Added support for Unicode text in Ledger files, thanks to 'utfcpp', which can

be located at http://utfcpp.sourceforge.net.

1 files changed, 51 insertions, 1 deletions

diff --git a/src/format.h b/src/format.h
index 13a2fff2..fbfe452e 100644
--- a/src/format.h
+++ b/src/format.h
@@ -32,13 +32,63 @@
 #ifndef _FORMAT_H
 #define _FORMAT_H
 
+#define SUPPORT_UNICODE 1
+
 #include "journal.h"
 #include "expr.h"
+#if defined(SUPPORT_UNICODE)
+#include "utf8.h"
+#endif
 
 namespace ledger {
 
 DECLARE_EXCEPTION(format_error, std::runtime_error);
 
+#if defined(SUPPORT_UNICODE)
+/**
+ * @class unistring
+ *
+ * @brief Abstract working with UTF-32 encoded Unicode strings
+ *
+ * The input to the string is a UTF8 encoded ledger::string, which can
+ * then have its true length be taken, or characters extracted.
+ */
+class unistring
+{
+  std::vector<uint32_t> utf32chars;
+
+public:
+  unistring(const string& input)
+  {
+    TRACE_CTOR(unistring, "");
+
+    const char * p   = input.c_str();
+    std::size_t	 len = input.length();
+
+    VERIFY(utf8::is_valid(p, p + len));
+
+    utf8::utf8to32(p, p + len, std::back_inserter(utf32chars));
+  }
+  ~unistring() {
+    TRACE_DTOR(unistring);
+  }
+
+  std::size_t length() const {
+    return utf32chars.size();
+  }
+
+  string extract(const std::size_t begin = 0,
+		 const std::size_t len   = 0) const
+  {
+    string utf8result;
+    utf8::utf32to8(utf32chars.begin() + begin,
+		   utf32chars.begin() + begin + (len ? len : length()),
+		   std::back_inserter(utf8result));
+    return utf8result;
+  }
+};
+#endif
+
 class report_t;
 
 class format_t : public noncopyable
@@ -135,7 +185,7 @@ public:
       elem->dump(out);
   }
 
-  static string truncate(const string& str, unsigned int width,
+  static string truncate(const unistring& str, unsigned int width,
 			 const bool is_account = false);
 };