Added support for unicode sheet titles

Fixes #378

source/detail/unicode.cpp

@@ -29,6 +29,7 @@
 #pragma clang diagnostic pop
 
 #include <detail/unicode.hpp>
+#include <xlnt/utils/exceptions.hpp>
 
 namespace xlnt {
 namespace detail {
@@ -69,5 +70,16 @@ std::string latin1_to_utf8(const std::string &latin1)
     return utf8;
 }
 
+size_t string_length(const std::string &utf8_string)
+{
+    auto end_it = utf8::find_invalid(utf8_string.begin(), utf8_string.end());
+    if (end_it != utf8_string.end())
+    {
+        throw xlnt::exception("Invalid UTF-8 encoding detected");
+    }
+
+    return utf8::distance(utf8_string.begin(), end_it);
+}
+
 } // namespace detail
 } // namespace xlnt

source/detail/unicode.hpp

@@ -29,6 +29,7 @@ namespace detail {
 std::u16string utf8_to_utf16(const std::string &utf8_string);
 std::string utf16_to_utf8(const std::u16string &utf16_string);
 std::string latin1_to_utf8(const std::string &latin1);
+size_t string_length(const std::string &utf8_string);
 
 } // namespace detail
 } // namespace xlnt

source/worksheet/worksheet.cpp

@@ -46,6 +46,7 @@
 #include <detail/implementations/cell_impl.hpp>
 #include <detail/implementations/workbook_impl.hpp>
 #include <detail/implementations/worksheet_impl.hpp>
+#include <detail/unicode.hpp>
 
 namespace {
 
@@ -234,7 +235,7 @@ void worksheet::title(const std::string &title)
         return;
     }
     // excel limits worksheet titles to 31 characters
-    if (title.empty() || title.length() > 31)
+    if (title.empty() || detail::string_length(title) > 31)
     {
         throw invalid_sheet_title(title);
     }

tests/worksheet/worksheet_test_suite.cpp

@@ -104,6 +104,7 @@ public:
         register_test(test_clear_cell);
         register_test(test_clear_row);
         register_test(test_set_title);
+        register_test(test_set_title_unicode);
     }
 
     void test_new_worksheet()
@@ -1259,5 +1260,17 @@ public:
         xlnt_assert(ws1_title == ws1.title());
         xlnt_assert(ws2_title == ws2.title());
     }
+
+    void test_set_title_unicode()
+    {
+        xlnt::workbook wb;
+        auto ws = wb.active_sheet();
+        // the 31 char limit also applies to 4-byte characters
+        const std::string test_long_utf8_title("巧みな外交は戦争を避ける助けとなる。");
+        xlnt_assert_throws_nothing(ws.title(test_long_utf8_title));
+        const std::string invalid_unicode("\xe6\x97\xa5\xd1\x88\xfa");
+        xlnt_assert_throws(ws.title(invalid_unicode),
+                           xlnt::exception);
+    }
 };
-static worksheet_test_suite x;
+static worksheet_test_suite x;