add mysql charset id -> encoding name mapping

Author: sidorares

lib/constants/charset_encodings.js

@@ -0,0 +1,37 @@
+// see tools/generate-charset-mapping.js
+// basicalliy result of "SHOW COLLATION" query
+
+module.exports = [
+  null,"big5","latin2","dec8","cp850","latin1","hp8","koi8r",
+  "latin1","latin2","swe7","ascii","ujis","sjis","cp1251","latin1",
+  "hebrew",null,"tis620","euckr","latin7","latin2","koi8u","cp1251",
+  "gb2312","greek","cp1250","latin2","gbk","cp1257","latin5","latin1",
+  "armscii8","cesu8","cp1250","ucs2","cp866","keybcs2","macintosh","macroman",
+  "cp852","latin7","latin7","macintosh","cp1250","cesu8","cesu8","latin1",
+  "latin1","latin1","cp1251","cp1251","cp1251","macroman","utf16","utf16",
+  "utf16le","cp1256","cp1257","cp1257","utf32","utf32","utf16le","binary",
+  "armscii8","ascii","cp1250","cp1256","cp866","dec8","greek","hebrew",
+  "hp8","keybcs2","koi8r","koi8u",null,"latin2","latin5","latin7",
+  "cp850","cp852","swe7","cesu8","big5","euckr","gb2312","gbk",
+  "sjis","tis620","ucs2","ujis","geostd8","geostd8","latin1","cp932",
+  "cp932","eucjp","eucjp","cp1250",null,"utf16","utf16","utf16",
+  "utf16","utf16","utf16","utf16","utf16","utf16","utf16","utf16",
+  "utf16","utf16","utf16","utf16","utf16","utf16","utf16","utf16",
+  "utf16","utf16","utf16","utf16","utf16",null,null,null,
+  "ucs2","ucs2","ucs2","ucs2","ucs2","ucs2","ucs2","ucs2",
+  "ucs2","ucs2","ucs2","ucs2","ucs2","ucs2","ucs2","ucs2",
+  "ucs2","ucs2","ucs2","ucs2","ucs2","ucs2","ucs2","ucs2",
+  null,null,null,null,null,null,null,"ucs2",
+  "utf32","utf32","utf32","utf32","utf32","utf32","utf32","utf32",
+  "utf32","utf32","utf32","utf32","utf32","utf32","utf32","utf32",
+  "utf32","utf32","utf32","utf32","utf32","utf32","utf32","utf32",
+  null,null,null,null,null,null,null,null,
+  "cesu8","cesu8","cesu8","cesu8","cesu8","cesu8","cesu8","cesu8",
+  "cesu8","cesu8","cesu8","cesu8","cesu8","cesu8","cesu8","cesu8",
+  "cesu8","cesu8","cesu8","cesu8","cesu8","cesu8","cesu8","cesu8",
+  null,null,null,null,null,null,null,"cesu8",
+  "cesu8","cesu8","cesu8","cesu8","cesu8","cesu8","cesu8","cesu8",
+  "cesu8","cesu8","cesu8","cesu8","cesu8","cesu8","cesu8","cesu8",
+  "cesu8","cesu8","cesu8","cesu8","cesu8","cesu8","cesu8","cesu8",
+  "gb18030","gb18030","gb18030"
+];

tools/generate-charset-mapping.js

@@ -0,0 +1,41 @@
+var mysql = require('../index.js');
+
+var conn = mysql.createConnection({
+});
+
+var iconv = require('iconv-lite');
+
+var charsets = []
+
+// TODO: add encodings missing in iconv-lite
+// "dec8","hp8","swe7","ujis","keybcs2","utf32","geostd8"
+
+// see also https://github.com/ashtuchkin/iconv-lite/issues/125
+// https://en.wikipedia.org/wiki/Kamenick%C3%BD_encoding
+// https://github.com/twitter/mysql/tree/master/sql/share/charsets
+
+var mysql2iconv = {
+  utf8: 'cesu8',
+  utf8mb4: 'cesu8',
+  // need to check that this is correct mapping
+  macce: 'macintosh', // Mac Central European
+  eucjpms: 'eucjp'    // UJIS for Windows Japanese
+};
+
+var missing = {};
+
+conn.query('show collation', function(err, res) {
+  res.forEach( r => {
+    var charset = r.Charset;
+    var iconvCharset = mysql2iconv[charset] || charset; // if there is manuall mapping, override
+    if (!iconv.encodingExists(iconvCharset)) {
+      missing[iconvCharset] = 1;
+    }
+    charsets[r.Id] = iconvCharset;
+  });
+  //console.log(JSON.stringify(missing, 4, null));
+  console.log(JSON.stringify(charsets, 4, null));
+});
+
+conn.end();
+