fix: support repository urls ending with .git

research-software-directory · Jul 5, 2024 · 2fdb94e · 2fdb94e
1 parent 2e2ec41
commit 2fdb94e
Show file tree

Hide file tree

Showing 6 changed files with 27 additions and 9 deletions.
diff --git a/frontend/components/software/edit/links/SoftwareLinksInfo.tsx b/frontend/components/software/edit/links/SoftwareLinksInfo.tsx
@@ -1,5 +1,6 @@
 // SPDX-FileCopyrightText: 2022 - 2023 Dusan Mijatovic (dv4all)
 // SPDX-FileCopyrightText: 2022 - 2023 dv4all
+// SPDX-FileCopyrightText: 2024 Christian Meeßen (GFZ) <christian.meessen@gfz-potsdam.de>
 // SPDX-FileCopyrightText: 2024 Dusan Mijatovic (Netherlands eScience Center)
 // SPDX-FileCopyrightText: 2024 Ewan Cahen (Netherlands eScience Center) <e.cahen@esciencecenter.nl>
 // SPDX-FileCopyrightText: 2024 Felix Mühlbauer (GFZ) <felix.muehlbauer@gfz-potsdam.de>
@@ -29,7 +30,7 @@ export default function SoftwareLinksInfo({reorderedCategories}: SoftwareLinksIn
       On this page you can provide links and metadata to better describe your software. Some of this metadata can be imported automatically.
 
       <p className="py-2"><strong>Software URLs</strong></p>
-      <p>Providing the URL of the source code repository of the software allows the RSD show a link to the source code on the software page, and automatically harvest information about the development activity.</p>
+      <p>Providing the URL of the source code repository of the software allows the RSD show a link to the source code on the software page, and automatically harvest information about the development activity. The RSD supports https URLs pointing to the git repository, or to the website of the repository.</p>
       <p className="py-2">The getting started URL can be used to link to a webpage describing how to install and use the software. This link will be prominently shown on the top of the software page.</p>
 
       <p className="py-2"><strong>Software DOI</strong></p>

diff --git a/scrapers/src/main/java/nl/esciencecenter/rsd/scraper/git/GithubScraper.java b/scrapers/src/main/java/nl/esciencecenter/rsd/scraper/git/GithubScraper.java
@@ -1,7 +1,7 @@
+// SPDX-FileCopyrightText: 2022 - 2024 Christian Meeßen (GFZ) <christian.meessen@gfz-potsdam.de>
 // SPDX-FileCopyrightText: 2022 - 2024 Ewan Cahen (Netherlands eScience Center) <e.cahen@esciencecenter.nl>
+// SPDX-FileCopyrightText: 2022 - 2024 Helmholtz Centre Potsdam - GFZ German Research Centre for Geosciences
 // SPDX-FileCopyrightText: 2022 - 2024 Netherlands eScience Center
-// SPDX-FileCopyrightText: 2022 Christian Meeßen (GFZ) <christian.meessen@gfz-potsdam.de>
-// SPDX-FileCopyrightText: 2022 Helmholtz Centre Potsdam - GFZ German Research Centre for Geosciences
 //
 // SPDX-License-Identifier: Apache-2.0
 
@@ -28,13 +28,16 @@
 public class GithubScraper implements GitScraper {
 
 	private final String BASE_API_URL = "https://api.github.com";
-	private final String organisation;
-	private final String repo;
+	public final String organisation;
+	public final String repo;
 	private static final Pattern LINK_PATTERN = Pattern.compile("<([^>]+page=(\\d+)[^>]*)>; rel=\"([^\"]+)\"");
 	public static final Pattern GITHUB_URL_PATTERN = Pattern.compile("^https?://github\\.com/([^\\s/]+)/([^\\s/]+)/?$");
 
 	private GithubScraper(String organisation, String repo) {
 		this.organisation = Objects.requireNonNull(organisation);
+		if (repo.endsWith(".git")) {
+			repo = repo.substring(0, repo.length() - 4);
+		}
 		this.repo = Objects.requireNonNull(repo);
 	}
 

diff --git a/scrapers/src/main/java/nl/esciencecenter/rsd/scraper/git/GitlabScraper.java b/scrapers/src/main/java/nl/esciencecenter/rsd/scraper/git/GitlabScraper.java
@@ -1,7 +1,7 @@
+// SPDX-FileCopyrightText: 2022 - 2024 Christian Meeßen (GFZ) <christian.meessen@gfz-potsdam.de>
 // SPDX-FileCopyrightText: 2022 - 2024 Ewan Cahen (Netherlands eScience Center) <e.cahen@esciencecenter.nl>
+// SPDX-FileCopyrightText: 2022 - 2024 Helmholtz Centre Potsdam - GFZ German Research Centre for Geosciences
 // SPDX-FileCopyrightText: 2022 - 2024 Netherlands eScience Center
-// SPDX-FileCopyrightText: 2022 Christian Meeßen (GFZ) <christian.meessen@gfz-potsdam.de>
-// SPDX-FileCopyrightText: 2022 Helmholtz Centre Potsdam - GFZ German Research Centre for Geosciences
 //
 // SPDX-License-Identifier: Apache-2.0
 
@@ -35,6 +35,9 @@ public class GitlabScraper implements GitScraper {
 	 */
 	public GitlabScraper(String gitLabApiUrl, String projectPath) {
 		this.projectPath = projectPath;
+		if (this.projectPath.endsWith(".git")) {
+			this.projectPath = this.projectPath.substring(0, this.projectPath.length() - 4);
+		}
 		this.apiUri = gitLabApiUrl + "/v4";
 	}
 

diff --git a/scrapers/src/test/java/nl/esciencecenter/rsd/scraper/git/GithubScaperIT.java b/scrapers/src/test/java/nl/esciencecenter/rsd/scraper/git/GithubScaperIT.java
@@ -1,4 +1,6 @@
+// SPDX-FileCopyrightText: 2024 Christian Meeßen (GFZ) <christian.meessen@gfz-potsdam.de>
 // SPDX-FileCopyrightText: 2024 Ewan Cahen (Netherlands eScience Center) <e.cahen@esciencecenter.nl>
+// SPDX-FileCopyrightText: 2024 Helmholtz Centre Potsdam - GFZ German Research Centre for Geosciences
 // SPDX-FileCopyrightText: 2024 Netherlands eScience Center
 //
 // SPDX-License-Identifier: Apache-2.0

diff --git a/scrapers/src/test/java/nl/esciencecenter/rsd/scraper/git/GithubScraperTest.java b/scrapers/src/test/java/nl/esciencecenter/rsd/scraper/git/GithubScraperTest.java
@@ -1,7 +1,7 @@
+// SPDX-FileCopyrightText: 2022 - 2024 Christian Meeßen (GFZ) <christian.meessen@gfz-potsdam.de>
 // SPDX-FileCopyrightText: 2022 - 2024 Ewan Cahen (Netherlands eScience Center) <e.cahen@esciencecenter.nl>
+// SPDX-FileCopyrightText: 2022 - 2024 Helmholtz Centre Potsdam - GFZ German Research Centre for Geosciences
 // SPDX-FileCopyrightText: 2022 - 2024 Netherlands eScience Center
-// SPDX-FileCopyrightText: 2022 Christian Meeßen (GFZ) <christian.meessen@gfz-potsdam.de>
-// SPDX-FileCopyrightText: 2022 Helmholtz Centre Potsdam - GFZ German Research Centre for Geosciences
 //
 // SPDX-License-Identifier: Apache-2.0
 
@@ -17,6 +17,7 @@ public class GithubScraperTest {
 
 	private final String githubUrlPrefix = "https://github.com/";
 	private final String repo = "research-software-directory/RSD-as-a-service";
+	private final String repoGit = "research-software-directory/RSD-as-a-service.git";
 	private final String repoEmpty = "cmeessen/empty";
 	private final String repoNonEx = "research-software-directory/does-not-exist";
 
@@ -60,4 +61,10 @@ void givenInValidGithubUrl_whenCreatingScraper_thenEmptyScraperReturned() {
 		Optional<GithubScraper> scraper3 = GithubScraper.create(githubUrlPrefix + "org-only/");
 		Assertions.assertTrue(scraper3.isEmpty());
 	}
+
+	@Test
+	void givenGitRepoUrl_whenCreatingScraper_thenRemoveGitSuffix() {
+		Optional<GithubScraper> scraper = GithubScraper.create(githubUrlPrefix + repoGit);
+		Assertions.assertEquals(repo, scraper.get().organisation + "/" + scraper.get().repo);
+	}
 }
diff --git a/scrapers/src/test/java/nl/esciencecenter/rsd/scraper/git/GitlabScraperIT.java b/scrapers/src/test/java/nl/esciencecenter/rsd/scraper/git/GitlabScraperIT.java
@@ -1,5 +1,7 @@
 // SPDX-FileCopyrightText: 2022 - 2024 Ewan Cahen (Netherlands eScience Center) <e.cahen@esciencecenter.nl>
 // SPDX-FileCopyrightText: 2022 - 2024 Netherlands eScience Center
+// SPDX-FileCopyrightText: 2024 Christian Meeßen (GFZ) <christian.meessen@gfz-potsdam.de>
+// SPDX-FileCopyrightText: 2024 Helmholtz Centre Potsdam - GFZ German Research Centre for Geosciences
 //
 // SPDX-License-Identifier: Apache-2.0