Refactor IDN handling and related unit tests

Most code is shared despite different available libraries now and
can more effeciently be tested.

midori/main.c

@@ -1776,22 +1776,8 @@ main (int    argc,
             i = 0;
             while (uris[i] != NULL)
             {
-                #ifdef HAVE_LIBSOUP_2_27_90
+                gchar* new_uri = sokoke_uri_to_ascii (uris[i]);
-                gchar* path;
+                katze_assign (uris[i], new_uri);
-                gchar* hostname = sokoke_hostname_from_uri (uris[i], &path);
-                gchar* encoded = g_hostname_to_ascii (hostname);
-
-                if (encoded)
-                {
-                    gchar* res = g_strconcat ("http://", encoded, path, NULL);
-                    g_free (uris[i]);
-                    g_free (encoded);
-                    uris[i] = res;
-                }
-                g_free (hostname);
-                #else
-                uris[i] = sokoke_idn_to_punycode (uris[i]);
-                #endif
                 i++;
             }
             result = midori_app_instance_send_uris (app, uris);

midori/sokoke.c

@@ -185,7 +185,6 @@ sokoke_spawn_program (const gchar* command,
     return TRUE;
 }
 
-#if defined (HAVE_LIBSOUP_2_27_90) || HAVE_LIBIDN
 /**
  * sokoke_hostname_from_uri:
  * @uri: an URI string
@@ -222,79 +221,88 @@ sokoke_hostname_from_uri (const gchar* uri,
         hostname = g_strdup (uri);
     return hostname;
 }
-#endif
 
 /**
- * sokoke_idn_to_punycode:
+ * sokoke_hostname_to_ascii:
+ * @uri: an URI string
+ *
+ * The specified hostname is encoded if it is not ASCII.
+ *
+ * If no IDN support is available at compile time,
+ * the hostname will be returned unaltered.
+ *
+ * Return value: a newly allocated hostname
+ **/
+static gchar*
+sokoke_hostname_to_ascii (const gchar* hostname)
+{
+    #ifdef HAVE_LIBSOUP_2_27_90
+    return g_hostname_to_ascii (hostname);
+    #elif HAVE_LIBIDN
+    uint32_t* q;
+    char* encoded;
+    int rc;
+
+    if ((q = stringprep_utf8_to_ucs4 (hostname, -1, NULL)))
+    {
+        rc = idna_to_ascii_4z (q, &encoded, IDNA_ALLOW_UNASSIGNED);
+        free (q);
+        if (rc == IDNA_SUCCESS)
+            return encoded;
+    }
+    #endif
+    return g_strdup (hostname);
+}
+
+/**
+ * sokoke_uri_to_ascii:
  * @uri: an URI string
  *
  * The specified URI is parsed and the hostname
  * part of it is encoded if it is not ASCII.
  *
- * If libIDN is not available at compile time,
+ * If no IDN support is available at compile time,
- * this code will pass the string unaltered.
+ * the URI will be returned unaltered.
  *
- * The called function owns the passed string.
+ * Return value: a newly allocated URI
- *
- * Return value: a newly allocated ASCII URI
  **/
 gchar*
-sokoke_idn_to_punycode (gchar* uri)
+sokoke_uri_to_ascii (const gchar* uri)
 {
-    #if HAVE_LIBIDN
     gchar* proto;
-    gchar* hostname;
-    gchar* path;
-    char *s;
-    uint32_t *q;
-    int rc;
-    gchar *result;
 
     if ((proto = g_utf8_strchr (uri, -1, ':')))
     {
         gulong offset;
         gchar* buffer;
 
-        /* 'file' URIs don't have a hostname */
-        if (!strcmp (proto, "file"))
-            return uri;
-
         offset = g_utf8_pointer_to_offset (uri, proto);
         buffer = g_malloc0 (offset + 1);
         g_utf8_strncpy (buffer, uri, offset);
         proto = buffer;
     }
 
-    hostname = sokoke_hostname_from_uri (uri, &path);
+    gchar* path;
+    gchar* hostname = sokoke_hostname_from_uri (uri, &path);
+    gchar* encoded = sokoke_hostname_to_ascii (hostname);
 
-    if (!(q = stringprep_utf8_to_ucs4 (hostname, -1, NULL)))
+    if (encoded)
     {
-        g_free (proto);
+        gchar* res = g_strconcat (proto ? proto : "", proto ? "://" : "",
+                                  encoded, path, NULL);
+        g_free (encoded);
+        return res;
+    }
     g_free (hostname);
-        return uri;
+    return g_strdup (uri);
 }
 
-    rc = idna_to_ascii_4z (q, &s, IDNA_ALLOW_UNASSIGNED);
+static gchar*
-    free (q);
+sokoke_idn_to_punycode (gchar* uri)
-    if (rc != IDNA_SUCCESS)
 {
-        g_free (proto);
+    #if HAVE_LIBIDN
-        g_free (hostname);
+    gchar* result = sokoke_uri_to_ascii (uri);
-        return uri;
-    }
-
-    if (proto)
-    {
-        result = g_strconcat (proto, "://", s, path ? path : "", NULL);
-        g_free (proto);
-        if (path)
-            g_free (hostname);
-    }
-    else
-        result = g_strdup (s);
     g_free (uri);
-    free (s);
-
     return result;
     #else
     return uri;

midori/sokoke.h

@@ -46,7 +46,7 @@ sokoke_hostname_from_uri                (const gchar*    uri,
                                          gchar**         path);
 
 gchar*
-sokoke_idn_to_punycode                  (gchar*          uri);
+sokoke_uri_to_ascii                     (const gchar*    uri);
 
 gchar*
 sokoke_magic_uri                        (const gchar*    uri,

tests/magic-uri.c

@@ -86,27 +86,46 @@ magic_uri_uri (void)
 static void
 magic_uri_idn (void)
 {
-    #if HAVE_LIBIDN
+    typedef struct
-    test_input ("http://www.münchhausen.at", "http://www.xn--mnchhausen-9db.at");
+    {
-    test_input ("http://www.خداوند.com/", "http://www.xn--mgbndb8il.com/");
+        const gchar* before;
-    test_input ("айкидо.com", "http://xn--80aildf0a.com");
+        const gchar* after;
-    test_input ("http://東京理科大学.jp", "http://xn--1lq68wkwbj6ugkpigi.jp");
+    } URIItem;
-    test_input ("https://青のネコ",  "https://xn--u9jthzcs263c");
+
+    static const URIItem items[] = {
+    #if HAVE_LIBIDN || defined (HAVE_LIBSOUP_2_27_90)
+     { "http://www.münchhausen.at", "http://www.xn--mnchhausen-9db.at" },
+     { "http://www.خداوند.com/", "http://www.xn--mgbndb8il.com/" },
+     { "айкидо.com", "xn--80aildf0a.com" },
+     { "http://東京理科大学.jp", "http://xn--1lq68wkwbj6ugkpigi.jp" },
+     { "https://青のネコ",  "https://xn--u9jthzcs263c" },
     #else
-    test_input ("http://www.münchhausen.at", "http://www.münchhausen.at");
+     { "http://www.münchhausen.at", NULL },
-    test_input ("http://www.خداوند.com/", "http://www.خداوند.com/");
+     { "http://www.خداوند.com/", NULL },
-    test_input ("айкидо.com", "http://айкидо.com");
+     { "айкидо.com", NULL },
-    test_input ("http://東京理科大学.jp", "http://東京理科大学.jp");
+     { "http://東京理科大学.jp", NULL },
-    test_input ("https://青のネコ.co.jp",  "https://青のネコ.co.jp");
+     { "https://青のネコ.co.jp",  NULL },
+    #endif
+    { "http://en.wikipedia.org/wiki/Kölsch_language", NULL },
+    { "file:///home/mark/frühstück", NULL },
+     };
+    guint i;
+
+    for (i = 0; i < G_N_ELEMENTS (items); i++)
+    {
+        gchar* result = sokoke_uri_to_ascii (items[i].before);
+        const gchar* after = items[i].after ? items[i].after : items[i].before;
+        sokoke_assert_str_equal (items[i].before, result, after);
+        g_free (result);
+    }
+
+    #if HAVE_LIBIDN
+    test_input ("айкидо.com", "http://xn--80aildf0a.com");
+    #else
+    test_input ("айкидо.com", "http://айкидо.com");
     #endif
-    test_input ("http://en.wikipedia.org/wiki/Kölsch_language",
-                "http://en.wikipedia.org/wiki/Kölsch_language");
-    test_input ("en.wikipedia.org/wiki/Kölsch_language",
-                "http://en.wikipedia.org/wiki/Kölsch_language");
     test_input ("sm Küchenzubehör", SM "Küchenzubehör");
     test_input ("sm 東京理科大学", SM "東京理科大学");
-    test_input ("file:///home/mark/frühstück",
-                "file:///home/mark/frühstück");
 }
 
 static void