remove trailing slash on paths

postrank-labs · Mar 31, 2011 · a92af06 · a92af06
1 parent 555e7ae
commit a92af06
Show file tree

Hide file tree

Showing 4 changed files with 22 additions and 10 deletions.
diff --git a/lib/postrank-uri.rb b/lib/postrank-uri.rb
@@ -125,6 +125,7 @@ def hash(uri)
     def normalize(uri)
       u = parse(uri)
       u.path = u.path.squeeze('/')
+      u.path = u.path.chomp('/') if u.path.size != 1
       u.query = nil if u.query && u.query.empty?
       u.fragment = nil
       u

diff --git a/lib/postrank-uri/version.rb b/lib/postrank-uri/version.rb
@@ -1,5 +1,5 @@
 module PostRank
   module URI
-    VERSION = "1.0.7"
+    VERSION = "1.0.8"
   end
 end
diff --git a/spec/c18n_hosts.yml b/spec/c18n_hosts.yml
@@ -12,7 +12,7 @@
   - http://www.nytimes.com/2011/03/15/business/media/15adco.html
 
 - - http://networkeffect.allthingsd.com/20110308/googles-approach-to-social/?mod=tweet
-  - http://networkeffect.allthingsd.com/20110308/googles-approach-to-social/
+  - http://networkeffect.allthingsd.com/20110308/googles-approach-to-social
 
 - - http://online.wsj.com/article/SB10001424052748704657704576150191661959856.html?mod=WSJ_hp_LEFTWhatsNewsCollection
   - http://online.wsj.com/article/SB10001424052748704657704576150191661959856.html
@@ -21,7 +21,7 @@
   - http://diepresse.com/home/wirtschaft/636448/Griechenland_Drachme-als-letzte-Rettung
 
 - - http://dotearth.blogs.nytimes.com/2010/12/14/beyond-political-science/?partner=rss&emc=rss
-  - http://dotearth.blogs.nytimes.com/2010/12/14/beyond-political-science/
+  - http://dotearth.blogs.nytimes.com/2010/12/14/beyond-political-science
 
 - - http://www.washingtonpost.com/wp-dyn/content/article/2010/12/14/AR2010121406045.html?nav=rss_email/components
   - http://www.washingtonpost.com/wp-dyn/content/article/2010/12/14/AR2010121406045.html
@@ -48,7 +48,7 @@
   - http://www.dw-world.de/dw/article/0,,6330472,00.html
 
 - - http://www.repubblica.it/rubriche/il-caso-del-giorno/2010/12/13/news/riscossa_aeffe-10153565/?rss
-  - http://www.repubblica.it/rubriche/il-caso-del-giorno/2010/12/13/news/riscossa_aeffe-10153565/
+  - http://www.repubblica.it/rubriche/il-caso-del-giorno/2010/12/13/news/riscossa_aeffe-10153565
 
 - - http://www.welt.de/sport/Der-Hoellenritt-des-Fussball-Profis-Jean-Marc-Bosman.html?wtmc=RSS.Sport.Fussball
   - http://www.welt.de/sport/Der-Hoellenritt-des-Fussball-Profis-Jean-Marc-Bosman.html

diff --git a/spec/postrank-uri_spec.rb b/spec/postrank-uri_spec.rb
@@ -86,6 +86,16 @@ def n(uri)
       n('IGVITA.COM/ABC').should == (igvita + "ABC")
     end
 
+    it "should remove trailing slash on paths" do
+      n('http://igvita.com/').should == 'http://igvita.com/'
+
+      n('http://igvita.com/a').should == 'http://igvita.com/a'
+      n('http://igvita.com/a/').should == 'http://igvita.com/a'
+
+      n('http://igvita.com/a/b').should == 'http://igvita.com/a/b'
+      n('http://igvita.com/a/b/').should == 'http://igvita.com/a/b'
+    end
+
   end
 
   context "canonicalization" do
@@ -125,7 +135,7 @@ def c(uri)
     context "embedded links" do
       it "should extract embedded redirects from Google News" do
         u = c('http://news.google.com/news/url?sa=t&fd=R&&url=http://www.ctv.ca/CTVNews/Politics/20110111/')
-        u.should == 'http://www.ctv.ca/CTVNews/Politics/20110111/'
+        u.should == 'http://www.ctv.ca/CTVNews/Politics/20110111'
       end
 
       it "should extract embedded redirects from xfruits.com" do
@@ -135,7 +145,7 @@ def c(uri)
 
       it "should extract embedded redirects from MySpace" do
         u = c('http://www.myspace.com/Modules/PostTo/Pages/?u=http%3A%2F%2Fghanaian-chronicle.com%2Fnews%2Fother-news%2Fcanadian-high-commissioner-urges-media%2F&t=Canadian%20High%20Commissioner%20urges%20media')
-        u.should == 'http://ghanaian-chronicle.com/news/other-news/canadian-high-commissioner-urges-media/'
+        u.should == 'http://ghanaian-chronicle.com/news/other-news/canadian-high-commissioner-urges-media'
       end
     end
   end
@@ -172,11 +182,12 @@ def h(uri)
     end
 
     it "should compute MD5 hash of the normalized URI" do
-      hash = '021a1032b1ea631a7c33d1a0ccc562bf'
+      hash = '55fae8910d312b7878a3201ed653b881'
 
-      h('http://EverBurnign.Com/feed/post/1').should == hash
-      h('Everburnign.com/feed/post/1').should == hash
-      h('everburnign.com/feed/post/1').should == hash
+      h('http://EverBurning.Com/feed/post/1').should == hash
+      h('Everburning.com/feed/post/1').should == hash
+      h('everburning.com/feed/post/1').should == hash
+      h('everburning.com/feed/post/1/').should == hash
     end
   end