Skip to content

Commit 990f0b2

Browse files
mvolzmvolz
authored
Fix tests (#124)
* Remove two tests failing because of changes in upstream metadata for parseGeneral with one test using example.com (the particular metadata being tested - lang and rtl tags - is already tested for in the static tests) * Reorganise remaining tests in a more sensible way.
1 parent dc40c32 commit 990f0b2

File tree

1 file changed

+73
-83
lines changed

1 file changed

+73
-83
lines changed

‎test/scraping.js‎

Lines changed: 73 additions & 83 deletions
Original file line numberDiff line numberDiff line change
@@ -27,65 +27,87 @@ describe( 'scraping', function () {
2727
}
2828

2929
describe( 'parseAll function', () => {
30-
it( 'should resolve promise from woorank with headers', () => {
31-
const url = 'https://www.woorank.com/en/blog/dublin-core-metadata-for-seo-and-usability';
32-
return meta( { uri: url, headers: { 'User-Agent': userAgent, Accept: acceptHeader } } )
33-
.then( ( result ) => {
34-
assert.ok( result, 'Expected result to be truthy' );
35-
} )
36-
.catch( ( e ) => {
37-
console.error( 'Error in woorank test:', e );
38-
throw e;
39-
} );
40-
} );
4130

42-
it( 'should resolve promise from blog.schema.org without headers', () => {
43-
const url = 'http://blog.schema.org';
44-
return meta( url )
45-
.then( ( result ) => {
46-
assert.ok( result, 'Expected result to be truthy' );
31+
describe( 'Promise style', () => {
32+
it( 'should resolve promise from woorank with headers', () => {
33+
const url = 'https://www.woorank.com/en/blog/dublin-core-metadata-for-seo-and-usability';
34+
return meta( { uri: url, headers: { 'User-Agent': userAgent, Accept: acceptHeader } } )
35+
.then( ( result ) => {
36+
assert.ok( result, 'Expected result to be truthy' );
37+
} )
38+
.catch( ( e ) => {
39+
console.error( 'Error in woorank test:', e );
40+
throw e;
41+
} );
42+
} );
43+
44+
it( 'should resolve promise from blog.schema.org without headers', () => {
45+
const url = 'http://blog.schema.org';
46+
return meta( url )
47+
.then( ( result ) => {
48+
assert.ok( result, 'Expected result to be truthy' );
49+
} )
50+
.catch( ( e ) => {
51+
console.error( 'Error in blog.schema.org test:', e );
52+
throw e;
53+
} );
54+
} );
55+
56+
it( 'should throw error if no uri supplied', () => meta()
57+
.then( () => {
58+
assert.fail( 'Should have rejected the promise' );
4759
} )
4860
.catch( ( e ) => {
49-
console.error( 'Error in blog.schema.org test:', e );
50-
throw e;
61+
assert.ok( e instanceof Error, 'Error should be an Error object' );
62+
assert.strictEqual( e.message, 'No uri supplied in argument', 'Error message should match expected message' );
63+
} )
64+
);
65+
66+
it( 'should not have any undefined values', () => {
67+
const url = 'http://web.archive.org/web/20220127144804/https://www.cnet.com/special-reports/vr101/';
68+
return getWithHeaders( url ).then( ( body ) => {
69+
const chtml = cheerio.load( body );
70+
return meta.parseAll( chtml )
71+
.then( ( results ) => {
72+
Object.keys( results ).forEach( ( metadataType ) => {
73+
Object.keys( results[ metadataType ] ).forEach( ( key ) => {
74+
assert.notStrictEqual( results[ metadataType ][ key ], undefined, `${ metadataType }.${ key } should not be undefined` );
75+
} );
76+
} );
77+
} );
5178
} );
52-
} );
79+
} );
5380

54-
it( 'should throw error if no uri supplied', () => meta()
55-
.then( () => {
56-
assert.fail( 'Should have rejected the promise' );
57-
} )
58-
.catch( ( e ) => {
59-
assert.ok( e instanceof Error, 'Error should be an Error object' );
60-
assert.strictEqual( e.message, 'No uri supplied in argument', 'Error message should match expected message' );
61-
} )
62-
);
63-
64-
it( 'should support await implementation with headers', async () => {
65-
const url = 'http://blog.schema.org';
66-
const result = await meta( { uri: url, headers: { 'User-Agent': userAgent, Accept: acceptHeader } } );
67-
assert.ok( result, 'Expected result to be truthy' );
6881
} );
6982

70-
it( 'should support await implementation without headers', async () => {
71-
const url = 'http://blog.schema.org';
72-
const result = await meta( url );
73-
assert.ok( result, 'Expected result to be truthy' );
74-
} );
83+
describe( 'Await style', () => {
7584

76-
it( 'should throw error if no uri is supplied with async/await', async () => {
77-
try {
78-
await meta();
79-
assert.fail( 'Should have thrown an error' );
80-
} catch ( e ) {
81-
assert.ok( e instanceof Error, 'Error should be an Error object' );
82-
assert.strictEqual( e.message, 'No uri supplied in argument', 'Error message should match expected message' );
83-
}
85+
it( 'should support await implementation with headers', async () => {
86+
const url = 'http://blog.schema.org';
87+
const result = await meta( { uri: url, headers: { 'User-Agent': userAgent, Accept: acceptHeader } } );
88+
assert.ok( result, 'Expected result to be truthy' );
89+
} );
90+
91+
it( 'should support await implementation without headers', async () => {
92+
const url = 'http://blog.schema.org';
93+
const result = await meta( url );
94+
assert.ok( result, 'Expected result to be truthy' );
95+
} );
96+
97+
it( 'should throw error if no uri is supplied with async/await', async () => {
98+
try {
99+
await meta();
100+
assert.fail( 'Should have thrown an error' );
101+
} catch ( e ) {
102+
assert.ok( e instanceof Error, 'Error should be an Error object' );
103+
assert.strictEqual( e.message, 'No uri supplied in argument', 'Error message should match expected message' );
104+
}
105+
} );
84106
} );
85107

86108
} );
87109

88-
describe( 'parseBEPress function', () => {
110+
describe( 'Individual metadata functions', () => {
89111
it( 'should get BE Press metadata tags', () => {
90112
const url = 'http://biostats.bepress.com/harvardbiostat/paper154/';
91113
return getWithHeaders( url ).then( ( body ) => {
@@ -106,9 +128,7 @@ describe( 'scraping', function () {
106128
} );
107129
} );
108130
} );
109-
} );
110131

111-
describe( 'parseCOinS function', () => {
112132
it( 'should get COinS metadata', () => {
113133
const url = 'https://en.wikipedia.org/wiki/Viral_phylodynamics';
114134
return getWithHeaders( url ).then( ( body ) => {
@@ -121,9 +141,7 @@ describe( 'scraping', function () {
121141
} );
122142
} );
123143
} );
124-
} );
125144

126-
describe( 'parseEPrints function', () => {
127145
it( 'should get EPrints metadata', () => {
128146
const url = 'http://eprints.gla.ac.uk/113711/';
129147
return getWithHeaders( url ).then( ( body ) => {
@@ -139,45 +157,17 @@ describe( 'scraping', function () {
139157
} );
140158
} );
141159
} );
142-
} );
143-
144-
describe( 'parseGeneral function', () => {
145-
it( 'should get html lang parameter', () => {
146-
const expected = 'fr';
147-
const url = 'http://www.lemonde.fr';
148-
return getWithHeaders( url ).then( ( body ) => {
149-
const chtml = cheerio.load( body );
150-
return meta.parseGeneral( chtml ).then( ( results ) => {
151-
assert.strictEqual( results.lang, expected );
152-
} );
153-
} );
154-
} );
155160

156-
it( 'should get html dir parameter', () => {
157-
const expected = 'rtl';
158-
const url = 'https://www.iranrights.org/fa/';
161+
it( 'should get general metadata', () => {
162+
const expected = 'Example Domain';
163+
const url = 'http://example.com';
159164
return getWithHeaders( url ).then( ( body ) => {
160165
const chtml = cheerio.load( body );
161166
return meta.parseGeneral( chtml ).then( ( results ) => {
162-
assert.strictEqual( results.dir, expected );
167+
assert.strictEqual( results.title, expected );
163168
} );
164169
} );
165170
} );
166171
} );
167172

168-
it( 'should not have any undefined values', () => {
169-
const url = 'http://web.archive.org/web/20220127144804/https://www.cnet.com/special-reports/vr101/';
170-
return getWithHeaders( url ).then( ( body ) => {
171-
const chtml = cheerio.load( body );
172-
return meta.parseAll( chtml )
173-
.then( ( results ) => {
174-
Object.keys( results ).forEach( ( metadataType ) => {
175-
Object.keys( results[ metadataType ] ).forEach( ( key ) => {
176-
assert.notStrictEqual( results[ metadataType ][ key ], undefined, `${ metadataType }.${ key } should not be undefined` );
177-
} );
178-
} );
179-
} );
180-
} );
181-
} );
182-
183173
} );

0 commit comments

Comments
 (0)