static unsigned long read_iso5426(yaz_iconv_t cd, yaz_iconv_decoder_t d,
unsigned char *inp,
size_t inbytesleft, size_t *no_read)
{
struct decoder_data *data = (struct decoder_data *) d->data;
unsigned long x;
if (data->comb_offset < data->comb_size)
{
*no_read = data->comb_no_read[data->comb_offset];
x = data->comb_x[data->comb_offset];
/* special case for double-diacritic combining characters,
INVERTED BREVE and DOUBLE TILDE.
We'll increment the no_read counter by 1, since we want to skip over
the processing of the closing ligature character
*/
/* this code is no longer necessary.. our handlers code in
yaz_iso5426_?_conv (generated by charconv.tcl) now returns
0 and no_read=1 when a sequence does not match the input.
The SECOND HALFs in codetables.xml produces a non-existant
entry in the conversion trie.. Hence when met, the input byte is
skipped as it should (in yaz_iconv)
*/
#if 0
if (x == 0x0361 || x == 0x0360)
*no_read += 1;
#endif
data->comb_offset++;
return x;
}
data->comb_offset = 0;
for (data->comb_size = 0; data->comb_size < 8; data->comb_size++)
{
int comb = 0;
if (inbytesleft == 0 && data->comb_size)
{
yaz_iconv_set_errno(cd, YAZ_ICONV_EINVAL);
x = 0;
*no_read = 0;
break;
}
x = yaz_read_iso5426_comb(cd, data, inp, inbytesleft, no_read, &comb);
if (!comb || !x)
break;
data->comb_x[data->comb_size] = x;
data->comb_no_read[data->comb_size] = *no_read;
inp += *no_read;
inbytesleft = inbytesleft - *no_read;
}
return x;
}
static size_t write_advancegreek(yaz_iconv_t cd, yaz_iconv_encoder_t w,
unsigned long x,
char **outbuf, size_t *outbytesleft)
{
size_t k = 0;
unsigned char *out = (unsigned char*) *outbuf;
if (*outbytesleft < 3)
{
yaz_iconv_set_errno(cd, YAZ_ICONV_E2BIG); /* not room for output */
return (size_t)(-1);
}
switch (x)
{
case 0x03ac : out[k++]=0x9d; out[k++]=0x81; break;
case 0x03ad : out[k++]=0x9d; out[k++]=0x85; break;
case 0x03ae : out[k++]=0x9d; out[k++]=0x87; break;
case 0x03af : out[k++]=0x9d; out[k++]=0x89; break;
case 0x03cc : out[k++]=0x9d; out[k++]=0x8f; break;
case 0x03cd : out[k++]=0x9d; out[k++]=0x95; break;
case 0x03ce : out[k++]=0x9d; out[k++]=0x99; break;
case 0x0390 : out[k++]=0x9d; out[k++]=0x9e; out[k++]=0x89; break;
case 0x03b0 : out[k++]=0x9d; out[k++]=0x9e; out[k++]=0x95; break;
case 0x0386 : out[k++]=0x9d; out[k++]=0x9f; out[k++]=0x81; break;
case 0x0388 : out[k++]=0x9d; out[k++]=0x9f; out[k++]=0x85; break;
case 0x0389 : out[k++]=0x9d; out[k++]=0x9f; out[k++]=0x87; break;
case 0x038a : out[k++]=0x9d; out[k++]=0x9f; out[k++]=0x89; break;
case 0x038c : out[k++]=0x9d; out[k++]=0x9f; out[k++]=0x8f; break;
case 0x038e : out[k++]=0x9d; out[k++]=0x9f; out[k++]=0x95; break;
case 0x038f : out[k++]=0x9d; out[k++]=0x9f; out[k++]=0x99; break;
case 0x03ca : out[k++]=0x9e; out[k++]=0x89; break;
case 0x03cb : out[k++]=0x9e; out[k++]=0x95; break;
case 0x03aa : out[k++]=0x9e; out[k++]=0x9f; out[k++]=0x89; break;
case 0x03ab : out[k++]=0x9e; out[k++]=0x9f; out[k++]=0x95; break;
case 0x0391 : out[k++]=0x9f; out[k++]=0x81; break;
case 0x0392 : out[k++]=0x9f; out[k++]=0x82; break;
case 0x0393 : out[k++]=0x9f; out[k++]=0x83; break;
case 0x0394 : out[k++]=0x9f; out[k++]=0x84; break;
case 0x0395 : out[k++]=0x9f; out[k++]=0x85; break;
case 0x0396 : out[k++]=0x9f; out[k++]=0x86; break;
case 0x0397 : out[k++]=0x9f; out[k++]=0x87; break;
case 0x0398 : out[k++]=0x9f; out[k++]=0x88; break;
case 0x0399 : out[k++]=0x9f; out[k++]=0x89; break;
case 0x039a : out[k++]=0x9f; out[k++]=0x8a; break;
case 0x039b : out[k++]=0x9f; out[k++]=0x8b; break;
case 0x039c : out[k++]=0x9f; out[k++]=0x8c; break;
case 0x039d : out[k++]=0x9f; out[k++]=0x8d; break;
case 0x039e : out[k++]=0x9f; out[k++]=0x8e; break;
case 0x039f : out[k++]=0x9f; out[k++]=0x8f; break;
case 0x03a0 : out[k++]=0x9f; out[k++]=0x90; break;
case 0x03a1 : out[k++]=0x9f; out[k++]=0x91; break;
case 0x03a3 : out[k++]=0x9f; out[k++]=0x93; break;
case 0x03a4 : out[k++]=0x9f; out[k++]=0x94; break;
case 0x03a5 : out[k++]=0x9f; out[k++]=0x95; break;
case 0x03a6 : out[k++]=0x9f; out[k++]=0x96; break;
case 0x03a7 : out[k++]=0x9f; out[k++]=0x97; break;
case 0x03a8 : out[k++]=0x9f; out[k++]=0x98; break;
case 0x03a9 : out[k++]=0x9f; out[k++]=0x99; break;
case 0x03b1 : out[k++]=0x81; break;
case 0x03b2 : out[k++]=0x82; break;
case 0x03b3 : out[k++]=0x83; break;
case 0x03b4 : out[k++]=0x84; break;
case 0x03b5 : out[k++]=0x85; break;
case 0x03b6 : out[k++]=0x86; break;
case 0x03b7 : out[k++]=0x87; break;
case 0x03b8 : out[k++]=0x88; break;
case 0x03b9 : out[k++]=0x89; break;
case 0x03ba : out[k++]=0x8a; break;
case 0x03bb : out[k++]=0x8b; break;
case 0x03bc : out[k++]=0x8c; break;
case 0x03bd : out[k++]=0x8d; break;
case 0x03be : out[k++]=0x8e; break;
case 0x03bf : out[k++]=0x8f; break;
case 0x03c0 : out[k++]=0x90; break;
case 0x03c1 : out[k++]=0x91; break;
case 0x03c2 : out[k++]=0x92; break;
case 0x03c3 : out[k++]=0x93; break;
case 0x03c4 : out[k++]=0x94; break;
case 0x03c5 : out[k++]=0x95; break;
case 0x03c6 : out[k++]=0x96; break;
case 0x03c7 : out[k++]=0x97; break;
case 0x03c8 : out[k++]=0x98; break;
case 0x03c9 : out[k++]=0x99; break;
default:
if (x > 255)
{
yaz_iconv_set_errno(cd, YAZ_ICONV_EILSEQ);
return (size_t) -1;
}
out[k++] = (unsigned char ) x;
break;
}
*outbytesleft -= k;
(*outbuf) += k;
return 0;
}
static unsigned long read_advancegreek(yaz_iconv_t cd, yaz_iconv_decoder_t d,
unsigned char *inp,
size_t inbytesleft, size_t *no_read)
{
unsigned long x = 0;
int shift = 0;
int tonos = 0;
int dialitika = 0;
*no_read = 0;
while (inbytesleft > 0)
{
if (*inp == 0x9d)
{
tonos = 1;
}
else if (*inp == 0x9e)
{
dialitika = 1;
}
else if (*inp == 0x9f)
{
shift = 1;
}
else
break;
inp++;
--inbytesleft;
(*no_read)++;
}
if (inbytesleft == 0)
{
yaz_iconv_set_errno(cd, YAZ_ICONV_EINVAL); /* incomplete input */
*no_read = 0;
return 0;
}
switch (*inp) {
case 0x81:
if (shift)
if (tonos)
x = 0x0386;
else
x = 0x0391;
else
if (tonos)
x = 0x03ac;
else
x = 0x03b1;
break;
case 0x82:
if (shift)
x = 0x0392;
else
x = 0x03b2;
break;
case 0x83:
if (shift)
x = 0x0393;
else
x = 0x03b3;
break;
case 0x84:
if (shift)
x = 0x0394;
else
x = 0x03b4;
break;
case 0x85:
if (shift)
if (tonos)
x = 0x0388;
else
x = 0x0395;
else
if (tonos)
x = 0x03ad;
else
x = 0x03b5;
break;
case 0x86:
if (shift)
x = 0x0396;
else
x = 0x03b6;
break;
case 0x87:
if (shift)
if (tonos)
x = 0x0389;
else
x = 0x0397;
else
if (tonos)
x = 0x03ae;
else
x = 0x03b7;
break;
case 0x88:
if (shift)
x = 0x0398;
else
x = 0x03b8;
break;
case 0x89:
if (shift)
if (tonos)
x = 0x038a;
else
if (dialitika)
x = 0x03aa;
else
x = 0x0399;
else
if (tonos)
if (dialitika)
x = 0x0390;
else
x = 0x03af;
else
if (dialitika)
x = 0x03ca;
else
x = 0x03b9;
break;
case 0x8a:
if (shift)
x = 0x039a;
else
x = 0x03ba;
break;
case 0x8b:
if (shift)
x = 0x039b;
else
x = 0x03bb;
break;
case 0x8c:
if (shift)
x = 0x039c;
else
x = 0x03bc;
break;
case 0x8d:
if (shift)
x = 0x039d;
else
x = 0x03bd;
break;
case 0x8e:
if (shift)
x = 0x039e;
else
x = 0x03be;
break;
case 0x8f:
if (shift)
if (tonos)
x = 0x038c;
else
x = 0x039f;
else
if (tonos)
x = 0x03cc;
else
x = 0x03bf;
break;
case 0x90:
if (shift)
x = 0x03a0;
else
x = 0x03c0;
break;
case 0x91:
if (shift)
x = 0x03a1;
else
x = 0x03c1;
break;
case 0x92:
x = 0x03c2;
break;
case 0x93:
if (shift)
x = 0x03a3;
else
x = 0x03c3;
break;
case 0x94:
if (shift)
x = 0x03a4;
else
x = 0x03c4;
break;
case 0x95:
if (shift)
if (tonos)
x = 0x038e;
else
if (dialitika)
x = 0x03ab;
else
x = 0x03a5;
else
if (tonos)
if (dialitika)
x = 0x03b0;
else
x = 0x03cd;
else
if (dialitika)
x = 0x03cb;
else
x = 0x03c5;
break;
case 0x96:
if (shift)
x = 0x03a6;
else
x = 0x03c6;
break;
case 0x97:
if (shift)
x = 0x03a7;
else
x = 0x03c7;
break;
case 0x98:
if (shift)
x = 0x03a8;
else
x = 0x03c8;
break;
case 0x99:
if (shift)
if (tonos)
x = 0x038f;
else
x = 0x03a9;
else
if (tonos)
x = 0x03ce;
else
x = 0x03c9;
break;
default:
x = *inp;
break;
}
(*no_read)++;
return x;
}
static unsigned long yaz_read_iso5426_comb(yaz_iconv_t cd,
struct decoder_data *data,
unsigned char *inp,
size_t inbytesleft, size_t *no_read,
int *comb)
{
*no_read = 0;
while (inbytesleft > 0 && *inp == 27)
{
int *modep = &data->g0_mode;
size_t inbytesleft0 = inbytesleft;
inbytesleft--;
inp++;
if (inbytesleft == 0)
goto incomplete;
if (*inp == '$') /* set with multiple bytes */
{
inbytesleft--;
inp++;
}
if (inbytesleft == 0)
goto incomplete;
if (*inp == '(' || *inp == ',') /* G0 */
{
inbytesleft--;
inp++;
}
else if (*inp == ')' || *inp == '-') /* G1 */
{
inbytesleft--;
inp++;
modep = &data->g1_mode;
}
if (inbytesleft == 0)
goto incomplete;
if (*inp == '!') /* ANSEL is a special case */
{
inbytesleft--;
inp++;
}
if (inbytesleft == 0)
goto incomplete;
*modep = *inp++; /* Final character */
inbytesleft--;
(*no_read) += inbytesleft0 - inbytesleft;
}
if (inbytesleft == 0)
return 0;
else if (*inp == ' ')
{
*no_read += 1;
return ' ';
}
else
{
unsigned long x;
size_t no_read_sub = 0;
int mode = *inp < 128 ? data->g0_mode : data->g1_mode;
*comb = 0;
switch(mode)
{
case 'B': /* Basic ASCII */
case 's': /* ASCII */
x = yaz_iso5426_42_conv(inp, inbytesleft, &no_read_sub, comb, 127, 0);
break;
case 'E': /* ANSEL */
x = yaz_iso5426_45_conv(inp, inbytesleft, &no_read_sub, comb, 127, 128);
break;
#if 0
case 'g': /* Greek */
x = yaz_iso5426_67_conv(inp, inbytesleft, &no_read_sub, comb, 127, 0);
break;
case 'b': /* Subscripts */
x = yaz_iso5426_62_conv(inp, inbytesleft, &no_read_sub, comb, 127, 0);
break;
case 'p': /* Superscripts */
x = yaz_iso5426_70_conv(inp, inbytesleft, &no_read_sub, comb, 127, 0);
break;
case '2': /* Basic Hebrew */
x = yaz_iso5426_32_conv(inp, inbytesleft, &no_read_sub, comb, 127, 0);
break;
case 'N': /* Basic Cyrillic */
x = yaz_iso5426_4E_conv(inp, inbytesleft, &no_read_sub, comb, 127, 0);
break;
case 'Q': /* Extended Cyrillic */
x = yaz_iso5426_51_conv(inp, inbytesleft, &no_read_sub, comb, 127, 0);
break;
case '3': /* Basic Arabic */
x = yaz_iso5426_33_conv(inp, inbytesleft, &no_read_sub, comb, 127, 0);
break;
case '4': /* Extended Arabic */
x = yaz_iso5426_34_conv(inp, inbytesleft, &no_read_sub, comb, 127, 0);
break;
case 'S': /* Greek */
x = yaz_iso5426_53_conv(inp, inbytesleft, &no_read_sub, comb, 127, 0);
break;
case '1': /* Chinese, Japanese, Korean (EACC) */
x = yaz_iso5426_31_conv(inp, inbytesleft, &no_read_sub, comb, 127, 0);
break;
#endif
default:
*no_read = 0;
yaz_iconv_set_errno(cd, YAZ_ICONV_EILSEQ);
return 0;
}
*no_read += no_read_sub;
return x;
}
incomplete:
*no_read = 0;
yaz_iconv_set_errno(cd, YAZ_ICONV_EINVAL);
return 0;
}
