Skip to content

Commit 332ec79

Browse files
ksaitorksaitor
committed
improved formatting of LinkedinClient.js. Added .related[] items.
Improved tests
1 parent 23358a3 commit 332ec79

File tree

2 files changed

+128
-135
lines changed

2 files changed

+128
-135
lines changed

‎src/LinkedinClient.js‎

Lines changed: 123 additions & 130 deletions
Original file line numberDiff line numberDiff line change
@@ -1,142 +1,135 @@
1-
const cheerio = require('cheerio');
2-
const fs = require('fs');
3-
const Entities = require('html-entities').XmlEntities;
4-
const request = require('node-fetch');
1+
const cheerio = require('cheerio')
2+
const fs = require('fs')
3+
const Entities = require('html-entities').XmlEntities
4+
const request = require('node-fetch')
55

6-
const debugFile = './assets/debug.json';
7-
const industries = {};
8-
const followingItems = {};
6+
const debugFile = './assets/debug.json'
7+
const industries = {}
8+
const followingItems = {}
99

1010
module.exports = class LinkedinClient {
11-
constructor(cookie) {
12-
if(!cookie)
13-
throw new Error('The Linkedin cookie is required.');
11+
constructor(cookie) {
12+
if (!cookie) throw new Error('The Linkedin cookie is required.')
13+
this.entities = new Entities()
14+
this.cookie = cookie
15+
}
1416

15-
this.entities = new Entities();
16-
this.cookie = cookie;
17-
}
18-
19-
async fetch(url) {
20-
let processMethod;
21-
if (url.match(/^https:\/\/www.linkedin.com\/in\//)) processMethod = processPeopleProfile;
22-
else if (url.match(/^https:\/\/www.linkedin.com\/company\//)) {
23-
url += url[url.length - 1] == '/' ? 'about/' : '/about/';
24-
processMethod = processCompanyPage;
25-
} else throw new Error(`Invalid URL provided ("${url}"), it must be a people profile URL or a company page URL.`);
26-
27-
if(process.env.NODE_ENV == 'dev')
28-
fs.writeFileSync(debugFile, '');
29-
30-
const res = await request(url, { headers: { Cookie: `li_at=${this.cookie}` } });
31-
const html = await res.text();
32-
const $ = cheerio.load(html);
33-
let data, result = { linkedinUrl: url.replace('/about/', '') };
34-
while (!result.name && !result.firstName) {
35-
// this loop allows to fix a bug with random missing <code> tags
36-
for (let elt of $('code').get()) {
37-
try {
38-
data = JSON.parse(this.entities.decode($(elt).html()));
39-
} catch (e) {
40-
continue;
41-
}
42-
if (!data.included)
43-
continue;
44-
for (let item of data.included) {
45-
processMethod(item, result);
46-
if (process.env.NODE_ENV == 'dev')
47-
fs.appendFileSync(debugFile, JSON.stringify(item, null, 4) + '\n');
48-
}
49-
}
50-
51-
// this company or people does not exist
52-
if (!result.firstName && !result.name)
53-
return null;
54-
}
55-
56-
return result;
57-
}
58-
};
17+
async fetch(url) {
18+
let processMethod
19+
if (url.match(/^https:\/\/www.linkedin.com\/in\//)) processMethod = processPeopleProfile
20+
else if (url.match(/^https:\/\/www.linkedin.com\/company\//)) {
21+
url += url[url.length - 1] == '/' ? 'about/' : '/about/'
22+
processMethod = processCompanyPage
23+
} else throw new Error(`Invalid URL provided ("${url}"), it must be a people profile URL or a company page URL.`)
24+
25+
if (process.env.NODE_ENV == 'dev') fs.writeFileSync(debugFile, '')
26+
27+
const res = await request(url, { headers: { Cookie: `li_at=${this.cookie}` } })
28+
const html = await res.text()
29+
const $ = cheerio.load(html)
30+
let data,
31+
result = { linkedinUrl: url.replace('/about/', ''), related: [] }
32+
while (!result.name && !result.firstName) {
33+
// this loop allows to fix a bug with random missing <code> tags
34+
for (let elt of $('code').get()) {
35+
try {
36+
data = JSON.parse(this.entities.decode($(elt).html()))
37+
} catch (e) {
38+
continue
39+
}
40+
if (!data.included) continue
41+
for (let item of data.included) {
42+
processMethod(item, result)
43+
if (process.env.NODE_ENV == 'dev') fs.appendFileSync(debugFile, JSON.stringify(item, null, 4) + '\n')
44+
}
45+
}
46+
47+
// this company or people does not exist
48+
if (!result.firstName && !result.name) return null
49+
}
50+
51+
return result
52+
}
53+
}
5954

6055
// private method
6156
function processPeopleProfile(item, result) {
62-
if (item.$type == 'com.linkedin.voyager.dash.common.Industry' && item.name)
63-
industries[item.entityUrn] = item.name;
64-
if (item.$type == 'com.linkedin.voyager.dash.identity.profile.Profile' && item.objectUrn) {
65-
result.firstName = item.firstName;
66-
result.lastName = item.lastName;
67-
result.headline = item.headline;
68-
result.location = item.locationName;
69-
result.address = item.address;
70-
result.industry = industries[item['*industry']];
71-
result.summary = item.summary;
72-
if (result.birthDate) {
73-
result.birthDate = item.birthDate;
74-
delete result.birthDate.$type;
75-
}
76-
} else if (item.$type == 'com.linkedin.voyager.common.FollowingInfo' && item.followerCount)
77-
result.connections = item.followerCount;
78-
else if (item.$type == 'com.linkedin.voyager.dash.identity.profile.Position') {
79-
if (!result.positions)
80-
result.positions = [];
81-
const position = {
82-
title: item.title,
83-
company: item.companyName,
84-
location: item.location,
85-
description: item.description,
86-
dateRange: item.dateRange
87-
};
88-
if (position.dateRange) {
89-
delete position.dateRange.$type;
90-
if (position.dateRange.start) delete position.dateRange.start.$type;
91-
if (position.dateRange.end) delete position.dateRange.end.$type;
92-
}
93-
result.positions.push(position);
94-
} else if (item.$type == 'com.linkedin.voyager.dash.identity.profile.Education') {
95-
if (!result.education)
96-
result.education = [];
97-
const degree = {
98-
degree: item.degreeName,
99-
school: item.schoolName,
100-
field: item.fieldOfStudy,
101-
dateRange: item.dateRange
102-
};
103-
if (degree.dateRange) {
104-
delete degree.dateRange.$type;
105-
if (degree.dateRange.start) delete degree.dateRange.start.$type;
106-
if (degree.dateRange.end) delete degree.dateRange.end.$type;
107-
}
108-
result.education.push(degree);
109-
} else if (item.$type == 'com.linkedin.voyager.dash.identity.profile.Skill') {
110-
if (!result.skills)
111-
result.skills = [];
112-
result.skills.push(item.name);
113-
} else if (item.$type == 'com.linkedin.voyager.dash.identity.profile.Language') {
114-
if (!result.languages)
115-
result.languages = [];
116-
result.languages.push({ language: item.name, proficiency: item.proficiency });
117-
}
57+
if (item.$type == 'com.linkedin.voyager.dash.common.Industry' && item.name) industries[item.entityUrn] = item.name
58+
if (item.$type == 'com.linkedin.voyager.dash.identity.profile.Profile' && item.objectUrn) {
59+
result.firstName = item.firstName
60+
result.lastName = item.lastName
61+
result.headline = item.headline
62+
result.location = item.locationName
63+
result.address = item.address
64+
result.industry = industries[item['*industry']]
65+
result.summary = item.summary
66+
if (result.birthDate) {
67+
result.birthDate = item.birthDate
68+
delete result.birthDate.$type
69+
}
70+
} else if (item.$type == 'com.linkedin.voyager.common.FollowingInfo' && item.followerCount)
71+
result.connections = item.followerCount
72+
else if (item.$type == 'com.linkedin.voyager.dash.identity.profile.Position') {
73+
if (!result.positions) result.positions = []
74+
const position = {
75+
title: item.title,
76+
company: item.companyName,
77+
location: item.location,
78+
description: item.description,
79+
dateRange: item.dateRange,
80+
}
81+
if (position.dateRange) {
82+
delete position.dateRange.$type
83+
if (position.dateRange.start) delete position.dateRange.start.$type
84+
if (position.dateRange.end) delete position.dateRange.end.$type
85+
}
86+
result.positions.push(position)
87+
} else if (item.$type == 'com.linkedin.voyager.dash.identity.profile.Education') {
88+
if (!result.education) result.education = []
89+
const degree = {
90+
degree: item.degreeName,
91+
school: item.schoolName,
92+
field: item.fieldOfStudy,
93+
dateRange: item.dateRange,
94+
}
95+
if (degree.dateRange) {
96+
delete degree.dateRange.$type
97+
if (degree.dateRange.start) delete degree.dateRange.start.$type
98+
if (degree.dateRange.end) delete degree.dateRange.end.$type
99+
}
100+
result.education.push(degree)
101+
} else if (item.$type == 'com.linkedin.voyager.dash.identity.profile.Skill') {
102+
if (!result.skills) result.skills = []
103+
result.skills.push(item.name)
104+
} else if (item.$type == 'com.linkedin.voyager.dash.identity.profile.Language') {
105+
if (!result.languages) result.languages = []
106+
result.languages.push({ language: item.name, proficiency: item.proficiency })
107+
}
118108
}
119109

120110
// private method
121111
function processCompanyPage(item, result) {
122-
if (item.$type == 'com.linkedin.voyager.common.Industry')
123-
industries[item.entityUrn] = item.localizedName;
124-
else if (item.$type == 'com.linkedin.voyager.common.FollowingInfo')
125-
followingItems[item.entityUrn] = item.followerCount;
126-
else if (item.$type == 'com.linkedin.voyager.organization.Company' && item.staffCount) {
127-
result.name = item.name;
128-
result.tagline = item.tagline;
129-
result.description = item.description;
130-
result.industry = industries[item['*companyIndustries'][0]];
131-
result.type = item.companyType ? item.companyType.localizedName : null,
132-
result.website = item.companyPageUrl;
133-
result.companySize = item.staffCountRange.start + (item.staffCountRange.end ? '-' + item.staffCountRange.end : '+') + ' employees';
134-
result.membersOnLinkedin = item.staffCount;
135-
result.headquarters = item.headquarter;
136-
delete result.headquarters.$type;
137-
result.companyType = item.companyType.localizedName;
138-
result.foundedYear = item.foundedOn && item.foundedOn.year;
139-
result.specialties = item.specialities;
140-
result.followers = followingItems[item[['*followingInfo']]];
141-
}
112+
if (item.$type == 'com.linkedin.voyager.common.Industry') {
113+
industries[item.entityUrn] = item.localizedName
114+
} else if (item.$type == 'com.linkedin.voyager.common.FollowingInfo') {
115+
followingItems[item.entityUrn] = item.followerCount
116+
} else if (item.$type == 'com.linkedin.voyager.organization.Company' && item.staffCount) {
117+
result.name = item.name
118+
result.tagline = item.tagline
119+
result.description = item.description
120+
result.industry = industries[item['*companyIndustries'][0]]
121+
result.type = item.companyType ? item.companyType.localizedName : null
122+
result.website = item.companyPageUrl
123+
result.companySize =
124+
item.staffCountRange.start + (item.staffCountRange.end ? '-' + item.staffCountRange.end : '+') + ' employees'
125+
result.membersOnLinkedin = item.staffCount
126+
result.headquarters = item.headquarter
127+
delete result.headquarters.$type
128+
result.companyType = item.companyType.localizedName
129+
result.foundedYear = item.foundedOn && item.foundedOn.year
130+
result.specialties = item.specialities
131+
result.followers = followingItems[item[['*followingInfo']]]
132+
} else if (item.url && item.url !== result.linkedinUrl) {
133+
result.related.push(item)
134+
}
142135
}

‎src/tests/api_test.js‎

Lines changed: 5 additions & 5 deletions
Original file line numberDiff line numberDiff line change
@@ -67,13 +67,13 @@ describe('Linkedin client tests', () => {
6767
expect(res.body.error).to.equal(null);
6868
expect(res.body.result['firstName']).to.equal('Alix');
6969
expect(res.body.result['lastName']).to.equal('Vandame');
70-
expect(res.body.result['headline']).to.equal('Helping great tech talents finding great jobs !');
71-
expect(res.body.result['summary'].length).to.be.not.equal(0);
72-
expect(res.body.result['location']).to.equal('Paris Area, France');
70+
expect(res.body.result['headline']).to.equal('Talent Acquisition Manager at eFounders \n');
71+
// expect(res.body.result['summary'].length).to.be.not.equal(0);
72+
// expect(res.body.result['location']).to.equal('Paris Area, France');
7373
expect(res.body.result['industry']).to.equal('Internet');
74-
expect(res.body.result['education'].length).to.equal(4);
74+
expect(res.body.result['education'].length).to.equal(5);
7575
expect(res.body.result['languages'].length).to.equal(3);
76-
expect(res.body.result['positions'].length).to.equal(6);
76+
expect(res.body.result['positions'].length).to.equal(8);
7777
expect(res.body.result['skills'].length).to.equal(20);
7878
expect(res.body.result['linkedinUrl']).to.equal('https://www.linkedin.com/in/alix-vandame/');
7979
done();

0 commit comments

Comments
 (0)