@opentermsarchive/engine 7.2.2 → 7.2.4

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@opentermsarchive/engine",
-  "version": "7.2.2",
+  "version": "7.2.4",
   "description": "Tracks and makes visible changes to the terms of online services",
   "homepage": "https://opentermsarchive.org",
   "bugs": {

package/scripts/declarations/utils/index.js

@@ -9,7 +9,7 @@ export default class DeclarationUtils {
   }
 
   static getServiceIdFromFilePath(filePath) {
-    return path.parse(filePath.replace(/\.history|\.filters/, '')).name;
+    return path.parse(filePath.replace(/\.history|\.filters/g, '')).name;
   }
 
   async getJSONFromFile(ref, filePath) {
@@ -47,7 +47,7 @@ export default class DeclarationUtils {
         return; // Assuming history modifications imply corresponding changes in the service declaration and that the analysis of which terms types of this service have changed will be done when analysing the related declaration, no further action is required here
       }
 
-      if (modifiedFilePath.endsWith('.filters.js')) {
+      if (modifiedFilePath.endsWith('.filters.js') || modifiedFilePath.endsWith('.filters.history.js')) {
         const declaration = await this.getJSONFromFile(this.defaultBranch, `declarations/${serviceId}.json`);
 
         servicesTermsTypes[serviceId] = Object.keys(declaration.terms); // Considering how rarely filters are used, simply return all term types that could potentially be impacted to spare implementing a function change check

package/scripts/declarations/utils/index.test.js

@@ -39,6 +39,24 @@ const removeLatestCommit = async () => {
 };
 
 describe('DeclarationUtils', () => {
+  describe('#getServiceIdFromFilePath', () => {
+    it('extracts service ID from regular file path', () => {
+      expect(DeclarationUtils.getServiceIdFromFilePath('declarations/ServiceA.json')).to.equal('ServiceA');
+    });
+
+    it('extracts service ID from history file path', () => {
+      expect(DeclarationUtils.getServiceIdFromFilePath('declarations/ServiceA.history.json')).to.equal('ServiceA');
+    });
+
+    it('extracts service ID from filters file path', () => {
+      expect(DeclarationUtils.getServiceIdFromFilePath('declarations/ServiceA.filters.js')).to.equal('ServiceA');
+    });
+
+    it('extracts service ID from filters history file path', () => {
+      expect(DeclarationUtils.getServiceIdFromFilePath('declarations/ServiceA.filters.history.js')).to.equal('ServiceA');
+    });
+  });
+
   describe('#getModifiedServicesAndTermsTypes', () => {
     before(async () => {
       await loadFixtures();
@@ -137,6 +155,50 @@ describe('DeclarationUtils', () => {
         });
       });
     });
+
+    context('when filters file is modified', () => {
+      before(async () => {
+        await fs.writeFile(path.resolve(SUBJECT_PATH, './declarations/ServiceA.filters.js'), 'module.exports = {};');
+        await declarationUtils.git.add('./declarations/ServiceA.filters.js');
+        await declarationUtils.git.commit('Add filters file', './declarations/ServiceA.filters.js');
+      });
+      after(removeLatestCommit);
+
+      it('returns all terms types from the service declaration', async () => {
+        const result = await declarationUtils.getModifiedServicesAndTermsTypes();
+
+        expect(result.services).to.include('ServiceA');
+        expect(result.servicesTermsTypes.ServiceA).to.have.members([ 'Privacy Policy', 'Terms of Service' ]);
+      });
+    });
+
+    context('when filters history file is modified', () => {
+      before(async () => {
+        await fs.writeFile(path.resolve(SUBJECT_PATH, './declarations/ServiceA.filters.history.js'), 'module.exports = {};');
+        await declarationUtils.git.add('./declarations/ServiceA.filters.history.js');
+        await declarationUtils.git.commit('Add filters history file', './declarations/ServiceA.filters.history.js');
+      });
+      after(removeLatestCommit);
+
+      it('returns all terms types from the service declaration', async () => {
+        const result = await declarationUtils.getModifiedServicesAndTermsTypes();
+
+        expect(result.services).to.include('ServiceA');
+        expect(result.servicesTermsTypes.ServiceA).to.have.members([ 'Privacy Policy', 'Terms of Service' ]);
+      });
+    });
+
+    context('when history file is modified without declaration changes', () => {
+      before(() => commitChanges(COMMIT_PATHS.serviceAHistory, FIXTURES.serviceATermsUpdatedHistory.content));
+      after(removeLatestCommit);
+
+      it('returns no services and no terms types', async () => {
+        expect(await declarationUtils.getModifiedServicesAndTermsTypes()).to.deep.equal({
+          services: [],
+          servicesTermsTypes: {},
+        });
+      });
+    });
   });
 });
 

package/src/archivist/extract/exposedFilters.js

@@ -1,25 +1,28 @@
 export function removeQueryParams(webPageDOM, paramsToRemove = []) {
-  if (typeof paramsToRemove === 'string') {
-    paramsToRemove = [paramsToRemove];
-  }
+  const normalizedParams = Array.isArray(paramsToRemove) ? paramsToRemove : [paramsToRemove];
 
-  if (!paramsToRemove.length) {
+  if (!normalizedParams.length) {
     return;
   }
 
-  const elements = [
-    ...webPageDOM.querySelectorAll('a[href]'),
-    ...webPageDOM.querySelectorAll('img[src]'),
-  ];
+  const elements = webPageDOM.querySelectorAll('a[href], img[src]');
 
-  elements.forEach(element => {
+  for (const element of elements) {
     try {
-      const url = new URL(element.href || element.src);
+      const urlString = element.href || element.src;
+      const url = new URL(urlString);
+
+      const hasTargetParams = normalizedParams.some(param => url.searchParams.has(param));
+
+      if (hasTargetParams) {
+        normalizedParams.forEach(param => url.searchParams.delete(param));
 
-      paramsToRemove.forEach(param => url.searchParams.delete(param));
-      element[element.tagName === 'A' ? 'href' : 'src'] = url.toString();
-    } catch (error) {
-      // ignore if the element has not a valid URL
+        const attributeName = element.tagName === 'A' ? 'href' : 'src';
+
+        element[attributeName] = url.toString();
+      }
+    } catch {
+      // Silently ignore invalid URLs
     }
-  });
+  }
 }

package/src/archivist/extract/exposedFilters.test.js

@@ -7,7 +7,7 @@ describe('exposedFilters', () => {
   let webPageDOM;
 
   before(() => {
-    webPageDOM = createWebPageDOM('<!DOCTYPE html><html><body></body></html>');
+    webPageDOM = createWebPageDOM('<!DOCTYPE html><html><body></body></html>', 'https://example.com');
   });
 
   describe('#removeQueryParams', () => {
@@ -16,7 +16,7 @@ describe('exposedFilters', () => {
 
       before(() => {
         link = webPageDOM.createElement('a');
-        link.href = 'https://example.com/page?utm_source=test&keep=value';
+        link.setAttribute('href', 'https://example.com/page?utm_source=test&keep=value');
         webPageDOM.body.appendChild(link);
       });
 
@@ -27,7 +27,7 @@ describe('exposedFilters', () => {
       it('removes the specified query parameters', () => {
         removeQueryParams(webPageDOM, ['utm_source']);
 
-        expect(link.href).to.equal('https://example.com/page?keep=value');
+        expect(link.getAttribute('href')).to.equal('https://example.com/page?keep=value');
       });
     });
 
@@ -36,7 +36,7 @@ describe('exposedFilters', () => {
 
       before(() => {
         img = webPageDOM.createElement('img');
-        img.src = 'https://example.com/image.jpg?width=100&keep=value';
+        img.setAttribute('src', 'https://example.com/image.jpg?width=100&keep=value');
         webPageDOM.body.appendChild(img);
       });
 
@@ -47,7 +47,7 @@ describe('exposedFilters', () => {
       it('removes the specified query parameters', () => {
         removeQueryParams(webPageDOM, ['width']);
 
-        expect(img.src).to.equal('https://example.com/image.jpg?keep=value');
+        expect(img.getAttribute('src')).to.equal('https://example.com/image.jpg?keep=value');
       });
     });
 
@@ -56,7 +56,7 @@ describe('exposedFilters', () => {
 
       before(() => {
         link = webPageDOM.createElement('a');
-        link.href = 'https://example.com/page?utm_source=test&keep=value';
+        link.setAttribute('href', 'https://example.com/page?utm_source=test&keep=value');
         webPageDOM.body.appendChild(link);
       });
 
@@ -67,7 +67,7 @@ describe('exposedFilters', () => {
       it('removes a single query parameter passed as string', () => {
         removeQueryParams(webPageDOM, 'utm_source');
 
-        expect(link.href).to.equal('https://example.com/page?keep=value');
+        expect(link.getAttribute('href')).to.equal('https://example.com/page?keep=value');
       });
     });
 
@@ -76,7 +76,7 @@ describe('exposedFilters', () => {
 
       before(() => {
         link = webPageDOM.createElement('a');
-        link.href = 'https://example.com/page?utm_source=test&keep=value';
+        link.setAttribute('href', 'https://example.com/page?utm_source=test&keep=value');
         webPageDOM.body.appendChild(link);
       });
 
@@ -87,17 +87,23 @@ describe('exposedFilters', () => {
       it('leaves the URL unchanged', () => {
         removeQueryParams(webPageDOM, []);
 
-        expect(link.href).to.equal('https://example.com/page?utm_source=test&keep=value');
+        expect(link.getAttribute('href')).to.equal('https://example.com/page?utm_source=test&keep=value');
       });
     });
 
     describe('with invalid URLs', () => {
       let link;
+      let webPageDOMWithBaseURL;
+      const invalidURL = 'ht^THIS_IS_WRONG^tp://example.com?utm_source=test';
 
       before(() => {
-        link = webPageDOM.createElement('a');
-        link.href = 'ht^THIS_IS_WRONG^tp://example.com?utm_source=test';
-        webPageDOM.body.appendChild(link);
+        webPageDOMWithBaseURL = createWebPageDOM('<!DOCTYPE html><html><body></body></html>');
+      });
+
+      before(() => {
+        link = webPageDOMWithBaseURL.createElement('a');
+        link.setAttribute('href', invalidURL);
+        webPageDOMWithBaseURL.body.appendChild(link);
       });
 
       after(() => {
@@ -105,9 +111,9 @@ describe('exposedFilters', () => {
       });
 
       it('ignores elements with invalid URLs', () => {
-        removeQueryParams(webPageDOM, ['utm_source']);
+        removeQueryParams(webPageDOMWithBaseURL, ['utm_source']);
 
-        expect(link.href).to.equal('ht^THIS_IS_WRONG^tp://example.com?utm_source=test');
+        expect(link.getAttribute('href')).to.equal(invalidURL);
       });
     });
 
@@ -116,7 +122,7 @@ describe('exposedFilters', () => {
 
       before(() => {
         link = webPageDOM.createElement('a');
-        link.href = 'https://example.com/page?utm_source=test&utm_medium=email&keep=value&remove=me';
+        link.setAttribute('href', 'https://example.com/page?utm_source=test&utm_medium=email&keep=value&remove=me');
         webPageDOM.body.appendChild(link);
       });
 
@@ -127,7 +133,7 @@ describe('exposedFilters', () => {
       it('removes all specified query parameters', () => {
         removeQueryParams(webPageDOM, [ 'utm_source', 'utm_medium', 'remove' ]);
 
-        expect(link.href).to.equal('https://example.com/page?keep=value');
+        expect(link.getAttribute('href')).to.equal('https://example.com/page?keep=value');
       });
     });
 
@@ -136,7 +142,7 @@ describe('exposedFilters', () => {
 
       before(() => {
         link = webPageDOM.createElement('a');
-        link.href = 'https://example.com/test?utm_source=to_remove_1&keep=true&utm_source=to_remove_2';
+        link.setAttribute('href', 'https://example.com/test?utm_source=to_remove_1&keep=true&utm_source=to_remove_2');
         webPageDOM.body.appendChild(link);
       });
 
@@ -147,7 +153,61 @@ describe('exposedFilters', () => {
       it('removes all instances of duplicate query parameters', () => {
         removeQueryParams(webPageDOM, ['utm_source']);
 
-        expect(link.href).to.equal('https://example.com/test?keep=true');
+        expect(link.getAttribute('href')).to.equal('https://example.com/test?keep=true');
+      });
+    });
+
+    describe('with URLs without target parameters', () => {
+      let absoluteLink;
+      let relativeLink;
+      let anchorLink;
+      let img;
+      const originalAbsoluteHref = 'https://example.com/page?keep=value&preserve=me';
+      const originalRelativeHref = './relative/path?existing=param';
+      const originalAnchorHref = '#section1';
+      const originalImgSrc = 'https://example.com/image.jpg?width=100&height=200';
+
+      before(() => {
+        absoluteLink = webPageDOM.createElement('a');
+        absoluteLink.setAttribute('href', originalAbsoluteHref);
+        webPageDOM.body.appendChild(absoluteLink);
+
+        relativeLink = webPageDOM.createElement('a');
+        relativeLink.setAttribute('href', originalRelativeHref);
+        webPageDOM.body.appendChild(relativeLink);
+
+        anchorLink = webPageDOM.createElement('a');
+        anchorLink.setAttribute('href', originalAnchorHref);
+        webPageDOM.body.appendChild(anchorLink);
+
+        img = webPageDOM.createElement('img');
+        img.setAttribute('src', originalImgSrc);
+        webPageDOM.body.appendChild(img);
+
+        removeQueryParams(webPageDOM, [ 'utm_source', 'utm_medium', 'session_id' ]);
+      });
+
+      after(() => {
+        absoluteLink.remove();
+        relativeLink.remove();
+        anchorLink.remove();
+        img.remove();
+      });
+
+      it('leaves absolute link URLs untouched', () => {
+        expect(absoluteLink.getAttribute('href')).to.equal(originalAbsoluteHref);
+      });
+
+      it('leaves relative link URLs untouched', () => {
+        expect(relativeLink.getAttribute('href')).to.equal(originalRelativeHref);
+      });
+
+      it('leaves anchor link URLs untouched', () => {
+        expect(anchorLink.getAttribute('href')).to.equal(originalAnchorHref);
+      });
+
+      it('leaves image source URLs untouched', () => {
+        expect(img.getAttribute('src')).to.equal(originalImgSrc);
       });
     });