Add GitLab functionalities

.env.example

@@ -1,3 +1,7 @@
 OTA_ENGINE_SENDINBLUE_API_KEY='xkeysib-3f51c…'
 OTA_ENGINE_SMTP_PASSWORD='password'
 OTA_ENGINE_GITHUB_TOKEN=ghp_XXXXXXXXX
+OTA_ENGINE_GITLAB_TOKEN=XXXXXXXXXX
+OTA_ENGINE_GITLAB_RELEASES_TOKEN=XXXXXXXXXX
+OTA_ENGINE_GITLAB_API_BASE_URL=https://gitlab.com/api/v4
+OTA_ENGINE_GITLAB_BASE_URL=https://gitlab.com

CHANGELOG.md

@@ -2,6 +2,12 @@
 
 All changes that impact users of this module are documented in this file, in the [Common Changelog](https://common-changelog.org) format with some additional specifications defined in the CONTRIBUTING file. This codebase adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0.html).
 
+## Unreleased [minor]
+
+### Added
+
+- Add GitLab functionalities
+
 ## 2.5.0 - 2024-10-29
 
 _Full changeset and discussions: [#1115](https://github.com/OpenTermsArchive/engine/pull/1115)._

config/default.json

@@ -59,6 +59,7 @@
     "dataset": {
       "title": "sandbox",
       "versionsRepositoryURL": "https://github.com/OpenTermsArchive/sandbox",
+      "versionsRepositoryURLGitLab": "https://gitlab.com/ota-sandbox-example/sandbox",
       "publishingSchedule": "30 8 * * MON"
     }
   }

scripts/dataset/assets/README.template.js

@@ -3,10 +3,10 @@ import config from 'config';
 const LOCALE = 'en-EN';
 const DATE_OPTIONS = { year: 'numeric', month: 'long', day: 'numeric' };
 
-export default function readme({ releaseDate, servicesCount, firstVersionDate, lastVersionDate }) {
+export default function readme({ releaseDate, servicesCount, firstVersionDate, lastVersionDate, versionsRepositoryURL }) {
   return `# Open Terms Archive — ${title({ releaseDate })}
 
-${body({ servicesCount, firstVersionDate, lastVersionDate })}`;
+${body({ servicesCount, firstVersionDate, lastVersionDate, versionsRepositoryURL })}`;
 }
 
 export function title({ releaseDate }) {
@@ -17,12 +17,10 @@ export function title({ releaseDate }) {
   return `${title} — ${releaseDate} dataset`;
 }
 
-export function body({ servicesCount, firstVersionDate, lastVersionDate }) {
+export function body({ servicesCount, firstVersionDate, lastVersionDate, versionsRepositoryURL }) {
   firstVersionDate = firstVersionDate.toLocaleDateString(LOCALE, DATE_OPTIONS);
   lastVersionDate = lastVersionDate.toLocaleDateString(LOCALE, DATE_OPTIONS);
 
-  const versionsRepositoryURL = config.get('@opentermsarchive/engine.dataset.versionsRepositoryURL');
-
   return `This dataset consolidates the contractual documents of ${servicesCount} service providers, in all their versions that were accessible online between ${firstVersionDate} and ${lastVersionDate}.
 
 This dataset is tailored for datascientists and other analysts. You can also explore all these versions interactively on [${versionsRepositoryURL}](${versionsRepositoryURL}).

scripts/dataset/export/index.js

@@ -16,7 +16,7 @@ const fs = fsApi.promises;
 
 const ARCHIVE_FORMAT = 'zip'; // for supported formats, see https://www.archiverjs.com/docs/archive-formats
 
-export default async function generate({ archivePath, releaseDate }) {
+export default async function generate({ archivePath, releaseDate, versionsRepositoryURL }) {
   const versionsRepository = await RepositoryFactory.create(config.get('@opentermsarchive/engine.recorder.versions.storage')).initialize();
 
   const archive = await initializeArchive(archivePath);
@@ -61,6 +61,7 @@ export default async function generate({ archivePath, releaseDate }) {
       releaseDate,
       firstVersionDate,
       lastVersionDate,
+      versionsRepositoryURL,
     }),
     { name: `${archive.basename}/README.md` },
   );

scripts/dataset/index.js

@@ -6,6 +6,7 @@ import config from 'config';
 import generateRelease from './export/index.js';
 import logger from './logger/index.js';
 import publishRelease from './publish/index.js';
+import publishReleaseGitLab from './publishGitLab/index.js';
 
 export async function release({ shouldPublish, shouldRemoveLocalCopy, fileName }) {
   const releaseDate = new Date();
@@ -14,7 +15,15 @@ export async function release({ shouldPublish, shouldRemoveLocalCopy, fileName }
 
   logger.info('Start exporting dataset…');
 
-  const stats = await generateRelease({ archivePath, releaseDate });
+  const usesGitHub = (typeof process.env.OTA_ENGINE_GITHUB_TOKEN !== 'undefined');
+  const usesGitLab = (typeof process.env.OTA_ENGINE_GITLAB_TOKEN !== 'undefined');
+
+  let versionsRepositoryURL = '';
+
+  if (usesGitHub) versionsRepositoryURL = config.get('@opentermsarchive/engine.dataset.versionsRepositoryURL');
+  if (usesGitLab) versionsRepositoryURL = config.get('@opentermsarchive/engine.dataset.versionsRepositoryURLGitLab');
+
+  const stats = await generateRelease({ archivePath, releaseDate, versionsRepositoryURL });
 
   logger.info(`Dataset exported in ${archivePath}`);
 
@@ -24,13 +33,25 @@ export async function release({ shouldPublish, shouldRemoveLocalCopy, fileName }
 
   logger.info('Start publishing dataset…');
 
-  const releaseUrl = await publishRelease({
-    archivePath,
-    releaseDate,
-    stats,
-  });
+  if (usesGitHub) {
+    const releaseUrl = await publishRelease({
+      archivePath,
+      releaseDate,
+      stats,
+    });
 
-  logger.info(`Dataset published to ${releaseUrl}`);
+    logger.info(`Dataset published to ${releaseUrl}`);
+  }
+
+  if (usesGitLab) {
+    const releaseUrl = await publishReleaseGitLab({
+      archivePath,
+      releaseDate,
+      stats,
+    });
+
+    logger.info(`Dataset published to ${releaseUrl}`);
+  }
 
   if (!shouldRemoveLocalCopy) {
     return;

scripts/dataset/publishGitLab/index.js

@@ -0,0 +1,135 @@
+import fsApi from 'fs';
+import path from 'path';
+import url from 'url';
+
+import config from 'config';
+import dotenv from 'dotenv';
+import FormData from 'form-data';
+import nodeFetch from 'node-fetch';
+
+import GitLab from '../../../src/reporterGitlab/gitlab.js';
+import * as readme from '../assets/README.template.js';
+import logger from '../logger/index.js';
+
+dotenv.config();
+
+const gitlabAPIUrl = process.env.OTA_ENGINE_GITLAB_API_BASE_URL;
+
+export default async function publishReleaseGitLab({
+  archivePath,
+  releaseDate,
+  stats,
+}) {
+  let projectId = null;
+
+  const [ owner, repo ] = url
+    .parse(config.get('@opentermsarchive/engine.dataset.versionsRepositoryURLGitLab'))
+    .pathname.split('/')
+    .filter(component => component);
+  const commonParams = { owner, repo };
+
+  try {
+    const repositoryPath = `${commonParams.owner}/${commonParams.repo}`;
+
+    const options = GitLab.baseOptionsHttpReq(process.env.OTA_ENGINE_GITLAB_RELEASES_TOKEN);
+
+    options.method = 'GET';
+    options.headers = {
+      'Content-Type': 'application/json',
+      ...options.headers,
+    };
+
+    const response = await nodeFetch(
+      `${gitlabAPIUrl}/projects/${encodeURIComponent(repositoryPath)}`,
+      options,
+    );
+    const res = await response.json();
+
+    projectId = res.id;
+  } catch (error) {
+    logger.error(`Error while obtaining projectId: ${error}`);
+    projectId = null;
+  }
+
+  const tagName = `${path.basename(archivePath, path.extname(archivePath))}`; // use archive filename as Git tag
+
+  try {
+    let options = GitLab.baseOptionsHttpReq(process.env.OTA_ENGINE_GITLAB_RELEASES_TOKEN);
+
+    options.method = 'POST';
+    options.body = {
+      ref: 'main',
+      tag_name: tagName,
+      name: readme.title({ releaseDate }),
+      description: readme.body(stats),
+    };
+    options.headers = {
+      'Content-Type': 'application/json',
+      ...options.headers,
+    };
+
+    options.body = JSON.stringify(options.body);
+
+    const releaseResponse = await nodeFetch(
+      `${gitlabAPIUrl}/projects/${projectId}/releases`,
+      options,
+    );
+    const releaseRes = await releaseResponse.json();
+
+    const releaseId = releaseRes.commit.id;
+
+    logger.info(`Created release with releaseId: ${releaseId}`);
+
+    // Upload the package
+    options = GitLab.baseOptionsHttpReq(process.env.OTA_ENGINE_GITLAB_RELEASES_TOKEN);
+    options.method = 'PUT';
+    options.body = fsApi.createReadStream(archivePath);
+
+    // restrict characters to the ones allowed by GitLab APIs
+    const packageName = config.get('@opentermsarchive/engine.dataset.title').replace(/[^a-zA-Z0-9.\-_]/g, '-');
+    const packageVersion = tagName.replace(/[^a-zA-Z0-9.\-_]/g, '-');
+    const packageFileName = archivePath.replace(/[^a-zA-Z0-9.\-_/]/g, '-');
+
+    logger.debug(`packageName: ${packageName}, packageVersion: ${packageVersion} packageFileName: ${packageFileName}`);
+
+    const packageResponse = await nodeFetch(
+      `${gitlabAPIUrl}/projects/${projectId}/packages/generic/${packageName}/${packageVersion}/${packageFileName}?status=default&select=package_file`,
+      options,
+    );
+    const packageRes = await packageResponse.json();
+
+    const packageFilesId = packageRes.id;
+
+    logger.debug(`package file id: ${packageFilesId}`);
+
+    // use the package id to build the download url for the release
+    const publishedPackageUrl = `${config.get('@opentermsarchive/engine.dataset.versionsRepositoryURLGitLab')}/-/package_files/${packageFilesId}/download`;
+
+    // Create the release and link the package
+    const formData = new FormData();
+
+    formData.append('name', archivePath);
+    formData.append('url', publishedPackageUrl);
+    formData.append('file', fsApi.createReadStream(archivePath), { filename: path.basename(archivePath) });
+
+    options = GitLab.baseOptionsHttpReq(process.env.OTA_ENGINE_GITLAB_RELEASES_TOKEN);
+    options.method = 'POST';
+    options.headers = {
+      ...formData.getHeaders(),
+      ...options.headers,
+    };
+    options.body = formData;
+
+    const uploadResponse = await nodeFetch(
+      `${gitlabAPIUrl}/projects/${projectId}/releases/${tagName}/assets/links`,
+      options,
+    );
+    const uploadRes = await uploadResponse.json();
+    const releaseUrl = uploadRes.direct_asset_url;
+
+    return releaseUrl;
+  } catch (error) {
+    logger.error('Failed to create release or upload ZIP file:', error);
+    throw error;
+  }
+}

src/index.js

@@ -8,6 +8,7 @@ import Archivist from './archivist/index.js';
 import logger from './logger/index.js';
 import Notifier from './notifier/index.js';
 import Reporter from './reporter/index.js';
+import ReporterGitlab from './reporterGitlab/index.js';
 
 const require = createRequire(import.meta.url);
 
@@ -72,6 +73,23 @@ export default async function track({ services, types, extractOnly, schedule })
     logger.warn('Environment variable "OTA_ENGINE_GITHUB_TOKEN" was not found; the Reporter module will be ignored');
   }
 
+  if (process.env.OTA_ENGINE_GITLAB_TOKEN) {
+    if (config.has('@opentermsarchive/engine.reporter.gitlabIssues.repositories.declarations')) {
+      try {
+        const reporter = new ReporterGitlab(config.get('@opentermsarchive/engine.reporter'));
+
+        await reporter.initialize();
+        archivist.attach(reporter);
+      } catch (error) {
+        logger.error('Cannot instantiate the ReporterGitlab module; it will be ignored:', error);
+      }
+    } else {
+      logger.warn('Configuration key "reporter.gitlabIssues.repositories.declarations" was not found; issues on the declarations repository cannot be created');
+    }
+  } else {
+    logger.warn('Environment variable "OTA_ENGINE_GITLAB_TOKEN" was not found; the ReporterGitlab module will be ignored');
+  }
+
   if (!schedule) {
     await archivist.track({ services, types });