X-Git-Url: http://git.indexdata.com/?a=blobdiff_plain;f=src%2Fsiconv.c;h=1eb66c455b943e255b5bc5cf1e118de538e88a30;hb=719e0dcf2d15c08086a06457701e21c6aff3d791;hp=36f2c2d3cdee09f5faa316f0a4dbb0b8e4675c87;hpb=c6e47cbbff56f39f6d81b079ebaeac41d793d4d9;p=yaz-moved-to-github.git

diff --git a/src/siconv.c b/src/siconv.c
index 36f2c2d..1eb66c4 100644
--- a/src/siconv.c
+++ b/src/siconv.c
@@ -1,8 +1,8 @@
 /*
- * Copyright (c) 1997-2003, Index Data
+ * Copyright (c) 1997-2004, Index Data
  * See the file LICENSE for details.
  *
- * $Id: siconv.c,v 1.1 2003-10-27 12:21:35 adam Exp $
+ * $Id: siconv.c,v 1.5 2004-03-16 13:12:43 adam Exp $
  */
 
 /* mini iconv and wrapper for system iconv library (if present) */
@@ -25,7 +25,23 @@
 #include <yaz/yaz-util.h>
 
 unsigned long yaz_marc8_conv (unsigned char *inp, size_t inbytesleft,
-                              size_t *no_read);
+			      size_t *no_read);
+unsigned long yaz_marc8_2_conv (unsigned char *inp, size_t inbytesleft,
+				size_t *no_read);
+unsigned long yaz_marc8_3_conv (unsigned char *inp, size_t inbytesleft,
+				size_t *no_read);
+unsigned long yaz_marc8_4_conv (unsigned char *inp, size_t inbytesleft,
+				size_t *no_read);
+unsigned long yaz_marc8_5_conv (unsigned char *inp, size_t inbytesleft,
+				size_t *no_read);
+unsigned long yaz_marc8_6_conv (unsigned char *inp, size_t inbytesleft,
+				size_t *no_read);
+unsigned long yaz_marc8_7_conv (unsigned char *inp, size_t inbytesleft,
+				size_t *no_read);
+unsigned long yaz_marc8_8_conv (unsigned char *inp, size_t inbytesleft,
+				size_t *no_read);
+unsigned long yaz_marc8_9_conv (unsigned char *inp, size_t inbytesleft,
+				size_t *no_read);
     
 struct yaz_iconv_struct {
     int my_errno;
@@ -36,6 +52,7 @@ struct yaz_iconv_struct {
                                  size_t inbytesleft, size_t *no_read);
     size_t (*write_handle)(yaz_iconv_t cd, unsigned long x,
                            char **outbuf, size_t *outbytesleft);
+    int marc8_esc_mode;
 #if HAVE_ICONV_H
     iconv_t iconv_cd;
 #endif
@@ -216,7 +233,75 @@ static unsigned long yaz_read_wchar_t (yaz_iconv_t cd, unsigned char *inp,
 static unsigned long yaz_read_marc8 (yaz_iconv_t cd, unsigned char *inp,
                                      size_t inbytesleft, size_t *no_read)
 {
-    return yaz_marc8_conv(inp, inbytesleft, no_read);
+    *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;
+	size_t no_read_sub = 0;
+
+	switch(cd->marc8_esc_mode)
+	{
+	case 'B':  /* Basic ASCII */
+	case 'E':  /* ANSEL */
+	case 's':  /* ASCII */
+	    x = yaz_marc8_conv(inp, inbytesleft, &no_read_sub);
+	    break;
+	case 'g':  /* Greek */
+	    x = yaz_marc8_2_conv(inp, inbytesleft, &no_read_sub);
+	    break;
+	case 'b':  /* Subscripts */
+	    x = yaz_marc8_3_conv(inp, inbytesleft, &no_read_sub);
+	    break;
+	case 'p':  /* Superscripts */
+	    x = yaz_marc8_4_conv(inp, inbytesleft, &no_read_sub);
+	    break;
+	case '2':  /* Basic Hebrew */
+	    x = yaz_marc8_5_conv(inp, inbytesleft, &no_read_sub);
+	    break;
+	case 'N':  /* Basic Cyrillic */
+	case 'Q':  /* Extended Cyrillic */
+	    x = yaz_marc8_6_conv(inp, inbytesleft, &no_read_sub);
+	    break;
+	case '3':  /* Basic Arabic */
+	case '4':  /* Extended Arabic */
+	    x = yaz_marc8_7_conv(inp, inbytesleft, &no_read_sub);
+	    break;
+	case 'S':  /* Greek */
+	    x = yaz_marc8_8_conv(inp, inbytesleft, &no_read_sub);
+	    break;
+	case '1':  /* Chinese, Japanese, Korean (EACC) */
+	    x = yaz_marc8_9_conv(inp, inbytesleft, &no_read_sub);
+	    break;
+	default:
+	    *no_read = 0;
+	    cd->my_errno = YAZ_ICONV_EILSEQ;
+	    return 0;
+	}
+	*no_read += no_read_sub;
+	return x;
+    }
 }
 
 static size_t yaz_write_UTF8 (yaz_iconv_t cd, unsigned long x,
@@ -307,9 +392,9 @@ static size_t yaz_write_UCS4 (yaz_iconv_t cd, unsigned long x,
     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>>24);
+        *outp++ = (unsigned char) (x>>16);
+        *outp++ = (unsigned char) (x>>8);
         *outp++ = (unsigned char) x;
         (*outbytesleft) -= 4;
     }
@@ -329,9 +414,9 @@ static size_t yaz_write_UCS4LE (yaz_iconv_t cd, unsigned long x,
     if (*outbytesleft >= 4)
     {
         *outp++ = (unsigned char) x;
-        *outp++ = (unsigned char) (x<<8);
-        *outp++ = (unsigned char) (x<<16);
-        *outp++ = (unsigned char) (x<<24);
+        *outp++ = (unsigned char) (x>>8);
+        *outp++ = (unsigned char) (x>>16);
+        *outp++ = (unsigned char) (x>>24);
         (*outbytesleft) -= 4;
     }
     else
@@ -379,6 +464,7 @@ yaz_iconv_t yaz_iconv_open (const char *tocode, const char *fromcode)
     cd->read_handle = 0;
     cd->init_handle = 0;
     cd->my_errno = YAZ_ICONV_UNKNOWN;
+    cd->marc8_esc_mode = 'B';
 
     /* a useful hack: if fromcode has leading @,
        the library not use YAZ's own conversions .. */
@@ -514,9 +600,12 @@ size_t yaz_iconv (yaz_iconv_t cd, char **inbuf, size_t *inbytesleft,
             r = (size_t)(-1);
             break;
         }
-        r = (cd->write_handle)(cd, x, outbuf, outbytesleft);
-        if (r)
-            break;
+	if (x)
+	{
+	    r = (cd->write_handle)(cd, x, outbuf, outbytesleft);
+	    if (r)
+		break;
+	}
         *inbytesleft -= no_read;
         (*inbuf) += no_read;
     }