PostgreSQL Source Code git master
Loading...
Searching...
No Matches
Functions
euc_tw_and_big5.c File Reference
#include "postgres.h"
#include "fmgr.h"
#include "mb/pg_wchar.h"
Include dependency graph for euc_tw_and_big5.c:

Go to the source code of this file.

Functions

 PG_MODULE_MAGIC_EXT (.name="euc_tw_and_big5",.version=PG_VERSION)
 
 PG_FUNCTION_INFO_V1 (euc_tw_to_big5)
 
 PG_FUNCTION_INFO_V1 (big5_to_euc_tw)
 
static int euc_tw2big5 (const unsigned char *euc, unsigned char *p, int len, bool noError)
 
static int big52euc_tw (const unsigned char *big5, unsigned char *p, int len, bool noError)
 
Datum euc_tw_to_big5 (PG_FUNCTION_ARGS)
 
Datum big5_to_euc_tw (PG_FUNCTION_ARGS)
 

Function Documentation

◆ big52euc_tw()

static int big52euc_tw ( const unsigned char *  big5,
unsigned char *  p,
int  len,
bool  noError 
)
static

Definition at line 154 of file euc_tw_and_big5.c.

155{
156 const unsigned char *start = big5;
157 unsigned short c1;
158 unsigned short big5buf,
159 cnsBuf;
160 unsigned char lc;
161 int l;
162
163 while (len > 0)
164 {
165 /* Verify and decode the next Big5 input character */
166 c1 = *big5;
167 if (IS_HIGHBIT_SET(c1))
168 {
169 l = pg_encoding_verifymbchar(PG_BIG5, (const char *) big5, len);
170 if (l < 0)
171 {
172 if (noError)
173 break;
174 report_invalid_encoding(PG_BIG5,
175 (const char *) big5, len);
176 }
177 big5buf = (c1 << 8) | big5[1];
178 cnsBuf = BIG5toCNS(big5buf, &lc);
179
180 if (lc == LC_CNS11643_1)
181 {
182 *p++ = (cnsBuf >> 8) & 0x00ff;
183 *p++ = cnsBuf & 0x00ff;
184 }
185 else if (lc == LC_CNS11643_2)
186 {
187 *p++ = SS2;
188 *p++ = 0xa2;
189 *p++ = (cnsBuf >> 8) & 0x00ff;
190 *p++ = cnsBuf & 0x00ff;
191 }
192 else if (lc >= LC_CNS11643_3 && lc <= LC_CNS11643_7)
193 {
194 *p++ = SS2;
195 *p++ = lc - LC_CNS11643_3 + 0xa3;
196 *p++ = (cnsBuf >> 8) & 0x00ff;
197 *p++ = cnsBuf & 0x00ff;
198 }
199 else
200 {
201 if (noError)
202 break;
203 report_untranslatable_char(PG_BIG5, PG_EUC_TW,
204 (const char *) big5, len);
205 }
206
207 big5 += l;
208 len -= l;
209 }
210 else
211 {
212 /* ASCII */
213 if (c1 == 0)
214 {
215 if (noError)
216 break;
217 report_invalid_encoding(PG_BIG5,
218 (const char *) big5, len);
219 }
220 *p++ = c1;
221 big5++;
222 len--;
223 continue;
224 }
225 }
226 *p = '\0';
227
228 return big5 - start;
229}
BIG5toCNS
unsigned short BIG5toCNS(unsigned short big5, unsigned char *lc)
Definition big5.c:291
IS_HIGHBIT_SET
#define IS_HIGHBIT_SET(ch)
Definition c.h:1244
start
return str start
Definition hashfn_unstable.h:289
report_untranslatable_char
void report_untranslatable_char(int src_encoding, int dest_encoding, const char *mbstr, int len)
Definition mbutils.c:1869
report_invalid_encoding
void report_invalid_encoding(int encoding, const char *mbstr, int len)
Definition mbutils.c:1824
len
const void size_t len
Definition pg_crc32c_sse42.c:31
LC_CNS11643_7
#define LC_CNS11643_7
Definition pg_wchar.h:50
PG_EUC_TW
@ PG_EUC_TW
Definition pg_wchar.h:80
PG_BIG5
@ PG_BIG5
Definition pg_wchar.h:115
LC_CNS11643_3
#define LC_CNS11643_3
Definition pg_wchar.h:46
SS2
#define SS2
Definition pg_wchar.h:38
LC_CNS11643_1
#define LC_CNS11643_1
Definition pg_wchar.h:44
LC_CNS11643_2
#define LC_CNS11643_2
Definition pg_wchar.h:45
fb
static int fb(int x)
Definition preproc-init.c:92
pg_encoding_verifymbchar
int pg_encoding_verifymbchar(int encoding, const char *mbstr, int len)
Definition wchar.c:1988

References BIG5toCNS(), fb(), IS_HIGHBIT_SET, LC_CNS11643_1, LC_CNS11643_2, LC_CNS11643_3, LC_CNS11643_7, len, PG_BIG5, pg_encoding_verifymbchar(), PG_EUC_TW, report_invalid_encoding(), report_untranslatable_char(), SS2, and start.

Referenced by big5_to_euc_tw().

◆ big5_to_euc_tw()

Datum big5_to_euc_tw ( PG_FUNCTION_ARGS  )

Definition at line 60 of file euc_tw_and_big5.c.

61{
62 unsigned char *src = (unsigned char *) PG_GETARG_CSTRING(2);
63 unsigned char *dest = (unsigned char *) PG_GETARG_CSTRING(3);
64 int len = PG_GETARG_INT32(4);
65 bool noError = PG_GETARG_BOOL(5);
66 int converted;
67
68 CHECK_ENCODING_CONVERSION_ARGS(PG_BIG5, PG_EUC_TW);
69
70 converted = big52euc_tw(src, dest, len, noError);
71
72 PG_RETURN_INT32(converted);
73}
big52euc_tw
static int big52euc_tw(const unsigned char *big5, unsigned char *p, int len, bool noError)
Definition euc_tw_and_big5.c:154
PG_GETARG_CSTRING
#define PG_GETARG_CSTRING(n)
Definition fmgr.h:278
PG_RETURN_INT32
#define PG_RETURN_INT32(x)
Definition fmgr.h:355
PG_GETARG_INT32
#define PG_GETARG_INT32(n)
Definition fmgr.h:269
PG_GETARG_BOOL
#define PG_GETARG_BOOL(n)
Definition fmgr.h:274
generate_unaccent_rules.dest
dest
Definition generate_unaccent_rules.py:285
CHECK_ENCODING_CONVERSION_ARGS
#define CHECK_ENCODING_CONVERSION_ARGS(srcencoding, destencoding)
Definition pg_wchar.h:360

References big52euc_tw(), CHECK_ENCODING_CONVERSION_ARGS, fb(), len, PG_BIG5, PG_EUC_TW, PG_GETARG_BOOL, PG_GETARG_CSTRING, PG_GETARG_INT32, and PG_RETURN_INT32.

◆ euc_tw2big5()

static int euc_tw2big5 ( const unsigned char *  euc,
unsigned char *  p,
int  len,
bool  noError 
)
static

Definition at line 76 of file euc_tw_and_big5.c.

77{
78 const unsigned char *start = euc;
79 unsigned char c1;
80 unsigned short big5buf,
81 cnsBuf;
82 unsigned char lc;
83 int l;
84
85 while (len > 0)
86 {
87 c1 = *euc;
88 if (IS_HIGHBIT_SET(c1))
89 {
90 /* Verify and decode the next EUC_TW input character */
91 l = pg_encoding_verifymbchar(PG_EUC_TW, (const char *) euc, len);
92 if (l < 0)
93 {
94 if (noError)
95 break;
96 report_invalid_encoding(PG_EUC_TW,
97 (const char *) euc, len);
98 }
99 if (c1 == SS2)
100 {
101 c1 = euc[1]; /* plane No. */
102 if (c1 == 0xa1)
103 lc = LC_CNS11643_1;
104 else if (c1 == 0xa2)
105 lc = LC_CNS11643_2;
106 else
107 lc = c1 - 0xa3 + LC_CNS11643_3;
108 cnsBuf = (euc[2] << 8) | euc[3];
109 }
110 else
111 { /* CNS11643-1 */
112 lc = LC_CNS11643_1;
113 cnsBuf = (c1 << 8) | euc[1];
114 }
115
116 /* Write it out in Big5 */
117 big5buf = CNStoBIG5(cnsBuf, lc);
118 if (big5buf == 0)
119 {
120 if (noError)
121 break;
122 report_untranslatable_char(PG_EUC_TW, PG_BIG5,
123 (const char *) euc, len);
124 }
125 *p++ = (big5buf >> 8) & 0x00ff;
126 *p++ = big5buf & 0x00ff;
127
128 euc += l;
129 len -= l;
130 }
131 else
132 { /* should be ASCII */
133 if (c1 == 0)
134 {
135 if (noError)
136 break;
137 report_invalid_encoding(PG_EUC_TW,
138 (const char *) euc, len);
139 }
140 *p++ = c1;
141 euc++;
142 len--;
143 }
144 }
145 *p = '\0';
146
147 return euc - start;
148}
CNStoBIG5
unsigned short CNStoBIG5(unsigned short cns, unsigned char lc)
Definition big5.c:344

References CNStoBIG5(), fb(), IS_HIGHBIT_SET, LC_CNS11643_1, LC_CNS11643_2, LC_CNS11643_3, len, PG_BIG5, pg_encoding_verifymbchar(), PG_EUC_TW, report_invalid_encoding(), report_untranslatable_char(), SS2, and start.

Referenced by euc_tw_to_big5().

◆ euc_tw_to_big5()

Datum euc_tw_to_big5 ( PG_FUNCTION_ARGS  )

Definition at line 44 of file euc_tw_and_big5.c.

45{
46 unsigned char *src = (unsigned char *) PG_GETARG_CSTRING(2);
47 unsigned char *dest = (unsigned char *) PG_GETARG_CSTRING(3);
48 int len = PG_GETARG_INT32(4);
49 bool noError = PG_GETARG_BOOL(5);
50 int converted;
51
52 CHECK_ENCODING_CONVERSION_ARGS(PG_EUC_TW, PG_BIG5);
53
54 converted = euc_tw2big5(src, dest, len, noError);
55
56 PG_RETURN_INT32(converted);
57}
euc_tw2big5
static int euc_tw2big5(const unsigned char *euc, unsigned char *p, int len, bool noError)
Definition euc_tw_and_big5.c:76

References CHECK_ENCODING_CONVERSION_ARGS, euc_tw2big5(), fb(), len, PG_BIG5, PG_EUC_TW, PG_GETARG_BOOL, PG_GETARG_CSTRING, PG_GETARG_INT32, and PG_RETURN_INT32.

◆ PG_FUNCTION_INFO_V1() [1/2]

PG_FUNCTION_INFO_V1 ( big5_to_euc_tw  )

◆ PG_FUNCTION_INFO_V1() [2/2]

PG_FUNCTION_INFO_V1 ( euc_tw_to_big5  )

◆ PG_MODULE_MAGIC_EXT()

PG_MODULE_MAGIC_EXT ( .  name = "euc_tw_and_big5",
.  version = PG_VERSION 
)