spec/postrank-uri_spec.rb in postrank-uri-1.0.17 vs spec/postrank-uri_spec.rb in postrank-uri-1.0.18

@@ -27,20 +27,24 @@
 
     context "accept improperly escaped PostRank::URI strings" do
       # See http://tools.ietf.org/html/rfc3986#section-2.3
 
       it "should unescape PostRank::URI with spaces encoded as '+'" do
-        PostRank::URI.unescape('id=+1').should == 'id= 1'
+        PostRank::URI.unescape('?id=+1').should == '?id= 1'
       end
 
       it "should unescape PostRank::URI with spaces encoded as '+'" do
-        PostRank::URI.unescape('id%3D+1').should == 'id= 1'
+        PostRank::URI.unescape('?id%3D+1').should == '?id= 1'
       end
 
       it "should unescape PostRank::URI with spaces encoded as %20" do
-        PostRank::URI.unescape('id=%201').should == 'id= 1'
+        PostRank::URI.unescape('?id=%201').should == '?id= 1'
       end
+
+      it "should not unescape '+' to spaces in paths" do
+        PostRank::URI.unescape('/foo+bar?id=foo+bar').should == '/foo+bar?id=foo bar'
+      end
     end
 
   end
 
   context "normalize" do
@@ -98,11 +102,11 @@
 
   end
 
   context "canonicalization" do
     def c(uri)
-      PostRank::URI.c18n(uri).to_s
+      PostRank::URI.c14n(uri).to_s
     end
 
     context "query parameters" do
       it "should handle nester parameters" do
         c('igvita.com/?id=a&utm_source=a').should == 'http://igvita.com/?id=a'
@@ -120,10 +124,16 @@
 
       it "should remove awesm/sms parameters" do
         c('igvita.com/?id=a&utm_source=a&awesm=b').should == 'http://igvita.com/?id=a'
         c('igvita.com/?id=a&sms_ss=a').should == 'http://igvita.com/?id=a'
       end
+
+      it "should remove PHPSESSID parameter" do
+        c('http://www.nachi.org/forum?PHPSESSID=9ee2fb10b7274ef2b15d1d4006b8c8dd').should == 'http://www.nachi.org/forum?'
+        c('http://www.nachi.org/forum/?PHPSESSID=9ee2fb10b7274ef2b15d1d4006b8c8dd').should == 'http://www.nachi.org/forum/?'
+        c('http://www.nachi.org/forum?id=123&PHPSESSID=9ee2fb10b7274ef2b15d1d4006b8c8dd').should == 'http://www.nachi.org/forum?id=123'
+      end
     end
 
     context "hashbang" do
       it "should rewrite twitter links to crawlable versions" do
         c('http://twitter.com/#!/igrigorik').should == 'http://twitter.com/igrigorik'
@@ -160,11 +170,11 @@
   context "clean" do
     def c(uri)
       PostRank::URI.clean(uri)
     end
 
-    it "should unescape, c18n and normalize" do
+    it "should unescape, c14n and normalize" do
       c('http://igvita.com/?id=1').should == 'http://igvita.com/?id=1'
       c('igvita.com/?id=1').should == 'http://igvita.com/?id=1'
 
       c('http://igvita.com/?id= 1').should == 'http://igvita.com/?id=%201'
       c('http://igvita.com/?id=+1').should == 'http://igvita.com/?id=%201'
@@ -177,11 +187,11 @@
 
       c('test.tumblr.com/post/23223/text-stub').should == 'http://test.tumblr.com/post/23223'
     end
 
     it "should clean host specific parameters" do
-      YAML.load_file('spec/c18n_hosts.yml').each do |orig, clean|
+      YAML.load_file('spec/c14n_hosts.yml').each do |orig, clean|
         c(orig).should == clean
       end
     end
   end
 
@@ -347,8 +357,16 @@
       PostRank::URI.valid?('http://www.test.c').should be_false
     end
 
     it 'marks www.test.com as valid' do
       PostRank::URI.valid?('http://www.test.com').should be_true
+    end
+
+    it 'marks Unicode domain as valid (NOTE: works only with a scheme)' do
+      PostRank::URI.valid?('http://президент.рф').should be_true
+    end
+
+    it 'marks punycode domain domain as valid' do
+      PostRank::URI.valid?('xn--d1abbgf6aiiy.xn--p1ai').should be_true
     end
   end
 end