Node/document saving is now performed via new xml_writer interface, save_file now works without STL

git-svn-id: http://pugixml.googlecode.com/svn/trunk@97 99668b35-9821-0410-8761-19e4c4f06640

2 files changed, 346 insertions, 161 deletions

diff --git a/src/pugixml.cpp b/src/pugixml.cpp
index 30ed7aa..b98bf4f 100644
--- a/src/pugixml.cpp
+++ b/src/pugixml.cpp
@@ -434,60 +434,6 @@ namespace
 	template <bool _1, bool _2, bool _3, bool _4> const bool opt4_to_type<_1, _2, _3, _4>::o3 = _3;

 	template <bool _1, bool _2, bool _3, bool _4> const bool opt4_to_type<_1, _2, _3, _4>::o4 = _4;

 

-#ifndef PUGIXML_NO_STL

-	template <typename opt1> void text_output_escaped(std::ostream& os, const char* s, opt1)

-	{

-		const bool attribute = opt1::o1;

-

-		while (*s)

-		{

-			const char* prev = s;

-			

-			// While *s is a usual symbol

-			while (*s && *s != '&' && *s != '<' && *s != '>' && (*s != '"' || !attribute)

-					&& (*s < 0 || *s >= 32 || (*s == '\r' && !attribute) || (*s == '\n' && !attribute) || *s == '\t'))

-				++s;

-		

-			if (prev != s) os.write(prev, static_cast<std::streamsize>(s - prev));

-

-			switch (*s)

-			{

-				case 0: break;

-				case '&':

-					os << "&amp;";

-					++s;

-					break;

-				case '<':

-					os << "&lt;";

-					++s;

-					break;

-				case '>':

-					os << "&gt;";

-					++s;

-					break;

-				case '"':

-					os << "&quot;";

-					++s;

-					break;

-				case '\r':

-					os << "&#13;";

-					++s;

-					break;

-				case '\n':

-					os << "&#10;";

-					++s;

-					break;

-				default: // s is not a usual symbol

-				{

-					unsigned int ch = (unsigned char)*s++;

-

-					os << "&#" << ch << ";";

-				}

-			}

-		}

-	}

-#endif

-

 	struct gap

 	{

 		char* end;

@@ -1361,6 +1307,236 @@ namespace
 			return find;

 		}

 	}

+

+	// Output facilities

+	struct xml_buffered_writer

+	{

+		xml_buffered_writer(const xml_buffered_writer&);

+		xml_buffered_writer& operator=(const xml_buffered_writer&);

+

+		xml_buffered_writer(xml_writer& writer): writer(writer), bufsize(0)

+		{

+		}

+

+		~xml_buffered_writer()

+		{

+			flush();

+		}

+

+		void flush()

+		{

+			if (bufsize > 0) writer.write(buffer, bufsize);

+			bufsize = 0;

+		}

+

+		void write(const void* data, size_t size)

+		{

+			if (bufsize + size > sizeof(buffer))

+			{

+				flush();

+

+				if (size > sizeof(buffer))

+				{

+					writer.write(data, size);

+					return;

+				}

+			}

+			else

+			{

+				memcpy(buffer + bufsize, data, size);

+				bufsize += size;

+			}

+		}

+

+		void write(const char* data)

+		{

+			write(data, strlen(data));

+		}

+

+		void write(char data)

+		{

+			if (bufsize + 1 > sizeof(buffer)) flush();

+

+			buffer[bufsize++] = data;

+		}

+

+		xml_writer& writer;

+		char buffer[8192];

+		size_t bufsize;

+	};

+

+	template <typename opt1> void text_output_escaped(xml_buffered_writer& writer, const char* s, opt1)

+	{

+		const bool attribute = opt1::o1;

+

+		while (*s)

+		{

+			const char* prev = s;

+			

+			// While *s is a usual symbol

+			while (*s && *s != '&' && *s != '<' && *s != '>' && (*s != '"' || !attribute)

+					&& (*s < 0 || *s >= 32 || (*s == '\r' && !attribute) || (*s == '\n' && !attribute) || *s == '\t'))

+				++s;

+		

+			writer.write(prev, static_cast<size_t>(s - prev));

+

+			switch (*s)

+			{

+				case 0: break;

+				case '&':

+					writer.write("&amp;");

+					++s;

+					break;

+				case '<':

+					writer.write("&lt;");

+					++s;

+					break;

+				case '>':

+					writer.write("&gt;");

+					++s;

+					break;

+				case '"':

+					writer.write("&quot;");

+					++s;

+					break;

+				case '\r':

+					writer.write("&#13;");

+					++s;

+					break;

+				case '\n':

+					writer.write("&#10;");

+					++s;

+					break;

+				default: // s is not a usual symbol

+				{

+					unsigned int ch = (unsigned char)*s++;

+

+					char buf[8];

+					sprintf(buf, "&#%d;", ch);

+

+					writer.write(buf);

+				}

+			}

+		}

+	}

+

+	void node_output(xml_buffered_writer& writer, const xml_node& node, const char* indent, unsigned int flags, unsigned int depth)

+	{

+		if ((flags & format_indent) != 0 && (flags & format_raw) == 0)

+			for (unsigned int i = 0; i < depth; ++i) writer.write(indent);

+

+		switch (node.type())

+		{

+		case node_document:

+		{

+			for (xml_node n = node.first_child(); n; n = n.next_sibling())

+				node_output(writer, n, indent, flags, depth);

+			break;

+		}

+			

+		case node_element:

+		{

+			writer.write('<');

+			writer.write(node.name());

+

+			for (xml_attribute a = node.first_attribute(); a; a = a.next_attribute())

+			{

+				writer.write(' ');

+				writer.write(a.name());

+				writer.write('=');

+				writer.write('"');

+

+				text_output_escaped(writer, a.value(), opt1_to_type<1>());

+

+				writer.write('"');

+			}

+

+			if (flags & format_raw)

+			{

+				if (!node.first_child())

+					writer.write(" />");

+				else

+				{

+					writer.write('>');

+

+					for (xml_node n = node.first_child(); n; n = n.next_sibling())

+						node_output(writer, n, indent, flags, depth + 1);

+

+					writer.write('<');

+					writer.write('/');

+					writer.write(node.name());

+					writer.write('>');

+				}

+			}

+			else if (!node.first_child())

+				writer.write(" />\n");

+			else if (node.first_child() == node.last_child() && node.first_child().type() == node_pcdata)

+			{

+				writer.write('>');

+

+				text_output_escaped(writer, node.first_child().value(), opt1_to_type<0>());

+

+				writer.write('<');

+				writer.write('/');

+				writer.write(node.name());

+				writer.write('>');

+				writer.write('\n');

+			}

+			else

+			{

+				writer.write('>');

+				writer.write('\n');

+				

+				for (xml_node n = node.first_child(); n; n = n.next_sibling())

+					node_output(writer, n, indent, flags, depth + 1);

+

+				if ((flags & format_indent) != 0 && (flags & format_raw) == 0)

+					for (unsigned int i = 0; i < depth; ++i) writer.write(indent);

+				

+				writer.write('<');

+				writer.write('/');

+				writer.write(node.name());

+				writer.write('>');

+				writer.write('\n');

+			}

+

+			break;

+		}

+		

+		case node_pcdata:

+			text_output_escaped(writer, node.value(), opt1_to_type<0>());

+			break;

+

+		case node_cdata:

+			writer.write("<![CDATA[");

+			writer.write(node.value());

+			writer.write("]]>");

+			if ((flags & format_raw) == 0) writer.write('\n');

+			break;

+

+		case node_comment:

+			writer.write("<!--");

+			writer.write(node.value());

+			writer.write("-->");

+			if ((flags & format_raw) == 0) writer.write('\n');

+			break;

+

+		case node_pi:

+			writer.write("<?");

+			writer.write(node.name());

+			if (node.value()[0])

+			{

+				writer.write(' ');

+				writer.write(node.value());

+			}

+			writer.write("?>");

+			if ((flags & format_raw) == 0) writer.write('\n');

+			break;

+		

+		default:

+			;

+		}

+	}

 }

 

 namespace pugi

@@ -1386,6 +1562,26 @@ namespace pugi
 		}

 	}

 

+	xml_writer_file::xml_writer_file(void* file): file(file)

+	{

+	}

+

+	void xml_writer_file::write(const void* data, size_t size)

+	{

+		fwrite(data, size, 1, static_cast<FILE*>(file));

+	}

+

+#ifndef PUGIXML_NO_STL

+	xml_writer_stream::xml_writer_stream(std::ostream& stream): stream(&stream)

+	{

+	}

+

+	void xml_writer_stream::write(const void* data, size_t size)

+	{

+		stream->write(reinterpret_cast<const char*>(data), size);

+	}

+#endif

+

 	xml_tree_walker::xml_tree_walker(): _depth(0)

 	{

 	}

@@ -2181,100 +2377,14 @@ namespace pugi
 		}

 	}

 

-#ifndef PUGIXML_NO_STL

-	void xml_node::print(std::ostream& os, const char* indent, unsigned int flags, unsigned int depth)

+	void xml_node::print(xml_writer& writer, const char* indent, unsigned int flags, unsigned int depth)

 	{

 		if (empty()) return;

 

-		if ((flags & format_indent) != 0 && (flags & format_raw) == 0)

-			for (unsigned int i = 0; i < depth; ++i) os << indent;

+		xml_buffered_writer buffered_writer(writer);

 

-		switch (type())

-		{

-		case node_document:

-		{

-			for (xml_node n = first_child(); n; n = n.next_sibling())

-				n.print(os, indent, flags, depth);

-			break;

-		}

-			

-		case node_element:

-		{

-			os << '<' << name();

-

-			for (xml_attribute a = first_attribute(); a; a = a.next_attribute())

-			{

-				os << ' ' << a.name() << "=\"";

-

-				text_output_escaped(os, a.value(), opt1_to_type<1>());

-

-				os << "\"";

-			}

-

-			if (flags & format_raw)

-			{

-				if (!_root->first_child) // 0 children

-					os << " />";

-				else

-				{

-					os << ">";

-					for (xml_node n = first_child(); n; n = n.next_sibling())

-						n.print(os, indent, flags, depth + 1);

-					os << "</" << name() << ">";

-				}

-			}

-			else if (!_root->first_child) // 0 children

-				os << " />\n";

-			else if (_root->first_child == _root->last_child && first_child().type() == node_pcdata)

-			{

-				os << ">";

-				

-				text_output_escaped(os, first_child().value(), opt1_to_type<0>());

-					

-				os << "</" << name() << ">\n";

-			}

-			else

-			{

-				os << ">\n";

-				

-				for (xml_node n = first_child(); n; n = n.next_sibling())

-					n.print(os, indent, flags, depth + 1);

-

-				if ((flags & format_indent) != 0 && (flags & format_raw) == 0)

-					for (unsigned int i = 0; i < depth; ++i) os << indent;

-				

-				os << "</" << name() << ">\n";

-			}

-

-			break;

-		}

-		

-		case node_pcdata:

-			text_output_escaped(os, value(), opt1_to_type<0>());

-			break;

-

-		case node_cdata:

-			os << "<![CDATA[" << value() << "]]>";

-			if ((flags & format_raw) == 0) os << "\n";

-			break;

-

-		case node_comment:

-			os << "<!--" << value() << "-->";

-			if ((flags & format_raw) == 0) os << "\n";

-			break;

-

-		case node_pi:

-			os << "<?" << name();

-			if (value()[0]) os << ' ' << value();

-			os << "?>";

-			if ((flags & format_raw) == 0) os << "\n";

-			break;

-		

-		default:

-			;

-		}

+		node_output(buffered_writer, *this, indent, flags, depth);

 	}

-#endif

 

 #ifdef __BORLANDC__

 	bool operator&&(const xml_node& lhs, bool rhs)

@@ -2599,25 +2709,34 @@ namespace pugi
 		return res;

 	}

 

-#ifndef PUGIXML_NO_STL

-	bool xml_document::save_file(const char* name, const char* indent, unsigned int flags)

+	void xml_document::save(xml_writer& writer, const char* indent, unsigned int flags)

 	{

-		std::ofstream out(name, std::ios::out);

-		if (!out) return false;

+		xml_buffered_writer buffered_writer(writer);

 

 		if (flags & format_write_bom_utf8)

 		{

 			static const unsigned char utf8_bom[] = {0xEF, 0xBB, 0xBF};

-			out.write(reinterpret_cast<const char*>(utf8_bom), 3);

+			buffered_writer.write(utf8_bom, 3);

 		}

 

-		out << "<?xml version=\"1.0\"?>";

-		if (!(flags & format_raw)) out << "\n";

-		print(out, indent, flags);

+		buffered_writer.write("<?xml version=\"1.0\"?>");

+		if (!(flags & format_raw)) buffered_writer.write("\n");

 		

+		node_output(buffered_writer, *this, indent, flags, 0);

+	}

+

+	bool xml_document::save_file(const char* name, const char* indent, unsigned int flags)

+	{

+		FILE* file = fopen(name, "wb");

+		if (!file) return false;

+

+		xml_writer_file writer(file);

+		save(writer, indent, flags);

+

+		fclose(file);

+

 		return true;

 	}

-#endif

 

 	void xml_document::precompute_document_order()

 	{

diff --git a/src/pugixml.hpp b/src/pugixml.hpp
index 927d18e..bebc246 100644
--- a/src/pugixml.hpp
+++ b/src/pugixml.hpp
@@ -289,6 +289,67 @@ namespace pugi
 	#endif
 	
 	/**
+	 * Abstract writer class
+	 * \see xml_node::print
+	 */
+	class xml_writer
+	{
+	public:
+		/**
+		 * Virtual dtor
+		 */
+		virtual ~xml_writer() {}
+
+		/**
+		 * Write memory chunk into stream/file/whatever
+		 *
+		 * \param data - data pointer
+		 * \param size - data size
+		 */
+		virtual void write(const void* data, size_t size) = 0;
+	};
+
+	/** xml_writer implementation for FILE*
+	 * \see xml_writer
+	 */
+	class xml_writer_file: public xml_writer
+	{
+	public:
+		/**
+		 * Construct writer instance
+		 *
+		 * \param file - this is FILE* object, void* is used to avoid header dependencies on stdio
+		 */
+		xml_writer_file(void* file);
+
+		virtual void write(const void* data, size_t size);
+
+	private:
+		void* file;
+	};
+
+	#ifndef PUGIXML_NO_STL
+	/** xml_writer implementation for streams
+	 * \see xml_writer
+	 */
+	class xml_writer_stream: public xml_writer
+	{
+	public:
+		/**
+		 * Construct writer instance
+		 *
+		 * \param stream - output stream object
+		 */
+		xml_writer_stream(std::ostream& stream);
+
+		virtual void write(const void* data, size_t size);
+
+	private:
+		std::ostream* stream;
+	};
+	#endif
+
+	/**
 	 * A light-weight wrapper for manipulating attributes in DOM tree.
 	 * Note: xml_attribute does not allocate any memory for the attribute it wraps; it only wraps a
 	 * pointer to existing attribute.
@@ -947,7 +1008,7 @@ namespace pugi
 		 */
 		template <typename Predicate> xml_node find_node(Predicate pred) const;
 
-#ifndef PUGIXML_NO_STL
+	#ifndef PUGIXML_NO_STL
 		/**
 		 * Get the absolute node path from root as a text string.
 		 *
@@ -955,7 +1016,7 @@ namespace pugi
 		 * \return path string (e.g. '/bookstore/book/author').
 		 */
 		std::string path(char delimiter = '/') const;
-#endif
+	#endif
 
 		/**
 		 * Search for a node by path.
@@ -1014,17 +1075,15 @@ namespace pugi
 		/// \internal Document order or 0 if not set
 		unsigned int document_order() const;
 
-	#ifndef PUGIXML_NO_STL
 		/**
-		 * Print subtree to stream
+		 * Print subtree to writer
 		 *
-		 * \param os - output stream
+		 * \param writer - writer object
 		 * \param indent - indentation string
 		 * \param flags - formatting flags
 		 * \param depth - starting depth (used for indentation)
 		 */
-		void print(std::ostream& os, const char* indent = "\t", unsigned int flags = format_default, unsigned int depth = 0);
-	#endif
+		void print(xml_writer& writer, const char* indent = "\t", unsigned int flags = format_default, unsigned int depth = 0);
 	};
 
 #ifdef __BORLANDC__
@@ -1381,7 +1440,15 @@ namespace pugi
 		 */
 		bool parse(const transfer_ownership_tag&, char* xmlstr, unsigned int options = parse_default);
 		
-#ifndef PUGIXML_NO_STL
+		/**
+		 * Save XML to writer
+		 *
+		 * \param writer - writer object
+		 * \param indent - indentation string
+		 * \param flags - formatting flags
+		 */
+		void save(xml_writer& writer, const char* indent = "\t", unsigned int flags = format_default);
+
 		/**
 		 * Save XML to file
 		 *
@@ -1391,7 +1458,6 @@ namespace pugi
 		 * \return success flag
 		 */
 		bool save_file(const char* name, const char* indent = "\t", unsigned int flags = format_default);
-#endif
 
 		/**
 		 * Compute document order for the whole tree