url_canonicalize 0.0.5 → 0.0.6

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA1:
-  metadata.gz: 219d397a462ff3656b983b11f219de82331c3bdd
-  data.tar.gz: 54a1b7916e2ed9bbc5bd0303a5f10a6d8bcd6cd8
+  metadata.gz: 6e3ce5f5d168512959c94e6af9a6482681e46ca9
+  data.tar.gz: e09877ca0b7488e62b1aca3d29923ed0ebf56f23
 SHA512:
-  metadata.gz: 2c1bace965b619a010d4ebecfe27960f5e09dfd71d26de2f06083eae2a5967a7f582bc5a0ca2012827c630f44e980aa368ff04e074220501a52f3ab45cc4b6fa
-  data.tar.gz: f73b70de8512cc161fdf89462b053e11dba67b4afecb048b7e06cc69067bb58cbb5d4353319958cf731306bae14ff7f8d39f193d1b27d8b3a25e9ca5bd7d027a
+  metadata.gz: c39e085b024f763ce99d37be169e7d979bf160d347e46254eab572fdc033f01c5cf490cda18cbff5961c3e62de6793548884d80b0bd83cd6fa06134ca16cac76
+  data.tar.gz: c6828ce1595c250313ad52d746ce8b337a5a4066c8c1e94af18af4bfd2434dbd991111e3ce2ad3fc62c3b3f69bde6d36b32290d2aa11548fc7d22abfd8cc5937

data/.rubocop.yml

@@ -2,6 +2,7 @@
 AllCops:
   Exclude:
     - '*.gemspec'
+    - 'spec/**/*'
 
 Style/TrailingCommaInArguments:
   EnforcedStyleForMultiline: no_comma
@@ -52,4 +53,4 @@ Lint/LiteralInInterpolation:
 
 Metrics/ClassLength:
   CountComments: false  # count full line comments?
-  Max: 120
+  Max: 150

data/lib/url_canonicalize/request.rb

@@ -47,12 +47,15 @@ module URLCanonicalize
     end
 
     def handle_redirection
+      puts response['location'] # debug
+
       case response
       when Net::HTTPFound, Net::HTTPMovedTemporarily, Net::HTTPTemporaryRedirect
         self.http_method = :get
         handle_success
       else
-        URLCanonicalize::Response::Redirect.new(response['location'])
+        location = relative_to_absolute(response['location'])
+        URLCanonicalize::Response::Redirect.new(location)
       end
     end
 
@@ -61,6 +64,8 @@ module URLCanonicalize
     end
 
     def enhanced_response
+      puts canonical_url # debug
+
       if canonical_url
         response_plus = URLCanonicalize::Response::Success.new(canonical_url, response, html)
         URLCanonicalize::Response::CanonicalFound.new(canonical_url, response_plus)
@@ -73,14 +78,18 @@ module URLCanonicalize
       @html ||= Nokogiri::HTML response.body
     end
 
-    def canonical_url_element
-      @canonical_url_element ||= html.xpath('//head/link[@rel="canonical"]').first
+    def canonical_url
+      @canonical_url ||= relative_to_absolute(canonical_url_raw)
     end
 
-    def canonical_url
+    def canonical_url_raw
       @canonical_url ||= canonical_url_element['href'] if canonical_url_element.is_a?(Nokogiri::XML::Element)
     end
 
+    def canonical_url_element
+      @canonical_url_element ||= html.xpath('//head/link[@rel="canonical"]').first
+    end
+
     def uri
       @uri ||= http.uri
     end
@@ -135,6 +144,20 @@ module URLCanonicalize
       @http_method = :get if host =~ /(linkedin|crunchbase).com/
     end
 
+    def relative_to_absolute(partial_url)
+      return unless partial_url
+      partial_uri = ::URI.parse(partial_url)
+
+      if partial_uri.host
+        partial_url # It's already absolute
+      else
+        base_uri = uri.dup || ::URI.parse(url)
+        base_uri.path = partial_url
+        puts base_uri.to_s # debug
+        base_uri.to_s
+      end
+    end
+
     NETWORK_EXCEPTIONS = [
       EOFError,
       Errno::ECONNREFUSED,

data/lib/url_canonicalize/version.rb

@@ -1,3 +1,3 @@
 module URLCanonicalize
-  VERSION = '0.0.5'.freeze
+  VERSION = '0.0.6'.freeze
 end

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: url_canonicalize
 version: !ruby/object:Gem::Version
-  version: 0.0.5
+  version: 0.0.6
 platform: ruby
 authors:
 - Dominic Sayers