add_cp_html_to_string: New function.

To be used when strings from gettext or strftime must be inserted
into an HTML document with a potentially different charset.

src/util/conv.c

@@ -288,7 +288,6 @@ add_html_to_string(struct string *string, const unsigned char *src, int len)
 				string->source[rollback_length] = '\0';
 				return NULL;
 			}
-			
 		} else {
 			if (!add_char_to_string(string, *src))
 				return NULL;
@@ -298,6 +297,50 @@ add_html_to_string(struct string *string, const unsigned char *src, int len)
 	return string;
 }
 
+struct string *
+add_cp_html_to_string(struct string *string, int src_codepage,
+		      const unsigned char *src, int len)
+{
+	const unsigned char *const end = src + len;
+	unicode_val_T unicode;
+
+	while (src != end) {
+		if (is_cp_utf8(src_codepage)) {
+#ifdef CONFIG_UTF8
+			unicode = utf8_to_unicode((unsigned char **) &src,
+						  end);
+			if (unicode == UCS_NO_CHAR)
+				break;
+#else  /* !CONFIG_UTF8 */
+			/* Cannot parse UTF-8 without CONFIG_UTF8.
+			 * Pretend the input is ISO-8859-1 instead.  */
+			unicode = *src++;
+#endif /* !CONFIG_UTF8 */
+		} else {
+			unicode = cp2u(src_codepage, *src++);
+		}
+
+		if (unicode < 0x20 || unicode >= 0x7F
+		    || unicode == '<' || unicode == '>' || unicode == '&'
+		    || unicode == '\"' || unicode == '\'') {
+			int rollback_length = string->length;
+
+			if (!add_bytes_to_string(string, "&#", 2)
+			    || !add_long_to_string(string, unicode)
+			    || !add_char_to_string(string, ';')) {
+				string->length = rollback_length;
+				string->source[rollback_length] = '\0';
+				return NULL;
+			}
+		} else {
+			if (!add_char_to_string(string, unicode))
+				return NULL;
+		}
+	}
+
+	return string;
+}
+
 /* TODO Optimize later --pasky */
 struct string *
 add_quoted_to_string(struct string *string, const unsigned char *src, int len)

src/util/conv.h

@@ -86,6 +86,12 @@ add_string_replace(struct string *string, unsigned char *src, int len,
  * know the charset of the input data.)  */
 struct string *add_html_to_string(struct string *string, const unsigned char *html, int htmllen);
 
+/* Convert reserved or non-ASCII chars to html &#xx;.  The resulting
+ * string can be correctly parsed in any charset where bytes
+ * 0x20...0x7E match ASCII.  */
+struct string *add_cp_html_to_string(struct string *string, int src_codepage,
+				     const unsigned char *html, int htmllen);
+
 /* Escapes \ and " with a \ */
 struct string *add_quoted_to_string(struct string *string, const unsigned char *q, int qlen);