/*
- * Copyright (C) 1995-2006, Index Data ApS
+ * Copyright (C) 1995-2008, Index Data ApS
* See the file LICENSE for details.
*
- * $Id: siconv.c,v 1.29 2006-08-31 18:19:53 adam Exp $
+ * $Id: siconv.c,v 1.50 2008-03-12 08:53:28 adam Exp $
*/
/**
* \file siconv.c
#include <iconv.h>
#endif
+#include <yaz/xmalloc.h>
+#include <yaz/nmem.h>
+#include "iconv-p.h"
-#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 (*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);
- int marc8_esc_mode;
+ char **outbuf, size_t *outbytesleft);
+ size_t (*flush_handle)(yaz_iconv_t cd,
+ char **outbuf, size_t *outbytesleft);
+ int g0_mode;
+ int g1_mode;
int comb_offset;
int comb_size;
#endif
unsigned long compose_char;
- 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;
+ const char *write_marc8_lpage;
+ const char *write_marc8_g0;
+ const char *write_marc8_g1;
};
+
static struct {
unsigned long x1, x2;
unsigned y;
{ 0, 0, 0}
};
-static unsigned long yaz_read_ISO8859_1 (yaz_iconv_t cd, unsigned char *inp,
- size_t inbytesleft, size_t *no_read)
+#define ESC "\033"
+
+static size_t yaz_write_marc8_page_chr(yaz_iconv_t cd,
+ char **outbuf, size_t *outbytesleft,
+ const char *page_chr);
+
+static unsigned long yaz_read_ISO8859_1(yaz_iconv_t cd, unsigned char *inp,
+ size_t inbytesleft, size_t *no_read)
{
unsigned long x = inp[0];
*no_read = 1;
}
-static size_t yaz_init_UTF8 (yaz_iconv_t cd, unsigned char *inp,
- size_t inbytesleft, size_t *no_read)
-{
- if (inp[0] != 0xef)
- {
- *no_read = 0;
- return 0;
- }
- if (inbytesleft < 3)
- {
- cd->my_errno = YAZ_ICONV_EINVAL;
- return (size_t) -1;
- }
- if (inp[1] != 0xbb && inp[2] == 0xbf)
- *no_read = 3;
- else
- *no_read = 0;
- return 0;
-}
-
-unsigned long yaz_read_UTF8_char(unsigned char *inp,
- size_t inbytesleft, size_t *no_read,
- int *error)
-{
- unsigned long x = 0;
-
- if (inp[0] <= 0x7f)
- {
- x = inp[0];
- *no_read = 1;
- }
- 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
- {
- *no_read = 0;
- *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
- {
- *no_read = 0;
- *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
- {
- *no_read = 0;
- *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
- {
- *no_read = 0;
- *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
- {
- *no_read = 0;
- *error = YAZ_ICONV_EILSEQ;
- }
- }
- else
- {
- *no_read = 0;
- *error = YAZ_ICONV_EINVAL;
- }
- return x;
-}
-
-static unsigned long yaz_read_UTF8 (yaz_iconv_t cd, unsigned char *inp,
- size_t inbytesleft, size_t *no_read)
-{
- return yaz_read_UTF8_char(inp, inbytesleft, no_read, &cd->my_errno);
-}
-
-static unsigned long yaz_read_UCS4 (yaz_iconv_t cd, unsigned char *inp,
- size_t inbytesleft, size_t *no_read)
-{
- unsigned long x = 0;
-
- if (inbytesleft < 4)
- {
- cd->my_errno = YAZ_ICONV_EINVAL; /* incomplete input */
- *no_read = 0;
- }
- else
- {
- x = (inp[0]<<24) | (inp[1]<<16) | (inp[2]<<8) | inp[3];
- *no_read = 4;
- }
- return x;
-}
-
-static unsigned long yaz_read_UCS4LE (yaz_iconv_t cd, unsigned char *inp,
- size_t inbytesleft, size_t *no_read)
-{
- unsigned long x = 0;
-
- if (inbytesleft < 4)
- {
- cd->my_errno = YAZ_ICONV_EINVAL; /* incomplete input */
- *no_read = 0;
- }
- else
- {
- x = (inp[3]<<24) | (inp[2]<<16) | (inp[1]<<8) | inp[0];
- *no_read = 4;
- }
- return x;
-}
#if HAVE_WCHAR_H
static unsigned long yaz_read_wchar_t (yaz_iconv_t cd, unsigned char *inp,
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;
int *comb)
{
*no_read = 0;
- while(inbytesleft >= 1 && inp[0] == 27)
+ while (inbytesleft > 0 && *inp == 27)
{
+ int *modep = &cd->g0_mode;
size_t inbytesleft0 = inbytesleft;
- inp++;
+
inbytesleft--;
- while(inbytesleft > 0 && strchr("(,$!", *inp))
+ inp++;
+ if (inbytesleft == 0)
+ goto incomplete;
+ if (*inp == '$') /* set with multiple bytes */
{
inbytesleft--;
inp++;
}
- if (inbytesleft <= 0)
+ if (inbytesleft == 0)
+ goto incomplete;
+ if (*inp == '(' || *inp == ',') /* G0 */
{
- *no_read = 0;
- cd->my_errno = YAZ_ICONV_EINVAL;
- return 0;
+ inbytesleft--;
+ inp++;
}
- cd->marc8_esc_mode = *inp++;
+ else if (*inp == ')' || *inp == '-') /* G1 */
+ {
+ inbytesleft--;
+ inp++;
+ modep = &cd->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)
+ 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 ? cd->g0_mode : cd->g1_mode;
*comb = 0;
- switch(cd->marc8_esc_mode)
+ switch(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;
*no_read += no_read_sub;
return x;
}
-}
-
-static size_t yaz_write_UTF8(yaz_iconv_t cd, unsigned long x,
- char **outbuf, size_t *outbytesleft,
- int last)
-{
- return yaz_write_UTF8_char(x, outbuf, outbytesleft, &cd->my_errno);
-}
-
-size_t yaz_write_UTF8_char(unsigned long x,
- char **outbuf, size_t *outbytesleft,
- int *error)
-{
- unsigned char *outp = (unsigned char *) *outbuf;
-
- if (x <= 0x7f && *outbytesleft >= 1)
- {
- *outp++ = (unsigned char) x;
- (*outbytesleft)--;
- }
- else if (x <= 0x7ff && *outbytesleft >= 2)
- {
- *outp++ = (unsigned char) ((x >> 6) | 0xc0);
- *outp++ = (unsigned char) ((x & 0x3f) | 0x80);
- (*outbytesleft) -= 2;
- }
- else if (x <= 0xffff && *outbytesleft >= 3)
- {
- *outp++ = (unsigned char) ((x >> 12) | 0xe0);
- *outp++ = (unsigned char) (((x >> 6) & 0x3f) | 0x80);
- *outp++ = (unsigned char) ((x & 0x3f) | 0x80);
- (*outbytesleft) -= 3;
- }
- else if (x <= 0x1fffff && *outbytesleft >= 4)
- {
- *outp++ = (unsigned char) ((x >> 18) | 0xf0);
- *outp++ = (unsigned char) (((x >> 12) & 0x3f) | 0x80);
- *outp++ = (unsigned char) (((x >> 6) & 0x3f) | 0x80);
- *outp++ = (unsigned char) ((x & 0x3f) | 0x80);
- (*outbytesleft) -= 4;
- }
- else if (x <= 0x3ffffff && *outbytesleft >= 5)
- {
- *outp++ = (unsigned char) ((x >> 24) | 0xf8);
- *outp++ = (unsigned char) (((x >> 18) & 0x3f) | 0x80);
- *outp++ = (unsigned char) (((x >> 12) & 0x3f) | 0x80);
- *outp++ = (unsigned char) (((x >> 6) & 0x3f) | 0x80);
- *outp++ = (unsigned char) ((x & 0x3f) | 0x80);
- (*outbytesleft) -= 5;
- }
- else if (*outbytesleft >= 6)
- {
- *outp++ = (unsigned char) ((x >> 30) | 0xfc);
- *outp++ = (unsigned char) (((x >> 24) & 0x3f) | 0x80);
- *outp++ = (unsigned char) (((x >> 18) & 0x3f) | 0x80);
- *outp++ = (unsigned char) (((x >> 12) & 0x3f) | 0x80);
- *outp++ = (unsigned char) (((x >> 6) & 0x3f) | 0x80);
- *outp++ = (unsigned char) ((x & 0x3f) | 0x80);
- (*outbytesleft) -= 6;
- }
- else
- {
- *error = YAZ_ICONV_E2BIG; /* not room for output */
- return (size_t)(-1);
- }
- *outbuf = (char *) outp;
+incomplete:
+ *no_read = 0;
+ cd->my_errno = YAZ_ICONV_EINVAL;
return 0;
}
-
-static size_t yaz_write_ISO8859_1 (yaz_iconv_t cd, unsigned long x,
- char **outbuf, size_t *outbytesleft,
- int last)
+static size_t yaz_write_ISO8859_1(yaz_iconv_t cd, unsigned long x,
+ 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_write_UCS4 (yaz_iconv_t cd, unsigned long x,
- char **outbuf, size_t *outbytesleft,
- int last)
+static size_t yaz_flush_ISO8859_1(yaz_iconv_t cd,
+ char **outbuf, size_t *outbytesleft)
{
- unsigned char *outp = (unsigned char *) *outbuf;
- if (*outbytesleft >= 4)
- {
- *outp++ = (unsigned char) (x>>24);
- *outp++ = (unsigned char) (x>>16);
- *outp++ = (unsigned char) (x>>8);
- *outp++ = (unsigned char) x;
- (*outbytesleft) -= 4;
- }
- else
- {
- cd->my_errno = YAZ_ICONV_E2BIG;
- return (size_t)(-1);
- }
- *outbuf = (char *) outp;
- return 0;
-}
-
-static size_t yaz_write_UCS4LE (yaz_iconv_t cd, unsigned long x,
- char **outbuf, size_t *outbytesleft,
- int last)
-{
- unsigned char *outp = (unsigned char *) *outbuf;
- if (*outbytesleft >= 4)
- {
- *outp++ = (unsigned char) x;
- *outp++ = (unsigned char) (x>>8);
- *outp++ = (unsigned char) (x>>16);
- *outp++ = (unsigned char) (x>>24);
- (*outbytesleft) -= 4;
- }
- else
+ if (cd->compose_char)
{
- cd->my_errno = YAZ_ICONV_E2BIG;
- return (size_t)(-1);
+ 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;
}
- *outbuf = (char *) outp;
return 0;
}
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;
*utf8_outbuf = '\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";
+ *page_chr = ESC "(B";
return x;
}
- x = yaz_marc8r_2_conv(inp, inbytesleft, &no_read_sub, comb);
+ x = yaz_marc8r_45_conv(inp, inbytesleft, &no_read_sub, comb);
if (x)
{
- *page_chr = "\033g";
+ *page_chr = ESC "(B";
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";
+ *page_chr = ESC "b";
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";
+ *page_chr = ESC "p";
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";
+ *page_chr = ESC "(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";
+ *page_chr = ESC "(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(3";
+ *page_chr = ESC "(Q";
return x;
}
- x = yaz_marc8r_8_conv(inp, inbytesleft, &no_read_sub, comb);
+ x = yaz_marc8r_33_conv(inp, inbytesleft, &no_read_sub, comb);
if (x)
{
- *page_chr = "\033(S";
+ *page_chr = ESC "(3";
return x;
}
- x = yaz_marc8r_9_conv(inp, inbytesleft, &no_read_sub, comb);
+ x = yaz_marc8r_34_conv(inp, inbytesleft, &no_read_sub, comb);
if (x)
{
- *page_chr = "\033$1";
+ *page_chr = ESC "(4";
+ return x;
+ }
+ x = yaz_marc8r_53_conv(inp, inbytesleft, &no_read_sub, comb);
+ if (x)
+ {
+ *page_chr = ESC "(S";
+ return x;
+ }
+ x = yaz_marc8r_31_conv(inp, inbytesleft, &no_read_sub, comb);
+ if (x)
+ {
+ *page_chr = ESC "$1";
return x;
}
cd->my_errno = YAZ_ICONV_EILSEQ;
char **outbuf, size_t *outbytesleft)
{
unsigned long y = cd->write_marc8_last;
- unsigned char byte, second_half = 0;
- char out_buf[10];
- size_t i, out_no = 0;
+ unsigned char byte;
+ char out_buf[4];
+ size_t out_no = 0;
if (!y)
return 0;
+ assert(cd->write_marc8_lpage);
+ if (cd->write_marc8_lpage)
+ {
+ size_t r = yaz_write_marc8_page_chr(cd, outbuf, outbytesleft,
+ cd->write_marc8_lpage);
+ if (r)
+ return r;
+ }
+
byte = (unsigned char )((y>>16) & 0xff);
if (byte)
out_buf[out_no++] = byte;
if (byte)
out_buf[out_no++] = byte;
- if (out_no + cd->write_marc8_comb_no + 1 > *outbytesleft)
+ if (out_no + 2 >= *outbytesleft)
{
cd->my_errno = YAZ_ICONV_E2BIG;
return (size_t) (-1);
}
- for (i = 0; i < cd->write_marc8_comb_no; i++)
- {
- /* 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_lpage = 0;
+ cd->write_marc8_second_half_char = 0;
return 0;
}
-static size_t yaz_write_marc8_2(yaz_iconv_t cd, unsigned long x,
- char **outbuf, size_t *outbytesleft,
- int last)
+static size_t yaz_write_marc8_page_chr(yaz_iconv_t cd,
+ char **outbuf, size_t *outbytesleft,
+ const char *page_chr)
{
- int comb = 0;
- const char *page_chr = 0;
- unsigned long y = lookup_marc8(cd, x, &comb, &page_chr);
+ const char **old_page_chr = &cd->write_marc8_g0;
- if (!y)
- return (size_t) (-1);
+ /* are we going to a G1-set (such as such as ESC ")!E") */
+ if (page_chr && page_chr[1] == ')')
+ old_page_chr = &cd->write_marc8_g1;
- if (comb)
- {
- if (cd->write_marc8_comb_no < 6)
- cd->write_marc8_comb_ch[cd->write_marc8_comb_no++] = y;
- }
- else
+ if (!*old_page_chr || strcmp(page_chr, *old_page_chr))
{
- 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)
{
- 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;
+ cd->my_errno = YAZ_ICONV_E2BIG;
+
+ return (size_t) (-1);
+ }
- if (!strcmp(old_page_chr, "\033p")
- || !strcmp(old_page_chr, "\033g")
- || !strcmp(old_page_chr, "\033b"))
+ if (*old_page_chr)
+ {
+ if (!strcmp(*old_page_chr, ESC "p")
+ || !strcmp(*old_page_chr, ESC "g")
+ || !strcmp(*old_page_chr, ESC "b"))
{
+ page_out = ESC "s";
/* Technique 1 leave */
- page_out = "\033s";
- if (strcmp(page_chr, "\033(B")) /* Not going ASCII page? */
+ if (strcmp(page_chr, ESC "(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;
+ page_out = ESC "(B";
}
}
- plen = strlen(page_out);
- memcpy(*outbuf, page_out, plen);
- (*outbuf) += plen;
- (*outbytesleft) -= plen;
}
- cd->write_marc8_last = y;
+ *old_page_chr = 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 comb = 0;
+ const char *page_chr = 0;
+ unsigned long y = lookup_marc8(cd, x, &comb, &page_chr);
+
+ if (!y)
+ return (size_t) (-1);
+
+ if (comb)
+ {
+ if (page_chr)
+ {
+ size_t r = yaz_write_marc8_page_chr(cd, outbuf, outbytesleft,
+ page_chr);
+ if (r)
+ return r;
+ }
+ if (x == 0x0361)
+ cd->write_marc8_second_half_char = 0xEC;
+ else if (x == 0x0360)
+ cd->write_marc8_second_half_char = 0xFB;
+
+ if (*outbytesleft <= 1)
+ {
+ cd->my_errno = YAZ_ICONV_E2BIG;
+ return (size_t) (-1);
+ }
+ *(*outbuf)++ = y;
+ (*outbytesleft)--;
}
- if (last)
+ else
{
size_t r = flush_combos(cd, outbuf, outbytesleft);
if (r)
- {
- if (comb)
- cd->write_marc8_comb_no--;
- else
- cd->write_marc8_last = 0;
return r;
- }
+
+ cd->write_marc8_last = y;
+ cd->write_marc8_lpage = page_chr;
}
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;
+ cd->write_marc8_g1 = 0;
+ return yaz_write_marc8_page_chr(cd, outbuf, outbytesleft, ESC "(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++)
char *outbuf0 = *outbuf;
size_t outbytesleft0 = *outbytesleft;
int last_ch = cd->write_marc8_last;
+ const char *lpage = cd->write_marc8_lpage;
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 */
*outbuf = outbuf0;
*outbytesleft = outbytesleft0;
cd->write_marc8_last = last_ch;
+ cd->write_marc8_lpage = lpage;
}
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->g0_mode = 'B';
+ cd->g1_mode = 'B';
+
+ cd->comb_offset = cd->comb_size = 0;
+ cd->compose_char = 0;
+
+ cd->write_marc8_second_half_char = 0;
+ cd->write_marc8_last = 0;
+ cd->write_marc8_lpage = 0;
+ cd->write_marc8_g0 = ESC "(B";
+ cd->write_marc8_g1 = 0;
+
+ 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
return 0;
}
+void yaz_iconv_set_errno(yaz_iconv_t cd, int no)
+{
+ cd->my_errno = no;
+}
+
/*
* Local variables:
* c-basic-offset: 4
* End:
* vim: shiftwidth=4 tabstop=8 expandtab
*/
-