From 8cfeb5fe08dd20d0fbeeaa0d2442ba6e8bff74a2 Mon Sep 17 00:00:00 2001
From: Patrik Oldsberg <poldsberg@gmail.com>
Date: Fri, 10 Sep 2021 12:08:08 +0200
Subject: [PATCH] backend-common: fix azure readTree subpath handling

Signed-off-by: Patrik Oldsberg <poldsberg@gmail.com>
---
 .../src/reading/AzureUrlReader.test.ts        | 10 ++++-----
 .../src/reading/AzureUrlReader.ts             | 22 +++++++++++++------
 2 files changed, 19 insertions(+), 13 deletions(-)

diff --git a/packages/backend-common/src/reading/AzureUrlReader.test.ts b/packages/backend-common/src/reading/AzureUrlReader.test.ts
index c7c8853929..e014fa2e75 100644
--- a/packages/backend-common/src/reading/AzureUrlReader.test.ts
+++ b/packages/backend-common/src/reading/AzureUrlReader.test.ts
@@ -81,14 +81,14 @@ describe('AzureUrlReader', () => {
         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/_apis/git/repositories/repo-name/items?api-version=6.0&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/org-name/project-name/_apis/git/repositories/repo-name/items?api-version=6.0&path=my-template.yaml',
         }),
       },
       {
@@ -125,14 +125,12 @@ describe('AzureUrlReader', () => {
       {
         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',
+        error: 'Azure URL must point to a git repository',
       },
       {
         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',
+        error: 'Invalid URL: com/a/b/blob/master/path/to/c.yaml',
       },
       {
         url: '',
diff --git a/packages/backend-common/src/reading/AzureUrlReader.ts b/packages/backend-common/src/reading/AzureUrlReader.ts
index 910e8e04a5..6015b044bc 100644
--- a/packages/backend-common/src/reading/AzureUrlReader.ts
+++ b/packages/backend-common/src/reading/AzureUrlReader.ts
@@ -23,11 +23,9 @@ import {
   ScmIntegrations,
 } from '@backstage/integration';
 import fetch from 'cross-fetch';
-import parseGitUrl from 'git-url-parse';
 import { Minimatch } from 'minimatch';
 import { Readable } from 'stream';
 import { NotFoundError, NotModifiedError } from '@backstage/errors';
-import { stripFirstDirectoryFromPath } from './tree/util';
 import {
   ReadTreeResponseFactory,
   ReaderFactory,
@@ -129,28 +127,38 @@ export class AzureUrlReader implements UrlReader {
       throw new Error(message);
     }
 
+    // When downloading a zip archive from azure on a subpath we get an extra directory
+    // layer added at the top. With for example the file /a/b/c.txt and a download of
+    // /a/b, we'll see /b/c.txt in the zip archive. This picks out /b so that we can remove it.
+    let subpath;
+    const path = new URL(url).searchParams.get('path');
+    if (path) {
+      subpath = path.split('/').filter(Boolean).slice(-1)[0];
+    }
+
     return await this.deps.treeResponseFactory.fromZipArchive({
       stream: archiveAzureResponse.body as unknown as Readable,
       etag: commitSha,
       filter: options?.filter,
+      subpath,
     });
   }
 
   async search(url: string, options?: SearchOptions): Promise<SearchResponse> {
-    const { filepath } = parseGitUrl(url);
-    const matcher = new Minimatch(filepath);
+    const treeUrl = new URL(url);
+
+    const path = treeUrl.searchParams.get('path');
+    const matcher = path && new Minimatch(path.replace(/^\/+/, ''));
 
     // TODO(freben): For now, read the entire repo and filter through that. In
     // a future improvement, we could be smart and try to deduce that non-glob
     // prefixes (like for filepaths such as some-prefix/**/a.yaml) can be used
     // to get just that part of the repo.
-    const treeUrl = new URL(url);
     treeUrl.searchParams.delete('path');
-    treeUrl.pathname = treeUrl.pathname.replace(/\/+$/, '');
 
     const tree = await this.readTree(treeUrl.toString(), {
       etag: options?.etag,
-      filter: path => matcher.match(stripFirstDirectoryFromPath(path)),
+      filter: p => (matcher ? matcher.match(p) : true),
     });
     const files = await tree.files();