Add yaz_utf8_check

A function that checks whether a string appears to be UTF-8 encoded.

diff --git a/include/yaz/yaz-iconv.h b/include/yaz/yaz-iconv.h
index a1a2dc4..96f326a 100644 (file)
--- a/include/yaz/yaz-iconv.h
+++ b/include/yaz/yaz-iconv.h
@@ -91,6 +91,13 @@ YAZ_EXPORT size_t yaz_write_UTF8_char(unsigned long x,
 #define yaz_tolower(x) ((x) + ('a' - 'A'))
 #define yaz_islower(x) ((x) >= 'a' && (x) <= 'z')
 
+/** \brief check whether string apppers to be UTF-8 encoded
+    \param cstr string to check
+    \retval 1 OK (appears to be UTF-8)
+    \retval 0 definitely not UTF-8
+*/
+YAZ_EXPORT int yaz_utf8_check(const char *cstr);
+
 YAZ_END_CDECL
 
 #endif

diff --git a/src/utf8.c b/src/utf8.c
index 8911683..7e294d4 100644 (file)
--- a/src/utf8.c
+++ b/src/utf8.c
@@ -239,7 +239,25 @@ 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: