@opentermsarchive/engine 0.34.0 → 0.34.2

package/README.md

@@ -6,6 +6,10 @@ For documentation, visit [docs.opentermsarchive.org](https://docs.opentermsarchi
 
 - - -
 
+## Contribute
+
+To contribute to the Open Terms Archive Engine, please refer to the [contributing guidelines](CONTRIBUTING.md) before submitting pull requests. Bugs can be reported or features requested by opening an issue.
+
 ## License
 
 The code for this software is distributed under the [European Union Public Licence (EUPL) v1.2](https://joinup.ec.europa.eu/collection/eupl/eupl-text-eupl-12). In short, this [means](https://choosealicense.com/licenses/eupl-1.2/) you are allowed to read, use, modify and redistribute this source code, as long as you as you credit “Open Terms Archive Contributors” and make available any change you make to it under similar conditions.

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@opentermsarchive/engine",
-  "version": "0.34.0",
+  "version": "0.34.2",
   "description": "Tracks and makes visible changes to the terms of online services",
   "homepage": "https://github.com/OpenTermsArchive/engine#readme",
   "bugs": {

package/src/archivist/extract/index.js

@@ -24,6 +24,7 @@ const ciceroMarkTransformer = new CiceroMarkTransformer();
 /**
  * Extract content from source document and convert it to Markdown
  *
+ * @function extract
  * @param {string} sourceDocument - Source document from which to extract content, see {@link ./src/archivist/services/sourceDocument.js}
  * @returns {Promise<string>} Promise which is fulfilled once the content is extracted and converted in Markdown. The promise will resolve into a string containing the extracted content in Markdown format
 */

package/src/archivist/fetcher/index.js

@@ -9,6 +9,7 @@ export { FetchDocumentError } from './errors.js';
 /**
  * Fetch a resource from the network, returning a promise which is fulfilled once the response is available
  *
+ * @function fetch
  * @param {Object} params - Fetcher parameters
  * @param {string} params.url - URL of the resource you want to fetch
  * @param {boolean} [params.executeClientScripts] - Enable execution of client scripts. When set to `true`, this property loads the page in a headless browser to load all assets and execute client scripts before returning its content

package/src/archivist/index.test.js

@@ -35,7 +35,7 @@ async function resetGitRepositories() {
 describe('Archivist', function () {
   this.timeout(10000);
 
-  const SERVICE_A_ID = 'service_A';
+  const SERVICE_A_ID = 'service·A';
   const SERVICE_A_TYPE = 'Terms of Service';
   const SERVICE_A_EXPECTED_SNAPSHOT_FILE_PATH = `${SNAPSHOTS_PATH}/${SERVICE_A_ID}/${SERVICE_A_TYPE}.html`;
   const SERVICE_A_EXPECTED_VERSION_FILE_PATH = `${VERSIONS_PATH}/${SERVICE_A_ID}/${SERVICE_A_TYPE}.md`;
@@ -49,7 +49,7 @@ describe('Archivist', function () {
   let serviceBSnapshotExpectedContent;
   let serviceBVersionExpectedContent;
 
-  const services = [ 'service_A', 'Service B!' ];
+  const services = [ 'service·A', 'Service B!' ];
 
   before(async () => {
     gitVersion = new Git({
@@ -61,8 +61,8 @@ describe('Archivist', function () {
     });
     await gitVersion.initialize();
 
-    serviceASnapshotExpectedContent = await fs.readFile(path.resolve(ROOT_PATH, 'test/fixtures/service_A_terms_snapshot.html'), { encoding: 'utf8' });
-    serviceAVersionExpectedContent = await fs.readFile(path.resolve(ROOT_PATH, 'test/fixtures/service_A_terms.md'), { encoding: 'utf8' });
+    serviceASnapshotExpectedContent = await fs.readFile(path.resolve(ROOT_PATH, 'test/fixtures/service·A_terms_snapshot.html'), { encoding: 'utf8' });
+    serviceAVersionExpectedContent = await fs.readFile(path.resolve(ROOT_PATH, 'test/fixtures/service·A_terms.md'), { encoding: 'utf8' });
     serviceBSnapshotExpectedContent = await fs.readFile(path.resolve(ROOT_PATH, 'test/fixtures/terms.pdf'));
     serviceBVersionExpectedContent = await fs.readFile(path.resolve(ROOT_PATH, 'test/fixtures/termsFromPDF.md'), { encoding: 'utf8' });
   });
@@ -166,7 +166,7 @@ describe('Archivist', function () {
 
             app.services[SERVICE_A_ID].getTerms({ type: SERVICE_A_TYPE }).sourceDocuments[0].contentSelectors = 'h1';
 
-            await app.track({ services: [ 'service_A', 'Service B!' ], extractOnly: true });
+            await app.track({ services: [ 'service·A', 'Service B!' ], extractOnly: true });
 
             const [reExtractedVersionCommit] = await gitVersion.log({ file: SERVICE_A_EXPECTED_VERSION_FILE_PATH });
 
@@ -282,7 +282,7 @@ describe('Archivist', function () {
       let snapshot;
 
       before(async () => {
-        terms = app.services.service_A.getTerms({ type: SERVICE_A_TYPE });
+        terms = app.services.service·A.getTerms({ type: SERVICE_A_TYPE });
         terms.fetchDate = FETCH_DATE;
         terms.sourceDocuments.forEach(async sourceDocument => {
           sourceDocument.content = serviceASnapshotExpectedContent;
@@ -364,7 +364,7 @@ describe('Archivist', function () {
       let version;
 
       before(async () => {
-        terms = app.services.service_A.getTerms({ type: SERVICE_A_TYPE });
+        terms = app.services.service·A.getTerms({ type: SERVICE_A_TYPE });
         terms.fetchDate = FETCH_DATE;
         terms.sourceDocuments.forEach(async sourceDocument => {
           sourceDocument.content = serviceASnapshotExpectedContent;

package/src/archivist/recorder/repositories/git/git.js

@@ -26,7 +26,8 @@ export default class Git {
       .addConfig('core.autocrlf', false)
       .addConfig('push.default', 'current')
       .addConfig('user.name', this.author.name)
-      .addConfig('user.email', this.author.email);
+      .addConfig('user.email', this.author.email)
+      .addConfig('core.quotePath', false); // disable Git's encoding of special characters in pathnames. For example, `service·A` will be encoded as `service\302\267A` without this setting, leading to issues. See https://git-scm.com/docs/git-config#Documentation/git-config.txt-corequotePath
   }
 
   async add(filePath) {

package/src/archivist/recorder/repositories/mongo/index.js

@@ -24,6 +24,8 @@ export default class MongoRepository extends RepositoryInterface {
 
     this.collection = db.collection(this.collectionName);
 
+    this.collection.createIndex({ serviceId: 1, termsType: 1, fetchDate: -1 });
+
     return this;
   }
 

package/src/archivist/services/index.test.js

@@ -97,8 +97,8 @@ describe('Services', () => {
       result = await services.load();
     });
 
-    describe('Service A', async () => {
-      await validateServiceWithoutHistory('service_A', expectedServices.service_A);
+    describe('Service·A', async () => {
+      await validateServiceWithoutHistory('service·A', expectedServices.service·A);
     });
 
     describe('Service B', async () => {
@@ -127,11 +127,11 @@ describe('Services', () => {
 
     context('when specifying services to load', async () => {
       before(async () => {
-        result = await services.load([ 'service_A', 'Service B!' ]);
+        result = await services.load([ 'service·A', 'Service B!' ]);
       });
 
       it('loads only the given services', async () => {
-        expect(result).to.have.all.keys('service_A', 'Service B!');
+        expect(result).to.have.all.keys('service·A', 'Service B!');
       });
     });
   });
@@ -266,8 +266,8 @@ describe('Services', () => {
       result = await services.loadWithHistory();
     });
 
-    describe('Service A', async () => {
-      await validateServiceWithHistory('service_A', expectedServices.service_A);
+    describe('Service·A', async () => {
+      await validateServiceWithHistory('service·A', expectedServices.service·A);
     });
 
     describe('Service B', async () => {
@@ -296,11 +296,11 @@ describe('Services', () => {
 
     context('when specifying services to load', async () => {
       before(async () => {
-        result = await services.loadWithHistory([ 'service_A', 'Service B!' ]);
+        result = await services.loadWithHistory([ 'service·A', 'Service B!' ]);
       });
 
       it('loads only the given services', async () => {
-        expect(result).to.have.all.keys('service_A', 'Service B!');
+        expect(result).to.have.all.keys('service·A', 'Service B!');
       });
     });
   });