Properly populate generic transcoder list

jruby · Dec 17, 2017 · 99c9098 · 99c9098
1 parent fdde919
commit 99c9098
Showing 2 changed files with 5 additions and 33 deletions.
diff --git a/scripts/generate.rb b/scripts/generate.rb
@@ -1,6 +1,6 @@
 # coding: utf-8
 
-REPO_PATH = ARGV.first || '/usr/src/ruby-2.4.2' # path to ruby repo
+REPO_PATH = ARGV.first || '/usr/src/ruby-2.4.3' # path to ruby repo
 SECTION_NAME = "rdata"
 UNICODE_VERSION = "9.0.0"
 SRC_DIR = "../src/org/jcodings"
@@ -96,9 +96,10 @@ def generate_transcoder_list
             src, dst, tree_start, table_info, iul, max_in, max_out, conv, state_size, state_init, state_fini, *funcs = body
             tree_start = trans_src[/#define\s+#{tree_start}\s+WORDINDEX2INFO\((\d+)\)/, 1].to_i << 2
             state_size = "0" if state_size == "sizeof(struct from_utf8_mac_status)"
+            generic = funcs.all?{|f|f == "NULL"}
 
-            generic_list << [src, dst, tree_start, "\"#{name}\"", iul, max_in, max_out, "AsciiCompatibility.#{conv.split('_').last.upcase}", state_size]
-            transcoder_list << [src, dst, t_name, !funcs.all?{|f|f == "NULL"}]
+            generic_list << [src, dst, tree_start, "\"#{name}\"", iul, max_in, max_out, "AsciiCompatibility.#{conv.split('_').last.upcase}", state_size] if generic
+            transcoder_list << [src, dst, t_name, !generic]
         end
 
     end

diff --git a/src/org/jcodings/transcode/TranscoderList.java b/src/org/jcodings/transcode/TranscoderList.java
@@ -251,8 +251,6 @@ static void load() {
         new GenericTranscoderEntry("UTF-8", "UTF8-SoftBank", 24060, "Emoji", 1, 4, 6, AsciiCompatibility.CONVERTER, 0),
         new GenericTranscoderEntry("stateless-ISO-2022-JP-KDDI", "UTF8-KDDI", 33996, "EmojiIso2022Kddi", 1, 3, 1, AsciiCompatibility.CONVERTER, 0),
         new GenericTranscoderEntry("UTF8-KDDI", "stateless-ISO-2022-JP-KDDI", 76404, "EmojiIso2022Kddi", 1, 4, 3, AsciiCompatibility.CONVERTER, 0),
-        new GenericTranscoderEntry("ISO-2022-JP-KDDI", "stateless-ISO-2022-JP-KDDI", 56, "EmojiIso2022Kddi", 1, 3, 3, AsciiCompatibility.DECODER, 1),
-        new GenericTranscoderEntry("stateless-ISO-2022-JP-KDDI", "ISO-2022-JP-KDDI", 108, "EmojiIso2022Kddi", 1, 3, 5, AsciiCompatibility.ENCODER, 1),
         new GenericTranscoderEntry("SJIS-DOCOMO", "UTF8-DOCOMO", 39616, "EmojiSjisDocomo", 1, 2, 1, AsciiCompatibility.CONVERTER, 0),
         new GenericTranscoderEntry("UTF8-DOCOMO", "SJIS-DOCOMO", 84704, "EmojiSjisDocomo", 1, 4, 2, AsciiCompatibility.CONVERTER, 0),
         new GenericTranscoderEntry("SJIS-KDDI", "UTF8-KDDI", 39616, "EmojiSjisKddi", 1, 2, 1, AsciiCompatibility.CONVERTER, 0),
@@ -262,21 +260,8 @@ static void load() {
         new GenericTranscoderEntry("", "amp_escape", 8, "Escape", 1, 1, 5, AsciiCompatibility.CONVERTER, 0),
         new GenericTranscoderEntry("", "xml_text_escape", 32, "Escape", 1, 1, 5, AsciiCompatibility.CONVERTER, 0),
         new GenericTranscoderEntry("", "xml_attr_content_escape", 60, "Escape", 1, 1, 6, AsciiCompatibility.CONVERTER, 0),
-        new GenericTranscoderEntry("", "xml_attr_quote", 72, "Escape", 1, 1, 7, AsciiCompatibility.ENCODER, 1),
-        new GenericTranscoderEntry("GB18030", "UTF-8", 57668, "Gb18030", 1, 4, 4, AsciiCompatibility.CONVERTER, 0),
-        new GenericTranscoderEntry("UTF-8", "GB18030", 115576, "Gb18030", 1, 4, 4, AsciiCompatibility.CONVERTER, 0),
         new GenericTranscoderEntry("GBK", "UTF-8", 89284, "Gbk", 1, 2, 1, AsciiCompatibility.CONVERTER, 0),
         new GenericTranscoderEntry("UTF-8", "GBK", 182912, "Gbk", 1, 4, 2, AsciiCompatibility.CONVERTER, 0),
-        new GenericTranscoderEntry("ISO-2022-JP", "stateless-ISO-2022-JP", 56, "Iso2022", 1, 3, 3, AsciiCompatibility.DECODER, 1),
-        new GenericTranscoderEntry("stateless-ISO-2022-JP", "ISO-2022-JP", 108, "Iso2022", 1, 3, 5, AsciiCompatibility.ENCODER, 1),
-        new GenericTranscoderEntry("stateless-ISO-2022-JP", "EUC-JP", 128, "Iso2022", 1, 3, 2, AsciiCompatibility.CONVERTER, 0),
-        new GenericTranscoderEntry("EUC-JP", "stateless-ISO-2022-JP", 192, "Iso2022", 1, 3, 3, AsciiCompatibility.CONVERTER, 0),
-        new GenericTranscoderEntry("CP50220", "cp51932", 244, "Iso2022", 1, 3, 3, AsciiCompatibility.DECODER, 1),
-        new GenericTranscoderEntry("CP50221", "cp51932", 244, "Iso2022", 1, 3, 3, AsciiCompatibility.DECODER, 1),
-        new GenericTranscoderEntry("CP51932", "CP50221", 268, "Iso2022", 1, 3, 5, AsciiCompatibility.ENCODER, 1),
-        new GenericTranscoderEntry("CP51932", "CP50220", 268, "Iso2022", 1, 3, 5, AsciiCompatibility.ENCODER, 3),
-        new GenericTranscoderEntry("EUC-JP", "Shift_JIS", 88, "Japanese", 1, 3, 2, AsciiCompatibility.CONVERTER, 0),
-        new GenericTranscoderEntry("Shift_JIS", "EUC-JP", 132, "Japanese", 1, 2, 2, AsciiCompatibility.CONVERTER, 0),
         new GenericTranscoderEntry("EUC-JP", "UTF-8", 54488, "JapaneseEuc", 1, 3, 1, AsciiCompatibility.CONVERTER, 0),
         new GenericTranscoderEntry("eucJP-ms", "UTF-8", 64480, "JapaneseEuc", 1, 3, 1, AsciiCompatibility.CONVERTER, 0),
         new GenericTranscoderEntry("CP51932", "UTF-8", 66380, "JapaneseEuc", 1, 2, 1, AsciiCompatibility.CONVERTER, 0),
@@ -293,7 +278,6 @@ static void load() {
         new GenericTranscoderEntry("EUC-KR", "UTF-8", 78500, "Korean", 1, 2, 1, AsciiCompatibility.CONVERTER, 0),
         new GenericTranscoderEntry("UTF-8", "CP949", 126176, "Korean", 1, 4, 2, AsciiCompatibility.CONVERTER, 0),
         new GenericTranscoderEntry("CP949", "UTF-8", 196644, "Korean", 1, 2, 1, AsciiCompatibility.CONVERTER, 0),
-        new GenericTranscoderEntry("", "universal_newline", 4, "Newline", 1, 1, 2, AsciiCompatibility.CONVERTER, 2),
         new GenericTranscoderEntry("", "crlf_newline", 20, "Newline", 1, 1, 2, AsciiCompatibility.CONVERTER, 0),
         new GenericTranscoderEntry("", "cr_newline", 36, "Newline", 1, 1, 1, AsciiCompatibility.CONVERTER, 0),
         new GenericTranscoderEntry("US-ASCII", "UTF-8", 8, "SingleByte", 1, 1, 1, AsciiCompatibility.CONVERTER, 0),
@@ -402,19 +386,6 @@ static void load() {
         new GenericTranscoderEntry("UTF-8", "CP852", 29656, "SingleByte", 1, 4, 1, AsciiCompatibility.CONVERTER, 0),
         new GenericTranscoderEntry("CP855", "UTF-8", 30180, "SingleByte", 1, 1, 1, AsciiCompatibility.CONVERTER, 0),
         new GenericTranscoderEntry("UTF-8", "CP855", 30732, "SingleByte", 1, 4, 1, AsciiCompatibility.CONVERTER, 0),
-        new GenericTranscoderEntry("UTF-8", "UTF8-MAC", 52420, "Utf8Mac", 1, 4, 9, AsciiCompatibility.CONVERTER, 0),
-        new GenericTranscoderEntry("UTF8-MAC", "UTF-8", 52544, "Utf8Mac", 1, 4, 10, AsciiCompatibility.ENCODER, 0),
-        new GenericTranscoderEntry("UTF-16BE", "UTF-8", 188, "Utf1632", 2, 4, 4, AsciiCompatibility.DECODER, 0),
-        new GenericTranscoderEntry("UTF-8", "UTF-16BE", 416, "Utf1632", 1, 4, 4, AsciiCompatibility.ENCODER, 0),
-        new GenericTranscoderEntry("UTF-16LE", "UTF-8", 52, "Utf1632", 2, 4, 4, AsciiCompatibility.DECODER, 0),
-        new GenericTranscoderEntry("UTF-8", "UTF-16LE", 416, "Utf1632", 1, 4, 4, AsciiCompatibility.ENCODER, 0),
-        new GenericTranscoderEntry("UTF-32BE", "UTF-8", 252, "Utf1632", 4, 4, 4, AsciiCompatibility.DECODER, 0),
-        new GenericTranscoderEntry("UTF-8", "UTF-32BE", 416, "Utf1632", 1, 4, 4, AsciiCompatibility.ENCODER, 0),
-        new GenericTranscoderEntry("UTF-32LE", "UTF-8", 128, "Utf1632", 4, 4, 4, AsciiCompatibility.DECODER, 0),
-        new GenericTranscoderEntry("UTF-8", "UTF-32LE", 416, "Utf1632", 1, 4, 4, AsciiCompatibility.ENCODER, 0),
-        new GenericTranscoderEntry("UTF-16", "UTF-8", 276, "Utf1632", 2, 4, 4, AsciiCompatibility.DECODER, 1),
-        new GenericTranscoderEntry("UTF-32", "UTF-8", 300, "Utf1632", 4, 4, 4, AsciiCompatibility.DECODER, 1),
-        new GenericTranscoderEntry("UTF-8", "UTF-16", 416, "Utf1632", 1, 4, 4, AsciiCompatibility.ENCODER, 1),
-        new GenericTranscoderEntry("UTF-8", "UTF-32", 416, "Utf1632", 1, 4, 4, AsciiCompatibility.ENCODER, 1)
+        new GenericTranscoderEntry("UTF-8", "UTF8-MAC", 52420, "Utf8Mac", 1, 4, 9, AsciiCompatibility.CONVERTER, 0)
     };
 }