- added some converting methods: esc_to_json(...), esc_to_xml(...), esc_to_csv() (convert/misc.h)

- BaseParser: added possibility to read from TextStream and WTextStream
- HTMLParser: added filter(const WTextStream & in, Stream & out, ...) method
- added utf8_stream.h with one method:
  template<typename StreamIteratorType>
  size_t utf8_to_int(
    StreamIteratorType & iterator_in,
    StreamIteratorType & iterator_end,
    int & res,
    bool & correct)

src/convert/misc.cpp

@@ -5,7 +5,7 @@
  */
 
 /*
- * Copyright (c) 2017, Tomasz Sowa
+ * Copyright (c) 2017-2021, Tomasz Sowa
  * All rights reserved.
  *
  * Redistribution and use in source and binary forms, with or without
@@ -36,6 +36,8 @@
  */
 
 #include "misc.h"
+#include "inttostr.h"
+#include "utf8/utf8.h"
 
 
 namespace pt
@@ -49,6 +51,287 @@ void SetOverflow(bool * was_overflow, bool val)
 }
 
 
+void esc_to_json(char val, Stream & out)
+{
+	if( (unsigned char)val < 32 )
+	{
+		char buf[10];
+		size_t len;
+		Toa((unsigned char)val, buf, sizeof(buf)/sizeof(char), 16, &len);
+
+		out << "\\u";
+
+		if( len < 4 )
+		{
+			for(size_t i=0 ; i < (4-len) ; ++i)
+			{
+				out << '0';
+			}
+		}
+
+		out << buf;
+	}
+	else
+	{
+	// CHECKME
+	// \r \n \t are <32 and will be serialized os \u.... above
+
+		switch( val )
+		{
+		case 0:		out << '\\';	out << '0';		break;	// may to skip this character is better?
+		case '\r':	out << '\\';	out << 'r';		break;
+		case '\n':	out << '\\';	out << 'n';		break;
+		case '\t':	out << '\\';	out << 't';		break;
+		case 0x08:	out << '\\';	out << 'b';		break;
+		case 0x0c:	out << '\\';	out << 'f';		break;
+		case '\\':	out << '\\';	out << '\\';		break;
+		case '"':	out << '\\';	out << '\"';		break;
+		default:
+			out << val;
+		}
+	}
+}
+
+
+void esc_to_json(wchar_t val, Stream & out)
+{
+	char utf8_buf[10];
+	std::size_t utf8_buf_len = sizeof(utf8_buf) / sizeof(char);
+
+	size_t len = int_to_utf8(static_cast<int>(val), utf8_buf, utf8_buf_len);
+
+	for(size_t a = 0 ; a < len ; ++a)
+	{
+		esc_to_json(utf8_buf[a], out);
+	}
+}
+
+
+void esc_to_json(const char * c, pt::Stream & out)
+{
+	for(size_t i = 0 ; c[i] != 0 ; ++i)
+	{
+		esc_to_json(c[i], out);
+	}
+}
+
+
+void esc_to_json(const char * c, std::size_t len, pt::Stream & out)
+{
+	for(size_t i = 0 ; i < len ; ++i)
+	{
+		esc_to_json(c[i], out);
+	}
+}
+
+
+void esc_to_json(const wchar_t * c, pt::Stream & out)
+{
+	for(size_t i = 0 ; c[i] != 0 ; ++i)
+	{
+		esc_to_json(c[i], out);
+	}
+}
+
+
+void esc_to_json(const wchar_t * c, size_t len, pt::Stream & out)
+{
+	for(size_t i = 0 ; i < len ; ++i)
+	{
+		esc_to_json(c[i], out);
+	}
+}
+
+
+void esc_to_json(const std::string & in, Stream & out)
+{
+	esc_to_json(in.c_str(), in.size(), out);
+}
+
+
+void esc_to_json(const std::wstring & in, Stream & out)
+{
+	esc_to_json(in.c_str(), in.size(), out);
+}
+
+
+
+
+
+
+void esc_to_xml(char val, Stream & out)
+{
+	switch(val)
+	{
+	case '<':
+		out << "&lt;";
+		break;
+
+	case '>':
+		out << "&gt;";
+		break;
+
+	case '&':
+		out << "&amp;";
+		break;
+
+	case '"':
+		out << "&quot;";
+		break;
+
+	default:
+		out << val;
+		break;
+
+	// what about zero (null) character?
+	}
+}
+
+void esc_to_xml(wchar_t val, Stream & out)
+{
+	char utf8_buf[10];
+	std::size_t utf8_buf_len = sizeof(utf8_buf) / sizeof(char);
+
+	size_t len = int_to_utf8(static_cast<int>(val), utf8_buf, utf8_buf_len);
+
+	for(size_t a = 0 ; a < len ; ++a)
+	{
+		esc_to_xml(utf8_buf[a], out);
+	}
+}
+
+
+void esc_to_xml(const char * c, pt::Stream & out)
+{
+	for(size_t i = 0 ; c[i] != 0 ; ++i)
+	{
+		esc_to_xml(c[i], out);
+	}
+}
+
+
+void esc_to_xml(const char * c, std::size_t len, pt::Stream & out)
+{
+	for(size_t i = 0 ; i < len ; ++i)
+	{
+		esc_to_xml(c[i], out);
+	}
+}
+
+
+void esc_to_xml(const wchar_t * c, pt::Stream & out)
+{
+	for(size_t i = 0 ; c[i] != 0 ; ++i)
+	{
+		esc_to_xml(c[i], out);
+	}
+}
+
+
+void esc_to_xml(const wchar_t * c, size_t len, pt::Stream & out)
+{
+	for(size_t i = 0 ; i < len ; ++i)
+	{
+		esc_to_xml(c[i], out);
+	}
+}
+
+
+void esc_to_xml(const std::string & in, Stream & out)
+{
+	esc_to_xml(in.c_str(), in.size(), out);
+}
+
+
+void esc_to_xml(const std::wstring & in, Stream & out)
+{
+	esc_to_xml(in.c_str(), in.size(), out);
+}
+
+
+
+
+
+void esc_to_csv(char c, pt::Stream & out)
+{
+	switch(c)
+	{
+	case '"':
+		out << "\"\"";
+		break;
+
+	default:
+		out << c;
+		break;
+
+	// what about zero (null) character?
+	}
+}
+
+
+void esc_to_csv(wchar_t val, Stream & out)
+{
+	char utf8_buf[10];
+	std::size_t utf8_buf_len = sizeof(utf8_buf) / sizeof(char);
+
+	size_t len = int_to_utf8(static_cast<int>(val), utf8_buf, utf8_buf_len);
+
+	for(size_t a = 0 ; a < len ; ++a)
+	{
+		esc_to_csv(utf8_buf[a], out);
+	}
+}
+
+
+
+void esc_to_csv(const char * c, pt::Stream & out)
+{
+	for(size_t i = 0 ; c[i] != 0 ; ++i)
+	{
+		esc_to_csv(c[i], out);
+	}
+}
+
+
+void esc_to_csv(const char * c, std::size_t len, pt::Stream & out)
+{
+	for(size_t i = 0 ; i < len ; ++i)
+	{
+		esc_to_csv(c[i], out);
+	}
+}
+
+
+void esc_to_csv(const wchar_t * c, pt::Stream & out)
+{
+	for(size_t i = 0 ; c[i] != 0 ; ++i)
+	{
+		esc_to_csv(c[i], out);
+	}
+}
+
+
+void esc_to_csv(const wchar_t * c, size_t len, pt::Stream & out)
+{
+	for(size_t i = 0 ; i < len ; ++i)
+	{
+		esc_to_csv(c[i], out);
+	}
+}
+
+
+void esc_to_csv(const std::string & in, Stream & out)
+{
+	esc_to_csv(in.c_str(), in.size(), out);
+}
+
+
+void esc_to_csv(const std::wstring & in, Stream & out)
+{
+	esc_to_csv(in.c_str(), in.size(), out);
+}
+
+
 
 
 }