From 6165a2ece3a0bce92b28ffcf5c4d407f199e4d3f Mon Sep 17 00:00:00 2001
From: Tomasz Sowa <t.sowa@ttmath.org>
Date: Wed, 19 Jun 2024 22:54:16 +0200
Subject: [PATCH] add a new utf8_to_output_function() method

add such a method:
template<typename OutputFunction>
bool utf8_to_output_function(const char * utf8, OutputFunction output_function, int mode = 1);
---
 src/utf8/utf8.h | 65 +++++++++++++++++++++++++++++++++++++++++--------
 1 file changed, 55 insertions(+), 10 deletions(-)

diff --git a/src/utf8/utf8.h b/src/utf8/utf8.h
index c2d3f90..edf812b 100644
--- a/src/utf8/utf8.h
+++ b/src/utf8/utf8.h
@@ -169,16 +169,6 @@ size_t int_to_wide(int c, wchar_t * res, size_t max_buf_len);
 bool int_to_wide(int c, std::wstring & res);
 
 
-/*
- * call a convert_function for each character from an utf8 string
- *
- * this function takes one int parameter:
- * output_function(int c)
- */
-template<typename OutputFunction>
-bool utf8_to_output_function(const char * utf8, size_t utf8_len, OutputFunction output_function, int mode = 1);
-
-
 
 /*
  * convert an UTF-8 string to a wide string
@@ -209,6 +199,18 @@ bool utf8_to_output_function_by_index(const Stream & stream, OutputFunction outp
 template<typename StreamIteratorType, typename StreamOrStringType>
 bool utf8_to_wide(StreamIteratorType & iterator_in, const StreamIteratorType & iterator_end, StreamOrStringType & out_stream, bool clear_stream = true, int mode = 1);
 
+/*
+ * call a convert_function for each character from an utf8 string
+ *
+ * this function takes one int parameter:
+ * output_function(int c)
+ */
+template<typename OutputFunction>
+bool utf8_to_output_function(const char * utf8, size_t utf8_len, OutputFunction output_function, int mode = 1);
+
+template<typename OutputFunction>
+bool utf8_to_output_function(const char * utf8, OutputFunction output_function, int mode = 1);
+
 template<typename StreamIteratorType, typename OutputFunction>
 bool utf8_to_output_function(StreamIteratorType & iterator_in, const StreamIteratorType & iterator_end, OutputFunction output_function, int mode = 1);
 
@@ -571,6 +573,46 @@ return !was_error;
 
 
 
+template<typename OutputFunction>
+bool utf8_to_output_function(const char * utf8, OutputFunction output_function, int mode)
+{
+int z;
+size_t len;
+bool correct, was_error = false;
+
+	while( *utf8 )
+	{
+		if( (unsigned char)*utf8 <= 0x7f )
+		{
+			// small optimization
+			len = 1;
+			correct = true;
+			z = static_cast<unsigned char>(*utf8);
+		}
+		else
+		{
+			len = utf8_to_int(utf8, z, correct); // the len will be different from zero
+		}
+
+		if( !correct )
+		{
+			if( mode == 1 )
+				output_function(0xFFFD); // U+FFFD "replacement character"
+
+			was_error = true;
+		}
+		else
+		{
+			output_function(z);
+		}
+
+		utf8 += len;
+	}
+
+	return !was_error;
+}
+
+
 template<typename OutputFunction>
 bool wide_to_output_function(const wchar_t * str, size_t len, OutputFunction output_function, int mode)
 {
@@ -976,6 +1018,9 @@ bool utf8_to_wide(StreamIteratorType & iterator_in, const StreamIteratorType & i
 
 
 
+
+
+
 template<typename StreamIteratorType, typename OutputFunction>
 bool utf8_to_output_function(StreamIteratorType & iterator_in, const StreamIteratorType & iterator_end, OutputFunction output_function, int mode)
 {