From 9f08256dff1766abe43ca2ec70db0ae9f981c286 Mon Sep 17 00:00:00 2001
From: Arvid Norberg <arvid@libtorrent.org>
Date: Tue, 7 Aug 2007 01:32:38 +0000
Subject: [PATCH] added support for attributes in the xml parser

---
 include/libtorrent/xml_parse.hpp | 79 +++++++++++++++++++++++++-------
 src/lsd.cpp                      |  1 -
 test/test_primitives.cpp         | 38 +++++++++++++++
 3 files changed, 101 insertions(+), 17 deletions(-)

diff --git a/include/libtorrent/xml_parse.hpp b/include/libtorrent/xml_parse.hpp
index c3aeddad7..ac8c350fa 100644
--- a/include/libtorrent/xml_parse.hpp
+++ b/include/libtorrent/xml_parse.hpp
@@ -35,9 +35,17 @@ POSSIBILITY OF SUCH DAMAGE.
 
 namespace libtorrent
 {
-	const int xml_start_tag = 0;
-	const int xml_end_tag = 1;
-	const int xml_string = 2;
+	enum
+	{
+		xml_start_tag = 0,
+		xml_end_tag = 1,
+		xml_empty_tag = 2,
+		xml_string = 3,
+		xml_attribute = 4
+	};
+
+	// callback(int type, char const* str, char const* str2)
+	// str2 is only used for attributes. str is name and str2 is value
 
 	template <class CallbackType>	
 	void xml_parse(char* p, char* end, CallbackType callback)
@@ -45,6 +53,8 @@ namespace libtorrent
 		for(;p != end; ++p)
 		{
 			char const* start = p;
+			char const* val_start = 0;
+			int token;
 			// look for tag start
 			for(; *p != '<' && p != end; ++p);
 
@@ -55,39 +65,76 @@ namespace libtorrent
 					assert(*p == '<');
 					*p = 0;
 				}
-				callback(xml_string, start);
+				token = xml_string;
+				callback(token, start, val_start);
 				if (p != end) *p = '<';
 			}
-		
+
 			if (p == end) break;
 		
 			// skip '<'
 			++p;	
 
-			// parse the name of the tag. Ignore attributes
-			for (start = p; p != end && *p != '>'; ++p)
-			{
-				// terminate the string at the first space
-				// to ignore tag attributes
-				if (*p == ' ') *p = 0;
-			}
+			// parse the name of the tag.
+			for (start = p; p != end && *p != '>' && *p != ' '; ++p);
+
+			char* tag_name_end = p;
+
+			// skip the attributes for now
+			for (; p != end && *p != '>'; ++p);
 
 			// parse error
 			if (p == end) break;
 			
 			assert(*p == '>');
-			*p = 0;
+			// save the character that terminated the tag name
+			// it could be both '>' and ' '.
+			char save = *tag_name_end;
+			*tag_name_end = 0;
 
+			char* tag_end = p;
 			if (*start == '/')
 			{
 				++start;
-				callback(xml_end_tag, start);
+				token = xml_end_tag;
+				callback(token, start, val_start);
+			}
+			else if (*(p-1) == '/')
+			{
+				*(p-1) = 0;
+				token = xml_empty_tag;
+				callback(token, start, val_start);
+				*(p-1) = '/';
+				tag_end = p - 1;
 			}
 			else
 			{
-				callback(xml_start_tag, start);
+				token = xml_start_tag;
+				callback(token, start, val_start);
+			}
+
+			*tag_name_end = save;
+
+			// parse attributes
+			start = tag_name_end;
+			for (char* i = tag_name_end; i < tag_end; ++i)
+			{
+				if (*i != '=') continue;
+				assert(*start == ' ');
+				++start;
+				val_start = i;
+				for (; i != tag_end && *i != ' '; ++i);
+				save = *i;
+				*i = 0;
+				const_cast<char&>(*val_start) = 0;
+				++val_start;
+				token = xml_attribute;
+				callback(token, start, val_start);
+				--val_start;
+				const_cast<char&>(*val_start) = '=';
+				*i = save;
+				start = i;
 			}
-			*p = '>';
 		}
 	
 	}
diff --git a/src/lsd.cpp b/src/lsd.cpp
index b73e407bc..76f25548d 100644
--- a/src/lsd.cpp
+++ b/src/lsd.cpp
@@ -35,7 +35,6 @@ POSSIBILITY OF SUCH DAMAGE.
 #include "libtorrent/lsd.hpp"
 #include "libtorrent/io.hpp"
 #include "libtorrent/http_tracker_connection.hpp"
-#include "libtorrent/xml_parse.hpp"
 #include <boost/bind.hpp>
 #include <boost/ref.hpp>
 #include <asio/ip/host_name.hpp>
diff --git a/test/test_primitives.cpp b/test/test_primitives.cpp
index 4c0e410be..b75e5125d 100644
--- a/test/test_primitives.cpp
+++ b/test/test_primitives.cpp
@@ -1,13 +1,16 @@
 #include "libtorrent/tracker_manager.hpp"
 #include "libtorrent/http_tracker_connection.hpp"
 #include "libtorrent/buffer.hpp"
+#include "libtorrent/xml_parse.hpp"
 #include <boost/tuple/tuple.hpp>
 #include <boost/tuple/tuple_comparison.hpp>
+#include <boost/bind.hpp>
 
 #include "test.hpp"
 
 using namespace libtorrent;
 using namespace boost::tuples;
+using boost::bind;
 
 tuple<int, int> feed_bytes(http_parser& parser, char const* str)
 {
@@ -24,6 +27,24 @@ tuple<int, int> feed_bytes(http_parser& parser, char const* str)
 	return ret;
 }
 
+void parser_callback(std::string& out, int token, char const* s, char const* val)
+{
+	switch (token)
+	{
+		case xml_start_tag: out += "B"; break;
+		case xml_end_tag: out += "F"; break;
+		case xml_empty_tag: out += "E"; break;
+		case xml_string: out += "S"; break;
+		case xml_attribute: out += "A"; break;
+	}
+	out += s;
+	if (token == xml_attribute)
+	{
+		out += "V";
+		out += val;
+	}
+}
+
 int test_main()
 {
 	using namespace libtorrent;
@@ -96,6 +117,23 @@ int test_main()
 	TEST_CHECK(parser.header<std::string>("ext") == "");
 	TEST_CHECK(parser.header<std::string>("date") == "Fri, 02 Jan 1970 08:10:38 GMT");
 
+	// test xml parser
+
+	char xml1[] = "<a>foo<b/>bar</a>";
+	std::string out1;
+
+	xml_parse(xml1, xml1 + sizeof(xml1) - 1, bind(&parser_callback
+		, boost::ref(out1), _1, _2, _3));
+	std::cerr << out1 << std::endl;
+	TEST_CHECK(out1 == "BaSfooEbSbarFa");
+
+	char xml2[] = "<c x=1 y=3/><d foo=bar></d boo=foo>";
+	std::string out2;
+
+	xml_parse(xml2, xml2 + sizeof(xml2) - 1, bind(&parser_callback
+		, boost::ref(out2), _1, _2, _3));
+	std::cerr << out2 << std::endl;
+	TEST_CHECK(out2 == "EcAxV1AyV3BdAfooVbarFdAbooVfoo");
 	return 0;
 }