Add a function to retrieve the decoded version of IDNA URLs

2 files changed, 59 insertions, 0 deletions

diff --git a/utils/nsurl.c b/utils/nsurl.c
index 8d53be84f..bb3054df1 100644
--- a/utils/nsurl.c
+++ b/utils/nsurl.c
@@ -1698,6 +1698,51 @@ const char *nsurl_access(const nsurl *url)
 	return url->string;
 }
 
+const char *nsurl_access_utf8(const nsurl *url)
+{
+	lwc_string *host;
+	char *idna_host;
+	size_t idna_host_len;
+	char *scheme;
+	size_t scheme_len;
+	char *path;
+	size_t path_len;
+	char *idna_url;
+	size_t idna_url_len;
+
+	assert(url != NULL);
+
+	host = nsurl_get_component(url, NSURL_HOST);
+	if (idna_decode(lwc_string_data(host), lwc_string_length(host),
+		&idna_host, &idna_host_len) != NSERROR_OK) {
+		lwc_string_unref(host);
+		return strdup(url->string);
+	}
+
+	lwc_string_unref(host);
+
+	if (nsurl_get(url, NSURL_SCHEME | NSURL_CREDENTIALS,
+		&scheme, &scheme_len) != NSERROR_OK) {
+		return strdup(url->string);
+	}
+
+	if (nsurl_get(url, NSURL_PORT | NSURL_PATH | NSURL_QUERY,
+		&path, &path_len) != NSERROR_OK) {
+		return strdup(url->string);
+	}
+
+	idna_url_len = scheme_len + idna_host_len + path_len + 1; /* +1 for \0 */
+	idna_url = malloc(idna_url_len); 
+
+	if (idna_url == NULL) {
+		return strdup(url->string);
+	}
+
+	snprintf(idna_url, idna_url_len, "%s%s%s", scheme, idna_host, path);
+
+	return(idna_url);
+}
+
 
 /* exported interface, documented in nsurl.h */
 const char *nsurl_access_leaf(const nsurl *url)
diff --git a/utils/nsurl.h b/utils/nsurl.h
index b84f55eed..4fbc17b49 100644
--- a/utils/nsurl.h
+++ b/utils/nsurl.h
@@ -181,6 +181,20 @@ const char *nsurl_access(const nsurl *url);
 
 
 /**
+ * Access a NetSurf URL object as a UTF-8 string (for human readable IDNA)
+ *
+ * \param url	  NetSurf URL to retrieve a string pointer for.
+ * \return the required string
+ *
+ * It is up to the client to free the returned string when they have
+ * finished with it.
+ *
+ * The returned string has a trailing '\0'.
+ */
+const char *nsurl_access_utf8(const nsurl *url);
+
+
+/**
  * Access a URL's path leaf as a string
  *
  * \param url	  NetSurf URL to retrieve a string pointer for.