History

Finish Statsmodels scraper

Thibaut Courouble committed 10 years ago

Commit 48479171 parent 06f5fae12 files · +14 −8Browse files

Showing the changes to lib/docs (5 of 12 files). Show all

Unified · Split

5 files
  1. lib/docs/filters/pandas/clean_html.rb +0 −2
  2. lib/docs/filters/sphinx/clean_html.rb +1 −1
  3. lib/docs/filters/statsmodels/clean_html.rb +4 −4
  4. lib/docs/filters/statsmodels/entries.rb +4 −1
  5. lib/docs/scrapers/statsmodels.rb +2 −3
lib/docs/filters/pandas/clean_html.rb+0−2
@@ -13,8 +13,6 @@ module Docs · 12 unmodified lines
1313 node.before(node.children).remove
1414 end
1515
16− css('.anchor-link').remove
17−
1816 doc
1917 end
2018 end
lib/docs/filters/sphinx/clean_html.rb+1−1
@@ -2,7 +2,7 @@ module Docs · 1 unmodified line
22 class Sphinx
33 class CleanHtmlFilter < Filter
44 def call
5− css('.headerlink', 'hr', '#contents .topic-title', '#topics .topic-title', 'colgroup', '.line-block').remove
5+ css('.headerlink', 'hr', '#contents .topic-title', '#topics .topic-title', 'colgroup', '.line-block', '.anchor-link').remove
66
77 css('.contents > ul:first-child:last-child.simple > li:first-child:last-child').each do |node|
88 node.parent.before(node.at_css('> ul')) if node.at_css('> ul')
lib/docs/filters/statsmodels/clean_html.rb+4−4
@@ -5,10 +5,10 @@ module Docs · 4 unmodified lines
55 @doc = at_css('.body')
66
77 if root_page?
8− at_css('h1').content = 'Statsmodels'
9− at_css('#basic-documentation').remove
10− at_css('#table-of-contents').remove
11− at_css('#indices-and-tables').remove
8+ at_css('h1').content = 'Statsmodels'
9+ at_css('#basic-documentation').remove
10+ at_css('#table-of-contents').remove
11+ at_css('#indices-and-tables').remove
1212 end
1313
1414 doc
lib/docs/filters/statsmodels/entries.rb+4−1
@@ -6,9 +6,12 @@ module Docs · 5 unmodified lines
66 name = at_css('dt').content.strip
77 name.sub! %r{\(.*}, '()' # Remove method arguments
88 name.remove! %r{[\=\[].*} # Remove "[source]"
9− name.remove! %r{\A(class(method)?) (statsmodels\.)?}
9+ name.remove! %r{\A(class(method)?) }
10+ name.remove! %r{\Astatsmodels\.}
1011 else
1112 name = at_css('h1').content.strip
13+ name.prepend 'Manual: ' if type == 'Manual'
14+ name.prepend 'Example: ' if type == 'Examples'
1215 end
1316 name.remove! "\u{00B6}" # Remove ¶
1417 name
lib/docs/scrapers/statsmodels.rb+2−3
@@ -1,9 +1,8 @@
11 module Docs
22 class Statsmodels < UrlScraper
3− self.name = 'Statsmodels'
43 self.type = 'sphinx'
54 self.release = '0.6.1'
6− self.base_url = "http://statsmodels.sourceforge.net/stable/"
5+ self.base_url = 'http://statsmodels.sourceforge.net/stable/'
76 self.root_path = 'index.html'
87 self.links = {
98 home: 'http://statsmodels.sourceforge.net/',
@@ -12,7 +11,7 @@ module Docs · 2 unmodified lines
1211
1312 html_filters.push 'statsmodels/entries', 'statsmodels/clean_html', 'sphinx/clean_html'
1413
15− options[:skip] = %w(about.html)
14+ options[:skip] = %w(about.html search.html genindex.html)
1615 options[:skip_patterns] = [/\Arelease/, /\Adev/, /\A_modules/, /\Adatasets/]
1716
1817 options[:attribution] = <<-HTML