PostgreSQL Source Code  git master
big5.c File Reference
#include "postgres_fe.h"
#include "mb/pg_wchar.h"
Include dependency graph for big5.c:

Go to the source code of this file.

Data Structures

struct  codes_t
 

Functions

static unsigned short BinarySearchRange (const codes_t *array, int high, unsigned short code)
 
unsigned short BIG5toCNS (unsigned short big5, unsigned char *lc)
 
unsigned short CNStoBIG5 (unsigned short cns, unsigned char lc)
 

Variables

static const codes_t big5Level1ToCnsPlane1 [25]
 
static const codes_t cnsPlane1ToBig5Level1 [26]
 
static const codes_t big5Level2ToCnsPlane2 [48]
 
static const codes_t cnsPlane2ToBig5Level2 [49]
 
static const unsigned short b1c4 [][2]
 
static const unsigned short b2c3 [][2]
 

Function Documentation

◆ BIG5toCNS()

unsigned short BIG5toCNS ( unsigned short  big5,
unsigned char *  lc 
)

Definition at line 292 of file big5.c.

References b1c4, b2c3, BinarySearchRange(), i, LC_CNS11643_1, LC_CNS11643_2, LC_CNS11643_3, and LC_CNS11643_4.

Referenced by big52mic().

293 {
294  unsigned short cns = 0;
295  int i;
296 
297  if (big5 < 0xc940U)
298  {
299  /* level 1 */
300 
301  for (i = 0; i < sizeof(b1c4) / (sizeof(unsigned short) * 2); i++)
302  {
303  if (b1c4[i][0] == big5)
304  {
305  *lc = LC_CNS11643_4;
306  return (b1c4[i][1] | 0x8080U);
307  }
308  }
309 
310  if (0 < (cns = BinarySearchRange(big5Level1ToCnsPlane1, 23, big5)))
311  *lc = LC_CNS11643_1;
312  }
313  else if (big5 == 0xc94aU)
314  {
315  /* level 2 */
316  *lc = LC_CNS11643_1;
317  cns = 0x4442;
318  }
319  else
320  {
321  /* level 2 */
322  for (i = 0; i < sizeof(b2c3) / (sizeof(unsigned short) * 2); i++)
323  {
324  if (b2c3[i][0] == big5)
325  {
326  *lc = LC_CNS11643_3;
327  return (b2c3[i][1] | 0x8080U);
328  }
329  }
330 
331  if (0 < (cns = BinarySearchRange(big5Level2ToCnsPlane2, 46, big5)))
332  *lc = LC_CNS11643_2;
333  }
334 
335  if (0 == cns)
336  { /* no mapping Big5 to CNS 11643-1992 */
337  *lc = 0;
338  return (unsigned short) '?';
339  }
340 
341  return cns | 0x8080;
342 }
#define LC_CNS11643_1
Definition: pg_wchar.h:134
static unsigned short BinarySearchRange(const codes_t *array, int high, unsigned short code)
Definition: big5.c:208
static const codes_t big5Level2ToCnsPlane2[48]
Definition: big5.c:84
#define LC_CNS11643_4
Definition: pg_wchar.h:205
static const codes_t big5Level1ToCnsPlane1[25]
Definition: big5.c:25
#define LC_CNS11643_2
Definition: pg_wchar.h:135
#define LC_CNS11643_3
Definition: pg_wchar.h:204
static const unsigned short b1c4[][2]
Definition: big5.c:189
int i
static const unsigned short b2c3[][2]
Definition: big5.c:197

◆ BinarySearchRange()

static unsigned short BinarySearchRange ( const codes_t array,
int  high,
unsigned short  code 
)
static

Definition at line 208 of file big5.c.

References codes_t::code, and codes_t::peer.

Referenced by BIG5toCNS(), and CNStoBIG5().

209 {
210  int low,
211  mid,
212  distance,
213  tmp;
214 
215  low = 0;
216  mid = high >> 1;
217 
218  for (; low <= high; mid = (low + high) >> 1)
219  {
220  if ((array[mid].code <= code) && (array[mid + 1].code > code))
221  {
222  if (0 == array[mid].peer)
223  return 0;
224  if (code >= 0xa140U)
225  {
226  /* big5 to cns */
227  tmp = ((code & 0xff00) - (array[mid].code & 0xff00)) >> 8;
228  high = code & 0x00ff;
229  low = array[mid].code & 0x00ff;
230 
231  /*
232  * NOTE: big5 high_byte: 0xa1-0xfe, low_byte: 0x40-0x7e,
233  * 0xa1-0xfe (radicals: 0x00-0x3e, 0x3f-0x9c) big5 radix is
234  * 0x9d. [region_low, region_high] We
235  * should remember big5 has two different regions (above).
236  * There is a bias for the distance between these regions.
237  * 0xa1 - 0x7e + bias = 1 (Distance between 0xa1 and 0x7e is
238  * 1.) bias = - 0x22.
239  */
240  distance = tmp * 0x9d + high - low +
241  (high >= 0xa1 ? (low >= 0xa1 ? 0 : -0x22)
242  : (low >= 0xa1 ? +0x22 : 0));
243 
244  /*
245  * NOTE: we have to convert the distance into a code point.
246  * The code point's low_byte is 0x21 plus mod_0x5e. In the
247  * first, we extract the mod_0x5e of the starting code point,
248  * subtracting 0x21, and add distance to it. Then we calculate
249  * again mod_0x5e of them, and restore the final codepoint,
250  * adding 0x21.
251  */
252  tmp = (array[mid].peer & 0x00ff) + distance - 0x21;
253  tmp = (array[mid].peer & 0xff00) + ((tmp / 0x5e) << 8)
254  + 0x21 + tmp % 0x5e;
255  return tmp;
256  }
257  else
258  {
259  /* cns to big5 */
260  tmp = ((code & 0xff00) - (array[mid].code & 0xff00)) >> 8;
261 
262  /*
263  * NOTE: ISO charsets ranges between 0x21-0xfe (94charset).
264  * Its radix is 0x5e. But there is no distance bias like big5.
265  */
266  distance = tmp * 0x5e
267  + ((int) (code & 0x00ff) - (int) (array[mid].code & 0x00ff));
268 
269  /*
270  * NOTE: Similar to big5 to cns conversion, we extract
271  * mod_0x9d and restore mod_0x9d into a code point.
272  */
273  low = array[mid].peer & 0x00ff;
274  tmp = low + distance - (low >= 0xa1 ? 0x62 : 0x40);
275  low = tmp % 0x9d;
276  tmp = (array[mid].peer & 0xff00) + ((tmp / 0x9d) << 8)
277  + (low > 0x3e ? 0x62 : 0x40) + low;
278  return tmp;
279  }
280  }
281  else if (array[mid].code > code)
282  high = mid - 1;
283  else
284  low = mid + 1;
285  }
286 
287  return 0;
288 }
unsigned short code
Definition: big5.c:20
unsigned short peer
Definition: big5.c:20

◆ CNStoBIG5()

unsigned short CNStoBIG5 ( unsigned short  cns,
unsigned char  lc 
)

Definition at line 345 of file big5.c.

References b1c4, b2c3, BinarySearchRange(), i, LC_CNS11643_1, LC_CNS11643_2, LC_CNS11643_3, and LC_CNS11643_4.

Referenced by mic2big5().

346 {
347  int i;
348  unsigned int big5 = 0;
349 
350  cns &= 0x7f7f;
351 
352  switch (lc)
353  {
354  case LC_CNS11643_1:
355  big5 = BinarySearchRange(cnsPlane1ToBig5Level1, 24, cns);
356  break;
357  case LC_CNS11643_2:
358  big5 = BinarySearchRange(cnsPlane2ToBig5Level2, 47, cns);
359  break;
360  case LC_CNS11643_3:
361  for (i = 0; i < sizeof(b2c3) / (sizeof(unsigned short) * 2); i++)
362  {
363  if (b2c3[i][1] == cns)
364  return b2c3[i][0];
365  }
366  break;
367  case LC_CNS11643_4:
368  for (i = 0; i < sizeof(b1c4) / (sizeof(unsigned short) * 2); i++)
369  {
370  if (b1c4[i][1] == cns)
371  return b1c4[i][0];
372  }
373  default:
374  break;
375  }
376  return big5;
377 }
#define LC_CNS11643_1
Definition: pg_wchar.h:134
static unsigned short BinarySearchRange(const codes_t *array, int high, unsigned short code)
Definition: big5.c:208
#define LC_CNS11643_4
Definition: pg_wchar.h:205
static const codes_t cnsPlane2ToBig5Level2[49]
Definition: big5.c:136
static const codes_t cnsPlane1ToBig5Level1[26]
Definition: big5.c:54
#define LC_CNS11643_2
Definition: pg_wchar.h:135
#define LC_CNS11643_3
Definition: pg_wchar.h:204
static const unsigned short b1c4[][2]
Definition: big5.c:189
int i
static const unsigned short b2c3[][2]
Definition: big5.c:197

Variable Documentation

◆ b1c4

const unsigned short b1c4[][2]
static
Initial value:
= {
{0xC879, 0x2123},
{0xC87B, 0x2124},
{0xC87D, 0x212A},
{0xC8A2, 0x2152}
}

Definition at line 189 of file big5.c.

Referenced by BIG5toCNS(), and CNStoBIG5().

◆ b2c3

const unsigned short b2c3[][2]
static
Initial value:
= {
{0xF9D6, 0x4337},
{0xF9D7, 0x4F50},
{0xF9D8, 0x444E},
{0xF9D9, 0x504A},
{0xF9DA, 0x2C5D},
{0xF9DB, 0x3D7E},
{0xF9DC, 0x4B5C}
}

Definition at line 197 of file big5.c.

Referenced by BIG5toCNS(), and CNStoBIG5().

◆ big5Level1ToCnsPlane1

const codes_t big5Level1ToCnsPlane1[25]
static
Initial value:
= {
{0xA140, 0x2121},
{0xA1F6, 0x2258},
{0xA1F7, 0x2257},
{0xA1F8, 0x2259},
{0xA2AF, 0x2421},
{0xA3C0, 0x4221},
{0xa3e1, 0x0000},
{0xA440, 0x4421},
{0xACFE, 0x5753},
{0xacff, 0x0000},
{0xAD40, 0x5323},
{0xAFD0, 0x5754},
{0xBBC8, 0x6B51},
{0xBE52, 0x6B50},
{0xBE53, 0x6F5C},
{0xC1AB, 0x7536},
{0xC2CB, 0x7535},
{0xC2CC, 0x7737},
{0xC361, 0x782E},
{0xC3B9, 0x7865},
{0xC3BA, 0x7864},
{0xC3BB, 0x7866},
{0xC456, 0x782D},
{0xC457, 0x7962},
{0xc67f, 0x0000}
}

Definition at line 25 of file big5.c.

◆ big5Level2ToCnsPlane2

const codes_t big5Level2ToCnsPlane2[48]
static

Definition at line 84 of file big5.c.

◆ cnsPlane1ToBig5Level1

const codes_t cnsPlane1ToBig5Level1[26]
static
Initial value:
= {
{0x2121, 0xA140},
{0x2257, 0xA1F7},
{0x2258, 0xA1F6},
{0x2259, 0xA1F8},
{0x234f, 0x0000},
{0x2421, 0xA2AF},
{0x2571, 0x0000},
{0x4221, 0xA3C0},
{0x4242, 0x0000},
{0x4421, 0xA440},
{0x5323, 0xAD40},
{0x5753, 0xACFE},
{0x5754, 0xAFD0},
{0x6B50, 0xBE52},
{0x6B51, 0xBBC8},
{0x6F5C, 0xBE53},
{0x7535, 0xC2CB},
{0x7536, 0xC1AB},
{0x7737, 0xC2CC},
{0x782D, 0xC456},
{0x782E, 0xC361},
{0x7864, 0xC3BA},
{0x7865, 0xC3B9},
{0x7866, 0xC3BB},
{0x7962, 0xC457},
{0x7d4c, 0x0000}
}

Definition at line 54 of file big5.c.

◆ cnsPlane2ToBig5Level2

const codes_t cnsPlane2ToBig5Level2[49]
static

Definition at line 136 of file big5.c.