Local UTF8 conversion code

src/dvb/dvb_support.c

@@ -29,8 +29,6 @@
 #include <stdlib.h>
 #include <string.h>
 
-#include <libavutil/common.h>
-
 #include <linux/dvb/frontend.h>
 
 #include "tvhead.h"
@@ -197,16 +195,18 @@ dvb_get_string_with_len(char *dst, size_t dstlen,
 void
 atsc_utf16_to_utf8(uint8_t *src, int len, char *buf, int buflen)
 {
-  int i, c;
-  char *q = buf;
+  int i, c, r;
 
   for(i = 0; i < len; i++) {
-    uint8_t tmp;
     c = (src[i * 2 + 0] << 8) | src[i * 2 + 1];
 
-    PUT_UTF8(c, tmp, if (q - buf < buflen - 1) *q++ = tmp;)
+    if(buflen >= 7) {
+      r = put_utf8(buf, c);
+      buf += r;
+      buflen -= r;
+    }
   }
-  *q = 0;
+  *buf = 0;
 }
 
 

src/htsmsg_xml.c

@@ -50,7 +50,7 @@
 #include <stdlib.h>
 #include <string.h>
 
-#include <libavutil/common.h>
+#include "tvhead.h"
 
 #include "htsmsg_xml.h"
 #include "htsbuf.h"
@@ -106,18 +106,19 @@ static void
 add_unicode(struct cdata_content_queue *ccq, int c)
 {
   cdata_content_t *cc;
-  char *q;
-  uint8_t tmp;
+  int r;
 
   cc = malloc(sizeof(cdata_content_t) + 6);
-  cc->cc_encoding = XML_ENCODING_UTF8;
-  q = cc->cc_buf;
+  r = put_utf8(cc->cc_buf, c);
+  if(r == 0) {
+    free(cc);
+    return;
+  }
 
+  cc->cc_encoding = XML_ENCODING_UTF8;
   TAILQ_INSERT_TAIL(ccq, cc, cc_link);
   cc->cc_start = cc->cc_buf;
-
-  PUT_UTF8(c, tmp, *q++ = tmp;)
-  cc->cc_end = q;
+  cc->cc_end   = cc->cc_buf + r;
 }
 
 /**
@@ -665,10 +666,8 @@ htsmsg_xml_parse_cd(xmlparser_t *xp, htsmsg_t *parent, char *src)
   htsmsg_field_t *f;
   cdata_content_t *cc;
   int c = 0, l, y = 0;
-  char *body;
-  char *x;
+  char *x, *body;
   htsmsg_t *tags = htsmsg_create_map();
-  uint8_t tmp;
   
   TAILQ_INIT(&ccq);
   src = htsmsg_xml_parse_cd0(xp, &ccq, tags, NULL, src, 0);
@@ -722,9 +721,8 @@ htsmsg_xml_parse_cd(xmlparser_t *xp, htsmsg_t *parent, char *src)
 	break;
 
       case XML_ENCODING_8859_1:
-	for(x = cc->cc_start; x < cc->cc_end; x++) {
-	  PUT_UTF8(*x, tmp, body[c++] = tmp;)
-	    }
+	for(x = cc->cc_start; x < cc->cc_end; x++)
+	  body += put_utf8(body, *x);
 	break;
       }
       

src/tvhead.h

@@ -840,4 +840,6 @@ uint32_t crc32(uint8_t *data, size_t datalen, uint32_t crc);
 
 int base64_decode(uint8_t *out, const char *in, int out_size);
 
+int put_utf8(char *out, int c);
+
 #endif /* TV_HEAD_H */

src/utils.c

@@ -136,3 +136,57 @@ base64_decode(uint8_t *out, const char *in, int out_size)
 
     return dst - out;
 }
+
+
+/**
+ *
+ */
+int
+put_utf8(char *out, int c)
+{
+  if(c == 0xfffe || c == 0xffff || (c >= 0xD800 && c < 0xE000))
+    return 0;
+  
+  if (c < 0x80) {
+    *out = c;
+    return 1;
+  }
+
+  if(c < 0x800) {
+    *out++ = 0xc0 | (0x1f & (c >>  6));
+    *out   = 0x80 | (0x3f &  c);
+    return 2;
+  }
+
+  if(c < 0x10000) {
+    *out++ = 0xe0 | (0x0f & (c >> 12));
+    *out++ = 0x80 | (0x3f & (c >> 6));
+    *out   = 0x80 | (0x3f &  c);
+    return 3;
+  }
+
+  if(c < 0x200000) {
+    *out++ = 0xf0 | (0x07 & (c >> 18));
+    *out++ = 0x80 | (0x3f & (c >> 12));
+    *out++ = 0x80 | (0x3f & (c >> 6));
+    *out   = 0x80 | (0x3f &  c);
+    return 4;
+  }
+  
+  if(c < 0x4000000) {
+    *out++ = 0xf8 | (0x03 & (c >> 24));
+    *out++ = 0x80 | (0x3f & (c >> 18));
+    *out++ = 0x80 | (0x3f & (c >> 12));
+    *out++ = 0x80 | (0x3f & (c >>  6));
+    *out++ = 0x80 | (0x3f &  c);
+    return 5;
+  }
+
+  *out++ = 0xfc | (0x01 & (c >> 30));
+  *out++ = 0x80 | (0x3f & (c >> 24));
+  *out++ = 0x80 | (0x3f & (c >> 18));
+  *out++ = 0x80 | (0x3f & (c >> 12));
+  *out++ = 0x80 | (0x3f & (c >>  6));
+  *out++ = 0x80 | (0x3f &  c);
+  return 6;
+}