YAZ  5.34.0
iso5428.c
Go to the documentation of this file.
1 /* This file is part of the YAZ toolkit.
2  * Copyright (C) Index Data
3  * See the file LICENSE for details.
4  */
10 #if HAVE_CONFIG_H
11 #include <config.h>
12 #endif
13 
14 #include <assert.h>
15 #include <errno.h>
16 #include <string.h>
17 
18 #include "iconv-p.h"
19 
20 static unsigned long read_iso_5428_1984(yaz_iconv_t cd, yaz_iconv_decoder_t d,
21  unsigned char *inp,
22  size_t inbytesleft, size_t *no_read)
23 {
24  unsigned long x = 0;
25  int tonos = 0;
26  int dialitika = 0;
27 
28  *no_read = 0;
29  while (inbytesleft > 0)
30  {
31  if (*inp == 0xa2)
32  {
33  tonos = 1;
34  }
35  else if (*inp == 0xa3)
36  {
37  dialitika = 1;
38  }
39  else
40  break;
41  inp++;
42  --inbytesleft;
43  (*no_read)++;
44  }
45  if (inbytesleft == 0)
46  {
47  yaz_iconv_set_errno(cd, YAZ_ICONV_EINVAL); /* incomplete input */
48  *no_read = 0;
49  return 0;
50  }
51  switch (*inp) {
52  case 0xe1: /* alpha small */
53  if (tonos)
54  x = 0x03ac;
55  else
56  x = 0x03b1;
57  break;
58  case 0xc1: /* alpha capital */
59  if (tonos)
60  x = 0x0386;
61  else
62  x = 0x0391;
63  break;
64 
65  case 0xe2: /* Beta small */
66  x = 0x03b2;
67  break;
68  case 0xc2: /* Beta capital */
69  x = 0x0392;
70  break;
71 
72  case 0xe4: /* Gamma small */
73  x = 0x03b3;
74  break;
75  case 0xc4: /* Gamma capital */
76  x = 0x0393;
77  break;
78 
79  case 0xe5: /* Delta small */
80  x = 0x03b4;
81  break;
82  case 0xc5: /* Delta capital */
83  x = 0x0394;
84  break;
85  case 0xe6: /* epsilon small */
86  if (tonos)
87  x = 0x03ad;
88  else
89  x = 0x03b5;
90  break;
91  case 0xc6: /* epsilon capital */
92  if (tonos)
93  x = 0x0388;
94  else
95  x = 0x0395;
96  break;
97  case 0xe9: /* Zeta small */
98  x = 0x03b6;
99  break;
100  case 0xc9: /* Zeta capital */
101  x = 0x0396;
102  break;
103  case 0xea: /* Eta small */
104  if (tonos)
105  x = 0x03ae;
106  else
107  x = 0x03b7;
108  break;
109  case 0xca: /* Eta capital */
110  if (tonos)
111  x = 0x0389;
112  else
113  x = 0x0397;
114  break;
115  case 0xeb: /* Theta small */
116  x = 0x03b8;
117  break;
118  case 0xcb: /* Theta capital */
119  x = 0x0398;
120  break;
121  case 0xec: /* Iota small */
122  if (tonos)
123  if (dialitika)
124  x = 0x0390;
125  else
126  x = 0x03af;
127  else
128  if (dialitika)
129  x = 0x03ca;
130  else
131  x = 0x03b9;
132  break;
133  case 0xcc: /* Iota capital */
134  if (tonos)
135  x = 0x038a;
136  else
137  if (dialitika)
138  x = 0x03aa;
139  else
140  x = 0x0399;
141  break;
142  case 0xed: /* Kappa small */
143  x = 0x03ba;
144  break;
145  case 0xcd: /* Kappa capital */
146  x = 0x039a;
147  break;
148  case 0xee: /* Lambda small */
149  x = 0x03bb;
150  break;
151  case 0xce: /* Lambda capital */
152  x = 0x039b;
153  break;
154  case 0xef: /* Mu small */
155  x = 0x03bc;
156  break;
157  case 0xcf: /* Mu capital */
158  x = 0x039c;
159  break;
160  case 0xf0: /* Nu small */
161  x = 0x03bd;
162  break;
163  case 0xd0: /* Nu capital */
164  x = 0x039d;
165  break;
166  case 0xf1: /* Xi small */
167  x = 0x03be;
168  break;
169  case 0xd1: /* Xi capital */
170  x = 0x039e;
171  break;
172  case 0xf2: /* Omicron small */
173  if (tonos)
174  x = 0x03cc;
175  else
176  x = 0x03bf;
177  break;
178  case 0xd2: /* Omicron capital */
179  if (tonos)
180  x = 0x038c;
181  else
182  x = 0x039f;
183  break;
184  case 0xf3: /* Pi small */
185  x = 0x03c0;
186  break;
187  case 0xd3: /* Pi capital */
188  x = 0x03a0;
189  break;
190  case 0xf5: /* Rho small */
191  x = 0x03c1;
192  break;
193  case 0xd5: /* Rho capital */
194  x = 0x03a1;
195  break;
196  case 0xf7: /* Sigma small (end of words) */
197  x = 0x03c2;
198  break;
199  case 0xf6: /* Sigma small */
200  x = 0x03c3;
201  break;
202  case 0xd6: /* Sigma capital */
203  x = 0x03a3;
204  break;
205  case 0xf8: /* Tau small */
206  x = 0x03c4;
207  break;
208  case 0xd8: /* Tau capital */
209  x = 0x03a4;
210  break;
211  case 0xf9: /* Upsilon small */
212  if (tonos)
213  if (dialitika)
214  x = 0x03b0;
215  else
216  x = 0x03cd;
217  else
218  if (dialitika)
219  x = 0x03cb;
220  else
221  x = 0x03c5;
222  break;
223  case 0xd9: /* Upsilon capital */
224  if (tonos)
225  x = 0x038e;
226  else
227  if (dialitika)
228  x = 0x03ab;
229  else
230  x = 0x03a5;
231  break;
232  case 0xfa: /* Phi small */
233  x = 0x03c6;
234  break;
235  case 0xda: /* Phi capital */
236  x = 0x03a6;
237  break;
238  case 0xfb: /* Chi small */
239  x = 0x03c7;
240  break;
241  case 0xdb: /* Chi capital */
242  x = 0x03a7;
243  break;
244  case 0xfc: /* Psi small */
245  x = 0x03c8;
246  break;
247  case 0xdc: /* Psi capital */
248  x = 0x03a8;
249  break;
250  case 0xfd: /* Omega small */
251  if (tonos)
252  x = 0x03ce;
253  else
254  x = 0x03c9;
255  break;
256  case 0xdd: /* Omega capital */
257  if (tonos)
258  x = 0x038f;
259  else
260  x = 0x03a9;
261  break;
262  default:
263  x = *inp;
264  break;
265  }
266  (*no_read)++;
267 
268  return x;
269 }
270 
271 static size_t write_iso_5428_1984(yaz_iconv_t cd, yaz_iconv_encoder_t en,
272  unsigned long x,
273  char **outbuf, size_t *outbytesleft)
274 {
275  size_t k = 0;
276  unsigned char *out = (unsigned char*) *outbuf;
277  if (*outbytesleft < 3)
278  {
279  yaz_iconv_set_errno(cd, YAZ_ICONV_E2BIG); /* not room for output */
280  return (size_t)(-1);
281  }
282  switch (x)
283  {
284  case 0x03ac : out[k++]=0xa2; out[k++]=0xe1; break;
285  case 0x03b1 : out[k++]=0xe1; break;
286  case 0x0386 : out[k++]=0xa2; out[k++]=0xc1; break;
287  case 0x0391 : out[k++]=0xc1; break;
288  case 0x03b2 : out[k++]=0xe2; break;
289  case 0x0392 : out[k++]=0xc2; break;
290  case 0x03b3 : out[k++]=0xe4; break;
291  case 0x0393 : out[k++]=0xc4; break;
292  case 0x03b4 : out[k++]=0xe5; break;
293  case 0x0394 : out[k++]=0xc5; break;
294  case 0x03ad : out[k++]=0xa2; out[k++]=0xe6; break;
295  case 0x03b5 : out[k++]=0xe6; break;
296  case 0x0388 : out[k++]=0xa2; out[k++]=0xc6; break;
297  case 0x0395 : out[k++]=0xc6; break;
298  case 0x03b6 : out[k++]=0xe9; break;
299  case 0x0396 : out[k++]=0xc9; break;
300  case 0x03ae : out[k++]=0xa2; out[k++]=0xea; break;
301  case 0x03b7 : out[k++]=0xea; break;
302  case 0x0389 : out[k++]=0xa2; out[k++]=0xca; break;
303  case 0x0397 : out[k++]=0xca; break;
304  case 0x03b8 : out[k++]=0xeb; break;
305  case 0x0398 : out[k++]=0xcb; break;
306  case 0x0390 : out[k++]=0xa2; out[k++]=0xa3; out[k++]=0xec; break;
307  case 0x03af : out[k++]=0xa2; out[k++]=0xec; break;
308  case 0x03ca : out[k++]=0xa3; out[k++]=0xec; break;
309  case 0x03b9 : out[k++]=0xec; break;
310  case 0x038a : out[k++]=0xa2; out[k++]=0xcc; break;
311  case 0x03aa : out[k++]=0xa3; out[k++]=0xcc; break;
312  case 0x0399 : out[k++]=0xcc; break;
313  case 0x03ba : out[k++]=0xed; break;
314  case 0x039a : out[k++]=0xcd; break;
315  case 0x03bb : out[k++]=0xee; break;
316  case 0x039b : out[k++]=0xce; break;
317  case 0x03bc : out[k++]=0xef; break;
318  case 0x039c : out[k++]=0xcf; break;
319  case 0x03bd : out[k++]=0xf0; break;
320  case 0x039d : out[k++]=0xd0; break;
321  case 0x03be : out[k++]=0xf1; break;
322  case 0x039e : out[k++]=0xd1; break;
323  case 0x03cc : out[k++]=0xa2; out[k++]=0xf2; break;
324  case 0x03bf : out[k++]=0xf2; break;
325  case 0x038c : out[k++]=0xa2; out[k++]=0xd2; break;
326  case 0x039f : out[k++]=0xd2; break;
327  case 0x03c0 : out[k++]=0xf3; break;
328  case 0x03a0 : out[k++]=0xd3; break;
329  case 0x03c1 : out[k++]=0xf5; break;
330  case 0x03a1 : out[k++]=0xd5; break;
331  case 0x03c2 : out[k++]=0xf7; break;
332  case 0x03c3 : out[k++]=0xf6; break;
333  case 0x03a3 : out[k++]=0xd6; break;
334  case 0x03c4 : out[k++]=0xf8; break;
335  case 0x03a4 : out[k++]=0xd8; break;
336  case 0x03b0 : out[k++]=0xa2; out[k++]=0xa3; out[k++]=0xf9; break;
337  case 0x03cd : out[k++]=0xa2; out[k++]=0xf9; break;
338  case 0x03cb : out[k++]=0xa3; out[k++]=0xf9; break;
339  case 0x03c5 : out[k++]=0xf9; break;
340  case 0x038e : out[k++]=0xa2; out[k++]=0xd9; break;
341  case 0x03ab : out[k++]=0xa3; out[k++]=0xd9; break;
342  case 0x03a5 : out[k++]=0xd9; break;
343  case 0x03c6 : out[k++]=0xfa; break;
344  case 0x03a6 : out[k++]=0xda; break;
345  case 0x03c7 : out[k++]=0xfb; break;
346  case 0x03a7 : out[k++]=0xdb; break;
347  case 0x03c8 : out[k++]=0xfc; break;
348  case 0x03a8 : out[k++]=0xdc; break;
349  case 0x03ce : out[k++]=0xa2; out[k++]=0xfd; break;
350  case 0x03c9 : out[k++]=0xfd; break;
351  case 0x038f : out[k++]=0xa2; out[k++]=0xdd; break;
352  case 0x03a9 : out[k++]=0xdd; break;
353  default:
354  if (x > 255)
355  {
356  yaz_iconv_set_errno(cd, YAZ_ICONV_EILSEQ);
357  return (size_t) -1;
358  }
359  out[k++] = (unsigned char ) x;
360  break;
361  }
362  *outbytesleft -= k;
363  (*outbuf) += k;
364  return 0;
365 }
366 
367 yaz_iconv_encoder_t yaz_iso_5428_encoder(const char *name,
368  yaz_iconv_encoder_t e)
369 {
370  if (!yaz_matchstr(name, "iso54281984")
371  || !yaz_matchstr(name, "iso5428:1984"))
372  {
373  e->write_handle = write_iso_5428_1984;
374  return e;
375  }
376  return 0;
377 }
378 
379 yaz_iconv_decoder_t yaz_iso_5428_decoder(const char *name,
380  yaz_iconv_decoder_t d)
381 {
382  if (!yaz_matchstr(name, "iso54281984")
383  || !yaz_matchstr(name, "iso5428:1984"))
384  {
385  d->read_handle = read_iso_5428_1984;
386  return d;
387  }
388  return 0;
389 }
390 
391 
392 
393 /*
394  * Local variables:
395  * c-basic-offset: 4
396  * c-file-style: "Stroustrup"
397  * indent-tabs-mode: nil
398  * End:
399  * vim: shiftwidth=4 tabstop=8 expandtab
400  */
401 
errno.h
Header for errno utilities.
iconv-p.h
Internal header for iconv.
yaz_iconv_set_errno
void yaz_iconv_set_errno(yaz_iconv_t cd, int no)
Definition: siconv.c:298
name
char * name
Definition: initopt.c:18
read_iso_5428_1984
static unsigned long read_iso_5428_1984(yaz_iconv_t cd, yaz_iconv_decoder_t d, unsigned char *inp, size_t inbytesleft, size_t *no_read)
Definition: iso5428.c:20
yaz_iso_5428_decoder
yaz_iconv_decoder_t yaz_iso_5428_decoder(const char *name, yaz_iconv_decoder_t d)
Definition: iso5428.c:379
yaz_iso_5428_encoder
yaz_iconv_encoder_t yaz_iso_5428_encoder(const char *name, yaz_iconv_encoder_t e)
Definition: iso5428.c:367
write_iso_5428_1984
static size_t write_iso_5428_1984(yaz_iconv_t cd, yaz_iconv_encoder_t en, unsigned long x, char **outbuf, size_t *outbytesleft)
Definition: iso5428.c:271
yaz_matchstr
int yaz_matchstr(const char *s1, const char *s2)
match strings - independent of case and '-'
Definition: matchstr.c:42
yaz_iconv_decoder_s
Definition: iconv-p.h:81
yaz_iconv_decoder_s::read_handle
unsigned long(* read_handle)(yaz_iconv_t cd, yaz_iconv_decoder_t d, unsigned char *inbuf, size_t inbytesleft, size_t *no_read)
Definition: iconv-p.h:86
yaz_iconv_encoder_s
Definition: iconv-p.h:43
yaz_iconv_encoder_s::write_handle
size_t(* write_handle)(yaz_iconv_t cd, yaz_iconv_encoder_t e, unsigned long x, char **outbuf, size_t *outbytesleft)
Definition: iconv-p.h:45
yaz_iconv_struct
Definition: siconv.c:32
YAZ_ICONV_EILSEQ
#define YAZ_ICONV_EILSEQ
error code: Invalid sequence
Definition: yaz-iconv.h:49
YAZ_ICONV_E2BIG
#define YAZ_ICONV_E2BIG
error code: Not sufficient room for output buffer
Definition: yaz-iconv.h:47
YAZ_ICONV_EINVAL
#define YAZ_ICONV_EINVAL
error code: An incomplete multibyte sequence is in input buffer
Definition: yaz-iconv.h:51
Generated by doxygen 1.9.1