Real-time collaboration for Jupyter Notebooks, Linux Terminals, LaTeX, VS Code, R IDE, and more,
all in one place.

1
# coding: ascii-8bit
2
# Copyright (C) 2008-2012 TOMITA Masahiro
3
# mailto:[email protected]
4

5
#
6
class RbMysql
7
  # @!attribute [r] number
8
  #   @private
9
  # @!attribute [r] name
10
  #   @return [String] charset name
11
  # @!attribute [r] csname
12
  #   @return [String] collation name
13
  class Charset
14
    # @private
15
    # @param [Integer] number
16
    # @param [String] name
17
    # @param [String] csname
18
    def initialize(number, name, csname)
19
      @number, @name, @csname = number, name, csname
20
      @unsafe = false
21
    end
22

23
    attr_reader :number, :name, :csname
24

25
    # @private
26
    attr_accessor :unsafe
27

28
    # [[charset_number, charset_name, collation_name, default], ...]
29
    # @private
30
    CHARSETS = [
31
      [  1, "big5",     "big5_chinese_ci",          true ],
32
      [  2, "latin2",   "latin2_czech_cs",          false],
33
      [  3, "dec8",     "dec8_swedish_ci",          true ],
34
      [  4, "cp850",    "cp850_general_ci",         true ],
35
      [  5, "latin1",   "latin1_german1_ci",        false],
36
      [  6, "hp8",      "hp8_english_ci",           true ],
37
      [  7, "koi8r",    "koi8r_general_ci",         true ],
38
      [  8, "latin1",   "latin1_swedish_ci",        true ],
39
      [  9, "latin2",   "latin2_general_ci",        true ],
40
      [ 10, "swe7",     "swe7_swedish_ci",          true ],
41
      [ 11, "ascii",    "ascii_general_ci",         true ],
42
      [ 12, "ujis",     "ujis_japanese_ci",         true ],
43
      [ 13, "sjis",     "sjis_japanese_ci",         true ],
44
      [ 14, "cp1251",   "cp1251_bulgarian_ci",      false],
45
      [ 15, "latin1",   "latin1_danish_ci",         false],
46
      [ 16, "hebrew",   "hebrew_general_ci",        true ],
47
      [ 17, "filename", "filename",                 true ],
48
      [ 18, "tis620",   "tis620_thai_ci",           true ],
49
      [ 19, "euckr",    "euckr_korean_ci",          true ],
50
      [ 20, "latin7",   "latin7_estonian_cs",       false],
51
      [ 21, "latin2",   "latin2_hungarian_ci",      false],
52
      [ 22, "koi8u",    "koi8u_general_ci",         true ],
53
      [ 23, "cp1251",   "cp1251_ukrainian_ci",      false],
54
      [ 24, "gb2312",   "gb2312_chinese_ci",        true ],
55
      [ 25, "greek",    "greek_general_ci",         true ],
56
      [ 26, "cp1250",   "cp1250_general_ci",        true ],
57
      [ 27, "latin2",   "latin2_croatian_ci",       false],
58
      [ 28, "gbk",      "gbk_chinese_ci",           true ],
59
      [ 29, "cp1257",   "cp1257_lithuanian_ci",     false],
60
      [ 30, "latin5",   "latin5_turkish_ci",        true ],
61
      [ 31, "latin1",   "latin1_german2_ci",        false],
62
      [ 32, "armscii8", "armscii8_general_ci",      true ],
63
      [ 33, "utf8",     "utf8_general_ci",          true ],
64
      [ 34, "cp1250",   "cp1250_czech_cs",          false],
65
      [ 35, "ucs2",     "ucs2_general_ci",          true ],
66
      [ 36, "cp866",    "cp866_general_ci",         true ],
67
      [ 37, "keybcs2",  "keybcs2_general_ci",       true ],
68
      [ 38, "macce",    "macce_general_ci",         true ],
69
      [ 39, "macroman", "macroman_general_ci",      true ],
70
      [ 40, "cp852",    "cp852_general_ci",         true ],
71
      [ 41, "latin7",   "latin7_general_ci",        true ],
72
      [ 42, "latin7",   "latin7_general_cs",        false],
73
      [ 43, "macce",    "macce_bin",                false],
74
      [ 44, "cp1250",   "cp1250_croatian_ci",       false],
75
      [ 45, "utf8mb4",  "utf8mb4_general_ci",       true ],
76
      [ 46, "utf8mb4",  "utf8mb4_bin",              false],
77
      [ 47, "latin1",   "latin1_bin",               false],
78
      [ 48, "latin1",   "latin1_general_ci",        false],
79
      [ 49, "latin1",   "latin1_general_cs",        false],
80
      [ 50, "cp1251",   "cp1251_bin",               false],
81
      [ 51, "cp1251",   "cp1251_general_ci",        true ],
82
      [ 52, "cp1251",   "cp1251_general_cs",        false],
83
      [ 53, "macroman", "macroman_bin",             false],
84
      [ 54, "utf16",    "utf16_general_ci",         true ],
85
      [ 55, "utf16",    "utf16_bin",                false],
86
      [ 56, "utf16le",  "utf16le_general_ci",       true ],
87
      [ 57, "cp1256",   "cp1256_general_ci",        true ],
88
      [ 58, "cp1257",   "cp1257_bin",               false],
89
      [ 59, "cp1257",   "cp1257_general_ci",        true ],
90
      [ 60, "utf32",    "utf32_general_ci",         true ],
91
      [ 61, "utf32",    "utf32_bin",                false],
92
      [ 62, "utf16le",  "utf16le_bin",              false],
93
      [ 63, "binary",   "binary",                   true ],
94
      [ 64, "armscii8", "armscii8_bin",             false],
95
      [ 65, "ascii",    "ascii_bin",                false],
96
      [ 66, "cp1250",   "cp1250_bin",               false],
97
      [ 67, "cp1256",   "cp1256_bin",               false],
98
      [ 68, "cp866",    "cp866_bin",                false],
99
      [ 69, "dec8",     "dec8_bin",                 false],
100
      [ 70, "greek",    "greek_bin",                false],
101
      [ 71, "hebrew",   "hebrew_bin",               false],
102
      [ 72, "hp8",      "hp8_bin",                  false],
103
      [ 73, "keybcs2",  "keybcs2_bin",              false],
104
      [ 74, "koi8r",    "koi8r_bin",                false],
105
      [ 75, "koi8u",    "koi8u_bin",                false],
106
      [ 77, "latin2",   "latin2_bin",               false],
107
      [ 78, "latin5",   "latin5_bin",               false],
108
      [ 79, "latin7",   "latin7_bin",               false],
109
      [ 80, "cp850",    "cp850_bin",                false],
110
      [ 81, "cp852",    "cp852_bin",                false],
111
      [ 82, "swe7",     "swe7_bin",                 false],
112
      [ 83, "utf8",     "utf8_bin",                 false],
113
      [ 84, "big5",     "big5_bin",                 false],
114
      [ 85, "euckr",    "euckr_bin",                false],
115
      [ 86, "gb2312",   "gb2312_bin",               false],
116
      [ 87, "gbk",      "gbk_bin",                  false],
117
      [ 88, "sjis",     "sjis_bin",                 false],
118
      [ 89, "tis620",   "tis620_bin",               false],
119
      [ 90, "ucs2",     "ucs2_bin",                 false],
120
      [ 91, "ujis",     "ujis_bin",                 false],
121
      [ 92, "geostd8",  "geostd8_general_ci",       true ],
122
      [ 93, "geostd8",  "geostd8_bin",              false],
123
      [ 94, "latin1",   "latin1_spanish_ci",        false],
124
      [ 95, "cp932",    "cp932_japanese_ci",        true ],
125
      [ 96, "cp932",    "cp932_bin",                false],
126
      [ 97, "eucjpms",  "eucjpms_japanese_ci",      true ],
127
      [ 98, "eucjpms",  "eucjpms_bin",              false],
128
      [ 99, "cp1250",   "cp1250_polish_ci",         false],
129
      [101, "utf16",    "utf16_unicode_ci",         false],
130
      [102, "utf16",    "utf16_icelandic_ci",       false],
131
      [103, "utf16",    "utf16_latvian_ci",         false],
132
      [104, "utf16",    "utf16_romanian_ci",        false],
133
      [105, "utf16",    "utf16_slovenian_ci",       false],
134
      [106, "utf16",    "utf16_polish_ci",          false],
135
      [107, "utf16",    "utf16_estonian_ci",        false],
136
      [108, "utf16",    "utf16_spanish_ci",         false],
137
      [109, "utf16",    "utf16_swedish_ci",         false],
138
      [110, "utf16",    "utf16_turkish_ci",         false],
139
      [111, "utf16",    "utf16_czech_ci",           false],
140
      [112, "utf16",    "utf16_danish_ci",          false],
141
      [113, "utf16",    "utf16_lithuanian_ci",      false],
142
      [114, "utf16",    "utf16_slovak_ci",          false],
143
      [115, "utf16",    "utf16_spanish2_ci",        false],
144
      [116, "utf16",    "utf16_roman_ci",           false],
145
      [117, "utf16",    "utf16_persian_ci",         false],
146
      [118, "utf16",    "utf16_esperanto_ci",       false],
147
      [119, "utf16",    "utf16_hungarian_ci",       false],
148
      [120, "utf16",    "utf16_sinhala_ci",         false],
149
      [121, "utf16",    "utf16_german2_ci",         false],
150
      [122, "utf16",    "utf16_croatian_ci",        false],
151
      [123, "utf16",    "utf16_unicode_520_ci",     false],
152
      [124, "utf16",    "utf16_vietnamese_ci",      false],
153
      [128, "ucs2",     "ucs2_unicode_ci",          false],
154
      [129, "ucs2",     "ucs2_icelandic_ci",        false],
155
      [130, "ucs2",     "ucs2_latvian_ci",          false],
156
      [131, "ucs2",     "ucs2_romanian_ci",         false],
157
      [132, "ucs2",     "ucs2_slovenian_ci",        false],
158
      [133, "ucs2",     "ucs2_polish_ci",           false],
159
      [134, "ucs2",     "ucs2_estonian_ci",         false],
160
      [135, "ucs2",     "ucs2_spanish_ci",          false],
161
      [136, "ucs2",     "ucs2_swedish_ci",          false],
162
      [137, "ucs2",     "ucs2_turkish_ci",          false],
163
      [138, "ucs2",     "ucs2_czech_ci",            false],
164
      [139, "ucs2",     "ucs2_danish_ci",           false],
165
      [140, "ucs2",     "ucs2_lithuanian_ci",       false],
166
      [141, "ucs2",     "ucs2_slovak_ci",           false],
167
      [142, "ucs2",     "ucs2_spanish2_ci",         false],
168
      [143, "ucs2",     "ucs2_roman_ci",            false],
169
      [144, "ucs2",     "ucs2_persian_ci",          false],
170
      [145, "ucs2",     "ucs2_esperanto_ci",        false],
171
      [146, "ucs2",     "ucs2_hungarian_ci",        false],
172
      [147, "ucs2",     "ucs2_sinhala_ci",          false],
173
      [148, "ucs2",     "ucs2_german2_ci",          false],
174
      [149, "ucs2",     "ucs2_croatian_ci",         false],
175
      [150, "ucs2",     "ucs2_unicode_520_ci",      false],
176
      [151, "ucs2",     "ucs2_vietnamese_ci",       false],
177
      [159, "ucs2",     "ucs2_general_mysql500_ci", false],
178
      [160, "utf32",    "utf32_unicode_ci",         false],
179
      [161, "utf32",    "utf32_icelandic_ci",       false],
180
      [162, "utf32",    "utf32_latvian_ci",         false],
181
      [163, "utf32",    "utf32_romanian_ci",        false],
182
      [164, "utf32",    "utf32_slovenian_ci",       false],
183
      [165, "utf32",    "utf32_polish_ci",          false],
184
      [166, "utf32",    "utf32_estonian_ci",        false],
185
      [167, "utf32",    "utf32_spanish_ci",         false],
186
      [168, "utf32",    "utf32_swedish_ci",         false],
187
      [169, "utf32",    "utf32_turkish_ci",         false],
188
      [170, "utf32",    "utf32_czech_ci",           false],
189
      [171, "utf32",    "utf32_danish_ci",          false],
190
      [172, "utf32",    "utf32_lithuanian_ci",      false],
191
      [173, "utf32",    "utf32_slovak_ci",          false],
192
      [174, "utf32",    "utf32_spanish2_ci",        false],
193
      [175, "utf32",    "utf32_roman_ci",           false],
194
      [176, "utf32",    "utf32_persian_ci",         false],
195
      [177, "utf32",    "utf32_esperanto_ci",       false],
196
      [178, "utf32",    "utf32_hungarian_ci",       false],
197
      [179, "utf32",    "utf32_sinhala_ci",         false],
198
      [180, "utf32",    "utf32_german2_ci",         false],
199
      [181, "utf32",    "utf32_croatian_ci",        false],
200
      [182, "utf32",    "utf32_unicode_520_ci",     false],
201
      [183, "utf32",    "utf32_vietnamese_ci",      false],
202
      [192, "utf8",     "utf8_unicode_ci",          false],
203
      [193, "utf8",     "utf8_icelandic_ci",        false],
204
      [194, "utf8",     "utf8_latvian_ci",          false],
205
      [195, "utf8",     "utf8_romanian_ci",         false],
206
      [196, "utf8",     "utf8_slovenian_ci",        false],
207
      [197, "utf8",     "utf8_polish_ci",           false],
208
      [198, "utf8",     "utf8_estonian_ci",         false],
209
      [199, "utf8",     "utf8_spanish_ci",          false],
210
      [200, "utf8",     "utf8_swedish_ci",          false],
211
      [201, "utf8",     "utf8_turkish_ci",          false],
212
      [202, "utf8",     "utf8_czech_ci",            false],
213
      [203, "utf8",     "utf8_danish_ci",           false],
214
      [204, "utf8",     "utf8_lithuanian_ci",       false],
215
      [205, "utf8",     "utf8_slovak_ci",           false],
216
      [206, "utf8",     "utf8_spanish2_ci",         false],
217
      [207, "utf8",     "utf8_roman_ci",            false],
218
      [208, "utf8",     "utf8_persian_ci",          false],
219
      [209, "utf8",     "utf8_esperanto_ci",        false],
220
      [210, "utf8",     "utf8_hungarian_ci",        false],
221
      [211, "utf8",     "utf8_sinhala_ci",          false],
222
      [212, "utf8",     "utf8_german2_ci",          false],
223
      [213, "utf8",     "utf8_croatian_ci",         false],
224
      [214, "utf8",     "utf8_unicode_520_ci",      false],
225
      [215, "utf8",     "utf8_vietnamese_ci",       false],
226
      [223, "utf8",     "utf8_general_mysql500_ci", false],
227
      [224, "utf8mb4",  "utf8mb4_unicode_ci",       false],
228
      [225, "utf8mb4",  "utf8mb4_icelandic_ci",     false],
229
      [226, "utf8mb4",  "utf8mb4_latvian_ci",       false],
230
      [227, "utf8mb4",  "utf8mb4_romanian_ci",      false],
231
      [228, "utf8mb4",  "utf8mb4_slovenian_ci",     false],
232
      [229, "utf8mb4",  "utf8mb4_polish_ci",        false],
233
      [230, "utf8mb4",  "utf8mb4_estonian_ci",      false],
234
      [231, "utf8mb4",  "utf8mb4_spanish_ci",       false],
235
      [232, "utf8mb4",  "utf8mb4_swedish_ci",       false],
236
      [233, "utf8mb4",  "utf8mb4_turkish_ci",       false],
237
      [234, "utf8mb4",  "utf8mb4_czech_ci",         false],
238
      [235, "utf8mb4",  "utf8mb4_danish_ci",        false],
239
      [236, "utf8mb4",  "utf8mb4_lithuanian_ci",    false],
240
      [237, "utf8mb4",  "utf8mb4_slovak_ci",        false],
241
      [238, "utf8mb4",  "utf8mb4_spanish2_ci",      false],
242
      [239, "utf8mb4",  "utf8mb4_roman_ci",         false],
243
      [240, "utf8mb4",  "utf8mb4_persian_ci",       false],
244
      [241, "utf8mb4",  "utf8mb4_esperanto_ci",     false],
245
      [242, "utf8mb4",  "utf8mb4_hungarian_ci",     false],
246
      [243, "utf8mb4",  "utf8mb4_sinhala_ci",       false],
247
      [244, "utf8mb4",  "utf8mb4_german2_ci",       false],
248
      [245, "utf8mb4",  "utf8mb4_croatian_ci",      false],
249
      [246, "utf8mb4",  "utf8mb4_unicode_520_ci",   false],
250
      [247, "utf8mb4",  "utf8mb4_vietnamese_ci",    false],
251
      [248, "gb18030",  "gb18030_chinese_ci",       true ],
252
      [249, "gb18030",  "gb18030_bin",              false],
253
      [250, "gb18030",  "gb18030_unicode_520_ci",   false],
254
      [254, "utf8",     "utf8_general_cs",          false],
255
    ]
256

257
    # @private
258
    UNSAFE_CHARSET = [
259
      "big5", "sjis", "filename", "gbk", "ucs2", "cp932",
260
    ]
261

262
    # @private
263
    NUMBER_TO_CHARSET = {}
264
    # @private
265
    COLLATION_TO_CHARSET = {}
266
    # @private
267
    CHARSET_DEFAULT = {}
268
    CHARSETS.each do |number, csname, clname, default|
269
      cs = Charset.new number, csname, clname
270
      cs.unsafe = true if UNSAFE_CHARSET.include? csname
271
      NUMBER_TO_CHARSET[number] = cs
272
      COLLATION_TO_CHARSET[clname] = cs
273
      CHARSET_DEFAULT[csname] = cs if default
274
    end
275

276
    # @private
277
    BINARY_CHARSET_NUMBER = CHARSET_DEFAULT['binary'].number
278

279
    # @private
280
    # @param [Integer] n
281
    # @return [RbMysql::Charset]
282
    def self.by_number(n)
283
      raise ClientError, "unknown charset number: #{n}" unless NUMBER_TO_CHARSET.key? n
284
      NUMBER_TO_CHARSET[n]
285
    end
286

287
    # @private
288
    # @param [String] str
289
    # @return [RbMysql::Charset]
290
    def self.by_name(str)
291
      ret = COLLATION_TO_CHARSET[str] || CHARSET_DEFAULT[str]
292
      raise ClientError, "unknown charset: #{str}" unless ret
293
      ret
294
    end
295

296
    if defined? Encoding
297

298
      # @private
299
      # MySQL Charset -> Ruby's Encoding
300
      CHARSET_ENCODING = {
301
        "armscii8" => nil,
302
        "ascii"    => Encoding::US_ASCII,
303
        "big5"     => Encoding::Big5,
304
        "binary"   => Encoding::ASCII_8BIT,
305
        "cp1250"   => Encoding::Windows_1250,
306
        "cp1251"   => Encoding::Windows_1251,
307
        "cp1256"   => Encoding::Windows_1256,
308
        "cp1257"   => Encoding::Windows_1257,
309
        "cp850"    => Encoding::CP850,
310
        "cp852"    => Encoding::CP852,
311
        "cp866"    => Encoding::IBM866,
312
        "cp932"    => Encoding::Windows_31J,
313
        "dec8"     => nil,
314
        "eucjpms"  => Encoding::EucJP_ms,
315
        "euckr"    => Encoding::EUC_KR,
316
        "gb2312"   => Encoding::EUC_CN,
317
        "gbk"      => Encoding::GBK,
318
        "geostd8"  => nil,
319
        "greek"    => Encoding::ISO_8859_7,
320
        "hebrew"   => Encoding::ISO_8859_8,
321
        "hp8"      => nil,
322
        "keybcs2"  => nil,
323
        "koi8r"    => Encoding::KOI8_R,
324
        "koi8u"    => Encoding::KOI8_U,
325
        "latin1"   => Encoding::ISO_8859_1,
326
        "latin2"   => Encoding::ISO_8859_2,
327
        "latin5"   => Encoding::ISO_8859_9,
328
        "latin7"   => Encoding::ISO_8859_13,
329
        "macce"    => Encoding::MacCentEuro,
330
        "macroman" => Encoding::MacRoman,
331
        "sjis"     => Encoding::SHIFT_JIS,
332
        "swe7"     => nil,
333
        "tis620"   => Encoding::TIS_620,
334
        "ucs2"     => Encoding::UTF_16BE,
335
        "ujis"     => Encoding::EucJP_ms,
336
        "utf8"     => Encoding::UTF_8,
337
        "utf8mb4"  => Encoding::UTF_8,
338
      }
339

340
      # @private
341
      # @param [String] value
342
      # @return [String]
343
      def self.to_binary(value)
344
        value.force_encoding Encoding::ASCII_8BIT
345
      end
346

347
      # @private
348
      # convert raw to encoding and convert to Encoding.default_internal
349
      # @param [String] raw
350
      # @param [Encoding] encoding
351
      # @return [String] result
352
      def self.convert_encoding(raw, encoding)
353
        raw.force_encoding(encoding).encode
354
      end
355

356
      # @private
357
      # return corresponding Ruby encoding
358
      # @return [Encoding] encoding
359
      def encoding
360
        enc = CHARSET_ENCODING[@name.downcase]
361
        raise RbMysql::ClientError, "unsupported charset: #{@name}" unless enc
362
        enc
363
      end
364

365
      # @private
366
      # convert encoding to corresponding to MySQL charset
367
      # @param [String] value
368
      # @return [String]
369
      def convert(value)
370
        if value.is_a? String and value.encoding != Encoding::ASCII_8BIT
371
          value = value.encode encoding
372
        end
373
        value
374
      end
375

376
    else
377
      # for Ruby 1.8
378

379
      def self.to_binary(value)
380
        value
381
      end
382

383
      def self.convert_encoding(raw, encoding)
384
        raw
385
      end
386

387
      def encoding
388
        nil
389
      end
390

391
      def convert(value)
392
        value
393
      end
394

395
    end
396
  end
397
end
398