* Copyright (C) 1995-2005, Index Data ApS
* See the file LICENSE for details.
*
- * $Id: siconv.c,v 1.13 2005-06-25 15:46:05 adam Exp $
+ * $Id: siconv.c,v 1.18 2006-03-25 14:41:53 adam Exp $
*/
/**
* \file siconv.c
unsigned long yaz_marc8_9_conv (unsigned char *inp, size_t inbytesleft,
size_t *no_read, int *combining);
-#define NEW_COMB 1
-
struct yaz_iconv_struct {
int my_errno;
int init_flag;
char **outbuf, size_t *outbytesleft,
int last);
int marc8_esc_mode;
-#if NEW_COMB
+
int comb_offset;
int comb_size;
unsigned long comb_x[8];
size_t comb_no_read[8];
-#else
- int marc8_comb_x;
- int marc8_comb_no_read;
-#endif
size_t no_read_x;
unsigned long unget_x;
#if HAVE_ICONV_H
else if (inp[0] <= 0xef && inbytesleft >= 3)
{
x = ((inp[0] & 0x0f) << 12) | ((inp[1] & 0x3f) << 6) |
- (inp[1] & 0x3f);
+ (inp[2] & 0x3f);
if (x >= 0x800)
*no_read = 3;
else
#endif
-#if NEW_COMB
static unsigned long yaz_read_marc8_comb (yaz_iconv_t cd, unsigned char *inp,
size_t inbytesleft, size_t *no_read,
int *comb);
{
*no_read = cd->comb_no_read[cd->comb_offset];
x = cd->comb_x[cd->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_marc8_?_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
cd->comb_offset++;
return x;
}
return x;
}
}
-#else
-static unsigned long yaz_read_marc8 (yaz_iconv_t cd, unsigned char *inp,
- size_t inbytesleft, size_t *no_read)
-{
- if (cd->marc8_comb_x)
- {
- unsigned long x = cd->marc8_comb_x;
- *no_read = cd->marc8_comb_no_read;
- cd->marc8_comb_x = 0;
- return x;
- }
- *no_read = 0;
- while(inbytesleft >= 1 && inp[0] == 27)
- {
- size_t inbytesleft0 = inbytesleft;
- inp++;
- inbytesleft--;
- while(inbytesleft > 0 && strchr("(,$!", *inp))
- {
- inbytesleft--;
- inp++;
- }
- if (inbytesleft <= 0)
- {
- *no_read = 0;
- cd->my_errno = YAZ_ICONV_EINVAL;
- return 0;
- }
- cd->marc8_esc_mode = *inp++;
- inbytesleft--;
- (*no_read) += inbytesleft0 - inbytesleft;
- }
- if (inbytesleft <= 0)
- return 0;
- else
- {
- unsigned long x;
- int comb = 0;
- size_t no_read_sub = 0;
-
- 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);
- break;
- case 'g': /* Greek */
- x = yaz_marc8_2_conv(inp, inbytesleft, &no_read_sub, &comb);
- break;
- case 'b': /* Subscripts */
- x = yaz_marc8_3_conv(inp, inbytesleft, &no_read_sub, &comb);
- break;
- case 'p': /* Superscripts */
- x = yaz_marc8_4_conv(inp, inbytesleft, &no_read_sub, &comb);
- break;
- case '2': /* Basic Hebrew */
- x = yaz_marc8_5_conv(inp, inbytesleft, &no_read_sub, &comb);
- break;
- case 'N': /* Basic Cyrillic */
- case 'Q': /* Extended Cyrillic */
- x = yaz_marc8_6_conv(inp, inbytesleft, &no_read_sub, &comb);
- break;
- case '3': /* Basic Arabic */
- case '4': /* Extended Arabic */
- x = yaz_marc8_7_conv(inp, inbytesleft, &no_read_sub, &comb);
- break;
- case 'S': /* Greek */
- x = yaz_marc8_8_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);
- break;
- default:
- *no_read = 0;
- cd->my_errno = YAZ_ICONV_EILSEQ;
- return 0;
- }
-#if 0
- printf ("esc mode=%c x=%04lX comb=%d\n", cd->marc8_esc_mode, x, comb);
-#endif
- *no_read += no_read_sub;
-
- if (comb && cd->marc8_comb_x == 0)
- {
- size_t tmp_read = 0;
- unsigned long next_x;
-
- /* read next char .. */
- next_x = yaz_read_marc8(cd, inp + *no_read,
- inbytesleft - *no_read, &tmp_read);
- /* save this x for later .. */
- cd->marc8_comb_x = x;
- /* save next read for later .. */
- cd->marc8_comb_no_read = tmp_read;
- /* return next x - thereby swap */
- x = next_x;
- }
- return x;
- }
-}
-#endif
static size_t yaz_write_UTF8 (yaz_iconv_t cd, unsigned long x,
char **outbuf, size_t *outbytesleft,
int last)
{
unsigned char *outp = (unsigned char *) *outbuf;
+
if (x <= 0x7f && *outbytesleft >= 1)
{
*outp++ = (unsigned char) x;
};
unsigned char *outp = (unsigned char *) *outbuf;
- if (!last && x > 32 && x < 127 && cd->compose_char == 0)
- {
- cd->compose_char = x;
- return 0;
- }
- else if (cd->compose_char)
+ if (cd->compose_char)
{
int i;
for (i = 0; comb[i].x1; i++)
x = comb[i].y;
break;
}
+ if (*outbytesleft < 1)
+ { /* no room. Retain compose_char and bail out */
+ cd->my_errno = YAZ_ICONV_E2BIG;
+ return (size_t)(-1);
+ }
if (!comb[i].x1)
- { /* not found */
- if (*outbytesleft >= 1)
- {
- *outp++ = (unsigned char) cd->compose_char;
- (*outbytesleft)--;
- *outbuf = (char *) outp;
- if (!last && x > 32 && x < 127)
- {
- cd->compose_char = x;
- return 0;
- }
- }
- else
- {
- cd->my_errno = YAZ_ICONV_E2BIG;
- return (size_t)(-1);
- }
+ { /* not found. Just write compose_char */
+ *outp++ = (unsigned char) cd->compose_char;
+ (*outbytesleft)--;
+ *outbuf = (char *) outp;
}
- /* compose_char and old x combined to one new char: x */
+ /* compose_char used so reset it. x now holds current char */
cd->compose_char = 0;
}
- if (x > 255 || x < 1)
+
+ if (!last && x > 32 && x < 127 && cd->compose_char == 0)
{
- cd->my_errno = YAZ_ICONV_EILSEQ;
- return (size_t) -1;
+ cd->compose_char = x;
+ return 0;
}
- else if (*outbytesleft >= 1)
+ else if (x > 255 || x < 1)
{
- *outp++ = (unsigned char) x;
- (*outbytesleft)--;
+ cd->my_errno = YAZ_ICONV_EILSEQ;
+ return (size_t) -1;
}
- else
+ else if (*outbytesleft < 1)
{
cd->my_errno = YAZ_ICONV_E2BIG;
return (size_t)(-1);
}
+ *outp++ = (unsigned char) x;
+ (*outbytesleft)--;
*outbuf = (char *) outp;
return 0;
}
cd->init_handle = 0;
cd->my_errno = YAZ_ICONV_UNKNOWN;
cd->marc8_esc_mode = 'B';
-#if NEW_COMB
cd->comb_offset = cd->comb_size = 0;
-#else
- cd->marc8_comb_x = 0;
-#endif
cd->compose_char = 0;
/* a useful hack: if fromcode has leading @,
{
char *inbuf0;
size_t r = 0;
+
#if HAVE_ICONV_H
if (cd->iconv_cd)
{
{
/* unable to write it. save it because read_handle cannot
rewind .. */
- cd->unget_x = x;
- cd->no_read_x = no_read;
- break;
+ if (cd->my_errno == YAZ_ICONV_E2BIG)
+ {
+ cd->unget_x = x;
+ cd->no_read_x = no_read;
+ break;
+ }
}
cd->unget_x = 0;
}