From 3cf987da8ba70d252af6384df283fbdf0f631212 Mon Sep 17 00:00:00 2001 From: Enoc Date: Wed, 5 May 2021 21:43:58 -0600 Subject: [PATCH] Fix mdn scrapers - xslt_xpath had an error with redirection. --- lib/docs/filters/xslt_xpath/entries.rb | 5 +++-- lib/docs/scrapers/mdn/css.rb | 2 +- lib/docs/scrapers/mdn/dom.rb | 2 +- lib/docs/scrapers/mdn/javascript.rb | 2 +- lib/docs/scrapers/mdn/svg.rb | 2 +- lib/docs/scrapers/mdn/xslt_xpath.rb | 6 +----- 6 files changed, 8 insertions(+), 11 deletions(-) diff --git a/lib/docs/filters/xslt_xpath/entries.rb b/lib/docs/filters/xslt_xpath/entries.rb index 5151246f..3480547f 100644 --- a/lib/docs/filters/xslt_xpath/entries.rb +++ b/lib/docs/filters/xslt_xpath/entries.rb @@ -6,13 +6,14 @@ module Docs name.remove! 'XPath.' name.remove! 'XSLT.' name.remove! 'Axes.' - name.prepend 'xsl:' if slug =~ /\AXSLT\/[a-z]/ + name.remove! 'Element.' + name.prepend 'xsl:' if slug =~ /XSLT\/Element/ name << '()' if name.gsub!('Functions.', '') name end def get_type - if slug =~ /\AXSLT\/[a-z]/ + if slug =~ /XSLT\/Element/ 'XSLT Elements' elsif slug.start_with?('XPath/Axes') 'XPath Axes' diff --git a/lib/docs/scrapers/mdn/css.rb b/lib/docs/scrapers/mdn/css.rb index 4c44f1f1..98a126ac 100644 --- a/lib/docs/scrapers/mdn/css.rb +++ b/lib/docs/scrapers/mdn/css.rb @@ -4,7 +4,7 @@ module Docs self.base_url = 'https://developer.mozilla.org/en-US/docs/Web/CSS' self.root_path = '/Reference' - html_filters.push 'css/clean_html', 'css/entries', 'title' + html_filters.push 'css/clean_html', 'css/entries' options[:root_title] = 'CSS' diff --git a/lib/docs/scrapers/mdn/dom.rb b/lib/docs/scrapers/mdn/dom.rb index 943aeb5b..57828df3 100644 --- a/lib/docs/scrapers/mdn/dom.rb +++ b/lib/docs/scrapers/mdn/dom.rb @@ -4,7 +4,7 @@ module Docs self.name = 'DOM' self.base_url = 'https://developer.mozilla.org/en-US/docs/Web/API' - html_filters.push 'dom/clean_html', 'dom/entries', 'title' + html_filters.push 'dom/clean_html', 'dom/entries' options[:root_title] = 'DOM' diff --git a/lib/docs/scrapers/mdn/javascript.rb b/lib/docs/scrapers/mdn/javascript.rb index 935df61c..8238d3b4 100644 --- a/lib/docs/scrapers/mdn/javascript.rb +++ b/lib/docs/scrapers/mdn/javascript.rb @@ -6,7 +6,7 @@ module Docs self.name = 'JavaScript' self.base_url = 'https://developer.mozilla.org/en-US/docs/Web/JavaScript/Reference' - html_filters.push 'javascript/clean_html', 'javascript/entries', 'title' + html_filters.push 'javascript/clean_html', 'javascript/entries' options[:root_title] = 'JavaScript' diff --git a/lib/docs/scrapers/mdn/svg.rb b/lib/docs/scrapers/mdn/svg.rb index db9de7a1..4a9de544 100644 --- a/lib/docs/scrapers/mdn/svg.rb +++ b/lib/docs/scrapers/mdn/svg.rb @@ -6,7 +6,7 @@ module Docs self.name = 'SVG' self.base_url = 'https://developer.mozilla.org/en-US/docs/Web/SVG' - html_filters.push 'svg/clean_html', 'svg/entries', 'title' + html_filters.push 'svg/clean_html', 'svg/entries' options[:root_title] = 'SVG' diff --git a/lib/docs/scrapers/mdn/xslt_xpath.rb b/lib/docs/scrapers/mdn/xslt_xpath.rb index 2407f0c0..286c48f4 100644 --- a/lib/docs/scrapers/mdn/xslt_xpath.rb +++ b/lib/docs/scrapers/mdn/xslt_xpath.rb @@ -6,15 +6,11 @@ module Docs self.root_path = '/XSLT' self.initial_paths = %w(/XPath) - html_filters.push 'xslt_xpath/clean_html', 'xslt_xpath/entries', 'title' + html_filters.push 'xslt_xpath/clean_html', 'xslt_xpath/entries' options[:root_title] = 'XSLT' options[:only_patterns] = [/\A\/XSLT/, /\A\/XPath/] - options[:fix_urls] = ->(url) do - url.sub! 'https://developer.mozilla.org/en-US/docs/Web/XSLT/Element', "#{XsltXpath.base_url}/XSLT" - url - end end end