* Copyright (C) 1995-2007, Index Data ApS
* See the file LICENSE for details.
*
- * $Id: siconv.c,v 1.39 2007-05-03 22:18:13 adam Exp $
+ * $Id: siconv.c,v 1.43 2007-09-17 19:18:27 adam Exp $
*/
/**
* \file siconv.c
#include <yaz/yaz-util.h>
-unsigned long yaz_marc8_1_conv(unsigned char *inp, size_t inbytesleft,
+unsigned long yaz_marc8_42_conv(unsigned char *inp, size_t inbytesleft,
size_t *no_read, int *combining);
-unsigned long yaz_marc8_2_conv(unsigned char *inp, size_t inbytesleft,
+unsigned long yaz_marc8_45_conv(unsigned char *inp, size_t inbytesleft,
size_t *no_read, int *combining);
-unsigned long yaz_marc8_3_conv(unsigned char *inp, size_t inbytesleft,
+unsigned long yaz_marc8_67_conv(unsigned char *inp, size_t inbytesleft,
size_t *no_read, int *combining);
-unsigned long yaz_marc8_4_conv(unsigned char *inp, size_t inbytesleft,
+unsigned long yaz_marc8_62_conv(unsigned char *inp, size_t inbytesleft,
size_t *no_read, int *combining);
-unsigned long yaz_marc8_5_conv(unsigned char *inp, size_t inbytesleft,
+unsigned long yaz_marc8_70_conv(unsigned char *inp, size_t inbytesleft,
size_t *no_read, int *combining);
-unsigned long yaz_marc8_6_conv(unsigned char *inp, size_t inbytesleft,
+unsigned long yaz_marc8_32_conv(unsigned char *inp, size_t inbytesleft,
size_t *no_read, int *combining);
-unsigned long yaz_marc8_7_conv(unsigned char *inp, size_t inbytesleft,
+unsigned long yaz_marc8_4E_conv(unsigned char *inp, size_t inbytesleft,
size_t *no_read, int *combining);
-unsigned long yaz_marc8_8_conv(unsigned char *inp, size_t inbytesleft,
+unsigned long yaz_marc8_51_conv(unsigned char *inp, size_t inbytesleft,
size_t *no_read, int *combining);
-unsigned long yaz_marc8_9_conv(unsigned char *inp, size_t inbytesleft,
+unsigned long yaz_marc8_33_conv(unsigned char *inp, size_t inbytesleft,
+ size_t *no_read, int *combining);
+unsigned long yaz_marc8_34_conv(unsigned char *inp, size_t inbytesleft,
+ size_t *no_read, int *combining);
+unsigned long yaz_marc8_53_conv(unsigned char *inp, size_t inbytesleft,
+ size_t *no_read, int *combining);
+unsigned long yaz_marc8_31_conv(unsigned char *inp, size_t inbytesleft,
size_t *no_read, int *combining);
-unsigned long yaz_marc8r_1_conv(unsigned char *inp, size_t inbytesleft,
- size_t *no_read, int *combining);
-unsigned long yaz_marc8r_2_conv(unsigned char *inp, size_t inbytesleft,
- size_t *no_read, int *combining);
-unsigned long yaz_marc8r_3_conv(unsigned char *inp, size_t inbytesleft,
- size_t *no_read, int *combining);
-unsigned long yaz_marc8r_4_conv(unsigned char *inp, size_t inbytesleft,
- size_t *no_read, int *combining);
-unsigned long yaz_marc8r_5_conv(unsigned char *inp, size_t inbytesleft,
- size_t *no_read, int *combining);
-unsigned long yaz_marc8r_6_conv(unsigned char *inp, size_t inbytesleft,
- size_t *no_read, int *combining);
-unsigned long yaz_marc8r_7_conv(unsigned char *inp, size_t inbytesleft,
- size_t *no_read, int *combining);
-unsigned long yaz_marc8r_8_conv(unsigned char *inp, size_t inbytesleft,
- size_t *no_read, int *combining);
-unsigned long yaz_marc8r_9_conv(unsigned char *inp, size_t inbytesleft,
- size_t *no_read, int *combining);
+unsigned long yaz_marc8r_42_conv(unsigned char *inp, size_t inbytesleft,
+ size_t *no_read, int *combining);
+unsigned long yaz_marc8r_45_conv(unsigned char *inp, size_t inbytesleft,
+ size_t *no_read, int *combining);
+unsigned long yaz_marc8r_67_conv(unsigned char *inp, size_t inbytesleft,
+ size_t *no_read, int *combining);
+unsigned long yaz_marc8r_62_conv(unsigned char *inp, size_t inbytesleft,
+ size_t *no_read, int *combining);
+unsigned long yaz_marc8r_70_conv(unsigned char *inp, size_t inbytesleft,
+ size_t *no_read, int *combining);
+unsigned long yaz_marc8r_32_conv(unsigned char *inp, size_t inbytesleft,
+ size_t *no_read, int *combining);
+unsigned long yaz_marc8r_4E_conv(unsigned char *inp, size_t inbytesleft,
+ size_t *no_read, int *combining);
+unsigned long yaz_marc8r_51_conv(unsigned char *inp, size_t inbytesleft,
+ size_t *no_read, int *combining);
+unsigned long yaz_marc8r_33_conv(unsigned char *inp, size_t inbytesleft,
+ size_t *no_read, int *combining);
+unsigned long yaz_marc8r_34_conv(unsigned char *inp, size_t inbytesleft,
+ size_t *no_read, int *combining);
+unsigned long yaz_marc8r_53_conv(unsigned char *inp, size_t inbytesleft,
+ size_t *no_read, int *combining);
+unsigned long yaz_marc8r_31_conv(unsigned char *inp, size_t inbytesleft,
+ size_t *no_read, int *combining);
struct yaz_iconv_struct {
int my_errno;
{
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;
}
return 0;
}
switch (*inp) {
- case 0xe1: // alpha small
+ case 0xe1: /* alpha small */
if (tonos)
x = 0x03ac;
else
x = 0x03b1;
break;
- case 0xc1: // alpha capital
+ case 0xc1: /* alpha capital */
if (tonos)
x = 0x0386;
else
x = 0x0391;
break;
- case 0xe2: // Beta small
+ case 0xe2: /* Beta small */
x = 0x03b2;
break;
- case 0xc2: // Beta capital
+ case 0xc2: /* Beta capital */
x = 0x0392;
break;
- case 0xe4: // Gamma small
+ case 0xe4: /* Gamma small */
x = 0x03b3;
break;
- case 0xc4: // Gamma capital
+ case 0xc4: /* Gamma capital */
x = 0x0393;
break;
- case 0xe5: // Delta small
+ case 0xe5: /* Delta small */
x = 0x03b4;
break;
- case 0xc5: // Delta capital
+ case 0xc5: /* Delta capital */
x = 0x0394;
break;
- case 0xe6: // epsilon small
+ case 0xe6: /* epsilon small */
if (tonos)
x = 0x03ad;
else
x = 0x03b5;
break;
- case 0xc6: // epsilon capital
+ case 0xc6: /* epsilon capital */
if (tonos)
x = 0x0388;
else
x = 0x0395;
break;
- case 0xe9: // Zeta small
+ case 0xe9: /* Zeta small */
x = 0x03b6;
break;
- case 0xc9: // Zeta capital
+ case 0xc9: /* Zeta capital */
x = 0x0396;
break;
- case 0xea: // Eta small
+ case 0xea: /* Eta small */
if (tonos)
x = 0x03ae;
else
x = 0x03b7;
break;
- case 0xca: // Eta capital
+ case 0xca: /* Eta capital */
if (tonos)
x = 0x0389;
else
x = 0x0397;
break;
- case 0xeb: // Theta small
+ case 0xeb: /* Theta small */
x = 0x03b8;
break;
- case 0xcb: // Theta capital
+ case 0xcb: /* Theta capital */
x = 0x0398;
break;
- case 0xec: // Iota small
+ case 0xec: /* Iota small */
if (tonos)
if (dialitika)
x = 0x0390;
else
x = 0x03b9;
break;
- case 0xcc: // Iota capital
+ case 0xcc: /* Iota capital */
if (tonos)
x = 0x038a;
else
else
x = 0x0399;
break;
- case 0xed: // Kappa small
+ case 0xed: /* Kappa small */
x = 0x03ba;
break;
- case 0xcd: // Kappa capital
+ case 0xcd: /* Kappa capital */
x = 0x039a;
break;
- case 0xee: // Lambda small
+ case 0xee: /* Lambda small */
x = 0x03bb;
break;
- case 0xce: // Lambda capital
+ case 0xce: /* Lambda capital */
x = 0x039b;
break;
- case 0xef: // Mu small
+ case 0xef: /* Mu small */
x = 0x03bc;
break;
- case 0xcf: // Mu capital
+ case 0xcf: /* Mu capital */
x = 0x039c;
break;
- case 0xf0: // Nu small
+ case 0xf0: /* Nu small */
x = 0x03bd;
break;
- case 0xd0: // Nu capital
+ case 0xd0: /* Nu capital */
x = 0x039d;
break;
- case 0xf1: // Xi small
+ case 0xf1: /* Xi small */
x = 0x03be;
break;
- case 0xd1: // Xi capital
+ case 0xd1: /* Xi capital */
x = 0x039e;
break;
- case 0xf2: // Omicron small
+ case 0xf2: /* Omicron small */
if (tonos)
x = 0x03cc;
else
x = 0x03bf;
break;
- case 0xd2: // Omicron capital
+ case 0xd2: /* Omicron capital */
if (tonos)
x = 0x038c;
else
x = 0x039f;
break;
- case 0xf3: // Pi small
+ case 0xf3: /* Pi small */
x = 0x03c0;
break;
- case 0xd3: // Pi capital
+ case 0xd3: /* Pi capital */
x = 0x03a0;
break;
- case 0xf5: // Rho small
+ case 0xf5: /* Rho small */
x = 0x03c1;
break;
- case 0xd5: // Rho capital
+ case 0xd5: /* Rho capital */
x = 0x03a1;
break;
- case 0xf7: // Sigma small (end of words)
+ case 0xf7: /* Sigma small (end of words) */
x = 0x03c2;
break;
- case 0xf6: // Sigma small
+ case 0xf6: /* Sigma small */
x = 0x03c3;
break;
- case 0xd6: // Sigma capital
+ case 0xd6: /* Sigma capital */
x = 0x03a3;
break;
- case 0xf8: // Tau small
+ case 0xf8: /* Tau small */
x = 0x03c4;
break;
- case 0xd8: // Tau capital
+ case 0xd8: /* Tau capital */
x = 0x03a4;
break;
- case 0xf9: // Upsilon small
+ case 0xf9: /* Upsilon small */
if (tonos)
if (dialitika)
x = 0x03b0;
else
x = 0x03c5;
break;
- case 0xd9: // Upsilon capital
+ case 0xd9: /* Upsilon capital */
if (tonos)
x = 0x038e;
else
else
x = 0x03a5;
break;
- case 0xfa: // Phi small
+ case 0xfa: /* Phi small */
x = 0x03c6;
break;
- case 0xda: // Phi capital
+ case 0xda: /* Phi capital */
x = 0x03a6;
break;
- case 0xfb: // Chi small
+ case 0xfb: /* Chi small */
x = 0x03c7;
break;
- case 0xdb: // Chi capital
+ case 0xdb: /* Chi capital */
x = 0x03a7;
break;
- case 0xfc: // Psi small
+ case 0xfc: /* Psi small */
x = 0x03c8;
break;
- case 0xdc: // Psi capital
+ case 0xdc: /* Psi capital */
x = 0x03a8;
break;
- case 0xfd: // Omega small
+ case 0xfd: /* Omega small */
if (tonos)
x = 0x03ce;
else
x = 0x03c9;
break;
- case 0xdd: // Omega capital
+ case 0xdd: /* Omega capital */
if (tonos)
x = 0x038f;
else
switch(cd->marc8_esc_mode)
{
case 'B': /* Basic ASCII */
- case 'E': /* ANSEL */
case 's': /* ASCII */
- x = yaz_marc8_1_conv(inp, inbytesleft, &no_read_sub, comb);
+ case 'E': /* ANSEL */
+ x = yaz_marc8_42_conv(inp, inbytesleft, &no_read_sub, comb);
+ if (!x)
+ {
+ no_read_sub = 0;
+ x = yaz_marc8_45_conv(inp, inbytesleft, &no_read_sub, comb);
+ }
break;
case 'g': /* Greek */
- x = yaz_marc8_2_conv(inp, inbytesleft, &no_read_sub, comb);
+ x = yaz_marc8_67_conv(inp, inbytesleft, &no_read_sub, comb);
break;
case 'b': /* Subscripts */
- x = yaz_marc8_3_conv(inp, inbytesleft, &no_read_sub, comb);
+ x = yaz_marc8_62_conv(inp, inbytesleft, &no_read_sub, comb);
break;
case 'p': /* Superscripts */
- x = yaz_marc8_4_conv(inp, inbytesleft, &no_read_sub, comb);
+ x = yaz_marc8_70_conv(inp, inbytesleft, &no_read_sub, comb);
break;
case '2': /* Basic Hebrew */
- x = yaz_marc8_5_conv(inp, inbytesleft, &no_read_sub, comb);
+ x = yaz_marc8_32_conv(inp, inbytesleft, &no_read_sub, comb);
break;
case 'N': /* Basic Cyrillic */
+ x = yaz_marc8_4E_conv(inp, inbytesleft, &no_read_sub, comb);
+ break;
case 'Q': /* Extended Cyrillic */
- x = yaz_marc8_6_conv(inp, inbytesleft, &no_read_sub, comb);
+ x = yaz_marc8_51_conv(inp, inbytesleft, &no_read_sub, comb);
break;
case '3': /* Basic Arabic */
+ x = yaz_marc8_33_conv(inp, inbytesleft, &no_read_sub, comb);
+ break;
case '4': /* Extended Arabic */
- x = yaz_marc8_7_conv(inp, inbytesleft, &no_read_sub, comb);
+ x = yaz_marc8_34_conv(inp, inbytesleft, &no_read_sub, comb);
break;
case 'S': /* Greek */
- x = yaz_marc8_8_conv(inp, inbytesleft, &no_read_sub, comb);
+ x = yaz_marc8_53_conv(inp, inbytesleft, &no_read_sub, comb);
break;
case '1': /* Chinese, Japanese, Korean (EACC) */
- x = yaz_marc8_9_conv(inp, inbytesleft, &no_read_sub, comb);
+ x = yaz_marc8_31_conv(inp, inbytesleft, &no_read_sub, comb);
break;
default:
*no_read = 0;
inp = (unsigned char *) utf8_buf;
inbytesleft = strlen(utf8_buf);
- x = yaz_marc8r_1_conv(inp, inbytesleft, &no_read_sub, comb);
+ x = yaz_marc8r_42_conv(inp, inbytesleft, &no_read_sub, comb);
+ if (x)
+ {
+ *page_chr = "\033(B";
+ return x;
+ }
+ x = yaz_marc8r_45_conv(inp, inbytesleft, &no_read_sub, comb);
if (x)
{
*page_chr = "\033(B";
return x;
}
- x = yaz_marc8r_2_conv(inp, inbytesleft, &no_read_sub, comb);
+ x = yaz_marc8r_67_conv(inp, inbytesleft, &no_read_sub, comb);
if (x)
{
*page_chr = "\033g";
return x;
}
- x = yaz_marc8r_3_conv(inp, inbytesleft, &no_read_sub, comb);
+ x = yaz_marc8r_62_conv(inp, inbytesleft, &no_read_sub, comb);
if (x)
{
*page_chr = "\033b";
return x;
}
- x = yaz_marc8r_4_conv(inp, inbytesleft, &no_read_sub, comb);
+ x = yaz_marc8r_70_conv(inp, inbytesleft, &no_read_sub, comb);
if (x)
{
*page_chr = "\033p";
return x;
}
- x = yaz_marc8r_5_conv(inp, inbytesleft, &no_read_sub, comb);
+ x = yaz_marc8r_32_conv(inp, inbytesleft, &no_read_sub, comb);
if (x)
{
*page_chr = "\033(2";
return x;
}
- x = yaz_marc8r_6_conv(inp, inbytesleft, &no_read_sub, comb);
+ x = yaz_marc8r_4E_conv(inp, inbytesleft, &no_read_sub, comb);
if (x)
{
*page_chr = "\033(N";
return x;
}
- x = yaz_marc8r_7_conv(inp, inbytesleft, &no_read_sub, comb);
+ x = yaz_marc8r_51_conv(inp, inbytesleft, &no_read_sub, comb);
+ if (x)
+ {
+ *page_chr = "\033(Q";
+ return x;
+ }
+ x = yaz_marc8r_33_conv(inp, inbytesleft, &no_read_sub, comb);
if (x)
{
*page_chr = "\033(3";
return x;
}
- x = yaz_marc8r_8_conv(inp, inbytesleft, &no_read_sub, comb);
+ x = yaz_marc8r_34_conv(inp, inbytesleft, &no_read_sub, comb);
+ if (x)
+ {
+ *page_chr = "\033(4";
+ return x;
+ }
+ x = yaz_marc8r_53_conv(inp, inbytesleft, &no_read_sub, comb);
if (x)
{
*page_chr = "\033(S";
return x;
}
- x = yaz_marc8r_9_conv(inp, inbytesleft, &no_read_sub, comb);
+ x = yaz_marc8r_31_conv(inp, inbytesleft, &no_read_sub, comb);
if (x)
{
*page_chr = "\033$1";
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;
{
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;
projects / yaz-moved-to-github.git / blobdiff