spec/erd_handler/label_spec.rb

   1 require 'spec_helper'
   2
   3 module ErdHandler
   4   describe Label do
   5     describe '#original' do
   6       it "reports the string it was initialised with" do
   7         test_label = "Test label"
   8         l1 = Label.new test_label
   9         l1.original.should == test_label
  10         l1 = Label.new
  11         l1.original.should == ""
  12       end
  13     end # original
  14
  15     describe '#processed' do
  16       it "reports the original if no processing has been done" do
  17         test_label = "Test label"
  18         l1 = Label.new test_label
  19         l1.processed.should == [test_label]
  20         l1.original.should == test_label
  21       end
  22     end # processed
  23
  24     describe '#split' do
  25       it "splits the original on the specified regexp" do
  26         l1 = Label.new "Test label"
  27         l1.split(/ /)
  28         l1.processed.should == ["Test", "label"]
  29         l1.original.should == "Test label"
  30
  31         l1 = Label.new "Test_label"
  32         l1.split(/_/)
  33         l1.processed.should == ["Test", "label"]
  34
  35         l1 = Label.new "Test label_string"
  36         l1.split(/[ _]/)
  37         l1.processed.should == ["Test", "label", "string"]
  38       end
  39
  40       it "splits the original on camel case" do
  41         l1 = Label.new "TestLabel"
  42         l1.split :camel_case => true
  43         l1.processed.should == ["Test", "Label"]
  44         l1.original.should == "TestLabel"
  45
  46         l2 = Label.new "testLabel"
  47         l2.split :camel_case => true
  48         l2.processed.should == ["test", "Label"]
  49         l2.original.should == "testLabel"
  50       end
  51
  52       it "doesn't split the original on camel case if asked not to" do
  53         l1 = Label.new "TestLabel"
  54         l1.split :camel_case => false
  55         l1.processed.should == ["TestLabel"]
  56         l1.original.should == "TestLabel"
  57
  58         l2 = Label.new "TestLabel"
  59         l2.split :camel_case => nil
  60         l2.processed.should == ["TestLabel"]
  61         l2.original.should == "TestLabel"
  62       end
  63
  64       it "splits the original on numbers" do
  65         l1 = Label.new "Test123Label"
  66         l1.split :numbers => true
  67         l1.processed.should == ["Test", "123", "Label"]
  68         l1.original.should == "Test123Label"
  69
  70         l2 = Label.new "test1label"
  71         l2.split :numbers => true
  72         l2.processed.should == ["test", "1", "label"]
  73         l2.original.should == "test1label"
  74       end
  75
  76       it "doesn't split the original on numbers if asked not to" do
  77         l1 = Label.new "Test123Label"
  78         l1.split :numbers => false
  79         l1.processed.should == ["Test123Label"]
  80         l1.original.should == "Test123Label"
  81
  82         l2 = Label.new "Test123Label"
  83         l2.split :numbers => nil
  84         l2.processed.should == ["Test123Label"]
  85         l2.original.should == "Test123Label"
  86       end
  87
  88       it "splits the original using a default regexp" do
  89         l1 = Label.new "Test label_string\tfred"
  90         l1.split
  91         l1.processed.should == ["Test", "label", "string", "fred"]
  92       end
  93
  94       it "splits the original on camel case by default" do
  95         l1 = Label.new "TestLabel"
  96         l1.split
  97         l1.processed.should == ["Test", "Label"]
  98         l1.original.should == "TestLabel"
  99       end
 100
 101       it "splits the original on numbers by default" do
 102         l1 = Label.new "Test123Label"
 103         l1.split
 104         l1.processed.should == ["Test", "123", "Label"]
 105         l1.original.should == "Test123Label"
 106       end
 107
 108       it "splits the original on punctuation, whitespace, camel case, and numbers by default" do
 109         l1 = Label.new "TestLabel is_split, 123 he,said456Fred"
 110         l1.split
 111         l1.processed.should == ["Test", "Label", "is", "split","123", "he", "said", "456", "Fred"]
 112         l1.original.should == "TestLabel is_split, 123 he,said456Fred"
 113       end
 114
 115       it "is idempotent" do
 116         l1 = Label.new "TestLabel is_split, 123 he,said456Fred"
 117         res1 = l1.split.dup
 118         res2 = l1.split
 119         res1.processed.should == res2.processed
 120         l1.original.should == "TestLabel is_split, 123 he,said456Fred"
 121       end
 122     end # split
 123
 124     describe "#downcase" do
 125       it "downcases all parts of the processed label" do
 126         l1 = Label.new "Test label_string"
 127         l1.split.downcase
 128         l1.processed.should == ["test", "label", "string"]
 129       end
 130     end # downcase
 131
 132     describe "#stem" do
 133       it "stems all parts of the processed label" do
 134         l1 = Label.new "testing labeller string pontificated"
 135         l1.split.stem
 136         l1.processed.should == ["test", "label", "string", "pontif"]
 137       end
 138     end # stem
 139
 140     describe "#tidy" do
 141       it "tidies a label" do
 142         l1 = Label.new "testingLabeller string, he_pontificated"
 143         l2 = Label.new l1.original
 144         l1.tidy
 145         l2.split.downcase.stem
 146         l1.processed.should == l2.processed
 147       end
 148     end # tidy
 149
 150     describe "#levenshtein" do
 151       it "calculates the Levenshtein distance of the processed string" do
 152         l1 = Label.new "Fred"
 153         l1.levenshtein("Fred").should == 0
 154         l1.levenshtein("Free").should == 1
 155         l1.levenshtein("").should == 4
 156         l2 = Label.new
 157         l2.levenshtein("Free").should == 4
 158         l2.levenshtein("").should == 0
 159         l3 = Label.new "meilenstein"
 160         l3.levenshtein("levenshtein").should == 4
 161         l4 = Label.new "testingLabeller string, he_pontificated"
 162         l4.tidy.levenshtein("testlabelstringhepontif").should == 0
 163         l4.tidy.levenshtein("testlabelXstringhepontif").should == 1
 164       end
 165
 166       it "calculates the Levenshtein distance between Labels" do
 167         l1 = Label.new "meilenstein"
 168         l2 = Label.new "levenshtein"
 169         l1.levenshtein(l2).should == 4
 170       end
 171     end
 172
 173   end
 174 end