13.tgz#04c96140b0a2d4ea01ebec4d4cfc655d5ed9a500"
integrity sha512-XIDl+Ny6kE1N8wpSPiOFrjPfAevs4GR4XmV6BT6NLMikkMFIbIVocWbA8pnKYyYXQe8Rccfli5o2zDfySw0FnQ==
swagger-ui-react@^5.32.12:
version "5.32.12"
resolved "https://registry.yarnpkg.com/swagger-ui-react/-/swagger-ui-react-5.32.12.tgz#47525a26774eb02db0e6203af72f5b32fa6205cc"
integrity sha512-WCdkNOQyMTZDu+z356FpwVWHf1dwZgQPUjdQPh1L4r7jULaJTKKlIItXq6WsZdYeXvsHndMdxxccEQXOAroUHQ==
dependencies:
"@babel/runtime-corejs3" "^7.27.1"
"@scarf/scarf" "=1.4.0"
@@ -15115,11 +15115,11 @@ swagger-ui-react@^5.32.13:
classnames "^2.5.1"
css.escape "1.5.1"
deep-extend "0.6.0"
dompurify "^3.4.13"
dompurify "^3.4.12"
ieee754 "^1.2.1"
immutable "^4.3.9"
js-file-download "^0.4.12"
js-yaml "=4.3.1"
js-yaml "=4.3.0"
lodash "^4.18.1"
prop-types "^15.8.1"
randexp "^0.5.3"
+20 -11
View File
@@ -99,7 +99,7 @@
"geostyler-openlayers-parser": "^5.7.1",
"geostyler-style": "11.0.2",
"geostyler-wfs-parser": "^3.0.1",
"google-auth-library": "^11.0.2",
"google-auth-library": "^11.0.1",
"immer": "^11.1.16",
"interweave": "^13.1.1",
"jquery": "^4.0.0",
@@ -186,7 +186,7 @@
"@storybook/test-runner": "0.24.4",
"@svgr/webpack": "^8.1.0",
"@swc/core": "^1.15.47",
"@swc/plugin-emotion": "^14.19.0",
"@swc/plugin-emotion": "^14.15.0",
"@swc/plugin-transform-imports": "^12.5.0",
"@testing-library/dom": "^10.4.1",
"@testing-library/jest-dom": "^7.0.1",
@@ -11808,9 +11808,9 @@
}
},
"node_modules/@swc/plugin-emotion": {
"version": "14.19.0",
"resolved": "https://registry.npmjs.org/@swc/plugin-emotion/-/plugin-emotion-14.19.0.tgz",
"integrity": "sha512-0/q84ro0a7kdjpYpn9Wmi5/RLHYuSwYjO638lE5ZBQfIvYpSLJxbEgLsObCmdH4KPe2stoN8plVKUpCsKPggaw==",
"version": "14.15.0",
"resolved": "https://registry.npmjs.org/@swc/plugin-emotion/-/plugin-emotion-14.15.0.tgz",
"integrity": "sha512-nCsTO7mOOPz2UnT3N6YWb014uI0CVxeKg53A/KM/CvuSIE6H3KPkhaziJQ3q2jI3u3LfFuDKEnU5ZmB1330Dqg==",
"dev": true,
"license": "Apache-2.0",
"dependencies": {
@@ -22783,9 +22783,9 @@
"license": "MIT"
},
"node_modules/google-auth-library": {
"version": "11.0.2",
"resolved": "https://registry.npmjs.org/google-auth-library/-/google-auth-library-11.0.2.tgz",
"integrity": "sha512-vzpgPutxrghPsnjrjpzLX2bdv8IOL719Rh0oEjGnQu8YCIbnbMuTTQ5zU9LcKvLdOPgCxBwppbvnhgW90Qna5Q==",
"version": "11.0.1",
"resolved": "https://registry.npmjs.org/google-auth-library/-/google-auth-library-11.0.1.tgz",
"integrity": "sha512-ZqfaYduu9ASUaFuUk5dF9g9QvufdhhSj7jFiEnCrTQcH57sFPKYetM0iU4dcKkQk6CqC1xpSrVr5uQ9NhqjNOg==",
"license": "Apache-2.0",
"dependencies": {
"base64-js": "^1.3.0",
@@ -43420,13 +43420,22 @@
"license": "Apache-2.0",
"dependencies": {
"d3": "^3.5.17",
"prop-types": "^15.8.1"
"prop-types": "^15.8.1",
"react": "^19.2.7"
},
"peerDependencies": {
"@apache-superset/core": "*",
"@superset-ui/chart-controls": "*",
"@superset-ui/core": "*",
"react": "^18.3.0"
"@superset-ui/core": "*"
}
},
"plugins/plugin-chart-chord/node_modules/react": {
"version": "19.2.8",
"resolved": "https://registry.npmjs.org/react/-/react-19.2.8.tgz",
"integrity": "sha512-PWaYA1L/q9u2u7xYQi+Y3L3Yfnie7XyLeaJICV1MGD6LprsBxcAqGjYyr0eY3p+QdsA+x/Irkt4Qif8D63+Sbw==",
"license": "MIT",
"engines": {
"node": ">=0.10.0"
}
},
"plugins/plugin-chart-country-map": {
+2 -2
View File
@@ -176,7 +176,7 @@
"geostyler-openlayers-parser": "^5.7.1",
"geostyler-style": "11.0.2",
"geostyler-wfs-parser": "^3.0.1",
"google-auth-library": "^11.0.2",
"google-auth-library": "^11.0.1",
"immer": "^11.1.16",
"interweave": "^13.1.1",
"jquery": "^4.0.0",
@@ -263,7 +263,7 @@
"@storybook/test-runner": "0.24.4",
"@svgr/webpack": "^8.1.0",
"@swc/core": "^1.15.47",
"@swc/plugin-emotion": "^14.19.0",
"@swc/plugin-emotion": "^14.15.0",
"@swc/plugin-transform-imports": "^12.5.0",
"@testing-library/dom": "^10.4.1",
"@testing-library/jest-dom": "^7.0.1",
@@ -130,7 +130,6 @@ export enum GenericDataType {
String = 1,
Temporal = 2,
Boolean = 3,
MultiValue = 4,
}
/**
@@ -28,7 +28,6 @@ import {
FieldBinaryOutlined,
FieldStringOutlined,
NumberOutlined,
UnorderedListOutlined,
} from '@ant-design/icons';
import { Icons } from '@superset-ui/core/components';
@@ -73,10 +72,6 @@ export function ColumnTypeLabel({ type }: ColumnTypeLabelProps) {
typeIcon = <FieldBinaryOutlined aria-label={t('boolean type icon')} />;
} else if (type === GenericDataType.Temporal) {
typeIcon = <ClockCircleOutlined aria-label={t('temporal type icon')} />;
} else if (type === GenericDataType.MultiValue) {
typeIcon = (
<UnorderedListOutlined aria-label={t('multi-value type icon')} />
);
}
return <TypeIconWrapper>{typeIcon}</TypeIconWrapper>;
@@ -64,21 +64,4 @@ describe('ColumnOption', () => {
renderColumnTypeLabel({ type: GenericDataType.Temporal });
expect(screen.getByLabelText('temporal type icon')).toBeVisible();
});
test('multi-value (array) type shows list icon', () => {
renderColumnTypeLabel({ type: GenericDataType.MultiValue });
expect(screen.getByLabelText('multi-value type icon')).toBeVisible();
});
});
describe('GenericDataType enum parity', () => {
// These numeric values are shared with the backend enum in
// superset/utils/core.py (GenericDataType). They must stay in sync because
// the backend serializes columns using these integers.
test('values match the backend contract', () => {
expect(GenericDataType.Numeric).toBe(0);
expect(GenericDataType.String).toBe(1);
expect(GenericDataType.Temporal).toBe(2);
expect(GenericDataType.Boolean).toBe(3);
expect(GenericDataType.MultiValue).toBe(4);
});
});
-4
View File
@@ -47,10 +47,6 @@ export default defineConfig({
// Retry logic - 2 retries in CI, 0 locally
retries: process.env.CI ? 2 : 0,
// Disable capturing Git commit info as the project's history is increasingly dense
// and breach Playwright's default 3-seconds `git` command timeout limit
captureGitInfo: { commit: false, diff: false },
// Reporter configuration - multiple reporters for better visibility
reporter: process.env.CI
? [
@@ -30,12 +30,12 @@
},
"dependencies": {
"d3": "^3.5.17",
"prop-types": "^15.8.1"
"prop-types": "^15.8.1",
"react": "^19.2.7"
},
"peerDependencies": {
"@apache-superset/core": "*",
"@superset-ui/chart-controls": "*",
"@superset-ui/core": "*",
"react": "^18.3.0"
"@superset-ui/core": "*"
}
}
@@ -17,10 +17,6 @@
* under the License.
*/
import { getNumberFormatter } from '@superset-ui/core';
import { render, fireEvent } from '../../../../spec/helpers/testing-library';
import BigNumberVis from './BigNumberViz';
/**
* Tests for the color threshold formatter logic in BigNumberViz.
*
@@ -87,33 +83,3 @@ describe('BigNumberViz color formatters', () => {
expect(getColorFromValue).not.toHaveBeenCalled();
});
});
describe('BigNumberViz context menu', () => {
test('invokes onContextMenu and stops the event bubbling to ancestor handlers', () => {
const onContextMenu = jest.fn();
const ancestorHandler = jest.fn();
const { container } = render(
<div onContextMenu={ancestorHandler}>
<BigNumberVis
width={200}
height={100}
bigNumber={42}
headerFormatter={getNumberFormatter()}
headerFontSize={0.3}
subheaderFontSize={0.125}
subtitleFontSize={0.125}
subtitle=""
refs={{}}
onContextMenu={onContextMenu}
/>
</div>,
);
const headerLine = container.querySelector('.header-line');
fireEvent.contextMenu(headerLine!, { clientX: 10, clientY: 20 });
expect(onContextMenu).toHaveBeenCalledWith(10, 20);
expect(ancestorHandler).not.toHaveBeenCalled();
});
});
@@ -224,7 +224,6 @@ function BigNumberVis({
const handleContextMenu = (e: MouseEvent<HTMLDivElement>) => {
if (onContextMenu) {
e.preventDefault();
e.stopPropagation();
onContextMenu(e.nativeEvent.clientX, e.nativeEvent.clientY);
}
};
@@ -390,7 +390,6 @@ export default function transformProps(chartProps: EchartsGanttChartProps) {
[GenericDataType.String]: undefined,
[GenericDataType.Temporal]: tooltipTimeFormatter,
[GenericDataType.Boolean]: undefined,
[GenericDataType.MultiValue]: undefined,
};
const echartOptions: EChartsCoreOption = {
@@ -16,7 +16,7 @@
* specific language governing permissions and limitations
* under the License.
*/
import { render, screen } from 'spec/helpers/testing-library';
import { render, screen, userEvent } from 'spec/helpers/testing-library';
import SaveDatasetActionButton from 'src/SqlLab/components/SaveDatasetActionButton';
// eslint-disable-next-line no-restricted-globals -- TODO: Migrate from describe blocks
@@ -41,4 +41,37 @@ describe('SaveDatasetActionButton', () => {
expect(saveBtn).toBeVisible();
expect(saveDatasetBtn).toBeVisible();
});
test('disables only the dataset button when saveDatasetDisabled is set', () => {
const onSaveAsExplore = jest.fn();
render(
<SaveDatasetActionButton
setShowSave={() => true}
onSaveAsExplore={onSaveAsExplore}
saveDatasetDisabled
/>,
);
// Saving the query needs no results.
expect(screen.getByRole('button', { name: 'Save' })).toBeEnabled();
expect(
screen.getByRole('button', { name: /save dataset/i }),
).toBeDisabled();
});
test('explains why the dataset button is unavailable', async () => {
render(
<SaveDatasetActionButton
setShowSave={() => true}
onSaveAsExplore={jest.fn()}
saveDatasetDisabled
/>,
);
userEvent.hover(screen.getByRole('button', { name: /save dataset/i }));
expect(
await screen.findByText('You must run the query successfully first'),
).toBeInTheDocument();
});
});
@@ -23,11 +23,14 @@ import { Button } from '@superset-ui/core/components';
interface SaveDatasetActionButtonProps {
setShowSave: (arg0: boolean) => void;
onSaveAsExplore?: () => void;
/** Set while the query has not run successfully. */
saveDatasetDisabled?: boolean;
}
const SaveDatasetActionButton = ({
setShowSave,
onSaveAsExplore,
saveDatasetDisabled = false,
}: SaveDatasetActionButtonProps) => (
<>
<Button
@@ -44,7 +47,12 @@ const SaveDatasetActionButton = ({
variant="text"
onClick={() => onSaveAsExplore?.()}
icon={<Icons.TableOutlined />}
tooltip={t('Save or Overwrite Dataset')}
tooltip={
saveDatasetDisabled
? t('You must run the query successfully first')
: t('Save or Overwrite Dataset')
}
disabled={saveDatasetDisabled}
aria-label={t('Save dataset')}
/>
)}
@@ -19,12 +19,14 @@
import { act, type ComponentProps } from 'react';
import {
cleanup,
createStore,
fireEvent,
render,
screen,
userEvent,
waitFor,
} from 'spec/helpers/testing-library';
import reducerIndex from 'spec/helpers/reducerIndex';
import fetchMock from 'fetch-mock';
import { SaveDatasetModal } from 'src/SqlLab/components/SaveDatasetModal';
import { createDatasource } from 'src/SqlLab/actions/sqlLab';
@@ -63,6 +65,12 @@ beforeEach(() => {
cleanup();
});
afterEach(() => {
// In-body restores are skipped when an assertion throws, leaking a
// configured spy into later tests.
jest.restoreAllMocks();
});
// Mock createDatasource to return a thunk that resolves with the dataset's
// new id. The test's mock store includes redux-thunk middleware (from RTK's
// getDefaultMiddleware), so dispatch(createDatasource(...)) properly unwraps
@@ -518,6 +526,39 @@ describe('SaveDatasetModal', () => {
});
});
test('surfaces the error and keeps the modal open when saving fails', async () => {
// The chart-payload step's toast was built but never dispatched, so a
// failure there was silent.
const postFormData = jest.spyOn(
require('src/explore/exploreUtils/formData'),
'postFormData',
);
postFormData.mockRejectedValue(new Error('Boom'));
const onHide = jest.fn();
const store = createStore({ user }, reducerIndex);
render(<SaveDatasetModal {...mockedProps} onHide={onHide} />, { store });
fireEvent.change(screen.getByDisplayValue(/unimportant/i), {
target: { value: 'my dataset' },
});
userEvent.click(screen.getByRole('button', { name: /save/i }));
// `createStore` builds its reducer map at runtime, so state isn't typed.
const toasts = () =>
(
store.getState() as unknown as {
messageToasts: { toastType: string }[];
}
).messageToasts;
await waitFor(() => {
expect(toasts()).toHaveLength(1);
});
expect(toasts()[0].toastType).toBe('DANGER_TOAST');
expect(onHide).not.toHaveBeenCalled();
});
test('clearDatasetCache is imported and available', () => {
const { clearDatasetCache } = require('src/utils/cachedSupersetGet');
@@ -61,6 +61,9 @@ import type Subject from 'src/types/Subject';
import { openInNewTab, redirect } from 'src/utils/navigationUtils';
import { mapSubjectValuesToIds } from 'src/features/subjects/SubjectPicker';
// Derived so it can't drift from what `getClientErrorObject` accepts.
type SaveErrorSource = Parameters<typeof getClientErrorObject>[0];
interface QueryDatabase {
id?: number;
}
@@ -391,9 +394,18 @@ export const SaveDatasetModal = ({
setDatasetName(getDefaultDatasetName());
onHide();
})
.catch(() => {
.catch((error?: SaveErrorSource) => {
setLoading(false);
addDangerToast(t('An error occurred saving dataset'));
// `createDatasource` already toasted the server's message and rejects
// with nothing; only the chart-payload step needs its own.
if (!error) {
return;
}
getClientErrorObject(error).then(e =>
dispatch(
addDangerToast(e.error || t('An error occurred saving dataset')),
),
);
});
};
@@ -61,6 +61,28 @@ const splitSaveBtnProps = {
},
};
const EDITOR_SQL = 'SELECT * FROM t';
const stateWithLatestQuery = ({
id,
state,
sql = EDITOR_SQL,
}: {
id: string;
state: string;
sql?: string;
}) => ({
...mockState,
sqlLab: {
...mockState.sqlLab,
queryEditors: mockState.sqlLab.queryEditors.map(qe => ({
...qe,
latestQueryId: id,
})),
queries: { [id]: { id, state, sql } },
},
});
const middlewares = [thunk];
const mockStore = configureStore(middlewares);
@@ -96,6 +118,59 @@ describe('SavedQuery', () => {
expect(saveBtn).toBeVisible();
});
test('blocks "Save dataset" until the query has run successfully', () => {
// Without a successful run the save can only fail server-side.
render(<SaveQuery {...splitSaveBtnProps} />, {
useRedux: true,
store: mockStore(stateWithLatestQuery({ id: 'qid-1', state: 'failed' })),
});
expect(
screen.getByRole('button', { name: /save dataset/i }),
).toBeDisabled();
// Saving the query itself is unaffected.
expect(screen.getByRole('button', { name: 'Save' })).toBeEnabled();
});
test('blocks "Save dataset" when no query has been run at all', () => {
render(<SaveQuery {...splitSaveBtnProps} />, {
useRedux: true,
store: mockStore(mockState),
});
expect(
screen.getByRole('button', { name: /save dataset/i }),
).toBeDisabled();
});
test('blocks "Save dataset" when the SQL changed after a successful run', () => {
// The run succeeded, but not for what is in the editor now -- and it is
// the editor's SQL that gets saved.
render(<SaveQuery {...splitSaveBtnProps} />, {
useRedux: true,
store: mockStore(
stateWithLatestQuery({
id: 'qid-1',
state: 'success',
sql: 'SELECT 1 AS ran_earlier',
}),
),
});
expect(
screen.getByRole('button', { name: /save dataset/i }),
).toBeDisabled();
});
test('enables "Save dataset" once the query has succeeded', () => {
render(<SaveQuery {...splitSaveBtnProps} />, {
useRedux: true,
store: mockStore(stateWithLatestQuery({ id: 'qid-1', state: 'success' })),
});
expect(screen.getByRole('button', { name: /save dataset/i })).toBeEnabled();
});
test('renders a save query modal when user clicks save button', () => {
render(<SaveQuery {...mockedProps} />, {
useRedux: true,
@@ -233,7 +308,7 @@ describe('SavedQuery', () => {
test('renders a save dataset modal when user clicks "save dataset" menu item', async () => {
render(<SaveQuery {...splitSaveBtnProps} />, {
useRedux: true,
store: mockStore(mockState),
store: mockStore(stateWithLatestQuery({ id: 'qid-1', state: 'success' })),
});
const saveDatasetMenuItem = await screen.findByLabelText(/save dataset/i);
@@ -247,7 +322,7 @@ describe('SavedQuery', () => {
test('renders the save dataset modal UI', async () => {
render(<SaveQuery {...splitSaveBtnProps} />, {
useRedux: true,
store: mockStore(mockState),
store: mockStore(stateWithLatestQuery({ id: 'qid-1', state: 'success' })),
});
const saveDatasetMenuItem = await screen.findByLabelText(/save dataset/i);
userEvent.click(saveDatasetMenuItem);
@@ -17,6 +17,8 @@
* under the License.
*/
import { useState, useEffect, useMemo, ChangeEvent } from 'react';
import { useSelector } from 'react-redux';
import { Query, QueryState } from '@superset-ui/core';
import type { DatabaseObject } from 'src/features/databases/types';
import { t } from '@apache-superset/core/translation';
import { styled } from '@apache-superset/core/theme';
@@ -37,7 +39,7 @@ import {
} from 'src/SqlLab/components/SaveDatasetModal';
import { getDatasourceAsSaveableDataset } from 'src/utils/datasourceUtils';
import useQueryEditor from 'src/SqlLab/hooks/useQueryEditor';
import { QueryEditor } from 'src/SqlLab/types';
import { QueryEditor, SqlLabRootState } from 'src/SqlLab/types';
import useLogAction from 'src/logger/useLogAction';
import {
LOG_ACTIONS_SQLLAB_CREATE_CHART,
@@ -111,6 +113,15 @@ const SaveQuery = ({
const [label, setLabel] = useState<string>(defaultLabel);
const [showSave, setShowSave] = useState<boolean>(false);
const [showSaveDatasetModal, setShowSaveDatasetModal] = useState(false);
// Saving a dataset runs the SQL to introspect columns, so it needs a
// successful run of the SQL being saved -- editing after a run invalidates
// it, and running a selection only validates that selection.
const latestQuery = useSelector<SqlLabRootState, Query | undefined>(
({ sqlLab }) => sqlLab.queries[queryEditor.latestQueryId || ''],
);
const hasSuccessfulQuery =
latestQuery?.state === QueryState.Success &&
latestQuery.sql === queryEditor.sql;
const isSaved = !!query.remoteId;
const isLabelEmpty = label.trim().length === 0;
const canExploreDatabase = !!database?.allows_virtual_table_explore;
@@ -207,6 +218,7 @@ const SaveQuery = ({
<SaveDatasetActionButton
setShowSave={setShowSave}
onSaveAsExplore={canExploreDatabase ? onSaveAsExplore : undefined}
saveDatasetDisabled={!hasSuccessfulQuery}
/>
)}
<SaveDatasetModal
@@ -1627,7 +1627,9 @@ function DatasourceEditor({
{t(
'Default URL to redirect to when accessing from the dataset list page. Accepts relative URLs such as',
)}{' '}
<Typography.Text code>/dashboard/{'{id}'}/</Typography.Text>
<Typography.Text code>
/superset/dashboard/{'{id}'}/
</Typography.Text>
</>
}
control={<TextControl controlId="default_endpoint" />}
@@ -71,17 +71,6 @@ test('renders Tabs', async () => {
expect(screen.getByTestId('edit-dataset-tabs')).toBeInTheDocument();
});
test('recommends a registered client route for the default URL', async () => {
await asyncRender(createProps());
userEvent.click(screen.getByRole('tab', { name: 'Settings' }));
expect(await screen.findByText('/dashboard/{id}/')).toBeInTheDocument();
expect(
screen.queryByText('/superset/dashboard/{id}/'),
).not.toBeInTheDocument();
});
test('can sync columns from source', async () => {
const testProps = createProps();
await asyncRender({
@@ -468,7 +468,7 @@ function SliceAdder({
<AutoSizer>
{({ height, width }: { height: number; width: number }) => (
<List
style={{ width, height, maxHeight: height }}
style={{ width, height }}
rowCount={filteredSlices.length}
rowHeight={DEFAULT_CELL_HEIGHT}
rowProps={listRowProps}
@@ -148,7 +148,7 @@ export const DatasourceItems = ({
return (
<List
style={{ width: width - BORDER_WIDTH, height, maxHeight: height }}
style={{ width: width - BORDER_WIDTH, height }}
rowHeight={rowHeight}
rowCount={flattenedItems.length}
rowProps={rowProps}
@@ -251,11 +251,4 @@ export const DEFAULT_CONFIG_FORM_LAYOUT: ColumnConfigFormLayout = {
{ name: 'horizontalAlign', override: { defaultValue: 'left' } },
],
],
[GenericDataType.MultiValue]: [
[
'columnWidth',
{ name: 'horizontalAlign', override: { defaultValue: 'left' } },
],
['truncateLongCells'],
],
};
@@ -29,9 +29,7 @@ import { ControlFormItemComponents } from './ControlForm';
* Column formatting configs.
*/
export type ColumnConfig = {
[
key in SharedColumnConfigProp
]?: (typeof SHARED_COLUMN_CONFIG_PROPS)[key]['value'];
[key in SharedColumnConfigProp]?: (typeof SHARED_COLUMN_CONFIG_PROPS)[key]['value'];
} & Record<string, StrictJsonValue>;
/**
@@ -270,74 +270,6 @@ describe('AdhocFilter', () => {
});
expect(adhocFilter.comparator).toBe(undefined);
});
// Charts saved before #32701 persisted `==` as the operation for IS_TRUE and
// IS_FALSE, alongside a boolean comparator. `translateToSql` and the backend
// both key off `operator`, so dropping the comparator would render such a
// filter as `col =` and query it as `col IS NULL`.
test('keeps the legacy boolean comparator for IS_TRUE', () => {
const adhocFilter = new AdhocFilter({
expressionType: ExpressionTypes.Simple,
subject: 'col',
operator: '==',
operatorId: Operators.IsTrue,
comparator: true,
clause: Clauses.Where,
});
expect(adhocFilter.operator).toBe('==');
expect(adhocFilter.comparator).toBe(true);
expect(adhocFilter.translateToSql()).toBe("col = 'TRUE'");
});
test('keeps the legacy boolean comparator for IS_FALSE', () => {
const adhocFilter = new AdhocFilter({
expressionType: ExpressionTypes.Simple,
subject: 'col',
operator: '==',
operatorId: Operators.IsFalse,
comparator: false,
clause: Clauses.Where,
});
expect(adhocFilter.operator).toBe('==');
expect(adhocFilter.comparator).toBe(false);
expect(adhocFilter.translateToSql()).toBe("col = 'FALSE'");
});
test('restores the boolean even when the stored comparator is missing', () => {
const adhocFilter = new AdhocFilter({
expressionType: ExpressionTypes.Simple,
subject: 'col',
operator: '==',
operatorId: Operators.IsTrue,
clause: Clauses.Where,
});
expect(adhocFilter.comparator).toBe(true);
});
test('keeps a legacy boolean filter intact when the control re-posts it', () => {
const stored = {
expressionType: ExpressionTypes.Simple,
subject: 'col',
operator: '==',
operatorId: Operators.IsTrue,
comparator: true,
clause: Clauses.Where,
};
// DndFilterSelect wraps props.value and hands those instances to onChange
const posted = JSON.parse(JSON.stringify(new AdhocFilter(stored)));
expect(posted.operator).toBe('==');
expect(posted.comparator).toBe(true);
expect(posted.operatorId).toBe(Operators.IsTrue);
});
test('leaves a genuine equality filter on a boolean value alone', () => {
const adhocFilter = new AdhocFilter({
expressionType: ExpressionTypes.Simple,
subject: 'col',
operator: '==',
operatorId: Operators.Equals,
comparator: true,
clause: Clauses.Where,
});
expect(adhocFilter.operator).toBe('==');
expect(adhocFilter.comparator).toBe(true);
expect(adhocFilter.translateToSql()).toBe("col = 'TRUE'");
});
test('sets the label properly if subject is a string', () => {
const adhocFilter = new AdhocFilter({
expressionType: ExpressionTypes.Simple,
@@ -30,15 +30,6 @@ const CUSTOM_OPERATIONS = [...CUSTOM_OPERATORS].map(
op => OPERATOR_ENUM_TO_OPERATOR_TYPE[op].operation,
);
// Charts saved before #32701 store `==` for IS_TRUE/IS_FALSE with the boolean
// in the comparator; blanking it makes them query `col IS NULL`. Restoring it
// leaves the emitted SQL untouched -- reconciling `operator` to `IS TRUE`
// would not, and Druid rejects that predicate on VARCHAR columns.
const LEGACY_BOOLEAN_COMPARATORS = new Map<string, boolean>([
[Operators.IsTrue, true],
[Operators.IsFalse, false],
]);
interface AdhocFilterInput {
expressionType?: string;
subject?: string | { column_name?: string; [key: string]: unknown } | null;
@@ -86,16 +77,6 @@ export default class AdhocFilter {
) {
this.comparator = undefined;
}
if (
this.operator ===
OPERATOR_ENUM_TO_OPERATOR_TYPE[Operators.Equals].operation &&
adhocFilter.operatorId &&
LEGACY_BOOLEAN_COMPARATORS.has(adhocFilter.operatorId)
) {
this.comparator = LEGACY_BOOLEAN_COMPARATORS.get(
adhocFilter.operatorId,
);
}
this.clause = adhocFilter.clause || Clauses.Where;
this.sqlExpression = null;
} else if (this.expressionType === ExpressionTypes.Sql) {
@@ -367,22 +367,8 @@ function AdhocFilterEditPopover({
</ErrorBoundary>
),
},
...(datasource?.type === 'semantic_view' ||
[
Operators.ContainsAny,
Operators.ContainsAll,
Operators.IsEmpty,
Operators.IsNotEmpty,
Operators.LengthEquals,
Operators.LengthGreaterThan,
Operators.LengthLessThan,
Operators.LengthGreaterThanOrEqual,
Operators.LengthLessThanOrEqual,
].includes(adhocFilter.operatorId as Operators)
? // Hide the Custom SQL tab for element-level array operators: they
// have no portable SQL representation, and converting one would
// silently turn the filter into invalid raw SQL.
[]
...(datasource?.type === 'semantic_view'
? []
: [
{
key: ExpressionTypes.Sql,
@@ -35,7 +35,6 @@ import {
} from 'src/explore/constants';
import AdhocMetric from 'src/explore/components/controls/MetricControl/AdhocMetric';
import { FeatureFlag, isFeatureEnabled } from '@superset-ui/core';
import { GenericDataType } from '@apache-superset/core/common';
import fetchMock from 'fetch-mock';
import { TestDataset, Dataset } from '@superset-ui/chart-controls';
@@ -253,78 +252,6 @@ test('shows boolean only operators when subject is number', () => {
].map(operator => expect(isOperatorRelevant(operator, 'value')).toBe(true));
});
test('shows array operators (tier 1 + tier 2) when subject is multi-value', () => {
const props = setup({
adhocFilter: new AdhocFilter({
expressionType: ExpressionTypes.Simple,
subject: 'skills',
operatorId: undefined,
operator: undefined,
comparator: undefined,
clause: undefined,
}),
datasource: {
columns: [
{
id: 3,
column_name: 'skills',
type: 'Array(String)',
type_generic: GenericDataType.MultiValue,
},
],
},
});
const { isOperatorRelevant } = useSimpleTabFilterProps(
props as unknown as Props,
);
// Tier 1 (whole-array) + Tier 2 (element-level) are all relevant.
[
Operators.Equals,
Operators.NotEquals,
Operators.In,
Operators.NotIn,
Operators.IsNull,
Operators.IsNotNull,
Operators.ContainsAny,
Operators.ContainsAll,
Operators.IsEmpty,
Operators.IsNotEmpty,
].forEach(operator =>
expect(isOperatorRelevant(operator, 'skills')).toBe(true),
);
// scalar-only operators are hidden for array columns
[Operators.GreaterThan, Operators.LessThan, Operators.Like].forEach(
operator => expect(isOperatorRelevant(operator, 'skills')).toBe(false),
);
});
test('hides element-level array operators for non multi-value columns', () => {
const props = setup({
adhocFilter: new AdhocFilter({
expressionType: ExpressionTypes.Simple,
subject: 'value',
operatorId: undefined,
operator: undefined,
comparator: undefined,
clause: undefined,
}),
datasource: {
columns: [{ id: 3, column_name: 'value', type: 'STRING' }],
},
});
const { isOperatorRelevant } = useSimpleTabFilterProps(
props as unknown as Props,
);
[
Operators.ContainsAny,
Operators.ContainsAll,
Operators.IsEmpty,
Operators.IsNotEmpty,
].forEach(operator =>
expect(isOperatorRelevant(operator, 'value')).toBe(false),
);
});
test('will convert from individual comparator to array if the operator changes to multi', () => {
const props = setup();
const { onOperatorChange } = useSimpleTabFilterProps(
@@ -382,49 +309,6 @@ test('will convert from array to individual comparators if the operator changes
);
});
test('resets the comparator when switching between array value families', () => {
// Equal to (whole-array literal) -> Contains all (individual elements):
// the value spaces are incompatible, so the stale value must be cleared.
const wholeArrayFilter = new AdhocFilter({
expressionType: ExpressionTypes.Simple,
subject: 'scores',
operatorId: Operators.Equals,
operator: OPERATOR_ENUM_TO_OPERATOR_TYPE[Operators.Equals].operation,
comparator: '[5,6,7]',
clause: Clauses.Where,
});
const props = setup({ adhocFilter: wholeArrayFilter });
const { onOperatorChange } = useSimpleTabFilterProps(
props as unknown as Props,
);
onOperatorChange(Operators.ContainsAll);
const lastCall =
props.onChange.mock.calls[props.onChange.mock.calls.length - 1][0];
expect(lastCall.operatorId).toEqual(Operators.ContainsAll);
expect(lastCall.comparator).toBeUndefined();
});
test('keeps the value when switching within the element family', () => {
// Contains any <-> Contains all both take individual elements, so the
// selected elements should carry over.
const elementFilter = new AdhocFilter({
expressionType: ExpressionTypes.Simple,
subject: 'scores',
operatorId: Operators.ContainsAny,
operator: OPERATOR_ENUM_TO_OPERATOR_TYPE[Operators.ContainsAny].operation,
comparator: ['5', '6'],
clause: Clauses.Where,
});
const props = setup({ adhocFilter: elementFilter });
const { onOperatorChange } = useSimpleTabFilterProps(
props as unknown as Props,
);
onOperatorChange(Operators.ContainsAll);
const lastCall =
props.onChange.mock.calls[props.onChange.mock.calls.length - 1][0];
expect(lastCall.comparator).toEqual(['5', '6']);
});
test('passes the new adhocFilter to onChange after onComparatorChange', () => {
const props = setup();
const { onComparatorChange } = useSimpleTabFilterProps(
@@ -515,28 +399,6 @@ test('will not display boolean operators when column type is string', () => {
});
});
test.each(['STRING', 'DATE'])(
'will not display boolean operators when an expression column declares type %s',
type => {
const props = setup({
datasource: {
type: 'table' as const,
datasource_name: 'table1',
schema: 'schema',
columns: [{ column_name: 'value', type, expression: '"value"' }],
},
adhocFilter: simpleAdhocFilter,
});
const { isOperatorRelevant } = useSimpleTabFilterProps(
props as unknown as Props,
);
const booleanOnlyOperators = [Operators.IsTrue, Operators.IsFalse];
booleanOnlyOperators.forEach(operator => {
expect(isOperatorRelevant(operator, 'value')).toBe(false);
});
},
);
test('will display boolean operators when column is an expression', () => {
const props = setup({
datasource: {
@@ -32,7 +32,6 @@ import {
isDefined,
SupersetClient,
} from '@superset-ui/core';
import { GenericDataType } from '@apache-superset/core/common';
import { styled, useTheme, css } from '@apache-superset/core/theme';
import {
Operators,
@@ -119,8 +118,6 @@ export const useSimpleTabFilterProps = (props: Props) => {
const isColumnNumber =
!!column && (column.type === 'INT' || column.type === 'INTEGER');
const isColumnFunction = !!column && !!column.expression;
const isColumnMultiValue =
!!column && column.type_generic === GenericDataType.MultiValue;
if (operator && operator === Operators.LatestPartition) {
const { partitionColumn } = props;
@@ -130,41 +127,8 @@ export const useSimpleTabFilterProps = (props: Props) => {
// hide the TEMPORAL_RANGE operator
return false;
}
// Element-level array operators only apply to multi-value columns.
const arrayElementOperators = [
Operators.ContainsAny,
Operators.ContainsAll,
Operators.IsEmpty,
Operators.IsNotEmpty,
Operators.LengthEquals,
Operators.LengthGreaterThan,
Operators.LengthLessThan,
Operators.LengthGreaterThanOrEqual,
Operators.LengthLessThanOrEqual,
];
if (arrayElementOperators.includes(operator)) {
return isColumnMultiValue;
}
if (isColumnMultiValue) {
// Array columns support whole-array operators (=, !=, In, Not in, null
// checks) plus the element-level operators above. Scalar-only operators
// (Like, <, >, <=, >=) are hidden because they aren't valid on an array.
return [
Operators.Equals,
Operators.NotEquals,
Operators.In,
Operators.NotIn,
Operators.IsNull,
Operators.IsNotNull,
...arrayElementOperators,
].includes(operator);
}
if (operator === Operators.IsTrue || operator === Operators.IsFalse) {
// An expression column may evaluate to a boolean, but that is only a
// safe assumption while its type is unknown; a declared type wins.
return (
isColumnBoolean || isColumnNumber || (isColumnFunction && !column?.type)
);
return isColumnBoolean || isColumnNumber || isColumnFunction;
}
if (isColumnBoolean) {
return operator === Operators.IsNull || operator === Operators.IsNotNull;
@@ -203,19 +167,9 @@ export const useSimpleTabFilterProps = (props: Props) => {
].operation
: null;
if (!isDefined(operator)) {
// The previous operator is not relevant for the new subject; pick a
// sensible default and reset the comparator. Multi-value (array) columns
// default to "Contains any" (element membership) rather than the
// scalar-only IN.
const newColumn = props.datasource.columns?.find(
col => col.column_name === subject,
);
const defaultOperator =
newColumn?.type_generic === GenericDataType.MultiValue
? Operators.ContainsAny
: Operators.In;
operator = defaultOperator;
operatorId = defaultOperator;
// if operator is `null`, use the `IN` and reset the comparator.
operator = Operators.In;
operatorId = Operators.In;
comparator = undefined;
}
@@ -239,38 +193,10 @@ export const useSimpleTabFilterProps = (props: Props) => {
};
const onOperatorChange = (operatorId: Operators) => {
const currentComparator = props.adhocFilter.comparator;
// The value space differs between operator families: element-level array
// ops (Contains any/all) take individual elements, whole-array/scalar ops
// (=, In, …) take whole arrays or scalars, Length ops take a count, and the
// unary ops take nothing. A value from one family is meaningless in another,
// so reset the value when the family changes (e.g. Equal to -> Contains all).
const comparatorKind = (op?: Operators): string => {
if (!op) return 'none';
if (op === Operators.ContainsAny || op === Operators.ContainsAll) {
return 'element';
}
if (
op === Operators.LengthEquals ||
op === Operators.LengthGreaterThan ||
op === Operators.LengthLessThan ||
op === Operators.LengthGreaterThanOrEqual ||
op === Operators.LengthLessThanOrEqual
) {
return 'length';
}
if (DISABLE_INPUT_OPERATORS.includes(op)) return 'none';
return 'value';
};
const valueFamilyChanged =
comparatorKind(props.adhocFilter.operatorId as Operators | undefined) !==
comparatorKind(operatorId);
let newComparator;
if (valueFamilyChanged) {
newComparator = undefined;
} else if (MULTI_OPERATORS.has(operatorId)) {
// convert between list of comparators and individual comparators
// (e.g. `in ('North America', 'Africa')` to `== 'North America'`)
// convert between list of comparators and individual comparators
// (e.g. `in ('North America', 'Africa')` to `== 'North America'`)
if (MULTI_OPERATORS.has(operatorId)) {
newComparator = Array.isArray(currentComparator)
? currentComparator
: [currentComparator].filter(element => element != null);
@@ -507,42 +433,19 @@ const AdhocFilterEditPopoverSimpleTabContent: FC<Props> = props => {
if (loadingComparatorSuggestions) {
controller.abort();
}
// Element-level array operators (Contains any / Contains all) search
// inside the array, so suggest individual elements; whole-array
// operators (=, In, …) keep the default distinct-array suggestions.
const { operatorId } = props.adhocFilter;
const arrayElements =
operatorId === Operators.ContainsAny ||
operatorId === Operators.ContainsAll;
setLoadingComparatorSuggestions(true);
SupersetClient.get({
signal,
endpoint: `/api/v1/datasource/${datasource.type}/${datasource.id}/column/${col}/values/${
arrayElements ? '?array_elements=true' : ''
}`,
endpoint: `/api/v1/datasource/${datasource.type}/${datasource.id}/column/${col}/values/`,
})
.then(({ json }) => {
setSuggestions(
json.result.map((suggestion: unknown) => {
// Complex column values arrive as JS arrays or objects: whole
// arrays for MULTI_VALUE columns (e.g. [5, 6, 7]) and Map/Tuple
// objects for nested-container columns (e.g. {"a": ["x","y"]}).
// A raw array/object is neither a valid single-select value
// (antd collapses an array to its first element) nor renderable
// as a React child (an object throws). Render it as its literal
// string, which is also exactly what the backend's
// parse_array_literal expects for the whole-array operators.
if (suggestion !== null && typeof suggestion === 'object') {
const literal = JSON.stringify(suggestion);
return { value: literal, label: literal };
}
return {
value: suggestion as null | number | boolean | string,
label: optionLabel(
suggestion as null | number | boolean | string,
),
};
}),
json.result.map(
(suggestion: null | number | boolean | string) => ({
value: suggestion,
label: optionLabel(suggestion),
}),
),
);
setLoadingComparatorSuggestions(false);
})
@@ -561,7 +464,6 @@ const AdhocFilterEditPopoverSimpleTabContent: FC<Props> = props => {
}, [
props.adhocFilter.subject,
props.adhocFilter.clause,
props.adhocFilter.operatorId,
props.datasource,
datePicker,
]);
@@ -44,17 +44,6 @@ export const OPERATORS_TO_SQL = {
'IS NULL': 'IS NULL',
'IS TRUE': 'IS TRUE',
'IS FALSE': 'IS FALSE',
// Element-level array operators (shown as filter labels; not executable SQL —
// the Custom SQL tab is hidden for these).
CONTAINS_ANY: 'CONTAINS ANY',
CONTAINS_ALL: 'CONTAINS ALL',
IS_EMPTY: 'IS EMPTY',
IS_NOT_EMPTY: 'IS NOT EMPTY',
LENGTH_EQUALS: 'LENGTH =',
LENGTH_GREATER_THAN: 'LENGTH >',
LENGTH_LESS_THAN: 'LENGTH <',
LENGTH_GREATER_THAN_OR_EQUALS: 'LENGTH >=',
LENGTH_LESS_THAN_OR_EQUALS: 'LENGTH <=',
'LATEST PARTITION': ({
datasource,
}: {
@@ -22,11 +22,10 @@ import FixedOrMetricControl from '.';
jest.mock(
'@superset-ui/core/components/Icons/AsyncIcon',
() =>
({ fileName }: { fileName: string }) =>
(
// eslint-disable-next-line jsx-a11y/prefer-tag-over-role -- mirrors AsyncIcon's real span+role="img" shape
<span role="img" aria-label={fileName.replace('_', '-')} />
),
({ fileName }: { fileName: string }) => (
// eslint-disable-next-line jsx-a11y/prefer-tag-over-role -- mirrors AsyncIcon's real span+role="img" shape
<span role="img" aria-label={fileName.replace('_', '-')} />
),
);
const createProps = () => ({
+1 -52
View File
@@ -45,17 +45,6 @@ export enum Operators {
IsTrue = 'IS_TRUE',
IsFalse = 'IS_FALSE',
TemporalRange = 'TEMPORAL_RANGE',
// Element-level operators for multi-value (array) columns
ContainsAny = 'CONTAINS_ANY',
ContainsAll = 'CONTAINS_ALL',
IsEmpty = 'IS_EMPTY',
IsNotEmpty = 'IS_NOT_EMPTY',
// Length (element-count) comparison operators for array columns
LengthEquals = 'LENGTH_EQUALS',
LengthGreaterThan = 'LENGTH_GREATER_THAN',
LengthLessThan = 'LENGTH_LESS_THAN',
LengthGreaterThanOrEqual = 'LENGTH_GREATER_THAN_OR_EQUALS',
LengthLessThanOrEqual = 'LENGTH_LESS_THAN_OR_EQUALS',
}
export interface OperatorType {
@@ -100,39 +89,6 @@ export const OPERATOR_ENUM_TO_OPERATOR_TYPE: {
display: t('TEMPORAL_RANGE'),
operation: 'TEMPORAL_RANGE',
},
[Operators.ContainsAny]: {
display: t('Contains any'),
operation: 'CONTAINS_ANY',
},
[Operators.ContainsAll]: {
display: t('Contains all'),
operation: 'CONTAINS_ALL',
},
[Operators.IsEmpty]: { display: t('Is empty'), operation: 'IS_EMPTY' },
[Operators.IsNotEmpty]: {
display: t('Is not empty'),
operation: 'IS_NOT_EMPTY',
},
[Operators.LengthEquals]: {
display: t('Length equals (=)'),
operation: 'LENGTH_EQUALS',
},
[Operators.LengthGreaterThan]: {
display: t('Length greater than (>)'),
operation: 'LENGTH_GREATER_THAN',
},
[Operators.LengthLessThan]: {
display: t('Length less than (<)'),
operation: 'LENGTH_LESS_THAN',
},
[Operators.LengthGreaterThanOrEqual]: {
display: t('Length greater or equal (>=)'),
operation: 'LENGTH_GREATER_THAN_OR_EQUALS',
},
[Operators.LengthLessThanOrEqual]: {
display: t('Length less or equal (<=)'),
operation: 'LENGTH_LESS_THAN_OR_EQUALS',
},
};
export const OPERATORS_OPTIONS = Object.values(Operators) as Operators[];
@@ -149,12 +105,7 @@ export const HAVING_OPERATORS = [
Operators.GreaterThan,
Operators.GreaterThanOrEqual,
];
export const MULTI_OPERATORS = new Set([
Operators.In,
Operators.NotIn,
Operators.ContainsAny,
Operators.ContainsAll,
]);
export const MULTI_OPERATORS = new Set([Operators.In, Operators.NotIn]);
// CUSTOM_OPERATORS will show operator in simple mode,
// but will generate customized sqlExpression
export const CUSTOM_OPERATORS = new Set([
@@ -169,8 +120,6 @@ export const DISABLE_INPUT_OPERATORS = [
Operators.LatestPartition,
Operators.IsTrue,
Operators.IsFalse,
Operators.IsEmpty,
Operators.IsNotEmpty,
];
export const sqlaAutoGeneratedMetricNameRegex =
@@ -82,14 +82,3 @@ test('Should handle boolean true comparator as a string value', () => {
"subject operator 'TRUE'",
);
});
test('Should render array-literal comparators as-is (not quoted)', () => {
// Whole-array = filter: the pasted array literal is shown unquoted.
expect(getSimpleSQLExpression('ingredients', '=', "['1 large egg']")).toBe(
"ingredients = ['1 large egg']",
);
// IN with multiple array literals.
expect(
getSimpleSQLExpression('ingredients', Operators.In, ["['a']", "['b']"]),
).toBe(`ingredients ${Operators.In} (['a'], ['b'])`);
});
@@ -461,15 +461,10 @@ export const getSimpleSQLExpression = (
if (comparatorArray.length > 0 && showComparator) {
const formattedComparators = comparatorArray
.map(val => optionLabel(val))
.map(val => {
// Array-literal values (e.g. ['a', 'b']) are shown as-is rather than
// quoted/escaped as a string, so array-column filters read naturally.
const asString = String(val);
if (asString.startsWith('[') && asString.endsWith(']')) {
return asString;
}
return `${quote}${isString ? asString.replace(/'/g, "''") : val}${quote}`;
});
.map(
val =>
`${quote}${isString ? String(val).replace(/'/g, "''") : val}${quote}`,
);
expression += ` ${prefix}${formattedComparators.join(', ')}${suffix}`;
}
}
@@ -38,11 +38,10 @@ import {
jest.mock(
'@superset-ui/core/components/Icons/AsyncIcon',
() =>
({ fileName }: { fileName: string }) =>
(
// eslint-disable-next-line jsx-a11y/prefer-tag-over-role -- mirrors AsyncIcon's real span+role="img" shape
<span role="img" aria-label={fileName.replace('_', '-')} />
),
({ fileName }: { fileName: string }) => (
// eslint-disable-next-line jsx-a11y/prefer-tag-over-role -- mirrors AsyncIcon's real span+role="img" shape
<span role="img" aria-label={fileName.replace('_', '-')} />
),
);
// eslint-disable-next-line no-restricted-globals -- TODO: Migrate from describe blocks
@@ -29,11 +29,10 @@ import DatasetPanelWrapper from 'src/features/datasets/AddDataset/DatasetPanel';
jest.mock(
'@superset-ui/core/components/Icons/AsyncIcon',
() =>
({ fileName }: { fileName: string }) =>
(
// eslint-disable-next-line jsx-a11y/prefer-tag-over-role -- mirrors AsyncIcon's real span+role="img" shape
<span role="img" aria-label={fileName.replace('_', '-')} />
),
({ fileName }: { fileName: string }) => (
// eslint-disable-next-line jsx-a11y/prefer-tag-over-role -- mirrors AsyncIcon's real span+role="img" shape
<span role="img" aria-label={fileName.replace('_', '-')} />
),
);
const errorMessageRegistry = getErrorMessageComponentRegistry();
@@ -31,12 +31,7 @@ import {
import { Group, Role, UserObject } from 'src/pages/UsersList/types';
import { Actions } from 'src/constants';
import { BaseUserListModalProps, FormValues } from './types';
import {
createUser,
updateUser,
atLeastOneRoleOrGroup,
handleUserError,
} from './utils';
import { createUser, updateUser, atLeastOneRoleOrGroup } from './utils';
export interface UserModalProps extends BaseUserListModalProps {
roles: Role[];
@@ -56,6 +51,36 @@ function UserListModal({
}: UserModalProps) {
const { addDangerToast, addSuccessToast } = useToasts();
const handleFormSubmit = async (values: FormValues) => {
const handleError = async (
err: any,
action: Actions.CREATE | Actions.UPDATE,
) => {
let errorMessage =
action === Actions.CREATE
? t('There was an error creating the user. Please, try again.')
: t('There was an error updating the user. Please, try again.');
if (err.status === 422) {
const errorData = await err.json();
const detail = errorData?.message || '';
if (detail.includes('duplicate key value')) {
if (detail.includes('ab_user_username_key')) {
errorMessage = t(
'This username is already taken. Please choose another one.',
);
} else if (detail.includes('ab_user_email_key')) {
errorMessage = t(
'This email is already associated with an account. Please choose another one.',
);
}
}
}
addDangerToast(errorMessage);
throw err;
};
if (isEditMode) {
if (!user) {
throw new Error('User is required in edit mode');
@@ -64,14 +89,14 @@ function UserListModal({
await updateUser(user.id, values);
addSuccessToast(t('The user has been updated successfully.'));
} catch (err) {
await handleUserError(err as Response, Actions.UPDATE, addDangerToast);
await handleError(err, Actions.UPDATE);
}
} else {
try {
await createUser(values);
addSuccessToast(t('The user has been created successfully.'));
} catch (err) {
await handleUserError(err as Response, Actions.CREATE, addDangerToast);
await handleError(err, Actions.CREATE);
}
}
};
@@ -1,99 +0,0 @@
/**
* Licensed to the Apache Software Foundation (ASF) under one
* or more contributor license agreements. See the NOTICE file
* distributed with this work for additional information
* regarding copyright ownership. The ASF licenses this file
* to you under the Apache License, Version 2.0 (the
* "License"); you may not use this file except in compliance
* with the License. You may obtain a copy of the License at
*
* http://www.apache.org/licenses/LICENSE-2.0
*
* Unless required by applicable law or agreed to in writing,
* software distributed under the License is distributed on an
* "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY
* KIND, either express or implied. See the License for the
* specific language governing permissions and limitations
* under the License.
*/
import { Actions } from 'src/constants';
import { handleUserError } from './utils';
test('shows the password validation message from a 400 response', async () => {
const error = new Response(
JSON.stringify({
message: {
password: ['Password must be at least 8 characters long.'],
},
}),
{ status: 400 },
);
const addDangerToast = jest.fn();
await expect(
handleUserError(error, Actions.CREATE, addDangerToast),
).rejects.toBe(error);
expect(addDangerToast).toHaveBeenCalledWith(
'Password must be at least 8 characters long.',
);
});
test('shows a plain string message from a 400 response', async () => {
const error = new Response(
JSON.stringify({ message: 'User must have at least one role or group!' }),
{ status: 400 },
);
const addDangerToast = jest.fn();
await expect(
handleUserError(error, Actions.UPDATE, addDangerToast),
).rejects.toBe(error);
expect(addDangerToast).toHaveBeenCalledWith(
'User must have at least one role or group!',
);
});
test('keeps the duplicate username message for a 422 response', async () => {
const error = new Response(
JSON.stringify({
message:
'duplicate key value violates unique constraint "ab_user_username_key"',
}),
{ status: 422 },
);
const addDangerToast = jest.fn();
await expect(
handleUserError(error, Actions.CREATE, addDangerToast),
).rejects.toBe(error);
expect(addDangerToast).toHaveBeenCalledWith(
'This username is already taken. Please choose another one.',
);
});
test('shows the generic message when a 422 response has no message', async () => {
const error = new Response(JSON.stringify({ foo: 'bar' }), { status: 422 });
const addDangerToast = jest.fn();
await expect(
handleUserError(error, Actions.CREATE, addDangerToast),
).rejects.toBe(error);
expect(addDangerToast).toHaveBeenCalledWith(
'There was an error creating the user. Please, try again.',
);
});
test('shows the generic message when a 400 response is not JSON', async () => {
const error = new Response('<html>Bad request</html>', {
status: 400,
headers: { 'Content-Type': 'text/html' },
});
const addDangerToast = jest.fn();
await expect(
handleUserError(error, Actions.CREATE, addDangerToast),
).rejects.toBe(error);
expect(addDangerToast).toHaveBeenCalledWith(
'There was an error creating the user. Please, try again.',
);
});
+1 -40
View File
@@ -17,49 +17,10 @@
* under the License.
*/
import { t } from '@apache-superset/core/translation';
import { getClientErrorObject, SupersetClient } from '@superset-ui/core';
import { SupersetClient } from '@superset-ui/core';
import { SelectOption } from 'src/components/ListView';
import { Actions } from 'src/constants';
import { FormValues } from './types';
type AddDangerToast = (message: string) => void;
export const handleUserError = async (
err: Response,
action: Actions.CREATE | Actions.UPDATE,
addDangerToast: AddDangerToast,
): Promise<never> => {
let errorMessage =
action === Actions.CREATE
? t('There was an error creating the user. Please, try again.')
: t('There was an error updating the user. Please, try again.');
if (err.status === 400 || err.status === 422) {
const errorData = await getClientErrorObject(err);
const message: unknown = errorData.message;
if (err.status === 400 && message && errorData.error) {
errorMessage = errorData.error;
} else if (
err.status === 422 &&
errorData.error?.includes('duplicate key value')
) {
if (errorData.error.includes('ab_user_username_key')) {
errorMessage = t(
'This username is already taken. Please choose another one.',
);
} else if (errorData.error.includes('ab_user_email_key')) {
errorMessage = t(
'This email is already associated with an account. Please choose another one.',
);
}
}
}
addDangerToast(errorMessage);
throw err;
};
export const createUser = async (values: FormValues) => {
const { confirmPassword: _confirmPassword, ...payload } = values;
if (payload.active == null) {
@@ -1157,34 +1157,6 @@ test('dataset links use internal routing when PREVENT_UNSAFE_DEFAULT_URLS_ON_DAT
});
});
test('legacy dashboard default URLs use the registered client route', async () => {
const dataset = {
...mockDatasets[0],
explore_url: '/superset/dashboard/123/?standalone=1#section',
};
mockDatasetListEndpoints({ result: [dataset], count: 1 });
renderDatasetList(
mockAdminUser,
{},
{
common: {
conf: {
PREVENT_UNSAFE_DEFAULT_URLS_ON_DATASET: true,
},
},
},
);
const datasetLink = await screen.findByRole('link', {
name: dataset.table_name,
});
expect(datasetLink).toHaveAttribute(
'href',
'/dashboard/123/?standalone=1#section',
);
});
// Note: These delete error tests verify that the modal doesn't open when fetching
// related_objects fails. The component's openDatasetDeleteModal error handler
// (index.tsx:262-268) returns a string but doesn't call addDangerToast(), so no
@@ -54,18 +54,10 @@ import {
const APP_ROOT = '/superset';
const renderUnderSubdirectory = (preventUnsafeDefaultUrls = false) => {
const defaultState = createDefaultStoreState(mockAdminUser);
const renderUnderSubdirectory = () => {
const store = createMockStore({
...defaultState,
...createDefaultStoreState(mockAdminUser),
user: mockAdminUser,
common: {
...defaultState.common,
conf: {
...defaultState.common?.conf,
PREVENT_UNSAFE_DEFAULT_URLS_ON_DATASET: preventUnsafeDefaultUrls,
},
},
});
return render(
<Provider store={store}>
@@ -123,31 +115,6 @@ test('explore link is single-prefixed under a subdirectory deployment', async ()
expect(exploreLink.getAttribute('href')).not.toContain('/superset/superset');
});
test('legacy dashboard default URL uses the router basename once', async () => {
// A subdirectory user pastes the full browser path, so the saved value
// carries both the application root and the legacy `/superset` prefix.
// stripAppRoot removes the root and the legacy normalization removes the
// prefix, leaving the basename to re-add the root exactly once.
const dataset = {
...mockDatasets[0],
explore_url: `${APP_ROOT}/superset/dashboard/123/?standalone=1#section`,
};
mockDatasetListEndpoints({ result: [dataset], count: 1 });
renderUnderSubdirectory(true);
const dashboardLink = await screen.findByRole('link', {
name: dataset.table_name,
});
expect(dashboardLink).toHaveAttribute(
'href',
`${APP_ROOT}/dashboard/123/?standalone=1#section`,
);
expect(dashboardLink.getAttribute('href')).not.toContain(
'/superset/superset',
);
});
test('external default_endpoint passes through unprefixed', async () => {
const dataset = {
...mockDatasets[0],
@@ -87,6 +87,7 @@ import withToasts from 'src/components/MessageToasts/withToasts';
import { Icons } from '@superset-ui/core/components/Icons';
import WarningIconWithTooltip from '@superset-ui/core/components/WarningIconWithTooltip';
import { isUserEditorOrAdmin } from 'src/dashboard/util/permissionUtils';
import {
PAGE_SIZE,
SORT_BY,
@@ -113,10 +114,6 @@ import type {
} from 'src/types/bootstrapTypes';
import type User from 'src/types/User';
// Keep saved Default URLs compatible with the prefix-free SPA route.
const normalizeLegacyDashboardUrl = (url: string) =>
url.replace(/^\/superset(?=\/dashboard(?:\/|$))/, '');
const SEMANTIC_LAYERS_FLAG = 'SEMANTIC_LAYERS' as FeatureFlag;
type DatasetExtra = {
certification?: {
@@ -725,9 +722,7 @@ const DatasetList: FunctionComponent<DatasetListProps> = ({
// Router basename, which re-prefixes the root — so strip it here to
// avoid a doubled `/superset/superset/...`. External
// `default_endpoint` URLs pass through unchanged.
const exploreTo = normalizeLegacyDashboardUrl(
stripAppRoot(exploreURL),
);
const exploreTo = stripAppRoot(exploreURL);
let titleLink: JSX.Element;
if (PREVENT_UNSAFE_DEFAULT_URLS_ON_DATASET) {
titleLink = (
+4 -4
View File
@@ -28,7 +28,7 @@
"@typescript-eslint/parser": "^8.67.0",
"eslint": "^10.8.1",
"eslint-config-prettier": "^10.1.8",
"globals": "^17.11.0",
"globals": "^17.9.0",
"oxfmt": "^0.63.0",
"tscw-config": "^1.1.2",
"typescript": "^6.0.3",
@@ -2053,9 +2053,9 @@
}
},
"node_modules/globals": {
"version": "17.11.0",
"resolved": "https://registry.npmjs.org/globals/-/globals-17.11.0.tgz",
"integrity": "sha512-Z2I8hM+PbJDXQDq3Icgpzv+mPdwr68iZUU9d5WW4FuXfDUQfkZaZuvjMv42/5crNyw154+9+VWXbYrUgDXbxNw==",
"version": "17.9.0",
"resolved": "https://registry.npmjs.org/globals/-/globals-17.9.0.tgz",
"integrity": "sha512-m/MvAW61QVU5VDNF1Vj8axt016h8w7L5TU1e9zlab7XIttAT2YAlCwl75K1fOqvMM9apmD7lbCIRhpfkhmxhCg==",
"dev": true,
"license": "MIT",
"engines": {
+1 -1
View File
@@ -36,7 +36,7 @@
"@typescript-eslint/parser": "^8.67.0",
"eslint": "^10.8.1",
"eslint-config-prettier": "^10.1.8",
"globals": "^17.11.0",
"globals": "^17.9.0",
"oxfmt": "^0.63.0",
"tscw-config": "^1.1.2",
"typescript": "^6.0.3",
+24 -2
View File
@@ -33,7 +33,11 @@ from superset.commands.dataset.exceptions import (
)
from superset.commands.utils import populate_subjects
from superset.daos.dataset import DatasetDAO
from superset.exceptions import SupersetParseError, SupersetSecurityException
from superset.exceptions import (
SupersetException,
SupersetParseError,
SupersetSecurityException,
)
from superset.extensions import security_manager
from superset.sql.parse import Table
from superset.utils.decorators import on_error, transaction
@@ -50,7 +54,25 @@ class CreateDatasetCommand(CreateMixin, BaseCommand):
self.validate()
dataset = DatasetDAO.create(attributes=self._properties)
dataset.fetch_metadata()
try:
dataset.fetch_metadata()
except SupersetException as ex:
# Not a SQLAlchemyError, so ``on_error`` re-raises it untouched and
# it escapes to FAB's ``@safe`` as an opaque 500 "Fatal error".
# Deliberately covers the 403 ``SupersetSecurityException`` raised
# for mutation/multi-statement SQL too: ``validate()`` already
# reports that class of rejection as a 422 on ``sql`` via
# ``DatasetDataAccessIsNotAllowed``.
raise DatasetInvalidError(
exceptions=[
ValidationError(
# ``lazy_gettext`` messages aren't ``str``, so
# marshmallow won't wrap them into a list on its own.
[str(ex.message)],
field_name="sql" if self._properties.get("sql") else "table",
)
]
) from ex
return dataset
def validate(self) -> None: # noqa: C901
+2 -81
View File
@@ -17,7 +17,6 @@
# pylint: disable=invalid-name
from __future__ import annotations
import inspect
import logging
from datetime import datetime
from pprint import pformat
@@ -206,86 +205,8 @@ class QueryObject: # pylint: disable=too-many-instance-attributes
def _set_post_processing(
self, post_processing: list[dict[str, Any] | None] | None
) -> None:
self.post_processing = [
self._drop_unsupported_options(post_proc)
for post_proc in post_processing or []
if post_proc
]
@staticmethod
def _drop_unsupported_options(post_proc: dict[str, Any]) -> dict[str, Any]:
"""
Drop options that the post-processing operation no longer accepts.
A chart's ``query_context`` is written when the chart is saved and is
never rewritten afterwards, while Explore rebuilds the query from
``form_data`` at every render. A chart saved by an older version of
Superset can therefore reference an option that has since been removed
from the operation. ``exec_post_processing`` passes the stored options
as keyword arguments, so that option raises a bare ``TypeError`` on
every path that replays the stored ``query_context`` -- the chart data
endpoint, alerts and reports, thumbnails, CSV export -- while the same
chart still renders correctly in Explore.
Comparing against the signature avoids a hard-coded list of removed
option names, which would need extending at each release.
"""
operation = post_proc.get("operation")
function = (
getattr(pandas_postprocessing, operation, None)
if isinstance(operation, str)
else None
)
if function is None:
# A missing or unknown operation is left untouched, so that
# exec_post_processing reports it as InvalidPostProcessingError.
return post_proc
parameters = inspect.signature(function).parameters
if any(
parameter.kind is inspect.Parameter.VAR_KEYWORD
for parameter in parameters.values()
):
return post_proc
# `exec_post_processing` calls the operation as `operation(df, **options)`,
# so an option can only reach a parameter that a caller may fill by
# keyword. That excludes the first parameter, which receives the
# DataFrame positionally, and any positional-only or `*args` parameter.
keyword_parameters = {
name
for position, (name, parameter) in enumerate(parameters.items())
if position > 0
and parameter.kind
in (
inspect.Parameter.POSITIONAL_OR_KEYWORD,
inspect.Parameter.KEYWORD_ONLY,
)
}
options = post_proc.get("options") or {}
unsupported = {key for key in options if key not in keyword_parameters}
if not unsupported:
return post_proc
# Logged at info: a chart saved before the option was removed hits this
# on every render, so a warning would repeat for as long as the chart
# is not resaved, without anything new to report.
logger.info(
"Dropping unsupported option(s) %s of post-processing operation "
"`%s`. The chart's stored query_context predates the current "
"signature of that operation.",
sorted(unsupported),
operation,
)
return {
**post_proc,
"options": {
key: value
for key, value in options.items()
if key in keyword_parameters
},
}
post_processing = post_processing or []
self.post_processing = [post_proc for post_proc in post_processing if post_proc]
def _init_series_columns(
self,
+1 -7
View File
@@ -957,13 +957,7 @@ class AnnotationDatasource(BaseDatasource):
def get_query_str(self, query_obj: QueryObjectDict) -> str:
raise NotImplementedError()
def values_for_column(
self,
column_name: str,
limit: int = 10000,
denormalize_column: bool = False,
array_elements: bool = False,
) -> list[Any]:
def values_for_column(self, column_name: str, limit: int = 10000) -> list[Any]:
raise NotImplementedError()
-5
View File
@@ -133,9 +133,6 @@ class DatasourceRestApi(BaseSupersetApi):
row_limit = apply_max_row_limit(app.config["FILTER_SELECT_ROW_LIMIT"])
denormalize_column = not datasource.normalize_columns
# Element-level operators (Contains any / Contains all) request the
# distinct array *elements* rather than distinct whole arrays.
array_elements = parse_boolean_string(request.args.get("array_elements"))
# Cache distinct column-value results so a dashboard with many filters
# backed by the same (often heavy) virtual dataset doesn't re-execute
@@ -168,7 +165,6 @@ class DatasourceRestApi(BaseSupersetApi):
"col": column_name,
"limit": row_limit,
"denorm": denormalize_column,
"elements": array_elements,
"rls": security_manager.get_rls_cache_key(datasource),
"changed_on": str(getattr(datasource, "changed_on", "")),
},
@@ -193,7 +189,6 @@ class DatasourceRestApi(BaseSupersetApi):
column_name=column_name,
limit=row_limit,
denormalize_column=denormalize_column,
array_elements=array_elements,
)
except KeyError:
return self.response(
+1 -111
View File
@@ -55,13 +55,7 @@ from sqlalchemy.engine.reflection import Inspector
from sqlalchemy.engine.url import URL
from sqlalchemy.ext.compiler import compiles
from sqlalchemy.sql import literal_column, quoted_name, text
from sqlalchemy.sql.expression import (
BinaryExpression,
ColumnClause,
ColumnElement,
Select,
TextClause,
)
from sqlalchemy.sql.expression import BinaryExpression, ColumnClause, Select, TextClause
from sqlalchemy.types import TypeEngine
from superset import db
@@ -534,11 +528,6 @@ class BaseEngineSpec: # pylint: disable=too-many-public-methods
time_groupby_inline = False
limit_method = LimitMethod.FORCE_LIMIT
supports_multivalues_insert = False
# Whether this engine supports first-class multi-value (array-typed) columns.
# When True, array columns are classified as ``GenericDataType.MULTI_VALUE`` and
# the ``array_*`` capability methods below must be implemented. Defaults to
# False so engines that have not opted in keep treating arrays as strings.
supports_multivalue_columns = False
allows_joins = True
allows_subqueries = True
allows_alias_in_select = True
@@ -2582,105 +2571,6 @@ class BaseEngineSpec: # pylint: disable=too-many-public-methods
logger.error(ex, exc_info=True)
raise
@classmethod
def array_contains_any(cls, col: ColumnElement, values: list[Any]) -> ColumnElement:
"""
Build a boolean expression testing whether array column ``col`` contains
**any** of ``values`` (element-level membership, like ``IN``). Engines
that set ``supports_multivalue_columns = True`` must override this with
their native function (e.g. ClickHouse ``hasAny``).
:param col: SQLAlchemy column element for the array column
:param values: element values to look for inside the array
:return: a SQLAlchemy boolean expression
"""
raise NotImplementedError(
f"{cls.engine} does not support multi-value (array) columns"
)
@classmethod
def array_contains_all(cls, col: ColumnElement, values: list[Any]) -> ColumnElement:
"""
Build a boolean expression testing whether array column ``col`` contains
**all** of ``values``. Engines that set
``supports_multivalue_columns = True`` must override this with their
native function (e.g. ClickHouse ``hasAll``).
:param col: SQLAlchemy column element for the array column
:param values: element values that must all be present
:return: a SQLAlchemy boolean expression
"""
raise NotImplementedError(
f"{cls.engine} does not support multi-value (array) columns"
)
@classmethod
def array_length(cls, col: ColumnElement) -> ColumnElement:
"""
Build a numeric expression returning the number of elements in array
column ``col``. Engines that set ``supports_multivalue_columns = True``
must override this with their native array-length function. Used both for
the ``Length`` filter and the ``Is empty`` / ``Is not empty`` operators.
:param col: SQLAlchemy column element for the array column
:return: a SQLAlchemy numeric expression
"""
raise NotImplementedError(
f"{cls.engine} does not support multi-value (array) columns"
)
@classmethod
def array_literal(cls, values: list[Any]) -> ColumnElement:
"""
Build an array-literal expression from ``values`` (e.g. ClickHouse
``array(v1, v2)`` == ``[v1, v2]``). Used for the whole-array (column-
level) operators ``=`` / ``!=`` / ``IN`` / ``NOT IN`` where the array is
compared as a single value. Engines that set
``supports_multivalue_columns = True`` must override this.
:param values: element values that make up the array
:return: a SQLAlchemy array-literal expression
"""
raise NotImplementedError(
f"{cls.engine} does not support multi-value (array) columns"
)
@classmethod
def array_explode(cls, col: ColumnElement) -> ColumnElement:
"""
Build an expression that expands array column ``col`` into one row per
element (e.g. ClickHouse ``arrayJoin``). Used to source **element-level**
value suggestions (``SELECT DISTINCT array_explode(col)``) for the
``Contains any`` / ``Contains all`` filter operators, so the picker offers
individual elements rather than whole arrays. Engines that set
``supports_multivalue_columns = True`` must override this.
:param col: SQLAlchemy column element for the array column
:return: a SQLAlchemy expression yielding one element per row
"""
raise NotImplementedError(
f"{cls.engine} does not support multi-value (array) columns"
)
@classmethod
def get_array_element_type( # pylint: disable=unused-argument
cls, native_type: str | None
) -> GenericDataType | None:
"""
Return the generic type of an array column's **element** type, derived
from its native type string (e.g. ClickHouse ``Array(Int32)`` ->
``NUMERIC``), or ``None`` when the engine has no array support or the
element type cannot be resolved.
Callers use this to coerce filter values to the element type before
building array expressions, so, for example, a ``Contains any`` filter on
a numeric array compares against numbers rather than quoted strings.
:param native_type: native column type string of the array column
:return: the element's :class:`GenericDataType`, or ``None``
"""
return None
@classmethod
def get_column_spec( # pylint: disable=unused-argument
cls,
+4 -62
View File
@@ -26,9 +26,8 @@ from flask import current_app as app
from flask_babel import gettext as __
from marshmallow import fields, Schema
from marshmallow.validate import Range
from sqlalchemy import func, types
from sqlalchemy import types
from sqlalchemy.engine.url import URL
from sqlalchemy.sql.expression import ColumnElement
from urllib3.exceptions import NewConnectionError
from superset.databases.utils import make_url_safe
@@ -56,7 +55,6 @@ class ClickHouseBaseEngineSpec(BaseEngineSpec):
time_groupby_inline = True
supports_multivalues_insert = True
supports_multivalue_columns = True
# ClickHouse doesn't support IS true/false syntax, use = true/false instead
use_equality_for_boolean_filters = True
@@ -130,18 +128,12 @@ class ClickHouseBaseEngineSpec(BaseEngineSpec):
column_type_mappings = (
(
# Anchor to the start so only top-level arrays match. This must be
# ordered before the ``Enum`` entry below: ``Array(Enum8(...))`` is a
# real array and should classify as MULTI_VALUE, not STRING. The
# anchor also prevents over-matching nested arrays such as
# ``Map(String, Array(String))`` or ``Tuple(Array(String))``, which
# are not themselves array columns and must keep their own type.
re.compile(r"^Array\(", re.IGNORECASE),
re.compile(r".*Enum.*", re.IGNORECASE),
types.String(),
GenericDataType.MULTI_VALUE,
GenericDataType.STRING,
),
(
re.compile(r".*Enum.*", re.IGNORECASE),
re.compile(r".*Array.*", re.IGNORECASE),
types.String(),
GenericDataType.STRING,
),
@@ -182,56 +174,6 @@ class ClickHouseBaseEngineSpec(BaseEngineSpec):
),
)
@classmethod
def array_contains_any(cls, col: ColumnElement, values: list[Any]) -> ColumnElement:
# ClickHouse: hasAny(arr, [v1, v2]) -> 1 if arr shares any element.
# func.array(*values) renders as array(v1, v2) == [v1, v2].
return func.hasAny(col, func.array(*values))
@classmethod
def array_contains_all(cls, col: ColumnElement, values: list[Any]) -> ColumnElement:
# ClickHouse: hasAll(arr, [v1, v2]) -> 1 if arr contains all elements.
return func.hasAll(col, func.array(*values))
@classmethod
def array_length(cls, col: ColumnElement) -> ColumnElement:
# ClickHouse: length(arr) -> number of elements
return func.length(col)
@classmethod
def array_literal(cls, values: list[Any]) -> ColumnElement:
# ClickHouse: array(v1, v2) is equivalent to the literal [v1, v2].
return func.array(*values)
@classmethod
def array_explode(cls, col: ColumnElement) -> ColumnElement:
# ClickHouse: arrayJoin(arr) yields one row per element, so
# SELECT DISTINCT arrayJoin(arr) returns the distinct elements.
return func.arrayJoin(col)
# Matches the element type inside a top-level ``Array(...)`` column, e.g.
# ``Array(Int32)`` -> ``Int32``, ``Array(Nullable(String))`` -> ``String``.
_ARRAY_ELEMENT_RE = re.compile(r"^Array\((?P<inner>.+)\)$", re.IGNORECASE)
# Element-type wrappers that don't change the underlying generic type.
_ELEMENT_WRAPPER_RE = re.compile(
r"^(?:Nullable|LowCardinality)\((?P<inner>.+)\)$", re.IGNORECASE
)
@classmethod
def get_array_element_type(cls, native_type: str | None) -> GenericDataType | None:
if not native_type:
return None
match = cls._ARRAY_ELEMENT_RE.match(native_type.strip())
if not match:
return None
inner = match.group("inner").strip()
# Peel wrappers (Nullable/LowCardinality) that don't alter the generic
# type so the inner scalar type drives classification.
while wrapper := cls._ELEMENT_WRAPPER_RE.match(inner):
inner = wrapper.group("inner").strip()
spec = cls.get_column_spec(inner)
return spec.generic_type if spec else None
@classmethod
def epoch_to_dttm(cls) -> str:
return "{col}"
+1 -16
View File
@@ -63,7 +63,6 @@ from superset.mcp_service.utils import (
sanitize_for_llm_context,
)
from superset.mcp_service.utils.response_utils import humanize_timestamp
from superset.sql.parse import has_aggregate
from superset.utils import json
@@ -387,27 +386,13 @@ class CreateDatasetMetric(BaseModel):
"""Metric definition for dataset creation."""
metric_name: str = Field(..., description="Name of the metric")
expression: str = Field(
...,
description="Aggregate SQL expression for the metric, e.g. SUM(amount)",
)
expression: str = Field(..., description="SQL expression for the metric")
verbose_name: str | None = None
description: str | None = None
metric_type: str | None = None
d3format: str | None = None
warning_text: str | None = None
@field_validator("expression")
@classmethod
def expression_must_aggregate(cls, value: str) -> str:
if not has_aggregate(value):
raise ValueError(
"saved metrics must aggregate rows; wrap a row-level column in "
"an aggregate such as MAX(column), or omit the saved metric and "
"use the dataset column directly"
)
return value
class CreateDatasetCalculatedColumn(BaseModel):
"""Calculated column definition for dataset creation."""
@@ -21,7 +21,6 @@ from typing import Any
from fastmcp import Context
from superset_core.mcp.decorators import tool, ToolAnnotations
from superset.exceptions import SupersetGenericDBErrorException
from superset.extensions import event_logger
from superset.mcp_service.dataset.schemas import (
CreateVirtualDatasetRequest,
@@ -68,17 +67,14 @@ def _cleanup_failed_dataset(dataset_id: int) -> None:
def _update_virtual_dataset(dataset_id: int, update_props: dict[str, Any]) -> Any:
from superset.commands.dataset.exceptions import (
DatasetInvalidError,
DatasetUpdateFailedError,
)
from superset.commands.dataset.exceptions import DatasetUpdateFailedError
from superset.commands.dataset.update import UpdateDatasetCommand
try:
return UpdateDatasetCommand(dataset_id, update_props).run()
except Exception as exc:
_cleanup_failed_dataset(dataset_id)
if not isinstance(exc, (DatasetInvalidError, DatasetUpdateFailedError)):
if not isinstance(exc, DatasetUpdateFailedError):
raise DatasetUpdateFailedError() from exc
raise
@@ -93,7 +89,7 @@ def _update_virtual_dataset(dataset_id: int, update_props: dict[str, Any]) -> An
destructiveHint=False,
),
)
async def create_virtual_dataset( # noqa: C901
async def create_virtual_dataset(
request: CreateVirtualDatasetRequest, ctx: Context
) -> CreateVirtualDatasetResponse:
"""Save a SQL query as a virtual dataset so it can be charted.
@@ -217,18 +213,6 @@ async def create_virtual_dataset( # noqa: C901
url=None,
error=f"Failed to update dataset metadata (creation rolled back): {exc}",
)
except SupersetGenericDBErrorException as exc:
logger.warning("Virtual dataset SQL validation failed", exc_info=True)
await ctx.warning(f"Virtual dataset SQL failed validation: {exc}")
return CreateVirtualDatasetResponse(
id=None,
dataset_name=request.dataset_name,
sql=request.sql,
database_id=request.database_id,
columns=[],
url=None,
error=f"Dataset SQL could not be executed: {exc}",
)
except Exception as exc:
await ctx.error(
f"Unexpected error creating virtual dataset: {type(exc).__name__}: {exc}"
+3 -207
View File
@@ -19,7 +19,6 @@
from __future__ import annotations
import ast
import builtins
import copy
import dataclasses
@@ -418,52 +417,6 @@ UUID_NATIVE_TYPE_RE: re.Pattern[str] = re.compile(
)
def parse_array_literal(value: Any) -> list[Any]:
"""
Parse a user-entered array literal (e.g. ``['a', 'b']`` or ``[1, 2]``) into a
list of elements, for the whole-array (column-level) array operators.
Accepts either an actual list/tuple, a bracketed literal string (parsed with
``ast.literal_eval``), or a plain scalar (wrapped into a single-element list).
Falls back to a single-element list when the string is not a valid literal.
"""
if isinstance(value, (list, tuple)):
return list(value)
if isinstance(value, str):
stripped = value.strip()
if stripped.startswith("[") and stripped.endswith("]"):
try:
parsed = ast.literal_eval(stripped)
except (ValueError, SyntaxError):
parsed = None
if isinstance(parsed, (list, tuple)):
return list(parsed)
return [value]
def coerce_array_values(
values: list[Any], element_type: Optional[utils.GenericDataType]
) -> list[Any]:
"""
Coerce array-element ``values`` to the array column's element type so the
emitted literal matches the column. Array columns map to a SQLAlchemy
``String`` type, so values arrive as strings and would otherwise build
string literals (e.g. ``array('5')``) that fail against a numeric array on
the server. Numeric elements are cast to numbers and boolean elements to
booleans; every other element type (string, temporal, enum, unknown) is left
untouched.
:param values: element values entered for an array filter
:param element_type: the array's element :class:`GenericDataType`, or None
:return: the coerced values
"""
if element_type == utils.GenericDataType.NUMERIC:
return [utils.cast_to_num(v) if isinstance(v, str) else v for v in values]
if element_type == utils.GenericDataType.BOOLEAN:
return [utils.cast_to_boolean(v) if isinstance(v, str) else v for v in values]
return values
def is_uuid_native_type(native_type: Optional[str]) -> bool:
"""
Return True if a native column type represents a UUID.
@@ -3699,7 +3652,6 @@ class ExploreMixin: # pylint: disable=too-many-public-methods
column_name: str,
limit: int = 10000,
denormalize_column: bool = False,
array_elements: bool = False,
) -> list[Any]:
# denormalize column name before querying for values
# unless disabled in the dataset configuration
@@ -3714,25 +3666,13 @@ class ExploreMixin: # pylint: disable=too-many-public-methods
tp = self.get_template_processor()
tbl, cte = self.get_from_clause(tp)
db_engine_spec = self.database.db_engine_spec
value_expr = target_col.get_sqla_col(template_processor=tp)
# For element-level operators (Contains any / Contains all) on a
# multi-value (array) column, suggest the distinct **elements** rather
# than distinct whole arrays by expanding the array first (e.g. ClickHouse
# arrayJoin). Only when the engine supports arrays and the column is
# actually an array column; otherwise fall back to whole-value suggestions.
if array_elements and db_engine_spec.supports_multivalue_columns:
col_spec = db_engine_spec.get_column_spec(native_type=target_col.type)
if col_spec and col_spec.generic_type == GenericDataType.MULTI_VALUE:
value_expr = db_engine_spec.array_explode(value_expr)
qry = (
sa.select(
# The alias (label) here is important because some dialects will
# automatically add a random alias to the projection because of the
# call to DISTINCT; others will uppercase the column names. This
# gives us a deterministic column name in the dataframe.
value_expr.label("column_values")
target_col.get_sqla_col(template_processor=tp).label("column_values")
)
.select_from(tbl)
.distinct()
@@ -4419,7 +4359,7 @@ class ExploreMixin: # pylint: disable=too-many-public-methods
elif is_adhoc_column(flt_col):
try:
sqla_col, adhoc_generic_type = self.adhoc_column_to_sqla(
cast("AdhocColumn", flt_col),
flt_col,
force_type_check=True,
template_processor=template_processor,
)
@@ -4493,21 +4433,9 @@ class ExploreMixin: # pylint: disable=too-many-public-methods
sqla_col = Grouping(sqla_col)
col_type = col_obj.type if col_obj else None
col_spec = db_engine_spec.get_column_spec(native_type=col_type)
is_multivalue_col = bool(
col_spec and col_spec.generic_type == GenericDataType.MULTI_VALUE
)
# Element type of an array column (e.g. Array(Int32) -> NUMERIC),
# used to coerce filter values before building array expressions.
array_element_type = (
db_engine_spec.get_array_element_type(col_type)
if is_multivalue_col
else None
)
is_list_target = op in (
utils.FilterOperator.IN,
utils.FilterOperator.NOT_IN,
utils.FilterOperator.CONTAINS_ANY,
utils.FilterOperator.CONTAINS_ALL,
)
col_advanced_data_type = col_obj.advanced_data_type if col_obj else ""
@@ -4562,56 +4490,7 @@ class ExploreMixin: # pylint: disable=too-many-public-methods
sqla_col, op, bus_resp["values"]
)
)
elif is_multivalue_col and op in {
utils.FilterOperator.EQUALS,
utils.FilterOperator.NOT_EQUALS,
utils.FilterOperator.IN,
utils.FilterOperator.NOT_IN,
}:
# Whole-array (column-level) comparison against array
# literal(s). The value is a pasted array literal like
# ``['a', 'b']`` (parsed into elements): ``col = ['a', 'b']``
# for = / !=; for IN / NOT IN each entered value is one such
# array literal (``col IN (['a'], ['b'])``).
if op in {
utils.FilterOperator.EQUALS,
utils.FilterOperator.NOT_EQUALS,
}:
literal = db_engine_spec.array_literal(
coerce_array_values(
parse_array_literal(val), array_element_type
)
)
cond = (
sqla_col != literal
if op == utils.FilterOperator.NOT_EQUALS
else sqla_col == literal
)
else:
candidates: list[Any] = (
list(val) if isinstance(val, (list, tuple)) else [val]
)
cond = sqla_col.in_(
[
db_engine_spec.array_literal(
coerce_array_values(
parse_array_literal(candidate),
array_element_type,
)
)
for candidate in candidates
]
)
if op == utils.FilterOperator.NOT_IN:
cond = ~cond
target_clause_list.append(cond)
elif op in {
utils.FilterOperator.IN,
utils.FilterOperator.NOT_IN,
}:
# CONTAINS_ANY/CONTAINS_ALL also produce a list ``eq`` (they
# are in ``is_list_target``), but are element-level array ops
# handled by their own branch below — not IN.
elif is_list_target:
assert isinstance(eq, (tuple, list))
if len(eq) == 0:
raise QueryObjectValidationError(
@@ -4650,57 +4529,6 @@ class ExploreMixin: # pylint: disable=too-many-public-methods
target_clause_list.append(
db_engine_spec.handle_null_filter(sqla_col, op)
)
elif op in {
utils.FilterOperator.IS_EMPTY,
utils.FilterOperator.IS_NOT_EMPTY,
}:
# Element-level array operators: length(col) == 0 / > 0.
if target_generic_type != GenericDataType.MULTI_VALUE:
raise QueryObjectValidationError(
_(
"The %(op)s operator is only supported for "
"multi-value (array) columns.",
op=op,
)
)
length_expr = db_engine_spec.array_length(sqla_col)
if op == utils.FilterOperator.IS_EMPTY:
target_clause_list.append(length_expr == 0)
else:
target_clause_list.append(length_expr > 0)
elif op in {
utils.FilterOperator.LENGTH_EQUALS,
utils.FilterOperator.LENGTH_GREATER_THAN,
utils.FilterOperator.LENGTH_LESS_THAN,
utils.FilterOperator.LENGTH_GREATER_THAN_OR_EQUALS,
utils.FilterOperator.LENGTH_LESS_THAN_OR_EQUALS,
}:
# Length filter: compare the array's element count to a
# number, e.g. length(col) > 2.
if target_generic_type != GenericDataType.MULTI_VALUE:
raise QueryObjectValidationError(
_(
"The %(op)s operator is only supported for "
"multi-value (array) columns.",
op=op,
)
)
number = utils.cast_to_num(eq) # type: ignore[arg-type]
if number is None:
raise QueryObjectValidationError(
_("The Length filter requires a numeric value.")
)
length_expr = db_engine_spec.array_length(sqla_col)
length_comparisons = {
utils.FilterOperator.LENGTH_EQUALS: length_expr == number,
utils.FilterOperator.LENGTH_GREATER_THAN: length_expr > number,
utils.FilterOperator.LENGTH_LESS_THAN: length_expr < number,
utils.FilterOperator.LENGTH_GREATER_THAN_OR_EQUALS: length_expr
>= number,
utils.FilterOperator.LENGTH_LESS_THAN_OR_EQUALS: length_expr
<= number,
}
target_clause_list.append(length_comparisons[op])
elif op == utils.FilterOperator.IS_TRUE:
target_clause_list.append(
db_engine_spec.handle_boolean_filter(sqla_col, op, True)
@@ -4758,38 +4586,6 @@ class ExploreMixin: # pylint: disable=too-many-public-methods
target_clause_list.append(sqla_col.not_like(eq))
else:
target_clause_list.append(sqla_col.not_ilike(eq))
elif op in {
utils.FilterOperator.CONTAINS_ANY,
utils.FilterOperator.CONTAINS_ALL,
}:
# Element-level array membership. Enforce the target is
# actually a multi-value (array) column (only classified
# MULTI_VALUE on an array-capable engine), guarding against
# payloads that bypass the UI gating.
if target_generic_type != GenericDataType.MULTI_VALUE:
raise QueryObjectValidationError(
_(
"The %(op)s operator is only supported for "
"multi-value (array) columns.",
op=op,
)
)
array_values: list[Any] = coerce_array_values(
list(eq) if isinstance(eq, (list, tuple)) else [eq],
array_element_type,
)
if op == utils.FilterOperator.CONTAINS_ANY:
target_clause_list.append(
db_engine_spec.array_contains_any(
sqla_col, array_values
)
)
else:
target_clause_list.append(
db_engine_spec.array_contains_all(
sqla_col, array_values
)
)
elif (
op == utils.FilterOperator.TEMPORAL_RANGE
and isinstance(eq, str)
+1 -1
View File
@@ -384,7 +384,7 @@ class SemanticView(AuditMixinNullable, Model):
for dimension in dimensions
},
}
column_formats: dict[str, str | None] = {
column_formats = {
metric.name: metric.d3format for metric in metrics if metric.d3format
}
+1 -21
View File
@@ -209,7 +209,7 @@ class GenericDataType(IntEnum):
STRING = 1
TEMPORAL = 2
BOOLEAN = 3
MULTI_VALUE = 4 # array-typed columns (e.g. ClickHouse Array, Postgres ARRAY)
# ARRAY = 4 # Mapping all the complex data types to STRING for now
# JSON = 5 # and leaving these as a reminder.
# MAP = 6
# ROW = 7
@@ -299,17 +299,6 @@ class FilterOperator(StrEnum):
IS_TRUE = "IS TRUE"
IS_FALSE = "IS FALSE"
TEMPORAL_RANGE = "TEMPORAL_RANGE"
# Element-level operators for MULTI_VALUE (array) columns
CONTAINS_ANY = "CONTAINS_ANY"
CONTAINS_ALL = "CONTAINS_ALL"
IS_EMPTY = "IS_EMPTY"
IS_NOT_EMPTY = "IS_NOT_EMPTY"
# Length (element-count) comparison operators for array columns
LENGTH_EQUALS = "LENGTH_EQUALS"
LENGTH_GREATER_THAN = "LENGTH_GREATER_THAN"
LENGTH_LESS_THAN = "LENGTH_LESS_THAN"
LENGTH_GREATER_THAN_OR_EQUALS = "LENGTH_GREATER_THAN_OR_EQUALS"
LENGTH_LESS_THAN_OR_EQUALS = "LENGTH_LESS_THAN_OR_EQUALS"
class FilterStringOperators(StrEnum):
@@ -328,15 +317,6 @@ class FilterStringOperators(StrEnum):
LATEST_PARTITION = ("LATEST_PARTITION",)
IS_TRUE = ("IS_TRUE",)
IS_FALSE = ("IS_FALSE",)
CONTAINS_ANY = ("CONTAINS_ANY",)
CONTAINS_ALL = ("CONTAINS_ALL",)
IS_EMPTY = ("IS_EMPTY",)
IS_NOT_EMPTY = ("IS_NOT_EMPTY",)
LENGTH_EQUALS = ("LENGTH_EQUALS",)
LENGTH_GREATER_THAN = ("LENGTH_GREATER_THAN",)
LENGTH_LESS_THAN = ("LENGTH_LESS_THAN",)
LENGTH_GREATER_THAN_OR_EQUALS = ("LENGTH_GREATER_THAN_OR_EQUALS",)
LENGTH_LESS_THAN_OR_EQUALS = ("LENGTH_LESS_THAN_OR_EQUALS",)
class PostProcessingBoxplotWhiskerType(StrEnum):
@@ -15,7 +15,7 @@
# specific language governing permissions and limitations
# under the License.
from collections.abc import Sequence
from functools import partial, wraps
from functools import partial
from typing import Any, Callable
import numpy as np
@@ -122,10 +122,6 @@ def scalar_to_sequence(val: Any) -> Sequence[str]:
def validate_column_args(*argnames: str) -> Callable[..., Any]:
def wrapper(func: Callable[..., Any]) -> Callable[..., Any]:
# `wraps` keeps `func` reachable through `__wrapped__`, so that
# `inspect.signature` reports the parameters of the decorated operation
# rather than the `(df, **options)` of this wrapper.
@wraps(func)
def wrapped(df: DataFrame, **options: Any) -> Any:
if _is_multi_index_on_columns(df):
# MultiIndex column validate first level
+8 -60
View File
@@ -159,49 +159,16 @@ CHART_ERROR_OR_EMPTY_SELECTOR = (
f"{ALERT_SELECTOR}, {EMPTY_SELECTOR}, {MISSING_CHART_SELECTOR}"
)
# Runtime contract with the dashboard frontend. Dispatching this window event
# forces every DashboardVirtualization row to render regardless of whether it
# intersects the headless viewport, mirroring the client-side "Download as
# Image/PDF" path (see FORCE_IN_VIEW_EVENT in
# superset-frontend/src/dashboard/constants.ts and forceLoadAllCharts in
# superset-frontend/src/utils/downloadUtils.ts). The non-tiled report capture
# takes a single full-page screenshot that includes below-the-fold holders, so
# those holders must be forced to render before the readiness wait -- otherwise
# a virtualized (or still-loading) off-screen holder is captured blank. A plain
# Event with no `detail.rowIds` means "force every row", matching the frontend's
# single-pass branch.
FORCE_ALL_CHART_HOLDERS_IN_VIEW_EVENT = "superset-force-all-in-view"
FORCE_ALL_CHART_HOLDERS_IN_VIEW_JS = (
f"() => window.dispatchEvent(new Event('{FORCE_ALL_CHART_HOLDERS_IN_VIEW_EVENT}'))"
)
def _unready_chart_holders_js_body(*, viewport_only: bool) -> str:
"""Return the shared holder-readiness scan body.
A holder is ready only after a terminal marker appears and its loading
marker disappears. When ``viewport_only`` is True the scan skips holders
that do not intersect the current viewport: correct for the tiled path,
which scrolls every region into view before capturing it, and for
thumbnails, which only ever capture the viewport. The non-tiled *report*
capture takes a single full-page screenshot that includes below-the-fold
holders, so it must scan every mounted holder (``viewport_only=False``) --
otherwise an off-screen holder that never rendered is captured blank and
silently delivered as a Success.
"""
viewport_skip = (
"""
const r = holder.getBoundingClientRect();
if (!(r.top < window.innerHeight && r.bottom > 0)) {
continue;
}"""
if viewport_only
else ""
)
return f"""
# Shared body for holder readiness and timeout diagnostics. A holder is ready
# only after a terminal marker appears and its loading marker disappears.
UNREADY_CHART_HOLDERS_JS_BODY = f"""
const holders = document.querySelectorAll('{CHART_HOLDER_SELECTOR}');
const unready = [];
for (const holder of holders) {{{viewport_skip}
for (const holder of holders) {{
const r = holder.getBoundingClientRect();
if (!(r.top < window.innerHeight && r.bottom > 0)) {{
continue;
}}
const hasSliceContainer = holder.querySelector(
'{SLICE_CONTAINER_SELECTOR}'
) !== null;
@@ -239,13 +206,6 @@ def _unready_chart_holders_js_body(*, viewport_only: bool) -> str:
}}
"""
# Viewport-scoped scan (tiled path + thumbnails).
UNREADY_CHART_HOLDERS_JS_BODY = _unready_chart_holders_js_body(viewport_only=True)
# Full-dashboard scan (non-tiled report capture, which screenshots the whole
# element in one shot and therefore cannot ignore below-the-fold holders).
UNREADY_ALL_CHART_HOLDERS_JS_BODY = _unready_chart_holders_js_body(viewport_only=False)
# Diagnostic query for every chart holder, including terminal and virtualized
# states. It interpolates the same selector constants as the predicates.
FIND_CHART_HOLDER_STATES_JS = f"""
@@ -296,24 +256,12 @@ REPORT_CHART_HOLDERS_READY_JS = (
f"() => {{ {UNREADY_CHART_HOLDERS_JS_BODY} "
"return holders.length > 0 && unready.length === 0; }"
)
# Report readiness for the non-tiled full-page capture: every mounted holder --
# including below-the-fold ones -- must be terminally rendered. Off-screen
# holders are forced to render first (FORCE_ALL_CHART_HOLDERS_IN_VIEW_JS); if any
# still fails to render within budget the wait times out and the report fails
# loudly rather than shipping a blank/partial screenshot as a Success.
REPORT_ALL_CHART_HOLDERS_READY_JS = (
f"() => {{ {UNREADY_ALL_CHART_HOLDERS_JS_BODY} "
"return holders.length > 0 && unready.length === 0; }"
)
CHART_HOLDERS_MOUNTED_JS = (
f"() => document.querySelectorAll('{CHART_HOLDER_SELECTOR}').length > 0"
)
FIND_UNREADY_CHART_HOLDERS_JS = (
f"() => {{ {UNREADY_CHART_HOLDERS_JS_BODY} return unready; }}"
)
FIND_ALL_UNREADY_CHART_HOLDERS_JS = (
f"() => {{ {UNREADY_ALL_CHART_HOLDERS_JS_BODY} return unready; }}"
)
# A chart capture has one target rather than dashboard holders, but needs the
# same positive terminal-state guarantee and loading exclusion.
+5 -43
View File
@@ -34,10 +34,8 @@ from superset.utils.screenshot_utils import (
CHART_CONTAINER_READY_JS,
CHART_CONTAINER_STATE_JS,
CHART_HOLDERS_READY_JS,
FIND_ALL_UNREADY_CHART_HOLDERS_JS,
FIND_CHART_HOLDER_STATES_JS,
FORCE_ALL_CHART_HOLDERS_IN_VIEW_JS,
REPORT_ALL_CHART_HOLDERS_READY_JS,
REPORT_CHART_HOLDERS_READY_JS,
resolve_screenshot_task_budget_seconds,
ScreenshotTaskBudgetExceededError,
take_tiled_screenshot,
@@ -384,19 +382,7 @@ class WebDriverPlaywright(WebDriverProxy):
if element_name == "chart-container":
readiness_predicate = CHART_CONTAINER_READY_JS
elif report_execution_context:
# This non-tiled path captures the whole element in one shot
# (`_get_screenshot` uses `full_page=True` / `element.screenshot()`),
# so below-the-fold holders end up in the image. Force every
# virtualized row to render up front -- mirroring the client-side
# "Download as Image/PDF" path -- and then require *all* mounted
# holders (not just the viewport-visible ones) to reach a terminal
# state. If an off-screen holder never renders, the wait times out
# and the report fails loudly instead of silently delivering a
# blank/partial screenshot as a Success. The tiled path keeps the
# viewport-scoped predicate because it scrolls each region into view
# before capturing it.
page.evaluate(FORCE_ALL_CHART_HOLDERS_IN_VIEW_JS)
readiness_predicate = REPORT_ALL_CHART_HOLDERS_READY_JS
readiness_predicate = REPORT_CHART_HOLDERS_READY_JS
else:
# Preserve the thumbnail behavior introduced by #42253. The
# stricter zero-holder gate is report-specific because an empty
@@ -444,16 +430,6 @@ class WebDriverPlaywright(WebDriverProxy):
ready_holders = sum(
holder.get("state") in ready_states for holder in chart_holder_states
)
# `FIND_CHART_HOLDER_STATES_JS` short-circuits off-screen holders to
# "virtualized" (counted as ready above), so on the report path -- a
# full-page capture that includes below-the-fold holders -- the real
# culprits (off-screen holders that never rendered) would be hidden.
# Surface them explicitly using the non-viewport-scoped scan.
below_fold_unready = (
page.evaluate(FIND_ALL_UNREADY_CHART_HOLDERS_JS)
if report_execution_context
else unready_chart_holders
)
deadline_elapsed = deadline.elapsed_seconds if deadline else elapsed
deadline_remaining = (
deadline.remaining_seconds if deadline else remaining_budget
@@ -462,8 +438,7 @@ class WebDriverPlaywright(WebDriverProxy):
"report_readiness_terminal url=%s expected_holders=%s "
"mounted_holders=%s ready_holders=%s elapsed_seconds=%.2f "
"remaining_seconds=%s effective_wait_seconds=%.2f%s "
"terminal_reason=readiness_timeout unready_holders=%s "
"all_unready_holders=%s states=%s; "
"terminal_reason=readiness_timeout unready_holders=%s states=%s; "
"aborting before capture or delivery",
url,
expected_holders,
@@ -478,7 +453,6 @@ class WebDriverPlaywright(WebDriverProxy):
effective_load_wait,
context_suffix,
unready_chart_holders,
below_fold_unready,
chart_holder_states,
)
raise
@@ -773,21 +747,9 @@ class WebDriverPlaywright(WebDriverProxy):
context_suffix,
)
# Use tiled screenshots for large dashboards. For scheduled
# reports a likely-large dashboard whose measured height is
# at or below a single tile is almost always mid-layout
# (charts still virtualized/collapsed at measurement time),
# not genuinely short -- a 52-chart dashboard is never really
# <one viewport tall. Routing it to the single-shot,
# full-page non-tiled capture risks shipping a windowed
# partial render. Prefer the tiled path, which scrolls every
# region into view and waits per tile; worst case it is a
# single tile. The tiled decision for thumbnails is
# unchanged.
# Use tiled screenshots for large dashboards
use_tiled = likely_large_dashboard and (
height_unknown
or dashboard_height > tile_height
or report_execution_context is not None
height_unknown or dashboard_height > tile_height
)
if use_tiled:
@@ -154,22 +154,8 @@ class TestDatasourceApi(SupersetTestCase):
column_name="col2",
limit=10000,
denormalize_column=False,
array_elements=False,
)
@pytest.mark.usefixtures("app_context", "virtual_dataset")
@patch("superset.models.helpers.ExploreMixin.values_for_column")
def test_get_column_values_array_elements_param(self, values_for_column_mock):
# The ?array_elements=true param (Contains any/all) is threaded through
# so array columns can suggest individual elements.
self.login(ADMIN_USERNAME)
table = self.get_virtual_dataset()
self.client.get(
f"api/v1/datasource/table/{table.id}/column/col2/values/"
"?array_elements=true"
)
assert values_for_column_mock.call_args.kwargs["array_elements"] is True
@pytest.mark.usefixtures("app_context", "virtual_dataset")
@patch("superset.db_engine_specs.base.BaseEngineSpec.denormalize_name")
def test_get_column_values_not_denormalize_column(self, denormalize_name_mock):
@@ -190,7 +176,6 @@ class TestDatasourceApi(SupersetTestCase):
column_name="col2",
limit=10000,
denormalize_column=True,
array_elements=False,
)
@pytest.mark.usefixtures("app_context", "virtual_dataset")
@@ -1308,152 +1308,3 @@ def test_column_ordering_without_chart_flag(login_as_admin):
finally:
db.session.delete(table)
db.session.commit()
def _multivalue_table() -> SqlaTable:
"""A dataset with an ``Array(String)`` column, for multi-value query tests.
Built over the example database but never executed the tests only compile
the generated SQL, so the backing table need not physically exist.
"""
columns = [
TableColumn(column_name="skills", type="Array(String)"),
TableColumn(column_name="city", type="VARCHAR(255)"),
]
return SqlaTable(
table_name="test_multivalue_jobs",
database=get_example_database(),
columns=columns,
metrics=[SqlMetric(metric_name="count", expression="COUNT(*)")],
)
def _multivalue_query(
*,
filters: list[dict[str, Any]] | None = None,
groupby: list[Any] | None = None,
) -> dict[str, Any]:
return {
"granularity": None,
"from_dttm": None,
"to_dttm": None,
"is_timeseries": False,
"groupby": groupby if groupby is not None else ["city"],
"metrics": ["count"],
"filter": filters or [],
"extras": {},
}
def _compile(table: SqlaTable, query_obj: dict[str, Any]) -> str:
from superset.db_engine_specs.clickhouse import ClickHouseEngineSpec
with patch.object(
SqlaTable, "db_engine_spec", property(lambda self: ClickHouseEngineSpec)
):
sqla_query = table.get_sqla_query(**query_obj)
return table.database.compile_sqla_query(sqla_query.sqla_query).lower()
@pytest.mark.usefixtures("app_context")
def test_multivalue_contains_any_generates_native_sql():
"""CONTAINS_ANY compiles to ``hasAny(col, array(...))``."""
table = _multivalue_table()
sql = _compile(
table,
_multivalue_query(
filters=[
{
"col": "skills",
"op": FilterOperator.CONTAINS_ANY.value,
"val": ["Driver", "Cook"],
}
]
),
)
assert "hasany(skills" in sql
@pytest.mark.usefixtures("app_context")
def test_multivalue_contains_all_generates_native_sql():
"""CONTAINS_ALL compiles to ``hasAll(col, array(...))``."""
table = _multivalue_table()
sql = _compile(
table,
_multivalue_query(
filters=[
{
"col": "skills",
"op": FilterOperator.CONTAINS_ALL.value,
"val": ["Driver", "Cook"],
}
]
),
)
assert "hasall(skills" in sql
@pytest.mark.usefixtures("app_context")
def test_multivalue_is_empty_generates_native_sql():
"""IS_EMPTY compiles to ``length(col) = 0``."""
table = _multivalue_table()
sql = _compile(
table,
_multivalue_query(
filters=[{"col": "skills", "op": FilterOperator.IS_EMPTY.value}]
),
)
assert "length(skills) = 0" in sql
@pytest.mark.usefixtures("app_context")
def test_multivalue_length_filter_generates_native_sql():
"""A LENGTH_GREATER_THAN filter compiles to ``length(col) > N``."""
table = _multivalue_table()
sql = _compile(
table,
_multivalue_query(
filters=[
{
"col": "skills",
"op": FilterOperator.LENGTH_GREATER_THAN.value,
"val": 2,
}
]
),
)
assert "length(skills) > 2" in sql
@pytest.mark.usefixtures("app_context")
def test_multivalue_contains_unsupported_engine_raises():
"""CONTAINS_ANY on an engine without array support is rejected."""
table = _multivalue_table()
query_obj = _multivalue_query(
filters=[
{
"col": "skills",
"op": FilterOperator.CONTAINS_ANY.value,
"val": ["Driver"],
}
]
)
with pytest.raises(QueryObjectValidationError):
table.get_sqla_query(**query_obj)
@pytest.mark.usefixtures("app_context")
def test_multivalue_length_filter_unsupported_engine_raises():
"""A Length filter on an engine without array support is rejected."""
table = _multivalue_table()
query_obj = _multivalue_query(
filters=[
{
"col": "skills",
"op": FilterOperator.LENGTH_GREATER_THAN.value,
"val": 2,
}
]
)
with pytest.raises(QueryObjectValidationError):
table.get_sqla_query(**query_obj)
@@ -18,11 +18,15 @@ from unittest.mock import Mock, patch
import pytest
from marshmallow import ValidationError
from pytest_mock import MockerFixture
from superset.commands.dataset.create import CreateDatasetCommand
from superset.commands.dataset.exceptions import DatasetInvalidError
from superset.errors import ErrorLevel, SupersetError, SupersetErrorType
from superset.exceptions import SupersetParseError
from superset.exceptions import (
SupersetGenericDBErrorException,
SupersetParseError,
)
from superset.models.core import Database
@@ -250,3 +254,83 @@ def test_create_dataset_generic_exists_error_when_no_twin() -> None:
)
with pytest.raises(DatasetInvalidError):
command.validate()
def test_create_dataset_metadata_fetch_error_is_structured(
mocker: MockerFixture,
) -> None:
"""A metadata-fetch failure must surface the engine's own message.
``run()`` executes the SQL to introspect columns; the resulting
``SupersetGenericDBErrorException`` used to escape as a 500 "Fatal error".
"""
mocker.patch.object(CreateDatasetCommand, "validate")
dataset = Mock()
dataset.fetch_metadata.side_effect = SupersetGenericDBErrorException(
message="Invalid SQL: Unable to parse: SELECT ...",
)
mocker.patch(
"superset.commands.dataset.create.DatasetDAO.create",
return_value=dataset,
)
command = CreateDatasetCommand(
{
"database": 1,
"table_name": "dataset wrong",
"sql": "SELECT ...",
}
)
with pytest.raises(DatasetInvalidError) as exc_info:
command.run()
validation_errors = exc_info.value._exceptions
assert len(validation_errors) == 1
assert validation_errors[0].field_name == "sql"
assert "Invalid SQL: Unable to parse: SELECT ..." in str(
validation_errors[0].messages[0]
)
def test_create_dataset_metadata_fetch_error_physical_table(
mocker: MockerFixture,
) -> None:
"""The same conversion applies to physical datasets, keyed on ``table``."""
mocker.patch.object(CreateDatasetCommand, "validate")
dataset = Mock()
dataset.fetch_metadata.side_effect = SupersetGenericDBErrorException(
message="(psycopg2.OperationalError) could not connect to server",
)
mocker.patch(
"superset.commands.dataset.create.DatasetDAO.create",
return_value=dataset,
)
command = CreateDatasetCommand({"database": 1, "table_name": "physical_table"})
with pytest.raises(DatasetInvalidError) as exc_info:
command.run()
validation_errors = exc_info.value._exceptions
assert validation_errors[0].field_name == "table"
assert "could not connect to server" in str(validation_errors[0].messages[0])
def test_create_dataset_run_succeeds_when_metadata_fetch_works(
mocker: MockerFixture,
) -> None:
"""Control: the happy path still returns the created dataset."""
mocker.patch.object(CreateDatasetCommand, "validate")
dataset = Mock()
mocker.patch(
"superset.commands.dataset.create.DatasetDAO.create",
return_value=dataset,
)
command = CreateDatasetCommand(
{"database": 1, "table_name": "good_dataset", "sql": "SELECT 1 AS a"}
)
assert command.run() is dataset
dataset.fetch_metadata.assert_called_once()
+43
View File
@@ -214,3 +214,46 @@ def test_handle_filters_args_returns_request_scoped_filters(
fresh_filters = api.datamodel.get_filters.return_value
assert fresh_filters.rest_add_filters.call_count == 2
assert fresh_filters.get_joined_filters.call_count == 2
def test_post_dataset_with_invalid_sql_returns_actionable_422(
session: Session,
client: Any,
full_api_access: None,
) -> None:
"""Saving a dataset over unrunnable SQL must explain what is wrong.
With blanket database access ``validate()`` never parses the SQL, so
``run()``'s column introspection is the first thing to reject it. That
used to surface as a bare 500 ``{"message": "Fatal error"}``.
"""
from superset.connectors.sqla.models import SqlaTable
from superset.models.core import Database
SqlaTable.metadata.create_all(db.session.get_bind())
database = Database(database_name="invalid_sql_db", sqlalchemy_uri="sqlite://")
db.session.add(database)
db.session.flush()
response = client.post(
"/api/v1/dataset/",
json={
"database": database.id,
"schema": "main",
"table_name": "dataset wrong",
"sql": "SELECT ...",
},
)
assert response.status_code == 422
message = response.json["message"]
assert "Fatal error" not in str(message)
# Not the parser's exact wording -- that would break on a sqlglot bump.
assert message["sql"][0].startswith("Invalid SQL")
# The failed create must not leave a half-built dataset behind.
assert (
db.session.query(SqlaTable).filter_by(table_name="dataset wrong").one_or_none()
is None
)
@@ -1490,21 +1490,3 @@ def test_get_public_information_exposes_ansi_identifier_quote() -> None:
"end": '"',
"escape_by_doubling": True,
}
def test_multivalue_columns_disabled_by_default() -> None:
"""Engines must opt in to multi-value support; base defaults to off."""
assert BaseEngineSpec.supports_multivalue_columns is False
@pytest.mark.parametrize(
"method", ["array_contains_any", "array_contains_all", "array_length"]
)
def test_array_capabilities_raise_when_unsupported(method: str) -> None:
"""Array capability methods raise NotImplementedError unless overridden."""
from sqlalchemy import column
fn = getattr(BaseEngineSpec, method)
args = (column("c"), ["v"]) if "contains" in method else (column("c"),)
with pytest.raises(NotImplementedError):
fn(*args)
@@ -143,30 +143,7 @@ def test_connect_convert_dttm(
GenericDataType.STRING,
False,
),
("Array(UInt8)", String, None, GenericDataType.MULTI_VALUE, False),
("Array(String)", String, None, GenericDataType.MULTI_VALUE, False),
("Array(UInt64)", String, None, GenericDataType.MULTI_VALUE, False),
(
"Array(LowCardinality(String))",
String,
None,
GenericDataType.MULTI_VALUE,
False,
),
# Array(Enum(...)) is a real array and must classify as MULTI_VALUE, not
# get short-circuited by the Enum rule (the anchored ^Array\( pattern is
# ordered before the Enum entry).
(
"Array(Enum8('a' = 1, 'b' = 2))",
String,
None,
GenericDataType.MULTI_VALUE,
False,
),
# Arrays nested inside Map/Tuple are not top-level array columns; the
# anchored pattern must not over-match them into MULTI_VALUE.
("Map(String, Array(String))", String, None, GenericDataType.STRING, False),
("Tuple(Array(String))", String, None, GenericDataType.STRING, False),
("Array(UInt8)", String, None, GenericDataType.STRING, False),
("Enum('hello', 'world')", String, None, GenericDataType.STRING, False),
("Enum('UInt32', 'Bool')", String, None, GenericDataType.STRING, False),
(
@@ -653,115 +630,3 @@ def test_use_equality_for_boolean_filters_property() -> None:
from superset.db_engine_specs.clickhouse import ClickHouseBaseEngineSpec
assert ClickHouseBaseEngineSpec.use_equality_for_boolean_filters is True
def _compile(expr) -> str:
return str(expr.compile(compile_kwargs={"literal_binds": True}))
def test_clickhouse_supports_multivalue_columns() -> None:
from superset.db_engine_specs.clickhouse import ( # noqa: N813
ClickHouseEngineSpec as spec,
)
assert spec.supports_multivalue_columns is True
def test_multivalue_contains_any_sql() -> None:
from sqlalchemy import column
from superset.db_engine_specs.clickhouse import ( # noqa: N813
ClickHouseEngineSpec as spec,
)
expr = spec.array_contains_any(column("skills"), ["Driver", "Cook"])
assert _compile(expr) == "hasAny(skills, array('Driver', 'Cook'))"
def test_multivalue_contains_all_sql() -> None:
from sqlalchemy import column
from superset.db_engine_specs.clickhouse import ( # noqa: N813
ClickHouseEngineSpec as spec,
)
expr = spec.array_contains_all(column("skills"), ["Driver", "Cook"])
assert _compile(expr) == "hasAll(skills, array('Driver', 'Cook'))"
def test_multivalue_contains_binds_parameters() -> None:
"""Values must be bound parameters, not inlined (SQL-injection safety)."""
from sqlalchemy import column
from superset.db_engine_specs.clickhouse import ( # noqa: N813
ClickHouseEngineSpec as spec,
)
expr = spec.array_contains_any(column("skills"), ["Driver"])
compiled = expr.compile()
assert "Driver" not in str(compiled)
assert "Driver" in compiled.params.values()
def test_multivalue_length_sql() -> None:
from sqlalchemy import column
from superset.db_engine_specs.clickhouse import ( # noqa: N813
ClickHouseEngineSpec as spec,
)
expr = spec.array_length(column("skills"))
assert _compile(expr) == "length(skills)"
@pytest.mark.parametrize(
"native_type,expected",
[
("Array(String)", GenericDataType.STRING),
("Array(Int32)", GenericDataType.NUMERIC),
("Array(UInt64)", GenericDataType.NUMERIC),
("Array(Decimal(10, 2))", GenericDataType.NUMERIC),
("Array(DateTime)", GenericDataType.TEMPORAL),
("Array(Enum8('a' = 1))", GenericDataType.STRING),
# Wrappers around the element type don't change the generic type.
("Array(Nullable(Int64))", GenericDataType.NUMERIC),
("Array(LowCardinality(String))", GenericDataType.STRING),
# Non-array / nested-array types have no array element type.
("String", None),
("Map(String, Array(String))", None),
],
)
def test_multivalue_get_array_element_type(
native_type: str, expected: GenericDataType | None
) -> None:
from superset.db_engine_specs.clickhouse import ( # noqa: N813
ClickHouseEngineSpec as spec,
)
assert spec.get_array_element_type(native_type) == expected
def test_multivalue_array_explode_sql() -> None:
"""array_explode compiles to ``arrayJoin(col)`` (element expansion)."""
from sqlalchemy import column
from superset.db_engine_specs.clickhouse import ( # noqa: N813
ClickHouseEngineSpec as spec,
)
expr = spec.array_explode(column("scores"))
assert _compile(expr) == "arrayJoin(scores)"
def test_multivalue_contains_any_numeric_coercion_sql() -> None:
"""Numeric-array element values must render as numbers, not quoted strings."""
from sqlalchemy import column
from superset.db_engine_specs.clickhouse import ( # noqa: N813
ClickHouseEngineSpec as spec,
)
# Simulate values already coerced to numbers (as helpers.py does via the
# element type) and confirm the emitted array literal is numeric.
expr = spec.array_contains_any(column("scores"), [5, 6])
assert _compile(expr) == "hasAny(scores, array(5, 6))"
@@ -1981,23 +1981,6 @@ def test_create_virtual_dataset_request_optional_fields() -> None:
assert req.description == "A virtual dataset"
def test_create_virtual_dataset_rejects_non_aggregate_saved_metric() -> None:
from pydantic import ValidationError
with pytest.raises(ValidationError, match="saved metrics must aggregate rows"):
CreateVirtualDatasetRequest(
database_id=1,
sql="SELECT needed_operators FROM staffing",
dataset_name="Staffing",
metrics=[
{
"metric_name": "needed_operators",
"expression": "needed_operators",
}
],
)
# --- Tool logic tests ---
@@ -2136,39 +2119,6 @@ async def test_create_virtual_dataset_create_failed(mcp_server: object) -> None:
assert "Failed to create dataset" in data["error"]
@pytest.mark.asyncio
async def test_create_virtual_dataset_sql_error_is_actionable(
mcp_server: object,
) -> None:
"""Warehouse SQL errors are recoverable tool results, not adapter crashes."""
from superset.exceptions import SupersetGenericDBErrorException
mock_command = MagicMock()
mock_command.run.side_effect = SupersetGenericDBErrorException(
"Invalid column name 'missing_value'"
)
with patch(
"superset.commands.dataset.create.CreateDatasetCommand",
return_value=mock_command,
):
async with Client(mcp_server) as client:
request = CreateVirtualDatasetRequest(
database_id=1,
sql="SELECT missing_value FROM sample_events",
dataset_name="Test",
)
result = await client.call_tool(
"create_virtual_dataset", {"request": request.model_dump()}
)
data = json.loads(result.content[0].text)
assert data["id"] is None
assert data["columns"] == []
assert data["error"] is not None
assert "Invalid column name" in data["error"]
@pytest.mark.asyncio
async def test_create_virtual_dataset_permission_denied(mcp_server: object) -> None:
"""SQL access denied surfaces as DatasetInvalidError with id=None."""
@@ -2339,13 +2289,7 @@ async def test_create_virtual_dataset_update_failure_rollback(
if exception_to_raise == "DatasetUpdateFailedError":
mock_update_instance.run.side_effect = DatasetUpdateFailedError()
else:
from superset.commands.dataset.exceptions import (
DatasetColumnsExistsValidationError,
)
invalid_error = DatasetInvalidError()
invalid_error.append(DatasetColumnsExistsValidationError())
mock_update_instance.run.side_effect = invalid_error
mock_update_instance.run.side_effect = DatasetInvalidError()
mock_update_cls = MagicMock(return_value=mock_update_instance)
mock_delete_instance = MagicMock()
@@ -2392,11 +2336,7 @@ async def test_create_virtual_dataset_update_failure_rollback(
# Verify the error response
data = json.loads(result.content[0].text)
assert data["id"] is None
if exception_to_raise == "DatasetInvalidError":
assert "columns" in data["error"]
assert "already exist" in data["error"]
else:
assert "creation rolled back" in data["error"]
assert "creation rolled back" in data["error"]
@pytest.mark.asyncio
@@ -1,274 +0,0 @@
# Licensed to the Apache Software Foundation (ASF) under one
# or more contributor license agreements. See the NOTICE file
# distributed with this work for additional information
# regarding copyright ownership. The ASF licenses this file
# to you under the Apache License, Version 2.0 (the
# "License"); you may not use this file except in compliance
# with the License. You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing,
# software distributed under the License is distributed on an
# "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY
# KIND, either express or implied. See the License for the
# specific language governing permissions and limitations
# under the License.
"""Element-level array filter operators (Contains Any/All, Is empty/not empty)."""
from __future__ import annotations
from typing import Any, cast
import pytest
from flask import Flask
from pytest_mock import MockerFixture
from superset.connectors.sqla.models import SqlaTable, SqlMetric, TableColumn
from superset.exceptions import QueryObjectValidationError
from superset.models.core import Database
from superset.superset_typing import QueryObjectDict
from superset.utils.core import FilterOperator
def _make_dataset(mocker: MockerFixture) -> SqlaTable:
database = Database(id=1, database_name="test_db", sqlalchemy_uri="sqlite://")
columns = [
TableColumn(column_name="skills", type="Array(String)"),
TableColumn(column_name="scores", type="Array(Int32)"),
TableColumn(column_name="city", type="VARCHAR(100)"),
]
dataset = SqlaTable(
table_name="jobs",
columns=columns,
database=database,
metrics=[SqlMetric(metric_name="count", expression="COUNT(*)")],
)
mocker.patch(
"superset.connectors.sqla.models.security_manager.get_guest_rls_filters",
return_value=[],
)
mocker.patch(
"superset.connectors.sqla.models.security_manager.is_guest_user",
return_value=False,
)
return dataset
def _clickhouse(mocker: MockerFixture, dataset: SqlaTable) -> None:
# Imported lazily: clickhouse.py touches app.config at import time.
from superset.db_engine_specs.clickhouse import ClickHouseEngineSpec
mocker.patch.object(
SqlaTable, "db_engine_spec", new=property(lambda self: ClickHouseEngineSpec)
)
def _filter_query(filters: list[dict[str, Any]]) -> QueryObjectDict:
return cast(
QueryObjectDict,
{
"granularity": None,
"from_dttm": None,
"to_dttm": None,
"is_timeseries": False,
"groupby": ["city"],
"metrics": ["count"],
"filter": filters,
"columns": [],
},
)
def _sql(dataset: SqlaTable, filters: list[dict[str, Any]]) -> str:
return dataset.get_query_str_extended(
_filter_query(filters), mutate=False
).sql.lower()
def test_contains_any_generates_hasany(mocker: MockerFixture, app: Flask) -> None:
dataset = _make_dataset(mocker)
_clickhouse(mocker, dataset)
with app.test_request_context():
sql = _sql(
dataset,
[
{
"col": "skills",
"op": FilterOperator.CONTAINS_ANY.value,
"val": ["Driver", "Cook"],
}
],
)
assert "hasany(skills, array('driver', 'cook'))" in sql
def test_contains_any_numeric_array_coerces_values(
mocker: MockerFixture, app: Flask
) -> None:
"""Values for a numeric array must render as numbers, not quoted strings."""
dataset = _make_dataset(mocker)
_clickhouse(mocker, dataset)
with app.test_request_context():
sql = _sql(
dataset,
[
{
"col": "scores",
"op": FilterOperator.CONTAINS_ANY.value,
"val": ["5", "6"],
}
],
)
assert "hasany(scores, array(5, 6))" in sql
assert "'5'" not in sql
def test_equals_numeric_array_coerces_values(mocker: MockerFixture, app: Flask) -> None:
"""Whole-array equality on a numeric array coerces the parsed literal."""
dataset = _make_dataset(mocker)
_clickhouse(mocker, dataset)
with app.test_request_context():
sql = _sql(
dataset,
[
{
"col": "scores",
"op": FilterOperator.EQUALS.value,
"val": "[5, 6]",
}
],
)
assert "scores = array(5, 6)" in sql
def test_contains_all_generates_hasall(mocker: MockerFixture, app: Flask) -> None:
dataset = _make_dataset(mocker)
_clickhouse(mocker, dataset)
with app.test_request_context():
sql = _sql(
dataset,
[
{
"col": "skills",
"op": FilterOperator.CONTAINS_ALL.value,
"val": ["Driver", "Cook"],
}
],
)
assert "hasall(skills, array('driver', 'cook'))" in sql
def test_is_empty_generates_length_zero(mocker: MockerFixture, app: Flask) -> None:
dataset = _make_dataset(mocker)
_clickhouse(mocker, dataset)
with app.test_request_context():
sql = _sql(dataset, [{"col": "skills", "op": FilterOperator.IS_EMPTY.value}])
assert "length(skills) = 0" in sql
def test_is_not_empty_generates_length_gt_zero(
mocker: MockerFixture, app: Flask
) -> None:
dataset = _make_dataset(mocker)
_clickhouse(mocker, dataset)
with app.test_request_context():
sql = _sql(
dataset, [{"col": "skills", "op": FilterOperator.IS_NOT_EMPTY.value}]
)
assert "length(skills) > 0" in sql
def test_contains_resolves_to_hasany(mocker: MockerFixture, app: Flask) -> None:
dataset = _make_dataset(mocker)
_clickhouse(mocker, dataset)
with app.test_request_context():
sql = _sql(
dataset,
[
{
"col": "skills",
"op": FilterOperator.CONTAINS_ANY.value,
"val": ["Driver"],
}
],
)
assert "hasany(skills" in sql
def test_element_op_on_scalar_column_raises(mocker: MockerFixture, app: Flask) -> None:
"""CONTAINS_ANY on a scalar column is rejected on an array-capable engine."""
dataset = _make_dataset(mocker)
_clickhouse(mocker, dataset)
with app.test_request_context(): # noqa: SIM117
with pytest.raises(QueryObjectValidationError):
_sql(
dataset,
[
{
"col": "city",
"op": FilterOperator.CONTAINS_ANY.value,
"val": ["NYC"],
}
],
)
def test_element_op_unsupported_engine_raises(
mocker: MockerFixture, app: Flask
) -> None:
"""On an engine without array support (sqlite) the array op is rejected."""
dataset = _make_dataset(mocker)
with app.test_request_context(): # noqa: SIM117
with pytest.raises(QueryObjectValidationError):
_sql(dataset, [{"col": "skills", "op": FilterOperator.IS_EMPTY.value}])
def test_equals_on_array_parses_literal(mocker: MockerFixture, app: Flask) -> None:
"""A pasted array literal for = is parsed into col = array(...)."""
dataset = _make_dataset(mocker)
_clickhouse(mocker, dataset)
with app.test_request_context():
sql = _sql(
dataset,
[
{
"col": "skills",
"op": FilterOperator.EQUALS.value,
"val": "['Driver', 'Cook']",
}
],
)
assert "skills = array('driver', 'cook')" in sql
def test_equals_on_array_plain_value_fallback(
mocker: MockerFixture, app: Flask
) -> None:
"""A plain (non-bracketed) value becomes a single-element array."""
dataset = _make_dataset(mocker)
_clickhouse(mocker, dataset)
with app.test_request_context():
sql = _sql(
dataset,
[{"col": "skills", "op": FilterOperator.EQUALS.value, "val": "Driver"}],
)
assert "skills = array('driver')" in sql
def test_in_on_array_parses_literals(mocker: MockerFixture, app: Flask) -> None:
"""Whole-array IN parses each pasted array literal into its own array."""
dataset = _make_dataset(mocker)
_clickhouse(mocker, dataset)
with app.test_request_context():
sql = _sql(
dataset,
[
{
"col": "skills",
"op": FilterOperator.IN.value,
"val": ["['Driver']", "['Cook']"],
}
],
)
assert "skills in (array('driver'), array('cook'))" in sql
@@ -1,130 +0,0 @@
# Licensed to the Apache Software Foundation (ASF) under one
# or more contributor license agreements. See the NOTICE file
# distributed with this work for additional information
# regarding copyright ownership. The ASF licenses this file
# to you under the Apache License, Version 2.0 (the
# "License"); you may not use this file except in compliance
# with the License. You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing,
# software distributed under the License is distributed on an
# "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY
# KIND, either express or implied. See the License for the
# specific language governing permissions and limitations
# under the License.
"""The Length array filter operators: length(col) compared to a number."""
from __future__ import annotations
from typing import Any
import pytest
from flask import Flask
from pytest_mock import MockerFixture
from superset.connectors.sqla.models import SqlaTable, SqlMetric, TableColumn
from superset.exceptions import QueryObjectValidationError
from superset.models.core import Database
from superset.superset_typing import QueryObjectDict
from superset.utils.core import FilterOperator
def _make_dataset(mocker: MockerFixture) -> SqlaTable:
database = Database(id=1, database_name="test_db", sqlalchemy_uri="sqlite://")
columns = [
TableColumn(column_name="skills", type="Array(String)"),
TableColumn(column_name="city", type="VARCHAR(100)"),
]
dataset = SqlaTable(
table_name="jobs",
columns=columns,
database=database,
metrics=[SqlMetric(metric_name="count", expression="COUNT(*)")],
)
mocker.patch(
"superset.connectors.sqla.models.security_manager.get_guest_rls_filters",
return_value=[],
)
mocker.patch(
"superset.connectors.sqla.models.security_manager.is_guest_user",
return_value=False,
)
return dataset
def _clickhouse(mocker: MockerFixture, dataset: SqlaTable) -> None:
from superset.db_engine_specs.clickhouse import ClickHouseEngineSpec
mocker.patch.object(
SqlaTable, "db_engine_spec", new=property(lambda self: ClickHouseEngineSpec)
)
def _sql(dataset: SqlaTable, op: str, val: Any, col: str = "skills") -> str:
query: QueryObjectDict = {
"granularity": None,
"from_dttm": None,
"to_dttm": None,
"is_timeseries": False,
"groupby": ["city"],
"metrics": ["count"],
"filter": [{"col": col, "op": op, "val": val}],
"columns": [],
}
return dataset.get_query_str_extended(query, mutate=False).sql.lower()
@pytest.mark.parametrize(
"op,expected",
[
(FilterOperator.LENGTH_EQUALS, "length(skills) = 3"),
(FilterOperator.LENGTH_GREATER_THAN, "length(skills) > 3"),
(FilterOperator.LENGTH_LESS_THAN, "length(skills) < 3"),
(FilterOperator.LENGTH_GREATER_THAN_OR_EQUALS, "length(skills) >= 3"),
(FilterOperator.LENGTH_LESS_THAN_OR_EQUALS, "length(skills) <= 3"),
],
)
def test_length_operators_generate_length_comparison(
mocker: MockerFixture, app: Flask, op: FilterOperator, expected: str
) -> None:
dataset = _make_dataset(mocker)
_clickhouse(mocker, dataset)
with app.test_request_context():
sql = _sql(dataset, op.value, 3)
assert expected in sql
def test_length_accepts_string_number(mocker: MockerFixture, app: Flask) -> None:
"""A numeric string value is coerced (e.g. '2' -> length(col) > 2)."""
dataset = _make_dataset(mocker)
_clickhouse(mocker, dataset)
with app.test_request_context():
sql = _sql(dataset, FilterOperator.LENGTH_GREATER_THAN.value, "2")
assert "length(skills) > 2" in sql
def test_length_non_numeric_value_raises(mocker: MockerFixture, app: Flask) -> None:
dataset = _make_dataset(mocker)
_clickhouse(mocker, dataset)
with app.test_request_context(): # noqa: SIM117
with pytest.raises(QueryObjectValidationError):
_sql(dataset, FilterOperator.LENGTH_EQUALS.value, "abc")
def test_length_on_scalar_column_raises(mocker: MockerFixture, app: Flask) -> None:
"""Length on a scalar column is rejected even on an array-capable engine."""
dataset = _make_dataset(mocker)
_clickhouse(mocker, dataset)
with app.test_request_context(): # noqa: SIM117
with pytest.raises(QueryObjectValidationError):
_sql(dataset, FilterOperator.LENGTH_GREATER_THAN.value, 1, col="city")
def test_length_unsupported_engine_raises(mocker: MockerFixture, app: Flask) -> None:
"""On an engine without array support (sqlite) the length op is rejected."""
dataset = _make_dataset(mocker)
with app.test_request_context(): # noqa: SIM117
with pytest.raises(QueryObjectValidationError):
_sql(dataset, FilterOperator.LENGTH_GREATER_THAN.value, 1)
@@ -14,13 +14,7 @@
# KIND, either express or implied. See the License for the
# specific language governing permissions and limitations
# under the License.
import inspect
from superset.utils.pandas_postprocessing import (
escape_separator,
pivot,
unescape_separator,
)
from superset.utils.pandas_postprocessing import escape_separator, unescape_separator
def test_escape_separator():
@@ -34,19 +28,3 @@ def test_escape_separator():
escape_string = escape_separator("hello,world")
assert escape_string == r"hello\,world"
assert unescape_separator(escape_string) == "hello,world"
def test_validate_column_args_preserves_signature():
"""
The decorator must not hide the signature of the operation it wraps.
`inspect.signature` follows `__wrapped__`, which `functools.wraps` sets.
Without it every decorated operation reports `(df, **options)`, and code
that inspects the signature -- see `QueryObject._drop_unsupported_options`
-- cannot tell a supported option from an unsupported one.
"""
parameters = inspect.signature(pivot).parameters
assert pivot.__name__ == "pivot"
assert "options" not in parameters
assert {"index", "aggregates", "columns"} <= set(parameters)
@@ -22,7 +22,6 @@ from superset.common.query_object import QueryObject
from superset.connectors.sqla.models import SqlaTable
from superset.models.core import Database
from superset.superset_typing import Metric
from superset.utils import pandas_postprocessing
from superset.utils.core import override_user
@@ -439,143 +438,3 @@ def test_cache_key_cache_impersonation_on_with_different_user_and_db_impersonati
],
any_order=True,
)
def test_post_processing_drops_unsupported_options():
"""
An option that the operation no longer accepts is dropped, not passed on.
A chart saved by an older version of Superset stores `flatten_columns` in
the options of its `pivot` operation. `pivot` lost that parameter when
flattening became its own operation, so replaying the stored query_context
raised `TypeError: pivot() got an unexpected keyword argument
'flatten_columns'`.
"""
query_object = QueryObject(
row_limit=1,
post_processing=[
{
"operation": "pivot",
"options": {
"index": ["__timestamp"],
"columns": ["genre"],
"aggregates": {"count": {"operator": "mean"}},
"drop_missing_columns": False,
"flatten_columns": True,
"reset_index": True,
},
}
],
)
options = query_object.post_processing[0]["options"]
assert "flatten_columns" not in options
assert "reset_index" not in options
assert options["drop_missing_columns"] is False
assert options["index"] == ["__timestamp"]
def test_post_processing_keeps_supported_options():
"""Options the operation accepts are left alone."""
post_processing = [
{
"operation": "pivot",
"options": {"index": ["__timestamp"], "aggregates": {}},
}
]
query_object = QueryObject(row_limit=1, post_processing=post_processing)
assert query_object.post_processing == post_processing
def test_post_processing_keeps_unknown_operation():
"""
An unknown operation is kept, so that `exec_post_processing` can report it
as an `InvalidPostProcessingError` rather than being silently dropped here.
"""
query_object = QueryObject(
row_limit=1,
post_processing=[{"operation": "does_not_exist", "options": {"a": 1}}, None],
)
assert query_object.post_processing == [
{"operation": "does_not_exist", "options": {"a": 1}}
]
def test_post_processing_drops_the_dataframe_parameter():
"""
The DataFrame parameter is not an option.
`exec_post_processing` calls `operation(df, **options)`, so an option named
after the first parameter would raise `TypeError: pivot() got multiple
values for argument 'df'`.
"""
query_object = QueryObject(
row_limit=1,
post_processing=[
{
"operation": "pivot",
"options": {"df": "malformed", "index": ["a"], "aggregates": {}},
}
],
)
options = query_object.post_processing[0]["options"]
assert "df" not in options
assert options["index"] == ["a"]
def test_post_processing_keeps_options_of_a_variadic_operation():
"""An operation that accepts `**kwargs` accepts every option."""
def variadic(df, **kwargs):
return df
post_processing = [{"operation": "variadic", "options": {"anything": 1}}]
with patch.object(pandas_postprocessing, "variadic", variadic, create=True):
query_object = QueryObject(row_limit=1, post_processing=post_processing)
assert query_object.post_processing == post_processing
def test_post_processing_drops_a_variadic_positional_option():
"""
A `*args` parameter cannot be filled by a keyword argument.
`exec_post_processing` calls the operation as `operation(df, **options)`,
so an option named after a `*args` parameter would raise `TypeError:
variadic_positional() got an unexpected keyword argument 'args'` even
though the name appears in the signature.
"""
def variadic_positional(df, *args, index=None): # pylint: disable=unused-argument
return df
with patch.object(
pandas_postprocessing, "variadic_positional", variadic_positional, create=True
):
query_object = QueryObject(
row_limit=1,
post_processing=[
{
"operation": "variadic_positional",
"options": {"args": [1], "index": ["a"]},
}
],
)
options = query_object.post_processing[0]["options"]
assert "args" not in options
assert options["index"] == ["a"]
def test_post_processing_keeps_an_entry_without_an_operation():
"""
An entry that names no operation is kept, so that `exec_post_processing`
reports it as an `InvalidPostProcessingError`.
"""
post_processing = [{"options": {"a": 1}}]
query_object = QueryObject(row_limit=1, post_processing=post_processing)
assert query_object.post_processing == post_processing
-182
View File
@@ -637,91 +637,6 @@ class TestWebDriverPlaywrightErrorHandling:
"",
)
@patch("superset.utils.webdriver.PLAYWRIGHT_AVAILABLE", True)
@patch("superset.utils.webdriver._browser_manager")
@patch("superset.utils.webdriver.take_tiled_screenshot")
def test_large_report_dashboard_tiles_even_when_measured_height_is_short(
self, mock_take_tiled, mock_browser_manager
):
"""Regression: the GOOD (full) report path is the tiled one; BAD
(blank/partial) runs mis-route a large dashboard to the single-shot
non-tiled capture because ``scrollHeight`` is measured while charts are
still virtualized/collapsed (<= one tile). A scheduled report whose
dashboard is large by chart count must take the tiled path regardless
of that stale height measurement, so every region is scrolled into
view and waited on instead of captured as a windowed partial.
"""
mock_user = MagicMock()
mock_user.username = "test_user"
mock_browser = MagicMock()
mock_context = MagicMock()
mock_page = MagicMock()
mock_element = MagicMock()
mock_chart_container = MagicMock()
mock_browser_manager.get_browser.return_value = mock_browser
mock_browser.new_context.return_value = mock_context
mock_context.new_page.return_value = mock_page
def locator_side_effect(selector):
if selector == ".chart-container":
locator = MagicMock()
locator.all.return_value = [mock_chart_container]
return locator
return mock_element
mock_page.locator.side_effect = locator_side_effect
mock_take_tiled.return_value = b"tiled_screenshot"
def evaluate_side_effect(script):
if script == 'document.querySelectorAll(".chart-container").length':
return 52 # mounted containers
if "const target = document.querySelector" in script:
# Non-zero but <= one tile: the classic mid-layout measurement
# that previously vetoed tiling and dropped to the non-tiled
# path.
return 1500
return None
mock_page.evaluate.side_effect = evaluate_side_effect
with patch("superset.utils.webdriver.app") as mock_app:
mock_app.config = {
"WEBDRIVER_OPTION_ARGS": [],
"WEBDRIVER_WINDOW": {"pixel_density": 1},
"SCREENSHOT_PLAYWRIGHT_DEFAULT_TIMEOUT": 30000,
"SCREENSHOT_PLAYWRIGHT_WAIT_EVENT": "networkidle",
"SCREENSHOT_SELENIUM_HEADSTART": 1,
"SCREENSHOT_SELENIUM_ANIMATION_WAIT": 1,
"SCREENSHOT_LOCATE_WAIT": 10,
"SCREENSHOT_LOAD_WAIT": 10,
"SCREENSHOT_REPLACE_UNEXPECTED_ERRORS": False,
"SCREENSHOT_TILED_ENABLED": True,
"SCREENSHOT_TILED_CHART_THRESHOLD": 20,
"SCREENSHOT_TILED_HEIGHT_THRESHOLD": 5000,
# Larger than the measured 1500px height, so only the new
# report-mode branch (not `dashboard_height > tile_height`) can
# select tiling here.
"SCREENSHOT_TILED_VIEWPORT_HEIGHT": 2000,
}
with patch.object(WebDriverPlaywright, "auth") as mock_auth:
mock_auth.return_value = mock_context
driver = WebDriverPlaywright("chrome")
result = driver.get_screenshot(
"http://example.com/dashboard/805",
"standalone",
mock_user,
report_execution_context=_report_context(),
)
assert result == b"tiled_screenshot"
mock_take_tiled.assert_called_once()
# The non-tiled single-shot capture must not run for this large report.
mock_page.screenshot.assert_not_called()
@patch("superset.utils.webdriver.PLAYWRIGHT_AVAILABLE", True)
@patch("superset.utils.webdriver._browser_manager")
@patch("superset.utils.webdriver.logger")
@@ -1253,103 +1168,6 @@ class TestWebDriverPlaywrightChartReadiness:
page.wait_for_function.assert_not_called()
page.screenshot.assert_not_called()
def test_report_readiness_forces_below_fold_render_and_waits_for_all_holders(
self,
):
"""Regression for blank/partial report PDFs.
The non-tiled report capture takes a single full-page screenshot that
includes below-the-fold holders, so the readiness gate must (a) force
every virtualized row to render up front and (b) require *all* mounted
holders -- not just the viewport-visible ones -- to reach a terminal
state. Otherwise an off-screen holder that never rendered is captured
blank and silently delivered as a Success.
"""
from superset.utils.screenshot_utils import (
FORCE_ALL_CHART_HOLDERS_IN_VIEW_JS,
REPORT_ALL_CHART_HOLDERS_READY_JS,
)
page = MagicMock()
page.evaluate.return_value = [{"chartId": "7", "state": "rendered"}]
WebDriverPlaywright._wait_for_charts_ready(
page,
"http://example.com/dashboard/805",
5,
"standalone",
report_execution_context=_report_context(),
)
# (a) Off-screen rows are forced to render before the wait.
assert any(
call.args and call.args[0] == FORCE_ALL_CHART_HOLDERS_IN_VIEW_JS
for call in page.evaluate.call_args_list
)
# (b) The readiness predicate is the all-holders variant: it must not
# skip below-the-fold holders (no viewport-intersection test), so a
# virtualized/unrendered off-screen holder cannot satisfy the gate.
predicate = page.wait_for_function.call_args.args[0]
assert predicate == REPORT_ALL_CHART_HOLDERS_READY_JS
assert "getBoundingClientRect" not in predicate
assert "window.innerHeight" not in predicate
@patch("superset.utils.webdriver.logger")
def test_report_readiness_below_fold_unrendered_fails_loudly(self, mock_logger):
"""When an off-screen holder never renders within budget the report
must fail loudly (raise) rather than capture/deliver a blank
screenshot, and the terminal log must surface the below-the-fold
unready holders that the viewport-scoped diagnostic hides as
'virtualized'.
"""
from superset.utils.screenshot_utils import (
FIND_ALL_UNREADY_CHART_HOLDERS_JS,
FIND_CHART_HOLDER_STATES_JS,
)
from superset.utils.webdriver import PlaywrightTimeout
page = MagicMock()
# 22 on-screen rendered holders + 30 off-screen holders that the
# viewport-scoped diagnostic labels 'virtualized' (and would otherwise
# count as "ready").
holder_states = [
{"chartId": str(i), "state": "rendered"} for i in range(22)
] + [{"chartId": str(i), "state": "virtualized"} for i in range(22, 52)]
below_fold_unready = [
{"chartId": str(i), "state": "nothing_mounted"} for i in range(22, 52)
]
def _evaluate(script, *args):
if script == FIND_ALL_UNREADY_CHART_HOLDERS_JS:
return below_fold_unready
if script == FIND_CHART_HOLDER_STATES_JS:
return holder_states
return None
page.evaluate.side_effect = _evaluate
page.wait_for_function.side_effect = PlaywrightTimeout(
"below-fold holders never rendered"
)
with pytest.raises(PlaywrightTimeout):
WebDriverPlaywright._wait_for_charts_ready(
page,
"http://example.com/dashboard/805",
5,
"standalone",
report_execution_context=_report_context(),
)
terminal_call = next(
call
for call in mock_logger.warning.call_args_list
if call.args and call.args[0].startswith("report_readiness_terminal")
)
assert "terminal_reason=readiness_timeout" in terminal_call.args[0]
# The below-the-fold offenders are surfaced explicitly.
assert "all_unready_holders=" in terminal_call.args[0]
assert below_fold_unready in terminal_call.args
@patch("superset.utils.webdriver.logger")
def test_chart_capture_ready_logs_container_state_not_holder_counts(
self, mock_logger