textract 0.0.9 → 0.0.10

data/spec/lib/textract_spec.rb

@@ -15,7 +15,7 @@ describe Textract do
       url = "http://gawker.com/1694508525"
       article = Textract.get_text(url)
       expect(article.text.include?("Import")).to eq true
-      expect(article.md5).to eq "c11a810a3e73f24aac78fd3e39e69f87"
+      expect(article.md5).to eq "9cc00fcdeb4bc41e0649d0776cbb2157"
       expect(article.author).to eq "Hamilton Nolan"
     end
   end
@@ -29,6 +29,19 @@ describe Textract do
     end
   end
 
+  it "returns the canonical url if available" do
+    VCR.use_cassette("hamno") do
+      url = "http://gawker.com/1694508525"
+      article = Textract.get_text(url)
+      expect(article.url).to eq "http://gawker.com/there-are-no-candidates-for-the-middle-class-1694508525"
+    end
+    VCR.use_cassette("buzzfeed hash") do
+      url = "http://www.buzzfeed.com/katenocera/rand-paul-is-on-his-own-this-time#.sseGm85KG"
+      article = Textract.get_text(url)
+      expect(article.url).to eq "http://www.buzzfeed.com/katenocera/rand-paul-is-on-his-own-this-time"
+    end
+  end
+
   it "returns article text based on opengraph description" do
     VCR.use_cassette('og') do
       url = "http://www.tedcruz.org/record/our-standard-the-constitution/"

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: textract
 version: !ruby/object:Gem::Version
-  version: 0.0.9
+  version: 0.0.10
 platform: ruby
 authors:
 - Adam Pash
 autorequire: 
 bindir: bin
 cert_chain: []
-date: 2015-04-27 00:00:00.000000000 Z
+date: 2015-05-21 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: opengraph_parser
@@ -166,6 +166,7 @@ files:
 - lib/textract.rb
 - lib/textract/version.rb
 - spec/fixtures/vcr_cassettes/bad_frisky.yml
+- spec/fixtures/vcr_cassettes/buzzfeed_hash.yml
 - spec/fixtures/vcr_cassettes/cruz.yml
 - spec/fixtures/vcr_cassettes/hamno.yml
 - spec/fixtures/vcr_cassettes/imgs.yml
@@ -201,6 +202,7 @@ specification_version: 4
 summary: Extracts article text from a URL
 test_files:
 - spec/fixtures/vcr_cassettes/bad_frisky.yml
+- spec/fixtures/vcr_cassettes/buzzfeed_hash.yml
 - spec/fixtures/vcr_cassettes/cruz.yml
 - spec/fixtures/vcr_cassettes/hamno.yml
 - spec/fixtures/vcr_cassettes/imgs.yml