X-Git-Url: http://git.indexdata.com/?p=yaz-moved-to-github.git;a=blobdiff_plain;f=src%2Futf8.c;h=0a1ccb37e59aa069404e8f2e85ca8ce60ecf4827;hp=b893e0d055d5476ff9d6a4242806843c8a94b12d;hb=5c1eb188e9cf6f0dd2e435a1f81938f258515edb;hpb=cccb7ecd623450d5b3ca2391327788c84aed71c8

diff --git a/src/utf8.c b/src/utf8.c
index b893e0d..0a1ccb3 100644
--- a/src/utf8.c
+++ b/src/utf8.c
@@ -1,12 +1,10 @@
-/*
- * Copyright (C) 1995-2008, Index Data ApS
+/* This file is part of the YAZ toolkit.
+ * Copyright (C) Index Data
  * See the file LICENSE for details.
- *
- * $Id: siconv.c,v 1.50 2008-03-12 08:53:28 adam Exp $
  */
 /**
  * \file
- * \brief ISO-5428 character mapping (iconv)
+ * \brief UTF-8 encoding / decoding
  */
 
 #if HAVE_CONFIG_H
@@ -16,14 +14,14 @@
 #include <assert.h>
 #include <errno.h>
 #include <string.h>
-#include <ctype.h>
 
 #include "iconv-p.h"
 
-size_t yaz_init_UTF8(yaz_iconv_t cd, unsigned char *inp,
-                     size_t inbytesleft, size_t *no_read)
+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;
@@ -40,7 +38,7 @@ size_t yaz_init_UTF8(yaz_iconv_t cd, unsigned char *inp,
     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)
 {
@@ -82,7 +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
@@ -137,8 +135,9 @@ unsigned long yaz_read_UTF8_char(unsigned char *inp,
     return x;
 }
 
-unsigned long yaz_read_UTF8(yaz_iconv_t cd, unsigned char *inp,
-                            size_t inbytesleft, size_t *no_read)
+static unsigned long read_utf8(yaz_iconv_t cd, yaz_iconv_decoder_t d,
+                               unsigned char *inp,
+                               size_t inbytesleft, size_t *no_read)
 {
     int err = 0;
     int r = yaz_read_UTF8_char(inp, inbytesleft, no_read, &err);
@@ -147,8 +146,9 @@ unsigned long yaz_read_UTF8(yaz_iconv_t cd, unsigned char *inp,
 }
 
 
-size_t yaz_write_UTF8(yaz_iconv_t cd, unsigned long x,
-                      char **outbuf, size_t *outbytesleft)
+static size_t write_UTF8(yaz_iconv_t cd, yaz_iconv_encoder_t en,
+                             unsigned long x,
+                             char **outbuf, size_t *outbytesleft)
 {
     int err = 0;
     int r = yaz_write_UTF8_char(x, outbuf, outbytesleft, &err);
@@ -166,7 +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);
@@ -207,7 +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);
@@ -216,11 +216,55 @@ size_t yaz_write_UTF8_char(unsigned long x,
     return 0;
 }
 
+yaz_iconv_encoder_t yaz_utf8_encoder(const char *tocode,
+                                     yaz_iconv_encoder_t e)
+
+{
+    if (!yaz_matchstr(tocode, "UTF8"))
+    {
+        e->write_handle = write_UTF8;
+        return e;
+    }
+    return 0;
+}
+
+yaz_iconv_decoder_t yaz_utf8_decoder(const char *fromcode,
+                                     yaz_iconv_decoder_t d)
+{
+    if (!yaz_matchstr(fromcode, "UTF8"))
+    {
+        d->init_handle = init_utf8;
+        d->read_handle = read_utf8;
+        return d;
+    }
+    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
  */
+