From a9fb6a18ecaff6b2f39a4710677385557af15924 Mon Sep 17 00:00:00 2001
From: Berke Viktor <berkeviktor@aol.com>
Date: Fri, 20 Jan 2012 00:38:57 +0100
Subject: [PATCH] initial patch for filtering malicious characters on win32
 (Khisanth)

---
 src/common/server.c | 17 +++++++++++++++
 src/common/text.c   | 51 ++++++++++++++++++++++++++++++++++++++++++++-
 src/common/text.h   |  3 +++
 3 files changed, 70 insertions(+), 1 deletion(-)

diff --git a/src/common/server.c b/src/common/server.c
index cdd67cda..53e877e3 100644
--- a/src/common/server.c
+++ b/src/common/server.c
@@ -310,6 +310,11 @@ server_inline (server *serv, char *line, int len)
 {
 	char *utf_line_allocated = NULL;
 
+#ifdef WIN32
+	char *cleaned_line;
+	int cleaned_len;
+#endif
+
 	/* Checks whether we're set to use UTF-8 charset */
 	if (serv->using_irc ||				/* 1. using CP1252/UTF-8 Hybrid */
 		(serv->encoding == NULL && prefs.utf8_locale) || /* OR 2. using system default->UTF-8 */
@@ -396,12 +401,24 @@ server_inline (server *serv, char *line, int len)
 		}
 	}
 
+#ifdef WIN32
+	cleaned_line = text_replace_non_bmp (line, len, &cleaned_len);
+	if (cleaned_line != NULL ) {
+		line = cleaned_line;
+		len = cleaned_len;
+	}
+#endif
+
 	fe_add_rawlog (serv, line, len, FALSE);
 	url_check_line (line, len);
 
 	/* let proto-irc.c handle it */
 	serv->p_inline (serv, line, len);
 
+#ifdef WIN32
+	g_free (cleaned_line);
+#endif
+
 	if (utf_line_allocated != NULL) /* only if a special copy was allocated */
 		g_free (utf_line_allocated);
 }
diff --git a/src/common/text.c b/src/common/text.c
index 9aa1bc85..93392d1a 100644
--- a/src/common/text.c
+++ b/src/common/text.c
@@ -275,7 +275,10 @@ scrollback_load (session *sess)
 	time_t stamp;
 	int lines;
 
-#ifndef WIN32
+#ifdef WIN32
+	char *cleaned_text;
+	int cleaned_len;
+#else
 	char *map, *end_map;
 	struct stat statbuf;
 	const char *begin, *eol;
@@ -371,6 +374,12 @@ scrollback_load (session *sess)
 			if (text)
 			{
 				text = strip_color (text + 1, -1, STRIP_COLOR);
+				cleaned_text = text_replace_non_bmp (text, -1, &cleaned_len);
+				if (cleaned_text != NULL)
+				{
+					g_free (text);
+					text = cleaned_text;
+				}
 				fe_print_text (sess, text, stamp);
 				g_free (text);
 			}
@@ -852,6 +861,46 @@ iso_8859_1_to_utf8 (unsigned char *text, int len, gsize *bytes_written)
 	return res;
 }
 
+#ifdef WIN32
+/* replace characters outside of the Basic Multilingual Plane with
+ * replacement characters (0xFFFD) */
+char *
+text_replace_non_bmp (char *utf8_input, int input_length, glong *output_length)
+{
+	gunichar *ucs4_text;
+	gunichar suspect;
+	gchar *utf8_text;
+	glong ucs4_length;
+	glong index;
+
+	ucs4_text = g_utf8_to_ucs4_fast (utf8_input, input_length, &ucs4_length);
+
+	/* replace anything not in the Basic Multilingual Plane
+	 * (code points above 0xFFFF) with the replacement
+	 * character */
+	for (index = 0; index < ucs4_length; index++)
+	{
+		suspect = ucs4_text[index];
+		if ((suspect >= 0x1D173 && suspect <= 0x1D17A)
+			|| (suspect >= 0xE0001 && suspect <= 0xE007F))
+		{
+			ucs4_text[index] = 0xFFFD; /* replacement character */
+		}
+	}
+
+	utf8_text = g_ucs4_to_utf8 (
+		ucs4_text,
+		ucs4_length,
+		NULL,
+		output_length,
+		NULL
+	);
+	g_free (ucs4_text);
+
+	return utf8_text;
+}
+#endif
+
 char *
 text_validate (char **text, int *len)
 {
diff --git a/src/common/text.h b/src/common/text.h
index 150821ae..6d5ac03e 100644
--- a/src/common/text.h
+++ b/src/common/text.h
@@ -28,6 +28,9 @@ int pevent_load (char *filename);
 void pevent_make_pntevts (void);
 void text_emit (int index, session *sess, char *a, char *b, char *c, char *d);
 int text_emit_by_name (char *name, session *sess, char *a, char *b, char *c, char *d);
+#ifdef WIN32
+char *text_replace_non_bmp (char *utf8_input, int input_length, glong *output_length);
+#endif
 char *text_validate (char **text, int *len);
 int get_stamp_str (char *fmt, time_t tim, char **ret);
 void format_event (session *sess, int index, char **args, char *o, int sizeofo, unsigned int stripcolor_args);