test.ts.ts ➔ isUrl - Code Metrics - seantomburke/sitemapper - Measure and Improve Code Quality continuously with Scrutinizer

test.ts.ts ➔ isUrl A
last analyzed 2025-06-01 19:32 UTC

↳ Parent: src/tests/test.ts.ts

Complexity

Conditions

Size

Total Lines	8
Code Lines	7

Duplication

Lines	0
Ratio	0 %

Importance

Changes

Metric	Value
eloc	7
dl	0
loc	8
rs	10
c	0
b	0
f	0
cc	2

import 'async';
import 'assert';
import 'should';

// Simple function to validate URLs using the URL object
function isUrl(url: string): boolean {
  try {
    new URL(url);
    return true;
  } catch {
    return false;
  }
}

import Sitemapper from '../../lib/assets/sitemapper.js';
import { SitemapperResponse } from '../../sitemapper';
let sitemapper: Sitemapper;

describe('Sitemapper', function () {
  beforeEach(() => {
    sitemapper = new Sitemapper();
  });

  describe('Sitemapper Class', function () {
    it('should have initializeTimeout method', () => {
      sitemapper.initializeTimeout.should.be.Function;
    });

    it('should have crawl method', () => {
      sitemapper.crawl.should.be.Function;
    });

    it('should have parse method', () => {
      sitemapper.parse.should.be.Function;
    });

    it('should have fetch method', () => {
      sitemapper.fetch.should.be.Function;
    });

    it('should construct with a url', () => {
      sitemapper = new Sitemapper({
        url: 'google.com',
      });
      sitemapper.url.should.equal('google.com');
    });

    it('should construct with a timeout', () => {
      sitemapper = new Sitemapper({
        timeout: 1000,
      });
      sitemapper.timeout.should.equal(1000);
    });

    it('should set timeout', () => {
      sitemapper.timeout = 1000;
      sitemapper.timeout.should.equal(1000);
    });

    it('should set url', () => {
      sitemapper.url = 1000;
      sitemapper.url.should.equal(1000);
    });

    it('should construct with specific fields', () => {
      sitemapper = new Sitemapper({
        fields: { loc: true, lastmod: true, priority: true, changefreq: true },
      });
      sitemapper.fields.should.be.Object &&
        sitemapper.fields.should.have.keys(
          'loc',
          'lastmod',
          'priority',
          'changefreq'
        );
    });
  });

  describe('fetch Method resolves sites to array', function () {
    it('https://wp.seantburke.com/sitemap.xml sitemaps should be an array', function (done) {
      this.timeout(30000);
      const url = 'https://wp.seantburke.com/sitemap.xml';
      sitemapper
        .fetch(url)
        .then((data) => {
          data.sites.should.be.Array;
          data.url.should.equal(url);
          data.sites.length.should.be.above(2);
          isUrl(data.sites[0]).should.be.true;
          done();
        })
        .catch((error) => {
          console.error('Test failed');
          done(error);
        });
    });

    it('gibberish.gibberish should fail silently with an empty array', function (done) {
      this.timeout(30000);
      const url = 'http://gibberish.gibberish';
      sitemapper.debug = true;
      sitemapper
        .fetch(url)
        .then((data) => {
          data.sites.should.be.Array;
          data.errors.should.be.Array;
          data.errors.length.should.be.greaterThan(0);
          console.log(data);
          done();
        })
        .catch((error) => {
          console.error('Test failed');
          done(error);
        });
    });

    it('https://webflow.com/sitemap.xml sitemaps should be an array', function (done) {
      this.timeout(30000);
      const url = 'https://webflow.com/sitemap.xml';
      sitemapper
        .fetch(url)
        .then((data) => {
          data.sites.should.be.Array;
          data.url.should.equal(url);
          data.sites.length.should.be.above(2);
          isUrl(data.sites[0]).should.be.true;
          done();
        })
        .catch((error) => {
          console.error('Test failed');
          done(error);
        });
    });

    it('https://wp.seantburke.com/sitemap.xml sitemaps should contain extra fields', function (done) {
      this.timeout(30000);
      const url = 'https://wp.seantburke.com/sitemap.xml';
      sitemapper = new Sitemapper({
        fields: {
          loc: true,
          lastmod: true,
          priority: true,
          changefreq: true,
          sitemap: true,
        },
      });
      sitemapper
        .fetch(url)
        .then((data) => {
          data.sites.should.be.Array;
          data.url.should.equal(url);
          data.sites.length.should.be.above(2);
          data.sites[0].loc.should.be.String;
          data.sites[0].lastmod.should.be.String;
          data.sites[0].priority.should.be.String;
          data.sites[0].changefreq.should.be.String;
          data.sites[0].should.have.property('sitemap').which.is.a.String();
          isUrl(data.sites[0].sitemap).should.be.true;
          done();
        })
        .catch((error) => {
          console.error('Test failed');
          done(error);
        });
    });

    it('https://www.golinks.io/sitemap.xml sitemaps should be an array', function (done) {
      this.timeout(30000);
      const url = 'https://www.golinks.io/sitemap.xml';
      sitemapper.timeout = 5000;
      sitemapper
        .fetch(url)
        .then((data) => {
          data.sites.should.be.Array;
          data.url.should.equal(url);
          data.sites.length.should.be.above(2);
          isUrl(data.sites[0]).should.be.true;
          done();
        })
        .catch((error) => {
          console.error('Test failed');
          done(error);
        });
    });

    it('https://www.golinks.io/sitemap.xml sitemaps should return an empty array when timing out', function (done) {
      this.timeout(30000);
      const url = 'https://www.golinks.io/sitemap.xml';
      sitemapper.timeout = 1;
      sitemapper
        .fetch(url)
        .then((data) => {
          data.sites.should.be.Array;
          data.errors.should.be.Array;
          console.log(data);
          done();
        })
        .catch((error) => {
          console.error('Test failed');
          done(error);
        });
    });

    it('https://www.golinks.com/blog/sitemap.xml sitemaps should return an empty array when timing out', function (done) {
      this.timeout(30000);
      const url = 'https://www.golinks.com/blog/sitemap.xml';
      sitemapper.timeout = 10000;
      sitemapper
        .fetch(url)
        .then((data) => {
          data.sites.should.be.Array;
          data.errors.should.be.Array;
          done();
        })
        .catch((error) => {
          console.error('Test failed');
          done(error);
        });
    });

    it('https://www.banggood.com/sitemap/category.xml.gz gzip should be a non-empty array', function (done) {
      this.timeout(30000);
      const url = 'https://www.banggood.com/sitemap/category.xml.gz';
      sitemapper.timeout = 10000;
      sitemapper
        .fetch(url)
        .then((data) => {
          data.sites.should.be.Array;
          data.sites.length.should.be.greaterThan(0);
          done();
        })
        .catch((error) => {
          console.error('Test failed');
          done(error);
        });
    });
  });

  describe('gzipped sitemaps', function () {
    beforeEach(() => {
      sitemapper = new Sitemapper({
        requestHeaders: {
          'Accept-Encoding': 'gzip,deflate,sdch',
        },
      });
    });

    it('https://www.banggood.com/sitemap/category.xml.gz gzip should be a non-empty array', function (done) {
      this.timeout(30000);
      const url = 'https://www.banggood.com/sitemap/category.xml.gz';
      sitemapper.timeout = 10000;
      sitemapper
        .fetch(url)
        .then((data) => {
          data.sites.should.be.Array;
          data.errors.should.be.Array;
          data.sites.length.should.be.greaterThan(0);
          done();
        })
        .catch((error) => {
          console.error('Test failed');
          done(error);
        });
    });

    it('https://example.com/sitemap.xml should not allow insecure request', function (done) {
      this.timeout(30000);
      const url = 'https://example.com/sitemap.xml';
      sitemapper.timeout = 10000;
      sitemapper.rejectUnauthorized = false;
      sitemapper
        .fetch(url)
        .then((data) => {
          data.sites.should.be.Array;
          data.errors.should.be.Array;
          data.errors.should.containEql({
            type: 'HTTPError',
            message: 'HTTP Error occurred: Response code 404 (Not Found)',
            url: 'https://example.com/sitemap.xml',
            retries: 0,
          });
          done();
        })
        .catch((error) => {
          console.error('Test failed');
          done(error);
        });
    });
  });

  describe('sitemaps with namespace prefix', function () {
    it('https://www.emerald.com/sitemap.xml sitemaps should be an array', function (done) {
      this.timeout(30000);
      const url = 'https://www.emerald.com/sitemap.xml';
      sitemapper
        .fetch(url)
        .then((data) => {
          data.sites.should.be.Array;
          data.url.should.equal(url);
          data.sites.length.should.be.above(2);
          isUrl(data.sites[0]).should.be.true;
          done();
        })
        .catch((error) => {
          console.error('Test failed');
          done(error);
        });
    });
  });

  describe('getSites method', function () {
    it('getSites should be backwards compatible', function (done) {
      this.timeout(30000);
      const url = 'https://wp.seantburke.com/sitemap.xml';
      sitemapper.getSites(url, (err, sites) => {
        sites.should.be.Array;
        isUrl(sites[0]).should.be.true;
        done();
      });
    });
  });

  describe('exclusions option', function () {
    it('should prevent false positive', function (done) {
      this.timeout(30000);
      const url = 'https://wp.seantburke.com/sitemap.xml';
      sitemapper.exclusions = [/video/, /image/];
      sitemapper
        .fetch(url)
        .then((data) => {
          data.sites.should.be.Array;
          data.sites.includes('https://wp.seantburke.com/?page_id=2').should.be
            .true;
          done();
        })
        .catch((error) => {
          console.error('Test failed');
          done(error);
        });
    });

    it('should filter out page_id urls', function (done) {
      this.timeout(30000);
      const url = 'https://wp.seantburke.com/sitemap.xml';
      sitemapper.exclusions = [/page_id/];
      sitemapper
        .fetch(url)
        .then((data) => {
          data.sites.should.be.Array;
          data.sites.includes('https://wp.seantburke.com/?page_id=2').should.be
            .false;
          done();
        })
        .catch((error) => {
          console.error('Test failed');
          done(error);
        });
    });
  });

  describe('isExcluded method', function () {
    it('should return false when no exclusions are set', function () {
      const result = sitemapper.isExcluded('https://example.com/page1');
      result.should.be.false();
    });

    it('should return false when url does not match any exclusion patterns', function () {
      sitemapper.exclusions = [/\.pdf$/, /private/];
      const result = sitemapper.isExcluded('https://example.com/page1');
      result.should.be.false();
    });

    it('should return false when url matches an exclusion pattern', function () {
      sitemapper.exclusions = [/\.pdf$/, /private/];
      const result = sitemapper.isExcluded('https://example.com/document.pdf');
      result.should.be.true();
    });

    it('should return true when url matches any of multiple exclusion patterns', function () {
      sitemapper.exclusions = [/\.pdf$/, /private/, /temp/];
      const result = sitemapper.isExcluded(
        'https://example.com/private/temp.html'
      );
      result.should.be.true();
    });

    it('should handle complex regex patterns correctly', function () {
      sitemapper.exclusions = [/^https:\/\/example\.com\/([a-z]{2})\/private/];
      const result1 = sitemapper.isExcluded(
        'https://example.com/en/private/page'
      );
      const result2 = sitemapper.isExcluded(
        'https://example.com/en/public/page'
      );
      result1.should.be.true();
      result2.should.be.false();
    });

    it('should handle case sensitivity correctly', function () {
      sitemapper.exclusions = [/private/i];
      const result1 = sitemapper.isExcluded('https://example.com/PRIVATE/page');
      const result2 = sitemapper.isExcluded('https://example.com/Private/page');
      result1.should.be.true();
      result2.should.be.true();
    });
  });
});


1			import 'async';
2			import 'assert';
3			import 'should';
4
5			// Simple function to validate URLs using the URL object
6			function isUrl(url: string): boolean {
7			try {
8			new URL(url);
9			return true;
10			} catch {
11			return false;
12			}
13			}
14
15			import Sitemapper from '../../lib/assets/sitemapper.js';
16			import { SitemapperResponse } from '../../sitemapper';
17			let sitemapper: Sitemapper;
18
19			describe('Sitemapper', function () {
20			beforeEach(() => {
21			sitemapper = new Sitemapper();
22			});
23
24			describe('Sitemapper Class', function () {
25			it('should have initializeTimeout method', () => {
26			sitemapper.initializeTimeout.should.be.Function;
27			});
28
29			it('should have crawl method', () => {
30			sitemapper.crawl.should.be.Function;
31			});
32
33			it('should have parse method', () => {
34			sitemapper.parse.should.be.Function;
35			});
36
37			it('should have fetch method', () => {
38			sitemapper.fetch.should.be.Function;
39			});
40
41			it('should construct with a url', () => {
42			sitemapper = new Sitemapper({
43			url: 'google.com',
44			});
45			sitemapper.url.should.equal('google.com');
46			});
47
48			it('should construct with a timeout', () => {
49			sitemapper = new Sitemapper({
50			timeout: 1000,
51			});
52			sitemapper.timeout.should.equal(1000);
53			});
54
55			it('should set timeout', () => {
56			sitemapper.timeout = 1000;
57			sitemapper.timeout.should.equal(1000);
58			});
59
60			it('should set url', () => {
61			sitemapper.url = 1000;
62			sitemapper.url.should.equal(1000);
63			});
64
65			it('should construct with specific fields', () => {
66			sitemapper = new Sitemapper({
67			fields: { loc: true, lastmod: true, priority: true, changefreq: true },
68			});
69			sitemapper.fields.should.be.Object &&
70			sitemapper.fields.should.have.keys(
71			'loc',
72			'lastmod',
73			'priority',
74			'changefreq'
75			);
76			});
77			});
78
79			describe('fetch Method resolves sites to array', function () {
80			it('https://wp.seantburke.com/sitemap.xml sitemaps should be an array', function (done) {
81			this.timeout(30000);
82			const url = 'https://wp.seantburke.com/sitemap.xml';
83			sitemapper
84			.fetch(url)
85			.then((data) => {
86			data.sites.should.be.Array;
87			data.url.should.equal(url);
88			data.sites.length.should.be.above(2);
89			isUrl(data.sites[0]).should.be.true;
90			done();
91			})
92			.catch((error) => {
93			console.error('Test failed');
94			done(error);
95			});
96			});
97
98			it('gibberish.gibberish should fail silently with an empty array', function (done) {
99			this.timeout(30000);
100			const url = 'http://gibberish.gibberish';
101			sitemapper.debug = true;
102			sitemapper
103			.fetch(url)
104			.then((data) => {
105			data.sites.should.be.Array;
106			data.errors.should.be.Array;
107			data.errors.length.should.be.greaterThan(0);
108			console.log(data);
109			done();
110			})
111			.catch((error) => {
112			console.error('Test failed');
113			done(error);
114			});
115			});
116
117			it('https://webflow.com/sitemap.xml sitemaps should be an array', function (done) {
118			this.timeout(30000);
119			const url = 'https://webflow.com/sitemap.xml';
120			sitemapper
121			.fetch(url)
122			.then((data) => {
123			data.sites.should.be.Array;
124			data.url.should.equal(url);
125			data.sites.length.should.be.above(2);
126			isUrl(data.sites[0]).should.be.true;
127			done();
128			})
129			.catch((error) => {
130			console.error('Test failed');
131			done(error);
132			});
133			});
134
135			it('https://wp.seantburke.com/sitemap.xml sitemaps should contain extra fields', function (done) {
136			this.timeout(30000);
137			const url = 'https://wp.seantburke.com/sitemap.xml';
138			sitemapper = new Sitemapper({
139			fields: {
140			loc: true,
141			lastmod: true,
142			priority: true,
143			changefreq: true,
144			sitemap: true,
145			},
146			});
147			sitemapper
148			.fetch(url)
149			.then((data) => {
150			data.sites.should.be.Array;
151			data.url.should.equal(url);
152			data.sites.length.should.be.above(2);
153			data.sites[0].loc.should.be.String;
154			data.sites[0].lastmod.should.be.String;
155			data.sites[0].priority.should.be.String;
156			data.sites[0].changefreq.should.be.String;
157			data.sites[0].should.have.property('sitemap').which.is.a.String();
158			isUrl(data.sites[0].sitemap).should.be.true;
159			done();
160			})
161			.catch((error) => {
162			console.error('Test failed');
163			done(error);
164			});
165			});
166
167			it('https://www.golinks.io/sitemap.xml sitemaps should be an array', function (done) {
168			this.timeout(30000);
169			const url = 'https://www.golinks.io/sitemap.xml';
170			sitemapper.timeout = 5000;
171			sitemapper
172			.fetch(url)
173			.then((data) => {
174			data.sites.should.be.Array;
175			data.url.should.equal(url);
176			data.sites.length.should.be.above(2);
177			isUrl(data.sites[0]).should.be.true;
178			done();
179			})
180			.catch((error) => {
181			console.error('Test failed');
182			done(error);
183			});
184			});
185
186			it('https://www.golinks.io/sitemap.xml sitemaps should return an empty array when timing out', function (done) {
187			this.timeout(30000);
188			const url = 'https://www.golinks.io/sitemap.xml';
189			sitemapper.timeout = 1;
190			sitemapper
191			.fetch(url)
192			.then((data) => {
193			data.sites.should.be.Array;
194			data.errors.should.be.Array;
195			console.log(data);
196			done();
197			})
198			.catch((error) => {
199			console.error('Test failed');
200			done(error);
201			});
202			});
203
204			it('https://www.golinks.com/blog/sitemap.xml sitemaps should return an empty array when timing out', function (done) {
205			this.timeout(30000);
206			const url = 'https://www.golinks.com/blog/sitemap.xml';
207			sitemapper.timeout = 10000;
208			sitemapper
209			.fetch(url)
210			.then((data) => {
211			data.sites.should.be.Array;
212			data.errors.should.be.Array;
213			done();
214			})
215			.catch((error) => {
216			console.error('Test failed');
217			done(error);
218			});
219			});
220
221			it('https://www.banggood.com/sitemap/category.xml.gz gzip should be a non-empty array', function (done) {
222			this.timeout(30000);
223			const url = 'https://www.banggood.com/sitemap/category.xml.gz';
224			sitemapper.timeout = 10000;
225			sitemapper
226			.fetch(url)
227			.then((data) => {
228			data.sites.should.be.Array;
229			data.sites.length.should.be.greaterThan(0);
230			done();
231			})
232			.catch((error) => {
233			console.error('Test failed');
234			done(error);
235			});
236			});
237			});
238
239			describe('gzipped sitemaps', function () {
240			beforeEach(() => {
241			sitemapper = new Sitemapper({
242			requestHeaders: {
243			'Accept-Encoding': 'gzip,deflate,sdch',
244			},
245			});
246			});
247
248			it('https://www.banggood.com/sitemap/category.xml.gz gzip should be a non-empty array', function (done) {
249			this.timeout(30000);
250			const url = 'https://www.banggood.com/sitemap/category.xml.gz';
251			sitemapper.timeout = 10000;
252			sitemapper
253			.fetch(url)
254			.then((data) => {
255			data.sites.should.be.Array;
256			data.errors.should.be.Array;
257			data.sites.length.should.be.greaterThan(0);
258			done();
259			})
260			.catch((error) => {
261			console.error('Test failed');
262			done(error);
263			});
264			});
265
266			it('https://example.com/sitemap.xml should not allow insecure request', function (done) {
267			this.timeout(30000);
268			const url = 'https://example.com/sitemap.xml';
269			sitemapper.timeout = 10000;
270			sitemapper.rejectUnauthorized = false;
271			sitemapper
272			.fetch(url)
273			.then((data) => {
274			data.sites.should.be.Array;
275			data.errors.should.be.Array;
276			data.errors.should.containEql({
277			type: 'HTTPError',
278			message: 'HTTP Error occurred: Response code 404 (Not Found)',
279			url: 'https://example.com/sitemap.xml',
280			retries: 0,
281			});
282			done();
283			})
284			.catch((error) => {
285			console.error('Test failed');
286			done(error);
287			});
288			});
289			});
290
291			describe('sitemaps with namespace prefix', function () {
292			it('https://www.emerald.com/sitemap.xml sitemaps should be an array', function (done) {
293			this.timeout(30000);
294			const url = 'https://www.emerald.com/sitemap.xml';
295			sitemapper
296			.fetch(url)
297			.then((data) => {
298			data.sites.should.be.Array;
299			data.url.should.equal(url);
300			data.sites.length.should.be.above(2);
301			isUrl(data.sites[0]).should.be.true;
302			done();
303			})
304			.catch((error) => {
305			console.error('Test failed');
306			done(error);
307			});
308			});
309			});
310
311			describe('getSites method', function () {
312			it('getSites should be backwards compatible', function (done) {
313			this.timeout(30000);
314			const url = 'https://wp.seantburke.com/sitemap.xml';
315			sitemapper.getSites(url, (err, sites) => {
316			sites.should.be.Array;
317			isUrl(sites[0]).should.be.true;
318			done();
319			});
320			});
321			});
322
323			describe('exclusions option', function () {
324			it('should prevent false positive', function (done) {
325			this.timeout(30000);
326			const url = 'https://wp.seantburke.com/sitemap.xml';
327			sitemapper.exclusions = [/video/, /image/];
328			sitemapper
329			.fetch(url)
330			.then((data) => {
331			data.sites.should.be.Array;
332			data.sites.includes('https://wp.seantburke.com/?page_id=2').should.be
333			.true;
334			done();
335			})
336			.catch((error) => {
337			console.error('Test failed');
338			done(error);
339			});
340			});
341
342			it('should filter out page_id urls', function (done) {
343			this.timeout(30000);
344			const url = 'https://wp.seantburke.com/sitemap.xml';
345			sitemapper.exclusions = [/page_id/];
346			sitemapper
347			.fetch(url)
348			.then((data) => {
349			data.sites.should.be.Array;
350			data.sites.includes('https://wp.seantburke.com/?page_id=2').should.be
351			.false;
352			done();
353			})
354			.catch((error) => {
355			console.error('Test failed');
356			done(error);
357			});
358			});
359			});
360
361			describe('isExcluded method', function () {
362			it('should return false when no exclusions are set', function () {
363			const result = sitemapper.isExcluded('https://example.com/page1');
364			result.should.be.false();
365			});
366
367			it('should return false when url does not match any exclusion patterns', function () {
368			sitemapper.exclusions = [/\.pdf$/, /private/];
369			const result = sitemapper.isExcluded('https://example.com/page1');
370			result.should.be.false();
371			});
372
373			it('should return false when url matches an exclusion pattern', function () {
374			sitemapper.exclusions = [/\.pdf$/, /private/];
375			const result = sitemapper.isExcluded('https://example.com/document.pdf');
376			result.should.be.true();
377			});
378
379			it('should return true when url matches any of multiple exclusion patterns', function () {
380			sitemapper.exclusions = [/\.pdf$/, /private/, /temp/];
381			const result = sitemapper.isExcluded(
382			'https://example.com/private/temp.html'
383			);
384			result.should.be.true();
385			});
386
387			it('should handle complex regex patterns correctly', function () {
388			sitemapper.exclusions = [/^https:\/\/example\.com\/([a-z]{2})\/private/];
389			const result1 = sitemapper.isExcluded(
390			'https://example.com/en/private/page'
391			);
392			const result2 = sitemapper.isExcluded(
393			'https://example.com/en/public/page'
394			);
395			result1.should.be.true();
396			result2.should.be.false();
397			});
398
399			it('should handle case sensitivity correctly', function () {
400			sitemapper.exclusions = [/private/i];
401			const result1 = sitemapper.isExcluded('https://example.com/PRIVATE/page');
402			const result2 = sitemapper.isExcluded('https://example.com/Private/page');
403			result1.should.be.true();
404			result2.should.be.true();
405			});
406			});
407			});
408

seantomburke / sitemapper

test.ts.ts ➔ isUrl A last analyzed 2025-06-01 19:32 UTC

Complexity

Size

Duplication

Importance

Duplication Side-by-Side

Filter issues like

test.ts.ts ➔ isUrl A
last analyzed 2025-06-01 19:32 UTC