ckaczor/azuredatastudio
1
0
Fork 0
mirror of https://github.com/ckaczor/azuredatastudio.git synced 2026-02-17 02:51:36 -05:00
Code Issues Packages Projects Releases Wiki Activity

Update python packages to use sparkmagic 0.12.9 (#7240)

Browse Source
This commit is contained in:
Cory Rivera
2019-09-17 15:05:59 -07:00
committed by GitHub
parent 290dd9531f
commit 094d6f2339
9 changed files with 59 additions and 108 deletions
Download Patch File Download Diff File Expand all files Collapse all files

16
extensions/integration-tests/src/notebook.test.ts
View File

@@ -10,7 +10,7 @@ import * as assert from 'assert';
import * as azdata from 'azdata'; import * as azdata from 'azdata';
import * as vscode from 'vscode'; import * as vscode from 'vscode';
import { context } from './testContext'; import { context } from './testContext';
import { sqlNotebookContent, writeNotebookToFile, sqlKernelMetadata, getFileName, pySparkNotebookContent, pySpark3KernelMetadata, pythonKernelMetadata, sqlNotebookMultipleCellsContent, notebookContentForCellLanguageTest, sqlKernelSpec, pythonKernelSpec, pySpark3KernelSpec, CellTypes } from './notebook.util'; import { sqlNotebookContent, writeNotebookToFile, sqlKernelMetadata, getFileName, pySparkNotebookContent, pySparkKernelMetadata, pythonKernelMetadata, sqlNotebookMultipleCellsContent, notebookContentForCellLanguageTest, sqlKernelSpec, pythonKernelSpec, pySparkKernelSpec, CellTypes } from './notebook.util';
import { getBdcServer, getConfigValue, EnvironmentVariable_PYTHON_PATH } from './testConfig'; import { getBdcServer, getConfigValue, EnvironmentVariable_PYTHON_PATH } from './testConfig';
import { connectToServer, sleep } from './utils'; import { connectToServer, sleep } from './utils';
import * as fs from 'fs'; import * as fs from 'fs';
@@ -79,14 +79,14 @@ if (context.RunTest) {
await (new NotebookTester()).pythonChangeKernelDifferentProviderTest(this.test.title); await (new NotebookTester()).pythonChangeKernelDifferentProviderTest(this.test.title);
}); });
test('Change kernel same provider Python to PySpark3 to Python', async function () { test('Change kernel same provider Python to PySpark to Python', async function () {
await (new NotebookTester()).pythonChangeKernelSameProviderTest(this.test.title); await (new NotebookTester()).pythonChangeKernelSameProviderTest(this.test.title);
}); });
} }
if (process.env['RUN_PYSPARK_TEST'] === '1') { if (process.env['RUN_PYSPARK_TEST'] === '1') {
test('PySpark3 notebook test', async function () { test('PySpark notebook test', async function () {
await (new NotebookTester()).pySpark3NbTest(this.test.title); await (new NotebookTester()).pySparkNbTest(this.test.title);
}); });
} }
@@ -112,8 +112,8 @@ class NotebookTester {
invocationCount: number = 0; invocationCount: number = 0;
@stressify({ dop: NotebookTester.ParallelCount }) @stressify({ dop: NotebookTester.ParallelCount })
async pySpark3NbTest(title: string): Promise<void> { async pySparkNbTest(title: string): Promise<void> {
let notebook = await this.openNotebook(pySparkNotebookContent, pySpark3KernelMetadata, title + this.invocationCount++); let notebook = await this.openNotebook(pySparkNotebookContent, pySparkKernelMetadata, title + this.invocationCount++);
await this.runCell(notebook); await this.runCell(notebook);
let cellOutputs = notebook.document.cells[0].contents.outputs; let cellOutputs = notebook.document.cells[0].contents.outputs;
let sparkResult = (<azdata.nb.IStreamResult>cellOutputs[3]).text; let sparkResult = (<azdata.nb.IStreamResult>cellOutputs[3]).text;
@@ -281,9 +281,9 @@ class NotebookTester {
assert(notebook.document.providerId === 'jupyter', `Expected providerId to be jupyter, Actual: ${notebook.document.providerId}`); assert(notebook.document.providerId === 'jupyter', `Expected providerId to be jupyter, Actual: ${notebook.document.providerId}`);
assert(notebook.document.kernelSpec.name === 'python3', `Expected first kernel name: python3, Actual: ${notebook.document.kernelSpec.name}`); assert(notebook.document.kernelSpec.name === 'python3', `Expected first kernel name: python3, Actual: ${notebook.document.kernelSpec.name}`);
let kernelChanged = await notebook.changeKernel(pySpark3KernelSpec); let kernelChanged = await notebook.changeKernel(pySparkKernelSpec);
assert(notebook.document.providerId === 'jupyter', `Expected providerId to be jupyter, Actual: ${notebook.document.providerId}`); assert(notebook.document.providerId === 'jupyter', `Expected providerId to be jupyter, Actual: ${notebook.document.providerId}`);
assert(kernelChanged && notebook.document.kernelSpec.name === 'pyspark3kernel', `Expected second kernel name: pyspark3kernel, Actual: ${notebook.document.kernelSpec.name}`); assert(kernelChanged && notebook.document.kernelSpec.name === 'pysparkkernel', `Expected second kernel name: pysparkkernel, Actual: ${notebook.document.kernelSpec.name}`);
kernelChanged = await notebook.changeKernel(pythonKernelSpec); kernelChanged = await notebook.changeKernel(pythonKernelSpec);
assert(notebook.document.providerId === 'jupyter', `Expected providerId to be jupyter, Actual: ${notebook.document.providerId}`); assert(notebook.document.providerId === 'jupyter', `Expected providerId to be jupyter, Actual: ${notebook.document.providerId}`);

16
extensions/integration-tests/src/notebook.util.ts
View File

@@ -27,8 +27,8 @@ export const pySparkNotebookContent: azdata.nb.INotebookContents = {
}], }],
metadata: { metadata: {
'kernelspec': { 'kernelspec': {
'name': 'pyspark3kernel', 'name': 'pysparkkernel',
'display_name': 'PySpark3' 'display_name': 'PySpark'
} }
}, },
nbformat: 4, nbformat: 4,
@@ -127,16 +127,16 @@ export const sqlNotebookMultipleCellsContent: azdata.nb.INotebookContents = {
nbformat_minor: 2 nbformat_minor: 2
}; };
export const pySpark3KernelMetadata = { export const pySparkKernelMetadata = {
'kernelspec': { 'kernelspec': {
'name': 'pyspark3kernel', 'name': 'pysparkkernel',
'display_name': 'PySpark3' 'display_name': 'PySpark'
} }
}; };
export const pySpark3KernelSpec = { export const pySparkKernelSpec = {
name: 'pyspark3', name: 'pyspark',
display_name: 'PySpark3' display_name: 'PySpark'
}; };
export const sqlKernelMetadata = { export const sqlKernelMetadata = {

10
extensions/notebook/kernels/pyspark3kernel/kernel.json
View File

@@ -1,10 +0,0 @@
{
"argv": [
"python",
"-m",
"sparkmagic.kernels.pyspark3kernel.pyspark3kernel",
"-f",
"{connection_file}"
],
"display_name": "PySpark3"
}

8
extensions/notebook/package.json
View File

@@ -310,14 +310,6 @@
"IPYNB" "IPYNB"
], ],
"standardKernels": [ "standardKernels": [
{
"name": "pyspark3kernel",
"displayName": "PySpark3",
"connectionProviderIds": [
"HADOOP_KNOX",
"MSSQL"
]
},
{ {
"name": "pysparkkernel", "name": "pysparkkernel",
"displayName": "PySpark", "displayName": "PySpark",

8
extensions/notebook/src/common/constants.ts
View File

@@ -11,7 +11,6 @@ export const extensionOutputChannel = 'Notebooks';
// JUPYTER CONFIG ////////////////////////////////////////////////////////// // JUPYTER CONFIG //////////////////////////////////////////////////////////
export const pythonBundleVersion = '0.0.1'; export const pythonBundleVersion = '0.0.1';
export const pythonVersion = '3.6.6'; export const pythonVersion = '3.6.6';
export const sparkMagicVersion = '0.12.6.1';
export const pythonPathConfigKey = 'pythonPath'; export const pythonPathConfigKey = 'pythonPath';
export const existingPythonConfigKey = 'useExistingPython'; export const existingPythonConfigKey = 'useExistingPython';
export const notebookConfigKey = 'notebook'; export const notebookConfigKey = 'notebook';
@@ -42,7 +41,6 @@ export enum PythonPkgType {
Anaconda = 'Anaconda' Anaconda = 'Anaconda'
} }
export const pythonOfflinePipPackagesUrl = 'https://go.microsoft.com/fwlink/?linkid=2092867'; export const pythonWindowsInstallUrl = 'https://go.microsoft.com/fwlink/?linkid=2103837';
export const pythonWindowsInstallUrl = 'https://go.microsoft.com/fwlink/?linkid=2092866'; export const pythonMacInstallUrl = 'https://go.microsoft.com/fwlink/?linkid=2103836';
export const pythonMacInstallUrl = 'https://go.microsoft.com/fwlink/?linkid=2092865'; export const pythonLinuxInstallUrl = 'https://go.microsoft.com/fwlink/?linkid=2103835';
export const pythonLinuxInstallUrl = 'https://go.microsoft.com/fwlink/?linkid=2092864';

4
extensions/notebook/src/extension.ts
View File

@@ -221,8 +221,8 @@ async function analyzeNotebook(oeContext?: azdata.ObjectExplorerContext): Promis
providerId: JUPYTER_NOTEBOOK_PROVIDER, providerId: JUPYTER_NOTEBOOK_PROVIDER,
preview: false, preview: false,
defaultKernel: { defaultKernel: {
name: 'pyspark3kernel', name: 'pysparkkernel',
display_name: 'PySpark3', display_name: 'PySpark',
language: 'python' language: 'python'
} }
}); });

4
extensions/notebook/src/jupyter/jupyterController.ts
View File

@@ -150,8 +150,8 @@ export class JupyterController implements vscode.Disposable {
providerId: constants.jupyterNotebookProviderId, providerId: constants.jupyterNotebookProviderId,
preview: false, preview: false,
defaultKernel: { defaultKernel: {
name: 'pyspark3kernel', name: 'pysparkkernel',
display_name: 'PySpark3', display_name: 'PySpark',
language: 'python' language: 'python'
} }
}); });

73
extensions/notebook/src/jupyter/jupyterServerInstallation.ts
View File

@@ -77,9 +77,6 @@ export class JupyterServerInstallation {
try { try {
await this.installPythonPackage(backgroundOperation); await this.installPythonPackage(backgroundOperation);
this.outputChannel.appendLine(msgPythonDownloadComplete);
backgroundOperation.updateStatus(azdata.TaskStatus.InProgress, msgPythonDownloadComplete);
if (this._usingConda) { if (this._usingConda) {
await this.installCondaDependencies(); await this.installCondaDependencies();
} else if (this._usingExistingPython) { } else if (this._usingExistingPython) {
@@ -87,19 +84,11 @@ export class JupyterServerInstallation {
} else { } else {
await this.installOfflinePipDependencies(); await this.installOfflinePipDependencies();
} }
let doOnlineInstall = this._usingExistingPython;
await this.installSparkMagic(doOnlineInstall);
} catch (err) { } catch (err) {
this.outputChannel.appendLine(msgDependenciesInstallationFailed(utils.getErrorMessage(err))); this.outputChannel.appendLine(msgDependenciesInstallationFailed(utils.getErrorMessage(err)));
throw err; throw err;
} }
fs.remove(this._pythonPackageDir, (err: Error) => {
if (err) {
this.outputChannel.appendLine(err.message);
}
});
this.outputChannel.appendLine(msgInstallPkgFinish); this.outputChannel.appendLine(msgInstallPkgFinish);
backgroundOperation.updateStatus(azdata.TaskStatus.Succeeded, msgInstallPkgFinish); backgroundOperation.updateStatus(azdata.TaskStatus.Succeeded, msgInstallPkgFinish);
window.showInformationMessage(msgInstallPkgFinish); window.showInformationMessage(msgInstallPkgFinish);
@@ -107,15 +96,16 @@ export class JupyterServerInstallation {
} }
private installPythonPackage(backgroundOperation: azdata.BackgroundOperation): Promise<void> { private installPythonPackage(backgroundOperation: azdata.BackgroundOperation): Promise<void> {
if (this._usingExistingPython) {
return Promise.resolve();
}
let bundleVersion = constants.pythonBundleVersion; let bundleVersion = constants.pythonBundleVersion;
let pythonVersion = constants.pythonVersion; let pythonVersion = constants.pythonVersion;
let platformId = utils.getOSPlatformId(); let platformId = utils.getOSPlatformId();
let packageName: string; let packageName: string;
let pythonDownloadUrl: string; let pythonDownloadUrl: string;
if (this._usingExistingPython) {
packageName = `python-${pythonVersion}-${bundleVersion}-offlinePackages.zip`;
pythonDownloadUrl = constants.pythonOfflinePipPackagesUrl;
} else {
let extension = process.platform === constants.winPlatform ? 'zip' : 'tar.gz'; let extension = process.platform === constants.winPlatform ? 'zip' : 'tar.gz';
packageName = `python-${pythonVersion}-${platformId}-${bundleVersion}.${extension}`; packageName = `python-${pythonVersion}-${platformId}-${bundleVersion}.${extension}`;
@@ -131,17 +121,9 @@ export class JupyterServerInstallation {
pythonDownloadUrl = constants.pythonLinuxInstallUrl; pythonDownloadUrl = constants.pythonLinuxInstallUrl;
break; break;
} }
}
let installPath: string;
if (this._usingExistingPython) {
installPath = utils.getUserHome();
} else {
installPath = this._pythonInstallationPath;
}
let pythonPackagePathLocal = path.join(installPath, packageName);
return new Promise((resolve, reject) => { return new Promise((resolve, reject) => {
let installPath = this._pythonInstallationPath;
backgroundOperation.updateStatus(azdata.TaskStatus.InProgress, msgDownloadPython(platformId, pythonDownloadUrl)); backgroundOperation.updateStatus(azdata.TaskStatus.InProgress, msgDownloadPython(platformId, pythonDownloadUrl));
fs.mkdirs(installPath, (err) => { fs.mkdirs(installPath, (err) => {
if (err) { if (err) {
@@ -179,12 +161,13 @@ export class JupyterServerInstallation {
} }
}); });
let pythonPackagePathLocal = path.join(installPath, packageName);
downloadRequest.pipe(fs.createWriteStream(pythonPackagePathLocal)) downloadRequest.pipe(fs.createWriteStream(pythonPackagePathLocal))
.on('close', async () => { .on('close', async () => {
//unpack python zip/tar file //unpack python zip/tar file
this.outputChannel.appendLine(msgPythonUnpackPending); this.outputChannel.appendLine(msgPythonUnpackPending);
let pythonSourcePath = path.join(installPath, constants.pythonBundleVersion); let pythonSourcePath = path.join(installPath, constants.pythonBundleVersion);
if (!this._usingExistingPython && await utils.exists(pythonSourcePath)) { if (await utils.exists(pythonSourcePath)) {
try { try {
fs.removeSync(pythonSourcePath); fs.removeSync(pythonSourcePath);
} catch (err) { } catch (err) {
@@ -201,6 +184,8 @@ export class JupyterServerInstallation {
} }
}); });
this.outputChannel.appendLine(msgPythonDownloadComplete);
backgroundOperation.updateStatus(azdata.TaskStatus.InProgress, msgPythonDownloadComplete);
resolve(); resolve();
}).catch(err => { }).catch(err => {
backgroundOperation.updateStatus(azdata.TaskStatus.InProgress, msgPythonUnpackError); backgroundOperation.updateStatus(azdata.TaskStatus.InProgress, msgPythonUnpackError);
@@ -223,7 +208,7 @@ export class JupyterServerInstallation {
: path.join(this._pythonInstallationPath, constants.pythonBundleVersion); : path.join(this._pythonInstallationPath, constants.pythonBundleVersion);
if (this._usingExistingPython) { if (this._usingExistingPython) {
this._pythonPackageDir = path.join(utils.getUserHome(), 'offlinePackages'); this._pythonPackageDir = undefined;
} else { } else {
this._pythonPackageDir = path.join(pythonSourcePath, 'offlinePackages'); this._pythonPackageDir = path.join(pythonSourcePath, 'offlinePackages');
} }
@@ -438,41 +423,27 @@ export class JupyterServerInstallation {
installJupyterCommand = `"${this._pythonExecutable}" -m pip install --force-reinstall --no-index pip --find-links "${this._pythonPackageDir}" --no-warn-script-location`; installJupyterCommand = `"${this._pythonExecutable}" -m pip install --force-reinstall --no-index pip --find-links "${this._pythonPackageDir}" --no-warn-script-location`;
await this.executeStreamedCommand(installJupyterCommand); await this.executeStreamedCommand(installJupyterCommand);
fs.remove(this._pythonPackageDir, (err: Error) => {
if (err) {
this.outputChannel.appendLine(err.message);
}
});
this.outputChannel.appendLine(localize('msgJupyterInstallDone', "... Jupyter installation complete.")); this.outputChannel.appendLine(localize('msgJupyterInstallDone', "... Jupyter installation complete."));
} else { } else {
return Promise.resolve(); return Promise.resolve();
} }
} }
private async installSparkMagic(doOnlineInstall: boolean): Promise<void> {
let installSparkMagic: string;
if (process.platform === constants.winPlatform || this._usingExistingPython) {
// Overwrite existing install of sparkmagic, since we use a custom version
let cmdOptions = this._usingExistingPython ? '--user --force-reinstall' : '--force-reinstall';
let sparkWheel = path.join(this._pythonPackageDir, `sparkmagic-${constants.sparkMagicVersion}-py3-none-any.whl`);
if (doOnlineInstall) {
installSparkMagic = `"${this._pythonExecutable}" -m pip install ${cmdOptions} "${sparkWheel}" --no-warn-script-location`;
} else {
installSparkMagic = `"${this._pythonExecutable}" -m pip install ${cmdOptions} --no-index "${sparkWheel}" --find-links "${this._pythonPackageDir}" --no-warn-script-location`;
}
}
if (installSparkMagic) {
this.outputChannel.show(true);
this.outputChannel.appendLine(localize('msgInstallingSpark', "Installing SparkMagic..."));
await this.executeStreamedCommand(installSparkMagic);
}
}
private async installPipDependencies(): Promise<void> { private async installPipDependencies(): Promise<void> {
this.outputChannel.show(true); this.outputChannel.show(true);
this.outputChannel.appendLine(localize('msgInstallStart', "Installing required packages to run Notebooks...")); this.outputChannel.appendLine(localize('msgInstallStart', "Installing required packages to run Notebooks..."));
let cmdOptions = this._usingExistingPython ? '--user' : ''; let cmdOptions = this._usingExistingPython ? '--user' : '';
let installCommand = `"${this._pythonExecutable}" -m pip install ${cmdOptions} jupyter==1.0.0 pandas==0.24.2`; let installCommand = `"${this._pythonExecutable}" -m pip install ${cmdOptions} jupyter>=1.0.0 pandas>=0.24.2 sparkmagic>=0.12.9`;
await this.executeStreamedCommand(installCommand); await this.executeStreamedCommand(installCommand);
installCommand = `"${this._pythonExecutable}" -m pip install ${cmdOptions} prose-codeaccelerator==1.3.0 --extra-index-url https://prose-python-packages.azurewebsites.net`; installCommand = `"${this._pythonExecutable}" -m pip install ${cmdOptions} prose-codeaccelerator>=1.3.0 --extra-index-url https://prose-python-packages.azurewebsites.net`;
await this.executeStreamedCommand(installCommand); await this.executeStreamedCommand(installCommand);
this.outputChannel.appendLine(localize('msgJupyterInstallDone', "... Jupyter installation complete.")); this.outputChannel.appendLine(localize('msgJupyterInstallDone', "... Jupyter installation complete."));
@@ -482,14 +453,14 @@ export class JupyterServerInstallation {
this.outputChannel.show(true); this.outputChannel.show(true);
this.outputChannel.appendLine(localize('msgInstallStart', "Installing required packages to run Notebooks...")); this.outputChannel.appendLine(localize('msgInstallStart', "Installing required packages to run Notebooks..."));
let installCommand = `"${this.getCondaExePath()}" install -y jupyter==1.0.0 pandas==0.24.2`; let installCommand = `"${this.getCondaExePath()}" install -y jupyter>=1.0.0 pandas>=0.24.2`;
if (process.platform !== constants.winPlatform) { if (process.platform !== constants.winPlatform) {
installCommand = `${installCommand} pykerberos==1.2.1`; installCommand = `${installCommand} pykerberos>=1.2.1`;
} }
await this.executeStreamedCommand(installCommand); await this.executeStreamedCommand(installCommand);
let cmdOptions = this._usingExistingPython ? '--user' : ''; let cmdOptions = this._usingExistingPython ? '--user' : '';
installCommand = `"${this._pythonExecutable}" -m pip install ${cmdOptions} prose-codeaccelerator==1.3.0 --extra-index-url https://prose-python-packages.azurewebsites.net`; installCommand = `"${this._pythonExecutable}" -m pip install ${cmdOptions} sparkmagic>=0.12.9 prose-codeaccelerator>=1.3.0 --extra-index-url https://prose-python-packages.azurewebsites.net`;
await this.executeStreamedCommand(installCommand); await this.executeStreamedCommand(installCommand);
this.outputChannel.appendLine(localize('msgJupyterInstallDone', "... Jupyter installation complete.")); this.outputChannel.appendLine(localize('msgJupyterInstallDone', "... Jupyter installation complete."));

2
src/sql/workbench/services/notebook/common/notebookRegistry.ts
View File

@@ -113,7 +113,7 @@ let notebookLanguageMagicType: IJSONSchema = {
type: 'string' type: 'string'
}, },
kernels: { kernels: {
description: localize('carbon.extension.contributes.notebook.kernels', "Optional set of kernels this is valid for, e.g. python3, pyspark3, sql"), description: localize('carbon.extension.contributes.notebook.kernels', "Optional set of kernels this is valid for, e.g. python3, pyspark, sql"),
oneOf: [ oneOf: [
{ type: 'string' }, { type: 'string' },
{ {