diff --git a/packages/backend-common/src/reading/AzureUrlReader.test.ts b/packages/backend-common/src/reading/AzureUrlReader.test.ts index 6efc7641b8..ded54cba62 100644 --- a/packages/backend-common/src/reading/AzureUrlReader.test.ts +++ b/packages/backend-common/src/reading/AzureUrlReader.test.ts @@ -14,11 +14,13 @@ * limitations under the License. */ +import fs from 'fs'; +import path from 'path'; import { rest } from 'msw'; import { setupServer } from 'msw/node'; import { ConfigReader } from '@backstage/config'; import { getVoidLogger } from '../logging'; -import { AzureUrlReader } from './AzureUrlReader'; +import { AzureUrlReader, getDownloadUrl } from './AzureUrlReader'; import { msw } from '@backstage/test-utils'; import { ReadTreeResponseFactory } from './tree'; @@ -32,104 +34,165 @@ describe('AzureUrlReader', () => { const worker = setupServer(); msw.setupDefaultHandlers(worker); - beforeEach(() => { - worker.use( - rest.get('*', (req, res, ctx) => - res( - ctx.status(200), - ctx.json({ - url: req.url.toString(), - headers: req.headers.getAllHeaders(), - }), + describe('read', () => { + beforeEach(() => { + worker.use( + rest.get('*', (req, res, ctx) => + res( + ctx.status(200), + ctx.json({ + url: req.url.toString(), + headers: req.headers.getAllHeaders(), + }), + ), ), - ), - ); - }); - - const createConfig = (token?: string) => - new ConfigReader( - { - integrations: { azure: [{ host: 'dev.azure.com', token }] }, - }, - 'test-config', - ); - - it.each([ - { - url: - 'https://dev.azure.com/org-name/project-name/_git/repo-name?path=my-template.yaml&version=GBmaster', - config: createConfig(), - response: expect.objectContaining({ - url: - 'https://dev.azure.com/org-name/project-name/_apis/git/repositories/repo-name/items?path=my-template.yaml&version=master', - }), - }, - { - url: - 'https://dev.azure.com/org-name/project-name/_git/repo-name?path=my-template.yaml', - config: createConfig(), - response: expect.objectContaining({ - url: - 'https://dev.azure.com/org-name/project-name/_apis/git/repositories/repo-name/items?path=my-template.yaml', - }), - }, - { - url: 'https://dev.azure.com/a/b/_git/repo-name?path=my-template.yaml', - config: createConfig('0123456789'), - response: expect.objectContaining({ - headers: expect.objectContaining({ - authorization: 'Basic OjAxMjM0NTY3ODk=', - }), - }), - }, - { - url: 'https://dev.azure.com/a/b/_git/repo-name?path=my-template.yaml', - config: createConfig(undefined), - response: expect.objectContaining({ - headers: expect.not.objectContaining({ - authorization: expect.anything(), - }), - }), - }, - ])('should handle happy path %#', async ({ url, config, response }) => { - const [{ reader }] = AzureUrlReader.factory({ - config, - logger, - treeResponseFactory, + ); }); - const data = await reader.read(url); - const res = await JSON.parse(data.toString('utf-8')); - expect(res).toEqual(response); - }); + const createConfig = (token?: string) => + new ConfigReader( + { + integrations: { azure: [{ host: 'dev.azure.com', token }] }, + }, + 'test-config', + ); - it.each([ - { - url: 'https://api.com/a/b/blob/master/path/to/c.yaml', - config: createConfig(), - error: - 'Incorrect url: https://api.com/a/b/blob/master/path/to/c.yaml, Error: Wrong Azure Devops URL or Invalid file path', - }, - { - url: 'com/a/b/blob/master/path/to/c.yaml', - config: createConfig(), - error: - 'Incorrect url: com/a/b/blob/master/path/to/c.yaml, TypeError: Invalid URL: com/a/b/blob/master/path/to/c.yaml', - }, - { - url: '', - config: createConfig(''), - error: - "Invalid type in config for key 'integrations.azure[0].token' in 'test-config', got empty-string, wanted string", - }, - ])('should handle error path %#', async ({ url, config, error }) => { - await expect(async () => { + it.each([ + { + url: + 'https://dev.azure.com/org-name/project-name/_git/repo-name?path=my-template.yaml&version=GBmaster', + config: createConfig(), + response: expect.objectContaining({ + url: + 'https://dev.azure.com/org-name/project-name/_apis/git/repositories/repo-name/items?path=my-template.yaml&version=master', + }), + }, + { + url: + 'https://dev.azure.com/org-name/project-name/_git/repo-name?path=my-template.yaml', + config: createConfig(), + response: expect.objectContaining({ + url: + 'https://dev.azure.com/org-name/project-name/_apis/git/repositories/repo-name/items?path=my-template.yaml', + }), + }, + { + url: 'https://dev.azure.com/a/b/_git/repo-name?path=my-template.yaml', + config: createConfig('0123456789'), + response: expect.objectContaining({ + headers: expect.objectContaining({ + authorization: 'Basic OjAxMjM0NTY3ODk=', + }), + }), + }, + { + url: 'https://dev.azure.com/a/b/_git/repo-name?path=my-template.yaml', + config: createConfig(undefined), + response: expect.objectContaining({ + headers: expect.not.objectContaining({ + authorization: expect.anything(), + }), + }), + }, + ])('should handle happy path %#', async ({ url, config, response }) => { const [{ reader }] = AzureUrlReader.factory({ config, logger, treeResponseFactory, }); - await reader.read(url); - }).rejects.toThrow(error); + + const data = await reader.read(url); + const res = await JSON.parse(data.toString('utf-8')); + expect(res).toEqual(response); + }); + + it.each([ + { + url: 'https://api.com/a/b/blob/master/path/to/c.yaml', + config: createConfig(), + error: + 'Incorrect url: https://api.com/a/b/blob/master/path/to/c.yaml, Error: Wrong Azure Devops URL or Invalid file path', + }, + { + url: 'com/a/b/blob/master/path/to/c.yaml', + config: createConfig(), + error: + 'Incorrect url: com/a/b/blob/master/path/to/c.yaml, TypeError: Invalid URL: com/a/b/blob/master/path/to/c.yaml', + }, + { + url: '', + config: createConfig(''), + error: + "Invalid type in config for key 'integrations.azure[0].token' in 'test-config', got empty-string, wanted string", + }, + ])('should handle error path %#', async ({ url, config, error }) => { + await expect(async () => { + const [{ reader }] = AzureUrlReader.factory({ + config, + logger, + treeResponseFactory, + }); + await reader.read(url); + }).rejects.toThrow(error); + }); + }); + + describe('readTree', () => { + const repoBuffer = fs.readFileSync( + path.resolve('src', 'reading', '__fixtures__', 'repo.zip'), + ); + + beforeEach(() => { + worker.use( + rest.get( + 'https://dev.azure.com/organization/project/_apis/git/repositories/repository/items', + (_, res, ctx) => + res( + ctx.status(200), + ctx.set('Content-Type', 'application/zip'), + ctx.body(repoBuffer), + ), + ), + ); + }); + + it('returns the wanted files from an archive', async () => { + const processor = new AzureUrlReader( + { + host: 'dev.azure.com', + }, + treeResponseFactory, + ); + + const response = await processor.readTree( + 'https://dev.azure.com/organization/project/_git/repository', + ); + + const files = await response.files(); + + expect(files.length).toBe(2); + const mkDocsFile = await files[1].content(); + const indexMarkdownFile = await files[0].content(); + + expect(mkDocsFile.toString()).toBe('site_name: Test\n'); + expect(indexMarkdownFile.toString()).toBe('# Test\n'); + }); + }); + + describe('getDownloadUrl', () => { + it('add no scopePath if no path is specified', async () => { + const result = getDownloadUrl( + 'https://dev.azure.com/organization/project/_git/repository', + ); + + expect(result.searchParams.get('scopePath')).toBeNull(); + }); + + it('add the scopePath if a path is specified', async () => { + const result = getDownloadUrl( + 'https://dev.azure.com/organization/project/_git/repository?path=%2Fdocs', + ); + expect(result.searchParams.get('scopePath')).toEqual('docs'); + }); }); }); diff --git a/packages/backend-common/src/reading/AzureUrlReader.ts b/packages/backend-common/src/reading/AzureUrlReader.ts index db8b738667..068f4d2409 100644 --- a/packages/backend-common/src/reading/AzureUrlReader.ts +++ b/packages/backend-common/src/reading/AzureUrlReader.ts @@ -19,22 +19,53 @@ import { readAzureIntegrationConfigs, } from '@backstage/integration'; import fetch from 'cross-fetch'; +import { Readable } from 'stream'; +import parseGitUri from 'git-url-parse'; import { NotFoundError } from '../errors'; -import { ReaderFactory, ReadTreeResponse, UrlReader } from './types'; +import { + ReaderFactory, + ReadTreeOptions, + ReadTreeResponse, + UrlReader, +} from './types'; +import { ReadTreeResponseFactory } from './tree'; + +export function getDownloadUrl(url: string): URL { + const { + name: repoName, + owner: project, + organization, + protocol, + resource, + filepath, + } = parseGitUri(url); + + // scopePath will limit the downloaded content + // /docs will only download the docs folder and everything below it + // /docs/index.md will only download index.md but put it in the root of the archive + const scopePath = filepath ? `&scopePath=${filepath}` : ''; + + return new URL( + `${protocol}://${resource}/${organization}/${project}/_apis/git/repositories/${repoName}/items?recursionLevel=full&download=true&api-version=6.0${scopePath}`, + ); +} export class AzureUrlReader implements UrlReader { - static factory: ReaderFactory = ({ config }) => { + static factory: ReaderFactory = ({ config, treeResponseFactory }) => { const configs = readAzureIntegrationConfigs( config.getOptionalConfigArray('integrations.azure') ?? [], ); return configs.map(options => { - const reader = new AzureUrlReader(options); + const reader = new AzureUrlReader(options, treeResponseFactory); const predicate = (url: URL) => url.host === options.host; return { reader, predicate }; }); }; - constructor(private readonly options: AzureIntegrationConfig) { + constructor( + private readonly options: AzureIntegrationConfig, + private readonly treeResponseFactory: ReadTreeResponseFactory, + ) { if (options.host !== 'dev.azure.com') { throw Error( `Azure integration currently only supports 'dev.azure.com', tried to use host '${options.host}'`, @@ -64,8 +95,26 @@ export class AzureUrlReader implements UrlReader { throw new Error(message); } - readTree(): Promise { - throw new Error('AzureUrlReader does not implement readTree'); + async readTree( + url: string, + options?: ReadTreeOptions, + ): Promise { + const response = await fetch(getDownloadUrl(url).toString(), { + ...this.getRequestOptions(), + headers: { Accept: 'application/zip' }, + }); + if (!response.ok) { + const message = `Failed to read tree from ${url}, ${response.status} ${response.statusText}`; + if (response.status === 404) { + throw new NotFoundError(message); + } + throw new Error(message); + } + + return this.treeResponseFactory.fromZipArchive({ + stream: (response.body as unknown) as Readable, + filter: options?.filter, + }); } // Converts diff --git a/packages/backend-common/src/reading/GithubUrlReader.ts b/packages/backend-common/src/reading/GithubUrlReader.ts index 2fbaa0b32b..907f2ada7a 100644 --- a/packages/backend-common/src/reading/GithubUrlReader.ts +++ b/packages/backend-common/src/reading/GithubUrlReader.ts @@ -207,7 +207,7 @@ export class GithubUrlReader implements UrlReader { const path = `${repoName}-${ref}/${filepath}`; - return this.deps.treeResponseFactory.fromArchive({ + return this.deps.treeResponseFactory.fromTarArchive({ // TODO(Rugvip): Underlying implementation of fetch will be node-fetch, we probably want // to stick to using that in exclusively backend code. stream: (response.body as unknown) as Readable, diff --git a/packages/backend-common/src/reading/tree/ReadTreeResponseFactory.ts b/packages/backend-common/src/reading/tree/ReadTreeResponseFactory.ts index 90d74e1f20..986a0302bc 100644 --- a/packages/backend-common/src/reading/tree/ReadTreeResponseFactory.ts +++ b/packages/backend-common/src/reading/tree/ReadTreeResponseFactory.ts @@ -19,6 +19,7 @@ import { Readable } from 'stream'; import { Config } from '@backstage/config'; import { ReadTreeResponse } from '../types'; import { TarArchiveResponse } from './TarArchiveResponse'; +import { ZipArchiveResponse } from './ZipArchiveResponse'; type FromArchiveOptions = { // A binary stream of a tar archive. @@ -39,7 +40,7 @@ export class ReadTreeResponseFactory { constructor(private readonly workDir: string) {} - async fromArchive(options: FromArchiveOptions): Promise { + async fromTarArchive(options: FromArchiveOptions): Promise { return new TarArchiveResponse( options.stream, options.path ?? '', @@ -47,4 +48,13 @@ export class ReadTreeResponseFactory { options.filter, ); } + + async fromZipArchive(options: FromArchiveOptions): Promise { + return new ZipArchiveResponse( + options.stream, + options.path ?? '', + this.workDir, + options.filter, + ); + } }