X-Git-Url: http://lists.indexdata.dk/cgi-bin?a=blobdiff_plain;f=src%2Fsiconv.c;h=525574843cb8c572dc5f7d5b251049c5ddcbb1bd;hb=6d61fec3c06d8ed1d648ec7da417a1a8aaa52691;hp=bdb20ee374efdc4619aecbe0bfb06870dbe999b9;hpb=cd6aeaa68dceee3c268dbb354fd32aadbc9fc942;p=yaz-moved-to-github.git

diff --git a/src/siconv.c b/src/siconv.c
index bdb20ee..5255748 100644
--- a/src/siconv.c
+++ b/src/siconv.c
@@ -2,7 +2,7 @@
  * Copyright (C) 1995-2007, Index Data ApS
  * See the file LICENSE for details.
  *
- * $Id: siconv.c,v 1.40 2007-05-03 22:20:45 adam Exp $
+ * $Id: siconv.c,v 1.44 2007-09-22 18:49:55 adam Exp $
  */
 /**
  * \file siconv.c
@@ -36,44 +36,56 @@
 
 #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;
@@ -212,6 +224,7 @@ unsigned long yaz_read_UTF8_char(unsigned char *inp,
 {
     unsigned long x = 0;
 
+    *no_read = 0; /* by default */
     if (inp[0] <= 0x7f)
     {
         x = inp[0];
@@ -219,75 +232,86 @@ unsigned long yaz_read_UTF8_char(unsigned char *inp,
     }
     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;
 }
 
@@ -1158,6 +1182,11 @@ static unsigned long yaz_read_marc8_comb(yaz_iconv_t cd, unsigned char *inp,
     }
     if (inbytesleft <= 0)
         return 0;
+    else if (*inp == ' ')
+    {
+        *no_read += 1;
+        return ' ';
+    }
     else
     {
         unsigned long x;
@@ -1167,35 +1196,44 @@ static unsigned long yaz_read_marc8_comb(yaz_iconv_t cd, unsigned char *inp,
         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;
@@ -1414,56 +1452,77 @@ static unsigned long lookup_marc8(yaz_iconv_t cd,
         *utf8_outbuf = '\0';        
         inp = (unsigned char *) utf8_buf;
         inbytesleft = strlen(utf8_buf);
+
+        if (x == ' ')
+            return x;
         
-        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";
@@ -1592,9 +1651,12 @@ static size_t yaz_write_marc8_2(yaz_iconv_t cd, unsigned long x,
         if (r)
             return r;
 
-        r = yaz_write_marc8_page_chr(cd, outbuf, outbytesleft, page_chr);
-        if (r)
-            return r;
+        if (page_chr)
+        {
+            r = yaz_write_marc8_page_chr(cd, outbuf, outbytesleft, page_chr);
+            if (r)
+                return r;
+        }
         cd->write_marc8_last = y;
     }
     return 0;
@@ -1706,6 +1768,8 @@ yaz_iconv_t yaz_iconv_open (const char *tocode, const char *fromcode)
             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;
@@ -1740,6 +1804,10 @@ yaz_iconv_t yaz_iconv_open (const char *tocode, const char *fromcode)
         {
             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;