spec/unibits_spec.rb in unibits-2.0.0 vs spec/unibits_spec.rb in unibits-2.1.0

@@ -65,28 +65,68 @@
       result.must_match "C"
       result.must_match "43"
       result.must_match "01000011"
     end
 
-    it "works with 'ISO-8859-' encodings" do
+    it "works with 'ISO-8859-X' encodings" do
       string = "\xBC Idiosyncr\xE4tic\n\x91".force_encoding("ISO-8859-1")
       result = Paint.unpaint(Unibits.visualize(string))
       result.must_match "BC"  # ¼
       result.must_match "E4"  # ä
       result.must_match "␊"   # \n
       result.must_match "PU1" # C1 name for \x91
     end
 
-    it "works with 'Windows-125' encodings" do
+    it "works with 'Windows-125X' encodings" do
       string = "\xBC Idiosyncr\xE4tic\n\x81".force_encoding("Windows-1252")
       result = Paint.unpaint(Unibits.visualize(string))
       result.must_match "BC"  # ¼
       result.must_match "E4"  # ä
       result.must_match "␊"   # \n
       result.must_match "n/a" # \x81 is not assigned
     end
 
+    it "works with 'IBMX' encodings" do
+      string = "\xFE Idiosyncr\x84tic\n".force_encoding("IBM437")
+      result = Paint.unpaint(Unibits.visualize(string))
+      result.must_match "FE"  # ■
+      result.must_match "84"  # ä
+      result.must_match "␊"   # \n
+    end
+
+    it "works with 'CP85X' encodings" do
+      string = "\xFE Idiosyncr\x84tic\n".force_encoding("CP850")
+      result = Paint.unpaint(Unibits.visualize(string))
+      result.must_match "FE"  # ■
+      result.must_match "84"  # ä
+      result.must_match "␊"   # \n
+    end
+
+    it "works with 'macX' encodings" do
+      string = "\xBD Idiosyncr\x8Atic \x11 \xF0\n".force_encoding("macRoman")
+      result = Paint.unpaint(Unibits.visualize(string))
+      result.must_match "BD"   # Ω
+      result.must_match "8A"   # ä
+      result.must_match "Logo" # \xF0
+      result.must_match "⌘"    # \x11
+      result.must_match "␊"    # \n
+    end
+
+    it "works with 'TIS-620/Windows-874' encodings" do
+      string = "\xA4 Idiosyncratic\n".force_encoding("TIS-620")
+      result = Paint.unpaint(Unibits.visualize(string))
+      result.must_match "A4"  # ค
+      result.must_match "␊"   # \n
+    end
+
+    it "works with 'KOI8-X' encodings" do
+      string = "\xE9\xE4\xE9\xEF\xF3\xF9\xEE\xE3\xF2\xE1\xF4\xE9\xE3\n".force_encoding("KOI8-R")
+      result = Paint.unpaint(Unibits.visualize(string))
+      result.must_match "F9"  # Ы
+      result.must_match "␊"   # \n
+    end
+
     describe "invalid UTF-8 encodings" do
       it "- unexpected continuation byte (1/2)" do
         string = "abc\x80efg"
         result = Paint.unpaint(Unibits.visualize(string))
         result.must_match "�"
@@ -216,9 +256,25 @@
         string = "🌫".b[0..-2].force_encoding("UTF-32LE")
         result = Paint.unpaint(Unibits.visualize(string))
         result.must_match "incompl."
         result.must_match "�"
       end
+
+      # TODO implement when https://bugs.ruby-lang.org/issues/13292 is released
+
+      # it "- too large codepoint" do
+      #   string = "\x00\x00\x11\x00".force_encoding("UTF-32LE")
+      #   result = Paint.unpaint(Unibits.visualize(string))
+      #   result.must_match "�"
+      #   result.must_match /toolarge.*toolarge.*toolarge.*toolarge/m
+      # end
+
+      # it "- has surrogate" do
+      #   string = "\x00\xD8\x00\x00".force_encoding("UTF-32LE")
+      #   result = Paint.unpaint(Unibits.visualize(string))
+      #   result.must_match "�"
+      #   result.must_match "sur.gate"
+      # end
     end
 
     describe "invalid ASCII encodings" do
       it "- contains bytes with 8th bit set" do
         string = "abc\x80efg".force_encoding("ASCII")