History

Finish scikit-image scraper

Thibaut Courouble committed 10 years ago

Commit 50f58321 parent 7f06c9c11 files · +30 −20Browse files

Showing the changes to lib/docs (4 of 11 files). Show all

Unified · Split

4 files
  1. lib/docs/filters/scikit_image/clean_html.rb +0 −11
  2. lib/docs/filters/scikit_image/entries.rb +19 −4
  3. lib/docs/filters/sphinx/clean_html.rb +1 −1
  4. lib/docs/scrapers/scikit_image.rb +7 −7
lib/docs/filters/scikit_image/clean_html.rb+0−11
@@ -1,11 +0,0 @@
1−module Docs
2− class ScikitImage
3− class CleanHtmlFilter < Filter
4− def call
5− css('h2').remove
6− css('h1 + table').remove
7− doc
8− end
9− end
10− end
11−end
lib/docs/filters/scikit_image/entries.rb+19−4
@@ -4,17 +4,27 @@ module Docs · 3 unmodified lines
44 def get_name
55 name = at_css('h1').content.strip
66 name.remove! "\u{00b6}"
7− name.remove! 'Module: '
8− name.remove! %r{ \(.*\)}
9− name.downcase!
7+
8+ if slug.start_with?('api')
9+ name.remove! 'Module: '
10+ name.remove! %r{ \(.*\)}
11+ name.downcase!
12+ end
13+
1014 name
1115 end
1216
1317 def get_type
14− name.split('.').first
18+ if slug.start_with?('api')
19+ name.split('.').first
20+ else
21+ 'Guide'
22+ end
1523 end
1624
1725 def additional_entries
26+ return [] unless slug.start_with?('api')
27+
1828 entries = []
1929
2030 css('.class > dt[id]', '.exception > dt[id]', '.attribute > dt[id]').each do |node|
@@ -30,8 +40,13 @@ module Docs · 9 unmodified lines
3040 css('.function > dt[id]', '.method > dt[id]', '.classmethod > dt[id]').each do |node|
3141 entries << [node['id'].remove('skimage.') + '()', node['id']]
3242 end
43+
3344 entries
3445 end
46+
47+ def include_default_entry?
48+ slug != 'api/api'
49+ end
3550 end
3651 end
3752 end
lib/docs/filters/sphinx/clean_html.rb+1−1
@@ -9,7 +9,7 @@ module Docs · 8 unmodified lines
99 node.remove
1010 end
1111
12− css('em.xref', 'tt').each do |node|
12+ css('em.xref', 'tt', 'cite').each do |node|
1313 node.name = 'code'
1414 end
1515
lib/docs/scrapers/scikit_image.rb+7−7
@@ -3,22 +3,22 @@ module Docs · 2 unmodified lines
33 self.name = 'scikit-image'
44 self.slug = 'scikit_image'
55 self.type = 'sphinx'
6− self.release = '0.12.2'
7− self.base_url = 'http://scikit-image.org/docs/0.12.x/api/'
8− self.root_path = 'api.html'
9−
6+ self.release = '0.12.3'
7+ self.base_url = 'http://scikit-image.org/docs/0.12.x/'
108 self.links = {
119 home: 'http://scikit-image.org/',
1210 code: 'https://github.com/scikit-image/scikit-image'
1311 }
1412
15− html_filters.push 'scikit_image/entries', 'scikit_image/clean_html', 'sphinx/clean_html'
13+ html_filters.push 'scikit_image/entries', 'sphinx/clean_html'
1614
1715 options[:container] = '.span9'
16+ options[:skip] = %w(api_changes.html)
17+ options[:only_patterns] = [/\Aapi/, /\Auser_guide/]
1818
1919 options[:attribution] = <<-HTML
20− &copy; 2011 the scikit-image development team<br>
21− Licensed under the scikit-image License.
20+ &copy; 2011 the scikit-image team<br>
21+ Licensed under the BSD 3-clause License.
2222 HTML
2323 end
2424 end