X-Git-Url: http://git.indexdata.com/?p=yaz-moved-to-github.git;a=blobdiff_plain;f=src%2Futf8.c;h=0a1ccb37e59aa069404e8f2e85ca8ce60ecf4827;hp=bf92cd32a2bff72e763bbb577620ae54ab42d1aa;hb=5c1eb188e9cf6f0dd2e435a1f81938f258515edb;hpb=96c6e58f286787106e4a7b3bb3900a36051968d6

diff --git a/src/utf8.c b/src/utf8.c
index bf92cd3..0a1ccb3 100644
--- a/src/utf8.c
+++ b/src/utf8.c
@@ -1,5 +1,5 @@
 /* This file is part of the YAZ toolkit.
- * Copyright (C) 1995-2008 Index Data
+ * Copyright (C) Index Data
  * See the file LICENSE for details.
  */
 /**
@@ -14,7 +14,6 @@
 #include <assert.h>
 #include <errno.h>
 #include <string.h>
-#include <ctype.h>
 
 #include "iconv-p.h"
 
@@ -22,7 +21,7 @@ static size_t init_utf8(yaz_iconv_t cd, yaz_iconv_decoder_t d,
                         unsigned char *inp,
                         size_t inbytesleft, size_t *no_read)
 {
-    if (inp[0] != 0xef)
+    if (!inp || inp[0] != 0xef)
     {
         *no_read = 0;
         return 0;
@@ -39,7 +38,7 @@ static size_t init_utf8(yaz_iconv_t cd, yaz_iconv_decoder_t d,
     return 0;
 }
 
-unsigned long yaz_read_UTF8_char(unsigned char *inp,
+unsigned long yaz_read_UTF8_char(const unsigned char *inp,
                                  size_t inbytesleft, size_t *no_read,
                                  int *error)
 {
@@ -81,7 +80,7 @@ unsigned long yaz_read_UTF8_char(unsigned char *inp,
         }
         else
             *error = YAZ_ICONV_EILSEQ;
-    }            
+    }
     else if (inp[0] <= 0xf7 && inbytesleft >= 4)
     {
         if ((inp[1] & 0xc0) == 0x80 && (inp[2] & 0xc0) == 0x80
@@ -167,7 +166,7 @@ size_t yaz_write_UTF8_char(unsigned long x,
     {
         *outp++ = (unsigned char) x;
         (*outbytesleft)--;
-    } 
+    }
     else if (x <= 0x7ff && *outbytesleft >= 2)
     {
         *outp++ = (unsigned char) ((x >> 6) | 0xc0);
@@ -208,7 +207,7 @@ size_t yaz_write_UTF8_char(unsigned long x,
         *outp++ = (unsigned char) ((x & 0x3f) | 0x80);
         (*outbytesleft) -= 6;
     }
-    else 
+    else
     {
         *error = YAZ_ICONV_E2BIG;  /* not room for output */
         return (size_t)(-1);
@@ -219,7 +218,7 @@ size_t yaz_write_UTF8_char(unsigned long x,
 
 yaz_iconv_encoder_t yaz_utf8_encoder(const char *tocode,
                                      yaz_iconv_encoder_t e)
-    
+
 {
     if (!yaz_matchstr(tocode, "UTF8"))
     {
@@ -240,12 +239,32 @@ yaz_iconv_decoder_t yaz_utf8_decoder(const char *fromcode,
     }
     return 0;
 }
-   
+
+int yaz_utf8_check(const char *str)
+{
+    /* cast OK: yaz_read_UTF8_char is read-only */
+    unsigned char *inp = (unsigned char *) str;
+    size_t inbytesleft = strlen(str);
+
+    while (inbytesleft)
+    {
+        int error = 0;
+        size_t no_read;
+        yaz_read_UTF8_char(inp, inbytesleft, &no_read, &error);
+        if (error)
+            return 0;
+        inp += no_read;
+        inbytesleft -= no_read;
+    }
+    return 1;
+}
 
 /*
  * Local variables:
  * c-basic-offset: 4
+ * c-file-style: "Stroustrup"
  * indent-tabs-mode: nil
  * End:
  * vim: shiftwidth=4 tabstop=8 expandtab
  */
+