/*
- * Copyright (C) 1995-2006, Index Data ApS
+ * Copyright (C) 1995-2007, Index Data ApS
* See the file LICENSE for details.
*
- * $Id: siconv.c,v 1.30 2006-12-19 22:41:28 adam Exp $
+ * $Id: siconv.c,v 1.42 2007-05-30 08:22:03 adam Exp $
*/
/**
* \file siconv.c
unsigned long (*read_handle)(yaz_iconv_t cd, unsigned char *inbuf,
size_t inbytesleft, size_t *no_read);
size_t (*write_handle)(yaz_iconv_t cd, unsigned long x,
- char **outbuf, size_t *outbytesleft,
- int last);
+ char **outbuf, size_t *outbytesleft);
+ size_t (*flush_handle)(yaz_iconv_t cd,
+ char **outbuf, size_t *outbytesleft);
int marc8_esc_mode;
int comb_offset;
unsigned long write_marc8_comb_ch[8];
size_t write_marc8_comb_no;
+ unsigned write_marc8_second_half_char;
unsigned long write_marc8_last;
const char *write_marc8_page_chr;
};
{
unsigned long x = 0;
+ *no_read = 0; /* by default */
if (inp[0] <= 0x7f)
{
x = inp[0];
}
else if (inp[0] <= 0xbf || inp[0] >= 0xfe)
{
- *no_read = 0;
*error = YAZ_ICONV_EILSEQ;
}
else if (inp[0] <= 0xdf && inbytesleft >= 2)
{
- x = ((inp[0] & 0x1f) << 6) | (inp[1] & 0x3f);
- if (x >= 0x80)
- *no_read = 2;
- else
+ if ((inp[1] & 0xc0) == 0x80)
{
- *no_read = 0;
- *error = YAZ_ICONV_EILSEQ;
+ x = ((inp[0] & 0x1f) << 6) | (inp[1] & 0x3f);
+ if (x >= 0x80)
+ *no_read = 2;
+ else
+ *error = YAZ_ICONV_EILSEQ;
}
+ else
+ *error = YAZ_ICONV_EILSEQ;
}
else if (inp[0] <= 0xef && inbytesleft >= 3)
{
- x = ((inp[0] & 0x0f) << 12) | ((inp[1] & 0x3f) << 6) |
- (inp[2] & 0x3f);
- if (x >= 0x800)
- *no_read = 3;
- else
+ if ((inp[1] & 0xc0) == 0x80 && (inp[2] & 0xc0) == 0x80)
{
- *no_read = 0;
- *error = YAZ_ICONV_EILSEQ;
+ x = ((inp[0] & 0x0f) << 12) | ((inp[1] & 0x3f) << 6) |
+ (inp[2] & 0x3f);
+ if (x >= 0x800)
+ *no_read = 3;
+ else
+ *error = YAZ_ICONV_EILSEQ;
}
- }
+ else
+ *error = YAZ_ICONV_EILSEQ;
+ }
else if (inp[0] <= 0xf7 && inbytesleft >= 4)
{
- x = ((inp[0] & 0x07) << 18) | ((inp[1] & 0x3f) << 12) |
- ((inp[2] & 0x3f) << 6) | (inp[3] & 0x3f);
- if (x >= 0x10000)
- *no_read = 4;
- else
+ if ((inp[1] & 0xc0) == 0x80 && (inp[2] & 0xc0) == 0x80
+ && (inp[3] & 0xc0) == 0x80)
{
- *no_read = 0;
- *error = YAZ_ICONV_EILSEQ;
+ x = ((inp[0] & 0x07) << 18) | ((inp[1] & 0x3f) << 12) |
+ ((inp[2] & 0x3f) << 6) | (inp[3] & 0x3f);
+ if (x >= 0x10000)
+ *no_read = 4;
+ else
+ *error = YAZ_ICONV_EILSEQ;
}
+ else
+ *error = YAZ_ICONV_EILSEQ;
}
else if (inp[0] <= 0xfb && inbytesleft >= 5)
{
- x = ((inp[0] & 0x03) << 24) | ((inp[1] & 0x3f) << 18) |
- ((inp[2] & 0x3f) << 12) | ((inp[3] & 0x3f) << 6) |
- (inp[4] & 0x3f);
- if (x >= 0x200000)
- *no_read = 5;
- else
+ if ((inp[1] & 0xc0) == 0x80 && (inp[2] & 0xc0) == 0x80
+ && (inp[3] & 0xc0) == 0x80 && (inp[4] & 0xc0) == 0x80)
{
- *no_read = 0;
- *error = YAZ_ICONV_EILSEQ;
+ x = ((inp[0] & 0x03) << 24) | ((inp[1] & 0x3f) << 18) |
+ ((inp[2] & 0x3f) << 12) | ((inp[3] & 0x3f) << 6) |
+ (inp[4] & 0x3f);
+ if (x >= 0x200000)
+ *no_read = 5;
+ else
+ *error = YAZ_ICONV_EILSEQ;
}
+ else
+ *error = YAZ_ICONV_EILSEQ;
}
else if (inp[0] <= 0xfd && inbytesleft >= 6)
{
- x = ((inp[0] & 0x01) << 30) | ((inp[1] & 0x3f) << 24) |
- ((inp[2] & 0x3f) << 18) | ((inp[3] & 0x3f) << 12) |
- ((inp[4] & 0x3f) << 6) | (inp[5] & 0x3f);
- if (x >= 0x4000000)
- *no_read = 6;
- else
+ if ((inp[1] & 0xc0) == 0x80 && (inp[2] & 0xc0) == 0x80
+ && (inp[3] & 0xc0) == 0x80 && (inp[4] & 0xc0) == 0x80
+ && (inp[5] & 0xc0) == 0x80)
{
- *no_read = 0;
- *error = YAZ_ICONV_EILSEQ;
+ x = ((inp[0] & 0x01) << 30) | ((inp[1] & 0x3f) << 24) |
+ ((inp[2] & 0x3f) << 18) | ((inp[3] & 0x3f) << 12) |
+ ((inp[4] & 0x3f) << 6) | (inp[5] & 0x3f);
+ if (x >= 0x4000000)
+ *no_read = 6;
+ else
+ *error = YAZ_ICONV_EILSEQ;
}
+ else
+ *error = YAZ_ICONV_EILSEQ;
}
else
- {
- *no_read = 0;
- *error = YAZ_ICONV_EINVAL;
- }
+ *error = YAZ_ICONV_EINVAL; /* incomplete sentence */
+
return x;
}
}
#endif
+static unsigned long yaz_read_iso5428_1984(yaz_iconv_t cd, unsigned char *inp,
+ size_t inbytesleft, size_t *no_read)
+{
+ unsigned long x = 0;
+ int tonos = 0;
+ int dialitika = 0;
+
+ *no_read = 0;
+ while (inbytesleft > 0)
+ {
+ if (*inp == 0xa2)
+ {
+ tonos = 1;
+ }
+ else if (*inp == 0xa3)
+ {
+ dialitika = 1;
+ }
+ else
+ break;
+ inp++;
+ --inbytesleft;
+ (*no_read)++;
+ }
+ if (inbytesleft == 0)
+ {
+ cd->my_errno = YAZ_ICONV_EINVAL; /* incomplete input */
+ *no_read = 0;
+ return 0;
+ }
+ switch (*inp) {
+ case 0xe1: /* alpha small */
+ if (tonos)
+ x = 0x03ac;
+ else
+ x = 0x03b1;
+ break;
+ case 0xc1: /* alpha capital */
+ if (tonos)
+ x = 0x0386;
+ else
+ x = 0x0391;
+ break;
+
+ case 0xe2: /* Beta small */
+ x = 0x03b2;
+ break;
+ case 0xc2: /* Beta capital */
+ x = 0x0392;
+ break;
+
+ case 0xe4: /* Gamma small */
+ x = 0x03b3;
+ break;
+ case 0xc4: /* Gamma capital */
+ x = 0x0393;
+ break;
+
+ case 0xe5: /* Delta small */
+ x = 0x03b4;
+ break;
+ case 0xc5: /* Delta capital */
+ x = 0x0394;
+ break;
+ case 0xe6: /* epsilon small */
+ if (tonos)
+ x = 0x03ad;
+ else
+ x = 0x03b5;
+ break;
+ case 0xc6: /* epsilon capital */
+ if (tonos)
+ x = 0x0388;
+ else
+ x = 0x0395;
+ break;
+ case 0xe9: /* Zeta small */
+ x = 0x03b6;
+ break;
+ case 0xc9: /* Zeta capital */
+ x = 0x0396;
+ break;
+ case 0xea: /* Eta small */
+ if (tonos)
+ x = 0x03ae;
+ else
+ x = 0x03b7;
+ break;
+ case 0xca: /* Eta capital */
+ if (tonos)
+ x = 0x0389;
+ else
+ x = 0x0397;
+ break;
+ case 0xeb: /* Theta small */
+ x = 0x03b8;
+ break;
+ case 0xcb: /* Theta capital */
+ x = 0x0398;
+ break;
+ case 0xec: /* Iota small */
+ if (tonos)
+ if (dialitika)
+ x = 0x0390;
+ else
+ x = 0x03af;
+ else
+ if (dialitika)
+ x = 0x03ca;
+ else
+ x = 0x03b9;
+ break;
+ case 0xcc: /* Iota capital */
+ if (tonos)
+ x = 0x038a;
+ else
+ if (dialitika)
+ x = 0x03aa;
+ else
+ x = 0x0399;
+ break;
+ case 0xed: /* Kappa small */
+ x = 0x03ba;
+ break;
+ case 0xcd: /* Kappa capital */
+ x = 0x039a;
+ break;
+ case 0xee: /* Lambda small */
+ x = 0x03bb;
+ break;
+ case 0xce: /* Lambda capital */
+ x = 0x039b;
+ break;
+ case 0xef: /* Mu small */
+ x = 0x03bc;
+ break;
+ case 0xcf: /* Mu capital */
+ x = 0x039c;
+ break;
+ case 0xf0: /* Nu small */
+ x = 0x03bd;
+ break;
+ case 0xd0: /* Nu capital */
+ x = 0x039d;
+ break;
+ case 0xf1: /* Xi small */
+ x = 0x03be;
+ break;
+ case 0xd1: /* Xi capital */
+ x = 0x039e;
+ break;
+ case 0xf2: /* Omicron small */
+ if (tonos)
+ x = 0x03cc;
+ else
+ x = 0x03bf;
+ break;
+ case 0xd2: /* Omicron capital */
+ if (tonos)
+ x = 0x038c;
+ else
+ x = 0x039f;
+ break;
+ case 0xf3: /* Pi small */
+ x = 0x03c0;
+ break;
+ case 0xd3: /* Pi capital */
+ x = 0x03a0;
+ break;
+ case 0xf5: /* Rho small */
+ x = 0x03c1;
+ break;
+ case 0xd5: /* Rho capital */
+ x = 0x03a1;
+ break;
+ case 0xf7: /* Sigma small (end of words) */
+ x = 0x03c2;
+ break;
+ case 0xf6: /* Sigma small */
+ x = 0x03c3;
+ break;
+ case 0xd6: /* Sigma capital */
+ x = 0x03a3;
+ break;
+ case 0xf8: /* Tau small */
+ x = 0x03c4;
+ break;
+ case 0xd8: /* Tau capital */
+ x = 0x03a4;
+ break;
+ case 0xf9: /* Upsilon small */
+ if (tonos)
+ if (dialitika)
+ x = 0x03b0;
+ else
+ x = 0x03cd;
+ else
+ if (dialitika)
+ x = 0x03cb;
+ else
+ x = 0x03c5;
+ break;
+ case 0xd9: /* Upsilon capital */
+ if (tonos)
+ x = 0x038e;
+ else
+ if (dialitika)
+ x = 0x03ab;
+ else
+ x = 0x03a5;
+ break;
+ case 0xfa: /* Phi small */
+ x = 0x03c6;
+ break;
+ case 0xda: /* Phi capital */
+ x = 0x03a6;
+ break;
+ case 0xfb: /* Chi small */
+ x = 0x03c7;
+ break;
+ case 0xdb: /* Chi capital */
+ x = 0x03a7;
+ break;
+ case 0xfc: /* Psi small */
+ x = 0x03c8;
+ break;
+ case 0xdc: /* Psi capital */
+ x = 0x03a8;
+ break;
+ case 0xfd: /* Omega small */
+ if (tonos)
+ x = 0x03ce;
+ else
+ x = 0x03c9;
+ break;
+ case 0xdd: /* Omega capital */
+ if (tonos)
+ x = 0x038f;
+ else
+ x = 0x03a9;
+ break;
+ default:
+ x = *inp;
+ break;
+ }
+ (*no_read)++;
+
+ return x;
+}
+
+static size_t yaz_write_iso5428_1984(yaz_iconv_t cd, unsigned long x,
+ char **outbuf, size_t *outbytesleft)
+{
+ size_t k = 0;
+ unsigned char *out = (unsigned char*) *outbuf;
+ if (*outbytesleft < 3)
+ {
+ cd->my_errno = YAZ_ICONV_E2BIG; /* not room for output */
+ return (size_t)(-1);
+ }
+ switch (x)
+ {
+ case 0x03ac : out[k++]=0xa2; out[k++]=0xe1; break;
+ case 0x03b1 : out[k++]=0xe1; break;
+ case 0x0386 : out[k++]=0xa2; out[k++]=0xc1; break;
+ case 0x0391 : out[k++]=0xc1; break;
+ case 0x03b2 : out[k++]=0xe2; break;
+ case 0x0392 : out[k++]=0xc2; break;
+ case 0x03b3 : out[k++]=0xe4; break;
+ case 0x0393 : out[k++]=0xc4; break;
+ case 0x03b4 : out[k++]=0xe5; break;
+ case 0x0394 : out[k++]=0xc5; break;
+ case 0x03ad : out[k++]=0xa2; out[k++]=0xe6; break;
+ case 0x03b5 : out[k++]=0xe6; break;
+ case 0x0388 : out[k++]=0xa2; out[k++]=0xc6; break;
+ case 0x0395 : out[k++]=0xc6; break;
+ case 0x03b6 : out[k++]=0xe9; break;
+ case 0x0396 : out[k++]=0xc9; break;
+ case 0x03ae : out[k++]=0xa2; out[k++]=0xea; break;
+ case 0x03b7 : out[k++]=0xea; break;
+ case 0x0389 : out[k++]=0xa2; out[k++]=0xca; break;
+ case 0x0397 : out[k++]=0xca; break;
+ case 0x03b8 : out[k++]=0xeb; break;
+ case 0x0398 : out[k++]=0xcb; break;
+ case 0x0390 : out[k++]=0xa2; out[k++]=0xa3; out[k++]=0xec; break;
+ case 0x03af : out[k++]=0xa2; out[k++]=0xec; break;
+ case 0x03ca : out[k++]=0xa3; out[k++]=0xec; break;
+ case 0x03b9 : out[k++]=0xec; break;
+ case 0x038a : out[k++]=0xa2; out[k++]=0xcc; break;
+ case 0x03aa : out[k++]=0xa3; out[k++]=0xcc; break;
+ case 0x0399 : out[k++]=0xcc; break;
+ case 0x03ba : out[k++]=0xed; break;
+ case 0x039a : out[k++]=0xcd; break;
+ case 0x03bb : out[k++]=0xee; break;
+ case 0x039b : out[k++]=0xce; break;
+ case 0x03bc : out[k++]=0xef; break;
+ case 0x039c : out[k++]=0xcf; break;
+ case 0x03bd : out[k++]=0xf0; break;
+ case 0x039d : out[k++]=0xd0; break;
+ case 0x03be : out[k++]=0xf1; break;
+ case 0x039e : out[k++]=0xd1; break;
+ case 0x03cc : out[k++]=0xa2; out[k++]=0xf2; break;
+ case 0x03bf : out[k++]=0xf2; break;
+ case 0x038c : out[k++]=0xa2; out[k++]=0xd2; break;
+ case 0x039f : out[k++]=0xd2; break;
+ case 0x03c0 : out[k++]=0xf3; break;
+ case 0x03a0 : out[k++]=0xd3; break;
+ case 0x03c1 : out[k++]=0xf5; break;
+ case 0x03a1 : out[k++]=0xd5; break;
+ case 0x03c2 : out[k++]=0xf7; break;
+ case 0x03c3 : out[k++]=0xf6; break;
+ case 0x03a3 : out[k++]=0xd6; break;
+ case 0x03c4 : out[k++]=0xf8; break;
+ case 0x03a4 : out[k++]=0xd8; break;
+ case 0x03b0 : out[k++]=0xa2; out[k++]=0xa3; out[k++]=0xf9; break;
+ case 0x03cd : out[k++]=0xa2; out[k++]=0xf9; break;
+ case 0x03cb : out[k++]=0xa3; out[k++]=0xf9; break;
+ case 0x03c5 : out[k++]=0xf9; break;
+ case 0x038e : out[k++]=0xa2; out[k++]=0xd9; break;
+ case 0x03ab : out[k++]=0xa3; out[k++]=0xd9; break;
+ case 0x03a5 : out[k++]=0xd9; break;
+ case 0x03c6 : out[k++]=0xfa; break;
+ case 0x03a6 : out[k++]=0xda; break;
+ case 0x03c7 : out[k++]=0xfb; break;
+ case 0x03a7 : out[k++]=0xdb; break;
+ case 0x03c8 : out[k++]=0xfc; break;
+ case 0x03a8 : out[k++]=0xdc; break;
+ case 0x03ce : out[k++]=0xa2; out[k++]=0xfd; break;
+ case 0x03c9 : out[k++]=0xfd; break;
+ case 0x038f : out[k++]=0xa2; out[k++]=0xdd; break;
+ case 0x03a9 : out[k++]=0xdd; break;
+ default:
+ if (x > 255)
+ {
+ cd->my_errno = YAZ_ICONV_EILSEQ;
+ return (size_t) -1;
+ }
+ out[k++] = x;
+ break;
+ }
+ *outbytesleft -= k;
+ (*outbuf) += k;
+ return 0;
+}
+
+static unsigned long yaz_read_advancegreek(yaz_iconv_t cd, 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)
+ {
+ cd->my_errno = 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 size_t yaz_write_advancegreek(yaz_iconv_t cd, unsigned long x,
+ char **outbuf, size_t *outbytesleft)
+{
+ size_t k = 0;
+ unsigned char *out = (unsigned char*) *outbuf;
+ if (*outbytesleft < 3)
+ {
+ cd->my_errno = 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++]=0x96; break;
+ case 0x03c8 : out[k++]=0x98; break;
+ case 0x03c9 : out[k++]=0x99; break;
+ default:
+ if (x > 255)
+ {
+ cd->my_errno = YAZ_ICONV_EILSEQ;
+ return (size_t) -1;
+ }
+ out[k++] = x;
+ break;
+ }
+ *outbytesleft -= k;
+ (*outbuf) += k;
+ return 0;
+}
+
static unsigned long yaz_read_marc8_comb (yaz_iconv_t cd, unsigned char *inp,
size_t inbytesleft, size_t *no_read,
for (cd->comb_size = 0; cd->comb_size < 8; cd->comb_size++)
{
int comb = 0;
+
+ if (inbytesleft == 0 && cd->comb_size)
+ {
+ cd->my_errno = YAZ_ICONV_EINVAL;
+ x = 0;
+ *no_read = 0;
+ break;
+ }
x = yaz_read_marc8_comb(cd, inp, inbytesleft, no_read, &comb);
if (!comb || !x)
break;
}
static size_t yaz_write_UTF8(yaz_iconv_t cd, unsigned long x,
- char **outbuf, size_t *outbytesleft,
- int last)
+ char **outbuf, size_t *outbytesleft)
{
return yaz_write_UTF8_char(x, outbuf, outbytesleft, &cd->my_errno);
}
return 0;
}
-
static size_t yaz_write_ISO8859_1 (yaz_iconv_t cd, unsigned long x,
- char **outbuf, size_t *outbytesleft,
- int last)
+ char **outbuf, size_t *outbytesleft)
{
/* list of two char unicode sequence that, when combined, are
equivalent to single unicode chars that can be represented in
cd->compose_char = 0;
}
- if (!last && x > 32 && x < 127 && cd->compose_char == 0)
+ if (x > 32 && x < 127 && cd->compose_char == 0)
{
cd->compose_char = x;
return 0;
return 0;
}
+static size_t yaz_flush_ISO8859_1(yaz_iconv_t cd,
+ char **outbuf, size_t *outbytesleft)
+{
+ if (cd->compose_char)
+ {
+ unsigned char *outp = (unsigned char *) *outbuf;
+ if (*outbytesleft < 1)
+ {
+ cd->my_errno = YAZ_ICONV_E2BIG;
+ return (size_t)(-1);
+ }
+ *outp++ = (unsigned char) cd->compose_char;
+ (*outbytesleft)--;
+ *outbuf = (char *) outp;
+ cd->compose_char = 0;
+ }
+ return 0;
+}
static size_t yaz_write_UCS4 (yaz_iconv_t cd, unsigned long x,
- char **outbuf, size_t *outbytesleft,
- int last)
+ char **outbuf, size_t *outbytesleft)
{
unsigned char *outp = (unsigned char *) *outbuf;
if (*outbytesleft >= 4)
}
static size_t yaz_write_UCS4LE (yaz_iconv_t cd, unsigned long x,
- char **outbuf, size_t *outbytesleft,
- int last)
+ char **outbuf, size_t *outbytesleft)
{
unsigned char *outp = (unsigned char *) *outbuf;
if (*outbytesleft >= 4)
char *utf8_outbuf = utf8_buf;
size_t utf8_outbytesleft = sizeof(utf8_buf)-1, r;
- r = yaz_write_UTF8(cd, x, &utf8_outbuf, &utf8_outbytesleft, 0);
+ r = yaz_write_UTF8(cd, x, &utf8_outbuf, &utf8_outbytesleft);
if (r == (size_t)(-1))
{
cd->my_errno = YAZ_ICONV_EILSEQ;
char **outbuf, size_t *outbytesleft)
{
unsigned long y = cd->write_marc8_last;
- unsigned char byte, second_half = 0;
+ unsigned char byte;
char out_buf[10];
size_t i, out_no = 0;
{
/* all MARC-8 combined characters are simple bytes */
byte = (unsigned char )(cd->write_marc8_comb_ch[i]);
- if (byte == 0xEB)
- second_half = 0xEC;
- else if (byte == 0xFA)
- second_half = 0xFB;
-
*(*outbuf)++ = byte;
(*outbytesleft)--;
}
memcpy(*outbuf, out_buf, out_no);
*outbuf += out_no;
(*outbytesleft) -= out_no;
- if (second_half)
+ if (cd->write_marc8_second_half_char)
{
- *(*outbuf)++ = second_half;
+ *(*outbuf)++ = cd->write_marc8_second_half_char;
(*outbytesleft)--;
}
cd->write_marc8_last = 0;
cd->write_marc8_comb_no = 0;
+ cd->write_marc8_second_half_char = 0;
return 0;
}
+static size_t yaz_write_marc8_page_chr(yaz_iconv_t cd,
+ char **outbuf, size_t *outbytesleft,
+ const char *page_chr)
+{
+ const char *old_page_chr = cd->write_marc8_page_chr;
+ if (strcmp(page_chr, old_page_chr))
+ {
+ size_t plen = 0;
+ const char *page_out = page_chr;
+
+ if (*outbytesleft < 8)
+ {
+ cd->my_errno = YAZ_ICONV_E2BIG;
+
+ return (size_t) (-1);
+ }
+ cd->write_marc8_page_chr = page_chr;
+
+ if (!strcmp(old_page_chr, "\033p")
+ || !strcmp(old_page_chr, "\033g")
+ || !strcmp(old_page_chr, "\033b"))
+ {
+ /* Technique 1 leave */
+ page_out = "\033s";
+ if (strcmp(page_chr, "\033(B")) /* Not going ASCII page? */
+ {
+ /* Must leave script + enter new page */
+ plen = strlen(page_out);
+ memcpy(*outbuf, page_out, plen);
+ (*outbuf) += plen;
+ (*outbytesleft) -= plen;
+ page_out = page_chr;
+ }
+ }
+ plen = strlen(page_out);
+ memcpy(*outbuf, page_out, plen);
+ (*outbuf) += plen;
+ (*outbytesleft) -= plen;
+ }
+ return 0;
+}
+
+
static size_t yaz_write_marc8_2(yaz_iconv_t cd, unsigned long x,
- char **outbuf, size_t *outbytesleft,
- int last)
+ char **outbuf, size_t *outbytesleft)
{
int comb = 0;
const char *page_chr = 0;
if (comb)
{
+ if (x == 0x0361)
+ cd->write_marc8_second_half_char = 0xEC;
+ else if (x == 0x0360)
+ cd->write_marc8_second_half_char = 0xFB;
+
if (cd->write_marc8_comb_no < 6)
cd->write_marc8_comb_ch[cd->write_marc8_comb_no++] = y;
}
else
{
size_t r = flush_combos(cd, outbuf, outbytesleft);
- const char *old_page_chr = cd->write_marc8_page_chr;
if (r)
return r;
- if (strcmp(page_chr, old_page_chr))
- {
- size_t plen = 0;
- const char *page_out = page_chr;
-
- if (*outbytesleft < 8)
- {
- cd->my_errno = YAZ_ICONV_E2BIG;
-
- return (size_t) (-1);
- }
- cd->write_marc8_page_chr = page_chr;
- if (!strcmp(old_page_chr, "\033p")
- || !strcmp(old_page_chr, "\033g")
- || !strcmp(old_page_chr, "\033b"))
- {
- /* Technique 1 leave */
- page_out = "\033s";
- if (strcmp(page_chr, "\033(B")) /* Not going ASCII page? */
- {
- /* Must leave script + enter new page */
- plen = strlen(page_out);
- memcpy(*outbuf, page_out, plen);
- (*outbuf) += plen;
- (*outbytesleft) -= plen;
- page_out = page_chr;
- }
- }
- plen = strlen(page_out);
- memcpy(*outbuf, page_out, plen);
- (*outbuf) += plen;
- (*outbytesleft) -= plen;
- }
- cd->write_marc8_last = y;
- }
- if (last)
- {
- size_t r = flush_combos(cd, outbuf, outbytesleft);
+ r = yaz_write_marc8_page_chr(cd, outbuf, outbytesleft, page_chr);
if (r)
- {
- if (comb)
- cd->write_marc8_comb_no--;
- else
- cd->write_marc8_last = 0;
return r;
- }
+ cd->write_marc8_last = y;
}
return 0;
}
+static size_t yaz_flush_marc8(yaz_iconv_t cd,
+ char **outbuf, size_t *outbytesleft)
+{
+ size_t r = flush_combos(cd, outbuf, outbytesleft);
+ if (r)
+ return r;
+ return yaz_write_marc8_page_chr(cd, outbuf, outbytesleft, "\033(B");
+}
+
static size_t yaz_write_marc8(yaz_iconv_t cd, unsigned long x,
- char **outbuf, size_t *outbytesleft,
- int last)
+ char **outbuf, size_t *outbytesleft)
{
int i;
for (i = 0; latin1_comb[i].x1; i++)
int last_ch = cd->write_marc8_last;
r = yaz_write_marc8_2(cd, latin1_comb[i].x1,
- outbuf, outbytesleft, 0);
+ outbuf, outbytesleft);
if (r)
return r;
r = yaz_write_marc8_2(cd, latin1_comb[i].x2,
- outbuf, outbytesleft, last);
+ outbuf, outbytesleft);
if (r && cd->my_errno == YAZ_ICONV_E2BIG)
{
/* not enough room. reset output to original values */
return r;
}
}
- return yaz_write_marc8_2(cd, x, outbuf, outbytesleft, last);
+ return yaz_write_marc8_2(cd, x, outbuf, outbytesleft);
}
#if HAVE_WCHAR_H
-static size_t yaz_write_wchar_t (yaz_iconv_t cd, unsigned long x,
- char **outbuf, size_t *outbytesleft,
- int last)
+static size_t yaz_write_wchar_t(yaz_iconv_t cd, unsigned long x,
+ char **outbuf, size_t *outbytesleft)
{
unsigned char *outp = (unsigned char *) *outbuf;
cd->write_handle = 0;
cd->read_handle = 0;
cd->init_handle = 0;
+ cd->flush_handle = 0;
cd->my_errno = YAZ_ICONV_UNKNOWN;
/* a useful hack: if fromcode has leading @,
cd->read_handle = yaz_read_marc8;
else if (!yaz_matchstr(fromcode, "MARC8s"))
cd->read_handle = yaz_read_marc8s;
+ else if (!yaz_matchstr(fromcode, "advancegreek"))
+ cd->read_handle = yaz_read_advancegreek;
+ else if (!yaz_matchstr(fromcode, "iso54281984"))
+ cd->read_handle = yaz_read_iso5428_1984;
+ else if (!yaz_matchstr(fromcode, "iso5428:1984"))
+ cd->read_handle = yaz_read_iso5428_1984;
#if HAVE_WCHAR_H
else if (!yaz_matchstr(fromcode, "WCHAR_T"))
cd->read_handle = yaz_read_wchar_t;
if (!yaz_matchstr(tocode, "UTF8"))
cd->write_handle = yaz_write_UTF8;
else if (!yaz_matchstr(tocode, "ISO88591"))
+ {
cd->write_handle = yaz_write_ISO8859_1;
+ cd->flush_handle = yaz_flush_ISO8859_1;
+ }
else if (!yaz_matchstr (tocode, "UCS4"))
cd->write_handle = yaz_write_UCS4;
else if (!yaz_matchstr(tocode, "UCS4LE"))
cd->write_handle = yaz_write_UCS4LE;
else if (!yaz_matchstr(tocode, "MARC8"))
+ {
cd->write_handle = yaz_write_marc8;
+ cd->flush_handle = yaz_flush_marc8;
+ }
else if (!yaz_matchstr(tocode, "MARC8s"))
+ {
cd->write_handle = yaz_write_marc8;
+ cd->flush_handle = yaz_flush_marc8;
+ }
+ else if (!yaz_matchstr(tocode, "advancegreek"))
+ {
+ cd->write_handle = yaz_write_advancegreek;
+ }
+ else if (!yaz_matchstr(tocode, "iso54281984"))
+ {
+ cd->write_handle = yaz_write_iso5428_1984;
+ }
+ else if (!yaz_matchstr(tocode, "iso5428:1984"))
+ {
+ cd->write_handle = yaz_write_iso5428_1984;
+ }
#if HAVE_WCHAR_H
else if (!yaz_matchstr(tocode, "WCHAR_T"))
cd->write_handle = yaz_write_wchar_t;
size_t yaz_iconv(yaz_iconv_t cd, char **inbuf, size_t *inbytesleft,
char **outbuf, size_t *outbytesleft)
{
- char *inbuf0;
+ char *inbuf0 = 0;
size_t r = 0;
#if HAVE_ICONV_H
return r;
}
#endif
- if (inbuf == 0 || *inbuf == 0)
+
+ if (inbuf)
+ inbuf0 = *inbuf;
+
+ if (cd->init_flag)
{
- cd->init_flag = 1;
cd->my_errno = YAZ_ICONV_UNKNOWN;
- return 0;
+ cd->marc8_esc_mode = 'B';
+
+ cd->comb_offset = cd->comb_size = 0;
+ cd->compose_char = 0;
+
+ cd->write_marc8_comb_no = 0;
+ cd->write_marc8_second_half_char = 0;
+ cd->write_marc8_last = 0;
+ cd->write_marc8_page_chr = "\033(B";
+
+ cd->unget_x = 0;
+ cd->no_read_x = 0;
}
- inbuf0 = *inbuf;
if (cd->init_flag)
{
- if (cd->init_handle)
+ if (cd->init_handle && inbuf && *inbuf)
{
size_t no_read = 0;
size_t r = (cd->init_handle)(cd, (unsigned char *) *inbuf,
*inbytesleft -= no_read;
*inbuf += no_read;
}
- cd->marc8_esc_mode = 'B';
-
- cd->comb_offset = cd->comb_size = 0;
- cd->compose_char = 0;
-
- cd->write_marc8_comb_no = 0;
- cd->write_marc8_last = 0;
- cd->write_marc8_page_chr = "\033(B";
-
- cd->init_flag = 0;
+ }
+ cd->init_flag = 0;
+
+ if (!inbuf || !*inbuf)
+ {
+ if (outbuf && *outbuf)
+ {
+ if (cd->unget_x)
+ r = (*cd->write_handle)(cd, cd->unget_x, outbuf, outbytesleft);
+ if (cd->flush_handle)
+ r = (*cd->flush_handle)(cd, outbuf, outbytesleft);
+ }
+ if (r == 0)
+ cd->init_flag = 1;
cd->unget_x = 0;
- cd->no_read_x = 0;
+ return r;
}
while (1)
{
unsigned long x;
size_t no_read;
- if (*inbytesleft == 0)
+ if (cd->unget_x)
{
- r = *inbuf - inbuf0;
- break;
+ x = cd->unget_x;
+ no_read = cd->no_read_x;
}
- if (!cd->unget_x)
+ else
{
- x = (cd->read_handle)(cd, (unsigned char *) *inbuf, *inbytesleft,
- &no_read);
+ if (*inbytesleft == 0)
+ {
+ r = *inbuf - inbuf0;
+ break;
+ }
+ x = (*cd->read_handle)(cd, (unsigned char *) *inbuf, *inbytesleft,
+ &no_read);
if (no_read == 0)
{
r = (size_t)(-1);
break;
}
}
- else
- {
- x = cd->unget_x;
- no_read = cd->no_read_x;
- }
if (x)
{
- r = (cd->write_handle)(cd, x, outbuf, outbytesleft,
- (*inbytesleft - no_read) == 0 ? 1 : 0);
+ r = (*cd->write_handle)(cd, x, outbuf, outbytesleft);
if (r)
{
/* unable to write it. save it because read_handle cannot
* End:
* vim: shiftwidth=4 tabstop=8 expandtab
*/
-