name-spotter_spec.rb in name-spotter-0.2.2

- old
+ new

@@ -8,10 +8,25 @@
     @neti = NameSpotter.new(neti_neti)
     @tf = NameSpotter.new(taxon_finder)
     @clients = [@neti, @tf]
   end
 
+  it "should find if a text is in english" do 
+    eng = open(File.join(File.dirname(__FILE__), 'files', 'english.txt'), 'r:utf-8').read
+    eng2 = open(File.join(File.dirname(__FILE__), 'files', 'english.txt'), 'r:utf-8').read
+    eng3 = open(File.join(File.dirname(__FILE__), 'files', 'journalofentomol13pomo_0018.txt'), 'r:utf-8').read
+    eng3 = open(File.join(File.dirname(__FILE__), 'files', 'journalofentomol13pomo_0063.txt'), 'r:utf-8').read
+    
+    not_eng = open(File.join(File.dirname(__FILE__), 'files', 'not_english.txt'), 'r:utf-8').read
+    100.times do
+      NameSpotter.english?(eng).should be_true
+      NameSpotter.english?(eng2).should be_true
+      NameSpotter.english?(eng3).should be_false
+      NameSpotter.english?(not_eng).should be_false
+    end
+  end
+
   it "should exist" do
     @neti.is_a?(NameSpotter).should be_true
     @tf.is_a?(NameSpotter).should be_true
   end
 
@@ -60,10 +75,12 @@
     res = @tf.find(text)[:names].map { |n| n[:scientificName] } 
     res.should == ['Xysticus canadensis', 'Pardosa moesta', 'X[ysticus] canadensis']
   end
 
   it "should get back correct names using offsets in utf-8 based text" do
-    text = "A text with multibyte characters नेति नेति:  Some text that has Betula\n alba and Mus musculus and \neven B. alba and even M. mus-\nculus. Also it has name unknown before: Varanus bitatawa species"
+    # this test depends on netineti tornado server, not on namespotter itself. Go and fix that!
+    # the issue and the fix: https://github.com/mbl-cli/NetiNeti/pull/1
+    text = "\r\r\n>':¥/. \r\nA text with multibyte characters नेति नेति:  Some text that has Betula\n alba and Mus musculus and \neven B. alba and even M. mus-\nculus. Also it has name unknown before: Varanus bitatawa species"
     res = @neti.find(text)[:names]
     res.map do |name|
       verbatim = name[:verbatim]
       found_name = text[name[:offsetStart]..name[:offsetEnd]]
       found_name.should == verbatim