@opentermsarchive/engine 7.2.3 → 7.2.4

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
Files changed (3) hide show
  1. package/package.json +1 -1
  2. package/src/archivist/extract/exposedFilters.js +18 -15
  3. package/src/archivist/extract/exposedFilters.test.js +78 -18
package/package.json CHANGED
@@ -1,6 +1,6 @@
1
1
  {
2
2
  "name": "@opentermsarchive/engine",
3
- "version": "7.2.3",
3
+ "version": "7.2.4",
4
4
  "description": "Tracks and makes visible changes to the terms of online services",
5
5
  "homepage": "https://opentermsarchive.org",
6
6
  "bugs": {
package/src/archivist/extract/exposedFilters.js CHANGED
@@ -1,25 +1,28 @@
1
1
  export function removeQueryParams(webPageDOM, paramsToRemove = []) {
2
- if (typeof paramsToRemove === 'string') {
3
- paramsToRemove = [paramsToRemove];
4
- }
2
+ const normalizedParams = Array.isArray(paramsToRemove) ? paramsToRemove : [paramsToRemove];
5
3
 
6
- if (!paramsToRemove.length) {
4
+ if (!normalizedParams.length) {
7
5
  return;
8
6
  }
9
7
 
10
- const elements = [
11
- ...webPageDOM.querySelectorAll('a[href]'),
12
- ...webPageDOM.querySelectorAll('img[src]'),
13
- ];
8
+ const elements = webPageDOM.querySelectorAll('a[href], img[src]');
14
9
 
15
- elements.forEach(element => {
10
+ for (const element of elements) {
16
11
  try {
17
- const url = new URL(element.href || element.src);
12
+ const urlString = element.href || element.src;
13
+ const url = new URL(urlString);
14
+
15
+ const hasTargetParams = normalizedParams.some(param => url.searchParams.has(param));
16
+
17
+ if (hasTargetParams) {
18
+ normalizedParams.forEach(param => url.searchParams.delete(param));
18
19
 
19
- paramsToRemove.forEach(param => url.searchParams.delete(param));
20
- element[element.tagName === 'A' ? 'href' : 'src'] = url.toString();
21
- } catch (error) {
22
- // ignore if the element has not a valid URL
20
+ const attributeName = element.tagName === 'A' ? 'href' : 'src';
21
+
22
+ element[attributeName] = url.toString();
23
+ }
24
+ } catch {
25
+ // Silently ignore invalid URLs
23
26
  }
24
- });
27
+ }
25
28
  }
package/src/archivist/extract/exposedFilters.test.js CHANGED
@@ -7,7 +7,7 @@ describe('exposedFilters', () => {
7
7
  let webPageDOM;
8
8
 
9
9
  before(() => {
10
- webPageDOM = createWebPageDOM('<!DOCTYPE html><html><body></body></html>');
10
+ webPageDOM = createWebPageDOM('<!DOCTYPE html><html><body></body></html>', 'https://example.com');
11
11
  });
12
12
 
13
13
  describe('#removeQueryParams', () => {
@@ -16,7 +16,7 @@ describe('exposedFilters', () => {
16
16
 
17
17
  before(() => {
18
18
  link = webPageDOM.createElement('a');
19
- link.href = 'https://example.com/page?utm_source=test&keep=value';
19
+ link.setAttribute('href', 'https://example.com/page?utm_source=test&keep=value');
20
20
  webPageDOM.body.appendChild(link);
21
21
  });
22
22
 
@@ -27,7 +27,7 @@ describe('exposedFilters', () => {
27
27
  it('removes the specified query parameters', () => {
28
28
  removeQueryParams(webPageDOM, ['utm_source']);
29
29
 
30
- expect(link.href).to.equal('https://example.com/page?keep=value');
30
+ expect(link.getAttribute('href')).to.equal('https://example.com/page?keep=value');
31
31
  });
32
32
  });
33
33
 
@@ -36,7 +36,7 @@ describe('exposedFilters', () => {
36
36
 
37
37
  before(() => {
38
38
  img = webPageDOM.createElement('img');
39
- img.src = 'https://example.com/image.jpg?width=100&keep=value';
39
+ img.setAttribute('src', 'https://example.com/image.jpg?width=100&keep=value');
40
40
  webPageDOM.body.appendChild(img);
41
41
  });
42
42
 
@@ -47,7 +47,7 @@ describe('exposedFilters', () => {
47
47
  it('removes the specified query parameters', () => {
48
48
  removeQueryParams(webPageDOM, ['width']);
49
49
 
50
- expect(img.src).to.equal('https://example.com/image.jpg?keep=value');
50
+ expect(img.getAttribute('src')).to.equal('https://example.com/image.jpg?keep=value');
51
51
  });
52
52
  });
53
53
 
@@ -56,7 +56,7 @@ describe('exposedFilters', () => {
56
56
 
57
57
  before(() => {
58
58
  link = webPageDOM.createElement('a');
59
- link.href = 'https://example.com/page?utm_source=test&keep=value';
59
+ link.setAttribute('href', 'https://example.com/page?utm_source=test&keep=value');
60
60
  webPageDOM.body.appendChild(link);
61
61
  });
62
62
 
@@ -67,7 +67,7 @@ describe('exposedFilters', () => {
67
67
  it('removes a single query parameter passed as string', () => {
68
68
  removeQueryParams(webPageDOM, 'utm_source');
69
69
 
70
- expect(link.href).to.equal('https://example.com/page?keep=value');
70
+ expect(link.getAttribute('href')).to.equal('https://example.com/page?keep=value');
71
71
  });
72
72
  });
73
73
 
@@ -76,7 +76,7 @@ describe('exposedFilters', () => {
76
76
 
77
77
  before(() => {
78
78
  link = webPageDOM.createElement('a');
79
- link.href = 'https://example.com/page?utm_source=test&keep=value';
79
+ link.setAttribute('href', 'https://example.com/page?utm_source=test&keep=value');
80
80
  webPageDOM.body.appendChild(link);
81
81
  });
82
82
 
@@ -87,17 +87,23 @@ describe('exposedFilters', () => {
87
87
  it('leaves the URL unchanged', () => {
88
88
  removeQueryParams(webPageDOM, []);
89
89
 
90
- expect(link.href).to.equal('https://example.com/page?utm_source=test&keep=value');
90
+ expect(link.getAttribute('href')).to.equal('https://example.com/page?utm_source=test&keep=value');
91
91
  });
92
92
  });
93
93
 
94
94
  describe('with invalid URLs', () => {
95
95
  let link;
96
+ let webPageDOMWithBaseURL;
97
+ const invalidURL = 'ht^THIS_IS_WRONG^tp://example.com?utm_source=test';
96
98
 
97
99
  before(() => {
98
- link = webPageDOM.createElement('a');
99
- link.href = 'ht^THIS_IS_WRONG^tp://example.com?utm_source=test';
100
- webPageDOM.body.appendChild(link);
100
+ webPageDOMWithBaseURL = createWebPageDOM('<!DOCTYPE html><html><body></body></html>');
101
+ });
102
+
103
+ before(() => {
104
+ link = webPageDOMWithBaseURL.createElement('a');
105
+ link.setAttribute('href', invalidURL);
106
+ webPageDOMWithBaseURL.body.appendChild(link);
101
107
  });
102
108
 
103
109
  after(() => {
@@ -105,9 +111,9 @@ describe('exposedFilters', () => {
105
111
  });
106
112
 
107
113
  it('ignores elements with invalid URLs', () => {
108
- removeQueryParams(webPageDOM, ['utm_source']);
114
+ removeQueryParams(webPageDOMWithBaseURL, ['utm_source']);
109
115
 
110
- expect(link.href).to.equal('ht^THIS_IS_WRONG^tp://example.com?utm_source=test');
116
+ expect(link.getAttribute('href')).to.equal(invalidURL);
111
117
  });
112
118
  });
113
119
 
@@ -116,7 +122,7 @@ describe('exposedFilters', () => {
116
122
 
117
123
  before(() => {
118
124
  link = webPageDOM.createElement('a');
119
- link.href = 'https://example.com/page?utm_source=test&utm_medium=email&keep=value&remove=me';
125
+ link.setAttribute('href', 'https://example.com/page?utm_source=test&utm_medium=email&keep=value&remove=me');
120
126
  webPageDOM.body.appendChild(link);
121
127
  });
122
128
 
@@ -127,7 +133,7 @@ describe('exposedFilters', () => {
127
133
  it('removes all specified query parameters', () => {
128
134
  removeQueryParams(webPageDOM, [ 'utm_source', 'utm_medium', 'remove' ]);
129
135
 
130
- expect(link.href).to.equal('https://example.com/page?keep=value');
136
+ expect(link.getAttribute('href')).to.equal('https://example.com/page?keep=value');
131
137
  });
132
138
  });
133
139
 
@@ -136,7 +142,7 @@ describe('exposedFilters', () => {
136
142
 
137
143
  before(() => {
138
144
  link = webPageDOM.createElement('a');
139
- link.href = 'https://example.com/test?utm_source=to_remove_1&keep=true&utm_source=to_remove_2';
145
+ link.setAttribute('href', 'https://example.com/test?utm_source=to_remove_1&keep=true&utm_source=to_remove_2');
140
146
  webPageDOM.body.appendChild(link);
141
147
  });
142
148
 
@@ -147,7 +153,61 @@ describe('exposedFilters', () => {
147
153
  it('removes all instances of duplicate query parameters', () => {
148
154
  removeQueryParams(webPageDOM, ['utm_source']);
149
155
 
150
- expect(link.href).to.equal('https://example.com/test?keep=true');
156
+ expect(link.getAttribute('href')).to.equal('https://example.com/test?keep=true');
157
+ });
158
+ });
159
+
160
+ describe('with URLs without target parameters', () => {
161
+ let absoluteLink;
162
+ let relativeLink;
163
+ let anchorLink;
164
+ let img;
165
+ const originalAbsoluteHref = 'https://example.com/page?keep=value&preserve=me';
166
+ const originalRelativeHref = './relative/path?existing=param';
167
+ const originalAnchorHref = '#section1';
168
+ const originalImgSrc = 'https://example.com/image.jpg?width=100&height=200';
169
+
170
+ before(() => {
171
+ absoluteLink = webPageDOM.createElement('a');
172
+ absoluteLink.setAttribute('href', originalAbsoluteHref);
173
+ webPageDOM.body.appendChild(absoluteLink);
174
+
175
+ relativeLink = webPageDOM.createElement('a');
176
+ relativeLink.setAttribute('href', originalRelativeHref);
177
+ webPageDOM.body.appendChild(relativeLink);
178
+
179
+ anchorLink = webPageDOM.createElement('a');
180
+ anchorLink.setAttribute('href', originalAnchorHref);
181
+ webPageDOM.body.appendChild(anchorLink);
182
+
183
+ img = webPageDOM.createElement('img');
184
+ img.setAttribute('src', originalImgSrc);
185
+ webPageDOM.body.appendChild(img);
186
+
187
+ removeQueryParams(webPageDOM, [ 'utm_source', 'utm_medium', 'session_id' ]);
188
+ });
189
+
190
+ after(() => {
191
+ absoluteLink.remove();
192
+ relativeLink.remove();
193
+ anchorLink.remove();
194
+ img.remove();
195
+ });
196
+
197
+ it('leaves absolute link URLs untouched', () => {
198
+ expect(absoluteLink.getAttribute('href')).to.equal(originalAbsoluteHref);
199
+ });
200
+
201
+ it('leaves relative link URLs untouched', () => {
202
+ expect(relativeLink.getAttribute('href')).to.equal(originalRelativeHref);
203
+ });
204
+
205
+ it('leaves anchor link URLs untouched', () => {
206
+ expect(anchorLink.getAttribute('href')).to.equal(originalAnchorHref);
207
+ });
208
+
209
+ it('leaves image source URLs untouched', () => {
210
+ expect(img.getAttribute('src')).to.equal(originalImgSrc);
151
211
  });
152
212
  });
153
213