indeed_scraper2022 0.1.3 → 0.3.0
Sign up to get free protection for your applications and to get access to all the features.
- checksums.yaml +4 -4
- checksums.yaml.gz.sig +0 -0
- data/lib/indeed_scraper2022.rb +157 -62
- data.tar.gz.sig +0 -0
- metadata +7 -7
- metadata.gz.sig +0 -0
checksums.yaml
CHANGED
@@ -1,7 +1,7 @@
|
|
1
1
|
---
|
2
2
|
SHA256:
|
3
|
-
metadata.gz:
|
4
|
-
data.tar.gz:
|
3
|
+
metadata.gz: 972f811430fae59121e39c9c4752b64fc43b37165a52dcec8c3eac42cf1e4555
|
4
|
+
data.tar.gz: 85e987eb264b098b4c892e2e05d2ab082e3b8968fff2bdd519552e889f014f9d
|
5
5
|
SHA512:
|
6
|
-
metadata.gz:
|
7
|
-
data.tar.gz:
|
6
|
+
metadata.gz: e7d3c2a13e315383248c557806dd0184d8831f46a2e314816395f62fcb886ba2e38a3e1f2deb180ceb33b614cf0b7be8a13379028fc56982e388948575bdb02c
|
7
|
+
data.tar.gz: 6ca4792d260c43b22fcee5ead8928525df17f1db112c49ac4cc7d7a5c9b29a8f483b94c18cd39bace1d4a4ee553ff83254a7fb445d76abf06c1705d19bac455c
|
checksums.yaml.gz.sig
CHANGED
Binary file
|
data/lib/indeed_scraper2022.rb
CHANGED
@@ -2,23 +2,31 @@
|
|
2
2
|
|
3
3
|
# file: indeed_scraper2022.rb
|
4
4
|
|
5
|
-
require '
|
5
|
+
require 'ferrumwizard'
|
6
6
|
require 'nokorexi'
|
7
|
+
require 'yaml'
|
7
8
|
|
8
9
|
# Given the nature of changes to jobsearch websites,
|
9
10
|
# don't rely upon this gem working in the near future.
|
10
11
|
|
11
12
|
|
13
|
+
|
12
14
|
class IndeedScraper2022Err < Exception
|
13
15
|
end
|
14
16
|
|
15
17
|
class IndeedScraper2022
|
16
18
|
|
17
|
-
|
19
|
+
attr_reader :browser
|
20
|
+
|
21
|
+
def initialize(url='https://uk.indeed.com/?r=us', q: '', location: '',
|
22
|
+
headless: true, cookies: nil, debug: false)
|
18
23
|
|
19
24
|
@debug = debug
|
20
25
|
@url_base, @q, @location = url, q, location
|
21
|
-
@
|
26
|
+
@headless, @cookies = headless, cookies
|
27
|
+
|
28
|
+
fw = FerrumWizard.new( headless: @headless, cookies: @cookies, debug: @debug)
|
29
|
+
@browser = fw.browser
|
22
30
|
|
23
31
|
end
|
24
32
|
|
@@ -28,6 +36,101 @@ class IndeedScraper2022
|
|
28
36
|
@results
|
29
37
|
end
|
30
38
|
|
39
|
+
def search(q: @q, location: @location, start: nil)
|
40
|
+
|
41
|
+
url = @url_base
|
42
|
+
url += 'start=' + start if start
|
43
|
+
|
44
|
+
@browser.goto(url)
|
45
|
+
#@browser.network.wait_for_idle
|
46
|
+
puts 'sleeping for 4 seconds' if @debug
|
47
|
+
sleep 4
|
48
|
+
|
49
|
+
if q.length > 1 then
|
50
|
+
|
51
|
+
input = @browser.at_xpath("//input[@name='q']")
|
52
|
+
|
53
|
+
# select any existing text and overwrite it
|
54
|
+
input.focus.type(:home); sleep 0.2
|
55
|
+
input.focus.type(:shift, :end); sleep 0.2
|
56
|
+
input.focus.type(q); sleep 0.2
|
57
|
+
end
|
58
|
+
|
59
|
+
if location.length > 1 then
|
60
|
+
|
61
|
+
input2 = @browser.at_xpath("//input[@name='l']")
|
62
|
+
|
63
|
+
# select any existing text and overwrite it
|
64
|
+
input2.focus.type(:home); sleep 0.2
|
65
|
+
input2.focus.type(:shift, :end); sleep 0.2
|
66
|
+
input2.focus.type(location); sleep 0.2
|
67
|
+
|
68
|
+
end
|
69
|
+
|
70
|
+
button = @browser.at_xpath("//button[@type='submit']")
|
71
|
+
button.click
|
72
|
+
#@browser.network.wait_for_idle
|
73
|
+
puts 'sleeping for 2 seconds' if @debug
|
74
|
+
sleep 2
|
75
|
+
|
76
|
+
doc2 = Nokogiri::XML(@browser.body)
|
77
|
+
|
78
|
+
a2 = doc2.xpath "//a[div/div/div/div/table/tbody/tr/td/div]"
|
79
|
+
puts 'a2: ' + a2.length.inspect if @debug
|
80
|
+
|
81
|
+
@a2 = a2.map {|x| Rexle.new x.to_s }
|
82
|
+
|
83
|
+
@results = @a2.map do |doc|
|
84
|
+
|
85
|
+
div = doc.element("a[@class='desktop']/div[@class='slider" \
|
86
|
+
"_container']/div[@class='slider_list']/div[@class='sl" \
|
87
|
+
"ider_item']/div[@class='job_seen_beacon']")
|
88
|
+
td = div.element("table[@class='jobCard_mainContent']/tbo" \
|
89
|
+
"dy/tr/td[@class='resultContent']")
|
90
|
+
|
91
|
+
# job title (e.g. Software Developer)
|
92
|
+
jobtitle = td.element("div[@class='tapItem-gutter']/h2[@" \
|
93
|
+
"class='jobTitle-color-purple']/span")&.text
|
94
|
+
puts 'jobtitle: ' + jobtitle.inspect if @debug
|
95
|
+
|
96
|
+
salary = td.element("div[@class='metadataContainer']/" \
|
97
|
+
"div[@class='salary-snippet-container']/div[@class='sa" \
|
98
|
+
"lary-snippet']/span")&.text
|
99
|
+
|
100
|
+
puts 'salary: ' + salary.inspect if @debug
|
101
|
+
div1 = td.element("div[@class='companyInfo']")
|
102
|
+
|
103
|
+
# company name (e.g. Coda Octopus Products Ltd)
|
104
|
+
company_name = div1.element("span[@class='companyName']")&.text
|
105
|
+
|
106
|
+
# company location (e.g. Edinburgh)
|
107
|
+
location = div1.element("div[@class='companyLocation']")&.text
|
108
|
+
tbody = div.element("table[@class='jobCardShelfContainer']/tbody")
|
109
|
+
|
110
|
+
div3 = tbody.element("tr[@class='underShelfFooter']/td/di" \
|
111
|
+
"v[@class='result-footer']")
|
112
|
+
|
113
|
+
# job (e.g. Our products are primarily written in C#, using...)
|
114
|
+
jobsnippet = div3.xpath("div[@class='job-snippet']/ul/li/text()").join("\n")
|
115
|
+
|
116
|
+
# visually (e.g. Posted 14 days ago)
|
117
|
+
dateposted = div3.element("span[@class='date']")&.texts
|
118
|
+
date = (Date.today - dateposted.first.to_i).to_s if dateposted
|
119
|
+
|
120
|
+
{
|
121
|
+
link: @url_base.sub(/\/[^\/]+$/,'') \
|
122
|
+
+ doc.root.attributes[:href].gsub(/&/,'&'),
|
123
|
+
title: jobtitle,
|
124
|
+
salary: salary,
|
125
|
+
company: company_name,
|
126
|
+
location: location,
|
127
|
+
jobsnippet: jobsnippet,
|
128
|
+
date: date
|
129
|
+
}
|
130
|
+
|
131
|
+
end
|
132
|
+
end
|
133
|
+
|
31
134
|
def page(n)
|
32
135
|
|
33
136
|
if n < 1 or n > @results.length then
|
@@ -64,16 +167,18 @@ class IndeedScraper2022
|
|
64
167
|
"ickyContainer-companyrating']/div/div[@class='icl-u-x" \
|
65
168
|
"s-mr--xs']")[1]
|
66
169
|
clink = div3.element('//a')
|
67
|
-
company = cname ? cname.text : clink.text
|
170
|
+
company = cname.text ? cname.text : clink.text
|
68
171
|
companylink = clink.attributes[:href] if clink
|
69
172
|
|
173
|
+
salary = div1.element("//span[@class='attribute_snippet']")&.text
|
174
|
+
type = div1.element("//span[@class='jobsearch-JobMetadataHeader-item']")&.texts&.last
|
70
175
|
div5 = div3.xpath("div/div")
|
71
176
|
location, worklocation = div5.map(&:text).compact
|
72
177
|
|
73
178
|
# icl (e.g. Full-time, Permanent)
|
74
179
|
jobtype = div1.element("div/div/div[@class='jobsearch-J" \
|
75
180
|
"obMetadataHeader-item']/span[@class='icl-u-xs-mt--xs']")
|
76
|
-
jobtype = jobtype
|
181
|
+
jobtype = jobtype&.texts.join if jobtype
|
77
182
|
|
78
183
|
# jobsearch (e.g. Urgently needed)
|
79
184
|
jobnote1 = e0.element("//div[@class='jobsearch-DesktopTag" \
|
@@ -81,98 +186,87 @@ class IndeedScraper2022
|
|
81
186
|
"h-DesktopTag-text']")&.text
|
82
187
|
|
83
188
|
# jobsearch (e.g. 10 days ago)
|
84
|
-
|
85
|
-
"tent']/div[@class='jobsearch-JobMetadataFooter']/div")&.text
|
189
|
+
days = e0.element("//div[@class='jobsearch-JobTab-con" \
|
190
|
+
"tent']/div[@class='jobsearch-JobMetadataFooter']/div[2]")&.text
|
191
|
+
d = Date.today - days.to_i
|
192
|
+
datepost = d.strftime("%Y-%m-%d")
|
193
|
+
|
86
194
|
|
87
195
|
jobdesc = e0.element("//div[@class='icl-u-xs-mt--md']/div[@cl" \
|
88
196
|
"ass='jobsearch-jobDescriptionText']").xml
|
89
197
|
|
90
198
|
{
|
91
199
|
title: jobtitle,
|
200
|
+
type: type,
|
92
201
|
company: company,
|
93
202
|
companylink: companylink,
|
94
203
|
location: location,
|
204
|
+
salary: salary,
|
95
205
|
worklocation: worklocation,
|
96
206
|
note: jobnote1,
|
97
|
-
date:
|
207
|
+
date: datepost,
|
98
208
|
desc: jobdesc
|
99
209
|
}
|
100
210
|
|
101
211
|
end
|
102
212
|
|
103
|
-
def search(q='', location='')
|
104
213
|
|
105
|
-
|
106
|
-
|
107
|
-
page = a.get(@url_base)
|
108
|
-
form = page.forms.first
|
109
|
-
form.fields[0].value = @q
|
110
|
-
form.fields[1].value = @location
|
111
|
-
pg = form.submit
|
214
|
+
end
|
112
215
|
|
113
|
-
|
216
|
+
class IS22Plus < IndeedScraper2022
|
114
217
|
|
115
|
-
|
116
|
-
|
218
|
+
def initialize(q: '', location: '', headless: true, cookies: nil, debug: false)
|
219
|
+
super(q: q, location: location, headless: headless, cookies: cookies,
|
220
|
+
debug: debug)
|
221
|
+
end
|
117
222
|
|
118
|
-
|
223
|
+
def archive(filepath='/tmp/indeed')
|
119
224
|
|
120
|
-
|
225
|
+
return unless @results
|
121
226
|
|
122
|
-
|
123
|
-
"_container']/div[@class='slider_list']/div[@class='sl" \
|
124
|
-
"ider_item']/div[@class='job_seen_beacon']")
|
125
|
-
td = div.element("table[@class='jobCard_mainContent']/tbo" \
|
126
|
-
"dy/tr/td[@class='resultContent']")
|
227
|
+
FileUtils.mkdir_p filepath
|
127
228
|
|
128
|
-
|
129
|
-
jobtitle = td.element("div[@class='tapItem-gutter']/h2[@" \
|
130
|
-
"class='jobTitle-color-purple']/span")&.text
|
131
|
-
puts 'jobtitle: ' + jobtitle.inspect if @debug
|
229
|
+
idxfile = File.join(filepath, 'index.yml')
|
132
230
|
|
133
|
-
|
134
|
-
|
135
|
-
|
231
|
+
index = if File.exists? idxfile then
|
232
|
+
YAML.load(File.read(idxfile))
|
233
|
+
else
|
234
|
+
{}
|
235
|
+
end
|
136
236
|
|
137
|
-
|
138
|
-
div1 = td.element("div[@class='companyInfo']")
|
237
|
+
@results.each.with_index do |item, i|
|
139
238
|
|
140
|
-
|
141
|
-
|
239
|
+
puts 'saving ' + item[:title] if @debug
|
240
|
+
puts 'link: ' + item[:link].inspect
|
241
|
+
links = RXFReader.reveal(item[:link])
|
242
|
+
puts 'links: ' + links.inspect
|
142
243
|
|
143
|
-
|
144
|
-
|
145
|
-
tbody = div.element("table[@class='jobCardShelfContainer']/tbody")
|
244
|
+
url = links.last
|
245
|
+
id = url[/(?<=\?jk=)[^&]+/]
|
146
246
|
|
147
|
-
|
148
|
-
|
247
|
+
if index[id.to_sym] then
|
248
|
+
next
|
249
|
+
else
|
149
250
|
|
150
|
-
|
151
|
-
jobsnippet = div3.xpath("div[@class='job-snippet']/ul/li/text()").join("\n")
|
251
|
+
File.write File.join(filepath, 'j' + id + '.txt'), page(i+1)
|
152
252
|
|
153
|
-
|
154
|
-
|
155
|
-
|
253
|
+
h = {
|
254
|
+
link: url[/^[^&]+/],
|
255
|
+
title: item[:title].to_s,
|
256
|
+
salary: item[:salary].to_s,
|
257
|
+
company: item[:company].to_s.strip,
|
258
|
+
location: item[:location].to_s,
|
259
|
+
jobsnippet: item[:jobsnippet],
|
260
|
+
date: item[:date]
|
261
|
+
}
|
156
262
|
|
157
|
-
|
158
|
-
|
159
|
-
+ doc.root.attributes[:href].gsub(/&/,'&'),
|
160
|
-
title: jobtitle,
|
161
|
-
salary: salary,
|
162
|
-
company: company_name,
|
163
|
-
location: location,
|
164
|
-
jobsnippet: jobsnippet,
|
165
|
-
date: date
|
166
|
-
}
|
263
|
+
index[id.to_sym] = h
|
264
|
+
end
|
167
265
|
|
168
266
|
end
|
169
|
-
end
|
170
|
-
end
|
171
267
|
|
172
|
-
|
268
|
+
File.write idxfile, index.to_yaml
|
173
269
|
|
174
|
-
def initialize(q: '', location: '', debug: false)
|
175
|
-
super(q: q, location: location, debug: debug)
|
176
270
|
end
|
177
271
|
|
178
272
|
def list()
|
@@ -183,4 +277,5 @@ class IS22Plus < IndeedScraper2022
|
|
183
277
|
|
184
278
|
end
|
185
279
|
|
280
|
+
|
186
281
|
end
|
data.tar.gz.sig
CHANGED
Binary file
|
metadata
CHANGED
@@ -1,7 +1,7 @@
|
|
1
1
|
--- !ruby/object:Gem::Specification
|
2
2
|
name: indeed_scraper2022
|
3
3
|
version: !ruby/object:Gem::Version
|
4
|
-
version: 0.
|
4
|
+
version: 0.3.0
|
5
5
|
platform: ruby
|
6
6
|
authors:
|
7
7
|
- James Robertson
|
@@ -35,7 +35,7 @@ cert_chain:
|
|
35
35
|
YLGLkwmbiUHX5dRU3RwOwOrZiyvND5BIj7S6dZ6jYHe0I727apgQNc3swTz5mW6I
|
36
36
|
SW/2zInu2bkj/meWm5eBoWHT
|
37
37
|
-----END CERTIFICATE-----
|
38
|
-
date: 2022-
|
38
|
+
date: 2022-04-01 00:00:00.000000000 Z
|
39
39
|
dependencies:
|
40
40
|
- !ruby/object:Gem::Dependency
|
41
41
|
name: nokorexi
|
@@ -58,25 +58,25 @@ dependencies:
|
|
58
58
|
- !ruby/object:Gem::Version
|
59
59
|
version: 0.7.0
|
60
60
|
- !ruby/object:Gem::Dependency
|
61
|
-
name:
|
61
|
+
name: ferrumwizard
|
62
62
|
requirement: !ruby/object:Gem::Requirement
|
63
63
|
requirements:
|
64
64
|
- - "~>"
|
65
65
|
- !ruby/object:Gem::Version
|
66
|
-
version: '2
|
66
|
+
version: '0.2'
|
67
67
|
- - ">="
|
68
68
|
- !ruby/object:Gem::Version
|
69
|
-
version: 2.
|
69
|
+
version: 0.2.2
|
70
70
|
type: :runtime
|
71
71
|
prerelease: false
|
72
72
|
version_requirements: !ruby/object:Gem::Requirement
|
73
73
|
requirements:
|
74
74
|
- - "~>"
|
75
75
|
- !ruby/object:Gem::Version
|
76
|
-
version: '2
|
76
|
+
version: '0.2'
|
77
77
|
- - ">="
|
78
78
|
- !ruby/object:Gem::Version
|
79
|
-
version: 2.
|
79
|
+
version: 0.2.2
|
80
80
|
description:
|
81
81
|
email: digital.robertson@gmail.com
|
82
82
|
executables: []
|
metadata.gz.sig
CHANGED
Binary file
|