RubyGems - grell - Versions diffs - 1.6.8 → 1.6.9 - Mend

grell 1.6.8 → 1.6.9

Files changed (6) hide show

checksums.yaml CHANGED

@@ -1,7 +1,7 @@
 ---
 SHA1:
-  metadata.gz: 4fcc56b727dfa90261c133037100f70617f15fed
-  data.tar.gz: 3e7d2a46160b4d8406a76b9190b8fdb64e4a0cf4
+  metadata.gz: 2a7136c81652b0260ee867e65380e69e5d3a2264
+  data.tar.gz: f95af6c4e4a99aa1216f8842c8829af521f7b1d9
 SHA512:
-  metadata.gz: 1cd4529624bd1d7d8482ab3d30c408c9a3452af73375ab2688e07d361b0c5cd1246dbcfcf0152b9e4ede8ab1b8e7d568733822471d2df8e815bc8e4e48cbad83
-  data.tar.gz: 95c2005f0edf3fcd191fa8f16e23d049f68eb4074cd720042eadbb982ffe7d1947b1bba9fe3737800f5252d504d0eaaed8d6ea1be51c4991231b0f2e453d8c5b
+  metadata.gz: 5b6c238c9894254531f5448dd5435603009b01c0c1fef015b748f3a91a662fa81fd0c7bbe882b95d88e8cbe91beee342222877f889931d5ed186cf0e2d03ef4e
+  data.tar.gz: 76a390dc30e53275ded279b27a8bd6ebabe4d9cefb0cc7367563467cd86fd37e6bb20440810d237137f400d72af6dd289602603ff6f70dce8f3bd6906f6d24e2

data/CHANGELOG.md CHANGED

@@ -1,3 +1,5 @@
+# 1.6.9
+  * Avoid following links when disabled by CSS (1.6.8 worked only for Javascript)
 # 1.6.8
   * Avoid following disabled links

data/lib/grell/page.rb CHANGED

@@ -43,7 +43,7 @@ module Grell
     # Number of times we have retried the current page
     def retries
-      [@times_visited -1, 0].max
+      [@times_visited - 1, 0].max
     end
     # The current URL, this may be different from the URL we asked for if there was some redirect
@@ -205,17 +205,9 @@ module Grell
       private
       def all_links
-        # <link> can only be used in the <head> as of: https://developer.mozilla.org/en/docs/Web/HTML/Element/link
-        anchors_in_body = @rawpage.all_anchors.reject { |anchor| anchor.tag_name == 'link' }
-        # Do not follow disabled links
-        enabled_links = anchors_in_body.reject { |anchor| anchor.disabled? }
-        unique_links = enabled_links.map do |anchor|
-         anchor['href'] || anchor['data-href']
-        end.compact
-        unique_links.map{|link| link_to_url(link)}.uniq.compact
+        links =  @rawpage.all_anchors.map { |anchor| Link.new(anchor) }
+        body_enabled_links = links.reject { |link| link.inside_header? || link.disabled? }
+        body_enabled_links.map { |link| link.to_url(host) }.uniq.compact
       rescue Capybara::Poltergeist::ObsoleteNode
         Grell.logger.warn "We found an obsolete node in #{@url}. Ignoring all links"
@@ -224,37 +216,54 @@ module Grell
         []
       end
-      # We only accept links in this same host that start with a path
-      # nil from this
-      def link_to_url(link)
-        uri = URI.parse(link)
-        if uri.absolute?
-          if uri.host != URI.parse(host).host
-            Grell.logger.debug "GRELL does not follow links to external hosts: #{link}"
-            nil
-          else
-            link # Absolute link to our own host
-          end
-        else
-          if uri.path.nil?
-            Grell.logger.debug "GRELL does not follow links without a path: #{uri}"
-            nil
-          end
-          if uri.path.start_with?('/')
-            host + link  #convert to full URL
-          else #links like href="google.com" the browser would go to http://google.com like "http://#{link}"
-            Grell.logger.debug "GRELL Bad formatted link: #{link}, assuming external"
-            nil
-          end
+      # Private class to group all the methods related to links.
+      class Link
+        def initialize(anchor)
+          @anchor = anchor
         end
-      rescue URI::InvalidURIError #We will have invalid links propagating till we navigate to them
-        link
-      end
-    end
+        # <link> can only be used in the <head> as of: https://developer.mozilla.org/en/docs/Web/HTML/Element/link
+        def inside_header?
+          @anchor.tag_name == 'link'
+        end
+        # Is the link disabled by either Javascript or CSS?
+        def disabled?
+          @anchor.disabled? || !!@anchor.native.attributes['disabled']
+        end
+        # Some links may use data-href + javascript to do interesting things
+        def href
+          @anchor['href'] || @anchor['data-href']
+        end
-  end
+        # We only accept links in this same host that start with a path
+        def to_url(host)
+          uri = URI.parse(href)
+          if uri.absolute?
+            if uri.host != URI.parse(host).host
+              Grell.logger.debug "GRELL does not follow links to external hosts: #{href}"
+              nil
+            else
+              href # Absolute link to our own host
+            end
+          else
+            if uri.path.nil?
+              Grell.logger.debug "GRELL does not follow links without a path: #{uri}"
+              nil
+            end
+            if uri.path.start_with?('/')
+              host + href  # convert to full URL
+            else # links like href="google.com" the browser would go to http://google.com like "http://#{link}"
+              Grell.logger.debug "GRELL Bad formatted link: #{href}, assuming external"
+              nil
+            end
+          end
+        rescue URI::InvalidURIError # Invalid links propagating till we navigate to them
+          href
+        end
+      end
+    end
+  end
 end

data/lib/grell/version.rb CHANGED

@@ -1,3 +1,3 @@
 module Grell
-  VERSION = "1.6.8".freeze
+  VERSION = "1.6.9".freeze
 end

data/spec/lib/page_spec.rb CHANGED

@@ -253,6 +253,7 @@ RSpec.describe Grell::Page do
       <a href=\"/trusmis.html\">trusmis</a>
       <a href=\"/help.html\">help</a>
       <a href=\"javascript: void(0)\">help</a>
+      <a href=\"/helpdisabled.html\" disabled=\"\">helpdisabled</a>
       </body></html>"
     end
     let(:links) { ['http://www.example.com/trusmis.html', 'http://www.example.com/help.html'] }

metadata CHANGED

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: grell
 version: !ruby/object:Gem::Version
-  version: 1.6.8
+  version: 1.6.9
 platform: ruby
 authors:
 - Jordi Polo Carres
 autorequire:
 bindir: bin
 cert_chain: []
-date: 2016-07-12 00:00:00.000000000 Z
+date: 2016-07-13 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: capybara