Rewrite ExCoder for 0.15, use tail call recursion

README.md

@@ -4,4 +4,4 @@ about the mess of parsing HTML) and I did not want to use a full XML library
 from the Erlang side.
 
 ExCoder supports both decoding named and numbered HTML entities. Encoding will
-always use the hexadecimal numbered format.
+use named entities where possible and fall back to hexadecimal encoded entities.

lib/excoder.ex

@@ -4,12 +4,11 @@ defmodule ExCoder do
     @max_codepoint 1114111
 
     # Regex for matching numeric entities
-    @numeric_decode_regex %R/^(?:\d+|x[\da-f]+);/i
+    @numeric_decode_regex ~R/^(?:\d+|x[\da-f]+);/i
 
     codec_table = [
         { "	", "  " },
-        {"
", "
-"},
+        { "
", "\n" },
         { "!", "!" },
         { """, "\"" },
         { """, "\"" },
@@ -2045,24 +2044,33 @@ defmodule ExCoder do
 
     # DECODING
 
+    def decode(str), do: decode(str, "")
+
     # Generate a decoding function for each element in the replacement table above
-    lc { entity, char } inlist codec_table do
-        def decode(<< unquote(entity), rest :: binary >>), do: unquote(char) <> decode rest
+    for {entity, char} <- codec_table do
+        def decode(<< unquote(entity), rest :: binary >>, out) do
+            decode rest, out <> unquote(char)
+        end
     end
 
-    def decode(""), do: ""
-
-    def decode(<< "&#", rest :: binary >>) do
+    def decode(<< "&#", rest :: binary >>, out) do
         if Regex.match? @numeric_decode_regex, rest do
-            decode rest, :numericmode
+            decode rest, out, :numericmode
         else
-            "&#" <> decode rest
+            decode rest, out <> "&#"
         end
     end
 
-    def decode(str), do: String.first(str) <> decode String.slice str, 1, String.length str
+    def decode("", out), do: out
+
+    # Just walk through any unrecognized characters
+    def decode(str, out) do
+        decode String.slice(str, 1, String.length str), out <> String.first(str)
+    end
 
 
+    # Decode a numeric character or return the original entity if the given
+    # number is out of range
     def decode(numeric, :numericmode, original) when is_number numeric do
         if numeric <= @max_codepoint do
             << numeric :: utf8 >>
@@ -2071,14 +2079,14 @@ defmodule ExCoder do
         end
     end
 
-    def decode(<< "x", rest :: binary >>, :numericmode) do
+    def decode(<< "x", rest :: binary >>, out, :numericmode) do
         [numeric | rest] = strip_numeric rest
-        decode(binary_to_integer(numeric, 16), :numericmode, "x" <> numeric) <> decode(Enum.join rest, ";")
+        decode Enum.join(rest, ";"), out <> decode(String.to_integer(numeric, 16), :numericmode, "x" <> numeric)
     end
 
-    def decode(str, :numericmode) do
+    def decode(str, out, :numericmode) do
         [numeric | rest] = strip_numeric str
-        decode(binary_to_integer(numeric), :numericmode, numeric) <> decode(Enum.join rest, ";")
+        decode Enum.join(rest, ";"), out <> decode(String.to_integer(numeric), :numericmode, numeric)
     end
 
 
@@ -2086,18 +2094,38 @@ defmodule ExCoder do
 
     # ENCODING
 
-    # Skip newline when encoding
-    def encode(<< "
-", rest :: binary >>), do: "
-" <> encode rest 
+    # Remove duplicate elements from replacement table when encoding
+    {encode_table, _} = Enum.reduce(codec_table, {[], HashSet.new},
+        fn ({entity, char}, {acc, passed_chars}) ->
+            unless Set.member?(passed_chars, char) do
+                passed_chars = Set.put passed_chars, char
+                {acc ++ [{entity, char}], passed_chars}
+            else
+                {acc, passed_chars}
+            end
+        end
+    )
 
     # Replace anything on the codec table with the named entity version
-    lc { entity, char } inlist codec_table do
-        def encode(<< unquote(char), rest :: binary >>), do: unquote(entity) <> encode rest
+    for {entity, char} <- encode_table do
+        # Except skip newlines
+        if char != "\n" do
+            def encode(<< unquote(char), rest :: binary >>, out) do
+                encode rest, out <> unquote(entity)
+            end
+        else
+            def encode(<< "\n", rest :: binary >>, out) do
+                encode rest, out <> "\n"
+            end
+        end
     end
 
-    def encode(<< codepoint :: utf8, rest :: binary >>) do
-        encode(codepoint) <> encode rest
+    def encode(<< codepoint :: utf8, rest :: binary >>, out) do
+        encode rest, out <> encode(codepoint)
+    end
+
+    def encode("", out) do
+        out
     end
 
     # Encode a-z as is
@@ -2106,18 +2134,16 @@ defmodule ExCoder do
     end
 
     def encode(codepoint) when is_number codepoint do
-        "&#x" <> integer_to_binary(codepoint, 16) <> ";"
+        "&#x" <> Integer.to_string(codepoint, 16) <> ";"
     end
 
-    def encode("") do
-        ""
-    end
+    def encode(str), do: encode(str, "")
 
 
 
 
     defp strip_numeric(str) do
-        String.split(str, ";")
+        String.split str, ";"
     end
 
 end

mix.exs

@@ -3,8 +3,8 @@ defmodule Excoder.Mixfile do
 
   def project do
     [ app: :excoder,
-      version: "1.1.0",
-      elixir: "~> 0.11",
+      version: "1.2.0",
+      elixir: "~> 0.15",
       deps: deps ]
   end
 

test/excoder_test.exs

@@ -23,6 +23,10 @@ defmodule ExcoderTest do
         assert(ExCoder.decode("&#453g3;") == "&#453g3;")
     end
 
+    test "Decode numeric that is too big" do
+        assert(ExCoder.decode("�") == "�")
+    end
+
     test "Decode HTML" do
         assert(ExCoder.decode("""
 <div class='line' id='LC9'>&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;<span class="ss">address:</span> <span class="s2">&quot;chat.eu.freenode.net&quot;</span><span class="p">,</span></div><div class='line' id='LC10'>&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;<span class="ss">
@@ -45,16 +49,8 @@ defmodule ExcoderTest do
         assert(ExCoder.encode("foo and bar") == "foo and bar")
     end
 
-    test "Encode linebreak" do
-        assert(ExCoder.encode("
-foo
-and
-bar
-") == "
-foo
-and
-bar
-")
+    test "Don't encode linebreaks" do
+        assert(ExCoder.encode("\nfoo\nand\nbar\n") == "\nfoo\nand\nbar\n")
     end
 
     test "Encode scandinavian characters" do