From 137b2b1b440b5594b4d371c10884acd0a90df6bf Mon Sep 17 00:00:00 2001
From: Carl Hetherington <cth@carlh.net>
Date: Fri, 22 Mar 2019 00:09:07 +0000
Subject: [PATCH] Make careful_string_filter a little better and use it for KDM
 filenames.

---
 src/lib/cinema_kdms.cc |  2 +-
 src/lib/screen_kdm.cc  |  2 +-
 src/lib/util.cc        | 29 +++++++++++++++++++++++++----
 test/util_test.cc      |  9 +++++++++
 4 files changed, 36 insertions(+), 6 deletions(-)

diff --git a/src/lib/cinema_kdms.cc b/src/lib/cinema_kdms.cc
index 2e1d03b40..558aee504 100644
--- a/src/lib/cinema_kdms.cc
+++ b/src/lib/cinema_kdms.cc
@@ -67,7 +67,7 @@ CinemaKDMs::make_zip_file (boost::filesystem::path zip_file, dcp::NameFormat nam
 
 		name_values['s'] = i.screen->name;
 		name_values['i'] = i.kdm.id ();
-		string const name = name_format.get(name_values, ".xml");
+		string const name = careful_string_filter(name_format.get(name_values, ".xml"));
 		if (zip_add (zip, name.c_str(), source) == -1) {
 			throw runtime_error ("failed to add KDM to ZIP archive");
 		}
diff --git a/src/lib/screen_kdm.cc b/src/lib/screen_kdm.cc
index df74cc7ef..5b7692469 100644
--- a/src/lib/screen_kdm.cc
+++ b/src/lib/screen_kdm.cc
@@ -65,7 +65,7 @@ ScreenKDM::write_files (
 		name_values['c'] = i.screen->cinema ? i.screen->cinema->name : "";
 		name_values['s'] = i.screen->name;
 		name_values['i'] = i.kdm.id ();
-		boost::filesystem::path out = directory / (name_format.get(name_values, ".xml"));
+		boost::filesystem::path out = directory / careful_string_filter(name_format.get(name_values, ".xml"));
 		if (!boost::filesystem::exists (out) || confirm_overwrite (out)) {
 			i.kdm.as_xml (out);
 			++written;
diff --git a/src/lib/util.cc b/src/lib/util.cc
index b52655194..c20891464 100644
--- a/src/lib/util.cc
+++ b/src/lib/util.cc
@@ -62,6 +62,7 @@ extern "C" {
 #include <boost/range/algorithm/replace_if.hpp>
 #include <boost/thread.hpp>
 #include <boost/filesystem.hpp>
+#include <boost/locale.hpp>
 #ifdef DCPOMATIC_WINDOWS
 #include <boost/locale.hpp>
 #include <dbghelp.h>
@@ -80,6 +81,7 @@ extern "C" {
 #include "i18n.h"
 
 using std::string;
+using std::wstring;
 using std::setfill;
 using std::ostream;
 using std::endl;
@@ -743,15 +745,34 @@ careful_string_filter (string s)
 	   Safety first and all that.
 	*/
 
+	wstring ws = boost::locale::conv::utf_to_utf<wchar_t>(s);
+
 	string out;
 	string const allowed = "0123456789ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz-_%.+";
-	for (size_t i = 0; i < s.size(); ++i) {
-		if (allowed.find (s[i]) != string::npos) {
-			out += s[i];
+	for (size_t i = 0; i < ws.size(); ++i) {
+
+		wchar_t c = ws[i];
+
+		/* Remove some accents */
+		if (wstring(L"áàâ").find(c) != string::npos) {
+			c = 'a';
+		}
+		if (wstring(L"éèêë").find(c) != string::npos) {
+			c = 'e';
+		}
+		if (wstring(L"ö").find(c) != string::npos) {
+			c = 'o';
+		}
+		if (wstring(L"ü").find(c) != string::npos) {
+			c = 'u';
+		}
+
+		if (allowed.find(c) != string::npos) {
+			out += c;
 		}
 	}
 
-	return out;
+	return boost::locale::conv::utf_to_utf<char>(out);
 }
 
 /** @param mapped List of mapped audio channels from a Film.
diff --git a/test/util_test.cc b/test/util_test.cc
index d8cb61fdc..1e13efa82 100644
--- a/test/util_test.cc
+++ b/test/util_test.cc
@@ -119,3 +119,12 @@ BOOST_AUTO_TEST_CASE (swaroop_chain_test)
 	BOOST_CHECK (cc->root_to_leaf() == back->root_to_leaf());
 }
 #endif
+
+BOOST_AUTO_TEST_CASE (careful_string_filter_test)
+{
+	BOOST_CHECK_EQUAL ("hello_world", careful_string_filter("hello_world"));
+	BOOST_CHECK_EQUAL ("hello_world", careful_string_filter("héllo_world"));
+	BOOST_CHECK_EQUAL ("hello_world", careful_string_filter("héllo_wörld"));
+	BOOST_CHECK_EQUAL ("hello_world", careful_string_filter("héllo_wörld"));
+	BOOST_CHECK_EQUAL ("hello_world_a", careful_string_filter("héllo_wörld_à"));
+}
-- 
2.30.2