1 /*-------------------------------------------------------------------------
2 *
3 * GB18030 <--> UTF8
4 *
5 * Portions Copyright (c) 1996-2018, PostgreSQL Global Development Group
6 * Portions Copyright (c) 1994, Regents of the University of California
7 *
8 * IDENTIFICATION
9 * src/backend/utils/mb/conversion_procs/utf8_and_gb18030/utf8_and_gb18030.c
10 *
11 *-------------------------------------------------------------------------
12 */
13
14 #include "postgres.h"
15 #include "fmgr.h"
16 #include "mb/pg_wchar.h"
17 #include "../../Unicode/gb18030_to_utf8.map"
18 #include "../../Unicode/utf8_to_gb18030.map"
19
20 PG_MODULE_MAGIC;
21
22 PG_FUNCTION_INFO_V1(gb18030_to_utf8);
23 PG_FUNCTION_INFO_V1(utf8_to_gb18030);
24
25 /*
26 * Convert 4-byte GB18030 characters to and from a linear code space
27 *
28 * The first and third bytes can range from 0x81 to 0xfe (126 values),
29 * while the second and fourth bytes can range from 0x30 to 0x39 (10 values).
30 */
31 static inline uint32
gb_linear(uint32 gb)32 gb_linear(uint32 gb)
33 {
34 uint32 b0 = (gb & 0xff000000) >> 24;
35 uint32 b1 = (gb & 0x00ff0000) >> 16;
36 uint32 b2 = (gb & 0x0000ff00) >> 8;
37 uint32 b3 = (gb & 0x000000ff);
38
39 return b0 * 12600 + b1 * 1260 + b2 * 10 + b3 -
40 (0x81 * 12600 + 0x30 * 1260 + 0x81 * 10 + 0x30);
41 }
42
43 static inline uint32
gb_unlinear(uint32 lin)44 gb_unlinear(uint32 lin)
45 {
46 uint32 r0 = 0x81 + lin / 12600;
47 uint32 r1 = 0x30 + (lin / 1260) % 10;
48 uint32 r2 = 0x81 + (lin / 10) % 126;
49 uint32 r3 = 0x30 + lin % 10;
50
51 return (r0 << 24) | (r1 << 16) | (r2 << 8) | r3;
52 }
53
54 /*
55 * Convert word-formatted UTF8 to and from Unicode code points
56 *
57 * Probably this should be somewhere else ...
58 */
59 static inline uint32
unicode_to_utf8word(uint32 c)60 unicode_to_utf8word(uint32 c)
61 {
62 uint32 word;
63
64 if (c <= 0x7F)
65 {
66 word = c;
67 }
68 else if (c <= 0x7FF)
69 {
70 word = (0xC0 | ((c >> 6) & 0x1F)) << 8;
71 word |= 0x80 | (c & 0x3F);
72 }
73 else if (c <= 0xFFFF)
74 {
75 word = (0xE0 | ((c >> 12) & 0x0F)) << 16;
76 word |= (0x80 | ((c >> 6) & 0x3F)) << 8;
77 word |= 0x80 | (c & 0x3F);
78 }
79 else
80 {
81 word = (0xF0 | ((c >> 18) & 0x07)) << 24;
82 word |= (0x80 | ((c >> 12) & 0x3F)) << 16;
83 word |= (0x80 | ((c >> 6) & 0x3F)) << 8;
84 word |= 0x80 | (c & 0x3F);
85 }
86
87 return word;
88 }
89
90 static inline uint32
utf8word_to_unicode(uint32 c)91 utf8word_to_unicode(uint32 c)
92 {
93 uint32 ucs;
94
95 if (c <= 0x7F)
96 {
97 ucs = c;
98 }
99 else if (c <= 0xFFFF)
100 {
101 ucs = ((c >> 8) & 0x1F) << 6;
102 ucs |= c & 0x3F;
103 }
104 else if (c <= 0xFFFFFF)
105 {
106 ucs = ((c >> 16) & 0x0F) << 12;
107 ucs |= ((c >> 8) & 0x3F) << 6;
108 ucs |= c & 0x3F;
109 }
110 else
111 {
112 ucs = ((c >> 24) & 0x07) << 18;
113 ucs |= ((c >> 16) & 0x3F) << 12;
114 ucs |= ((c >> 8) & 0x3F) << 6;
115 ucs |= c & 0x3F;
116 }
117
118 return ucs;
119 }
120
121 /*
122 * Perform mapping of GB18030 ranges to UTF8
123 *
124 * The ranges we need to convert are specified in gb-18030-2000.xml.
125 * All are ranges of 4-byte GB18030 codes.
126 */
127 static uint32
conv_18030_to_utf8(uint32 code)128 conv_18030_to_utf8(uint32 code)
129 {
130 #define conv18030(minunicode, mincode, maxcode) \
131 if (code >= mincode && code <= maxcode) \
132 return unicode_to_utf8word(gb_linear(code) - gb_linear(mincode) + minunicode)
133
134 conv18030(0x0452, 0x8130D330, 0x8136A531);
135 conv18030(0x2643, 0x8137A839, 0x8138FD38);
136 conv18030(0x361B, 0x8230A633, 0x8230F237);
137 conv18030(0x3CE1, 0x8231D438, 0x8232AF32);
138 conv18030(0x4160, 0x8232C937, 0x8232F837);
139 conv18030(0x44D7, 0x8233A339, 0x8233C931);
140 conv18030(0x478E, 0x8233E838, 0x82349638);
141 conv18030(0x49B8, 0x8234A131, 0x8234E733);
142 conv18030(0x9FA6, 0x82358F33, 0x8336C738);
143 conv18030(0xE865, 0x8336D030, 0x84308534);
144 conv18030(0xFA2A, 0x84309C38, 0x84318537);
145 conv18030(0xFFE6, 0x8431A234, 0x8431A439);
146 conv18030(0x10000, 0x90308130, 0xE3329A35);
147 /* No mapping exists */
148 return 0;
149 }
150
151 /*
152 * Perform mapping of UTF8 ranges to GB18030
153 */
154 static uint32
conv_utf8_to_18030(uint32 code)155 conv_utf8_to_18030(uint32 code)
156 {
157 uint32 ucs = utf8word_to_unicode(code);
158
159 #define convutf8(minunicode, maxunicode, mincode) \
160 if (ucs >= minunicode && ucs <= maxunicode) \
161 return gb_unlinear(ucs - minunicode + gb_linear(mincode))
162
163 convutf8(0x0452, 0x200F, 0x8130D330);
164 convutf8(0x2643, 0x2E80, 0x8137A839);
165 convutf8(0x361B, 0x3917, 0x8230A633);
166 convutf8(0x3CE1, 0x4055, 0x8231D438);
167 convutf8(0x4160, 0x4336, 0x8232C937);
168 convutf8(0x44D7, 0x464B, 0x8233A339);
169 convutf8(0x478E, 0x4946, 0x8233E838);
170 convutf8(0x49B8, 0x4C76, 0x8234A131);
171 convutf8(0x9FA6, 0xD7FF, 0x82358F33);
172 convutf8(0xE865, 0xF92B, 0x8336D030);
173 convutf8(0xFA2A, 0xFE2F, 0x84309C38);
174 convutf8(0xFFE6, 0xFFFF, 0x8431A234);
175 convutf8(0x10000, 0x10FFFF, 0x90308130);
176 /* No mapping exists */
177 return 0;
178 }
179
180 /* ----------
181 * conv_proc(
182 * INTEGER, -- source encoding id
183 * INTEGER, -- destination encoding id
184 * CSTRING, -- source string (null terminated C string)
185 * CSTRING, -- destination string (null terminated C string)
186 * INTEGER -- source string length
187 * ) returns VOID;
188 * ----------
189 */
190 Datum
gb18030_to_utf8(PG_FUNCTION_ARGS)191 gb18030_to_utf8(PG_FUNCTION_ARGS)
192 {
193 unsigned char *src = (unsigned char *) PG_GETARG_CSTRING(2);
194 unsigned char *dest = (unsigned char *) PG_GETARG_CSTRING(3);
195 int len = PG_GETARG_INT32(4);
196
197 CHECK_ENCODING_CONVERSION_ARGS(PG_GB18030, PG_UTF8);
198
199 LocalToUtf(src, len, dest,
200 &gb18030_to_unicode_tree,
201 NULL, 0,
202 conv_18030_to_utf8,
203 PG_GB18030);
204
205 PG_RETURN_VOID();
206 }
207
208 Datum
utf8_to_gb18030(PG_FUNCTION_ARGS)209 utf8_to_gb18030(PG_FUNCTION_ARGS)
210 {
211 unsigned char *src = (unsigned char *) PG_GETARG_CSTRING(2);
212 unsigned char *dest = (unsigned char *) PG_GETARG_CSTRING(3);
213 int len = PG_GETARG_INT32(4);
214
215 CHECK_ENCODING_CONVERSION_ARGS(PG_UTF8, PG_GB18030);
216
217 UtfToLocal(src, len, dest,
218 &gb18030_from_unicode_tree,
219 NULL, 0,
220 conv_utf8_to_18030,
221 PG_GB18030);
222
223 PG_RETURN_VOID();
224 }
225