From a16d415ee4a4b8a48f01c80919a11a026911c957 Mon Sep 17 00:00:00 2001
From: special <special@e03df62e-2008-0410-955e-edbf42e46eb7>
Date: Sat, 20 Jan 2007 23:34:42 +0000
Subject: Fixed (rewrote) m_http_client's URL parsing, it is now more flexible
 and.. actually works.

git-svn-id: http://svn.inspircd.org/repository/trunk/inspircd@6416 e03df62e-2008-0410-955e-edbf42e46eb7
---
 include/hashcomp.h            |   5 ++
 src/hashcomp.cpp              |   5 ++
 src/modules/m_http_client.cpp | 112 ++++++++++++++++++++++++------------------
 3 files changed, 73 insertions(+), 49 deletions(-)

diff --git a/include/hashcomp.h b/include/hashcomp.h
index 2d6be15a9..abcc1e94e 100644
--- a/include/hashcomp.h
+++ b/include/hashcomp.h
@@ -285,6 +285,11 @@ namespace irc
 		 * @return The next token is returned, or an empty string if none remain
 		 */
 		virtual const std::string GetToken();
+		
+		/** Returns true if the end of the stream has been reached
+		 * @return True if the end of the stream has been reached, otherwise false
+		 */
+		virtual bool StreamEnd();
 	};
 
 	/** A derived form of sepstream, which seperates on commas
diff --git a/src/hashcomp.cpp b/src/hashcomp.cpp
index 5c9cac7bb..55b3f9b58 100644
--- a/src/hashcomp.cpp
+++ b/src/hashcomp.cpp
@@ -284,6 +284,11 @@ const std::string irc::sepstream::GetToken()
 	return "";
 }
 
+bool irc::sepstream::StreamEnd()
+{
+	return ((n + 1) == tokens.end());
+}
+
 irc::sepstream::~sepstream()
 {
 }
diff --git a/src/modules/m_http_client.cpp b/src/modules/m_http_client.cpp
index 7e929a8c5..706008104 100644
--- a/src/modules/m_http_client.cpp
+++ b/src/modules/m_http_client.cpp
@@ -175,65 +175,79 @@ bool HTTPSocket::ParseURL(const std::string &iurl)
 {
 	url.url = iurl;
 	url.port = 80;
+	url.protocol = "http";
 
-	Instance->Log(DEBUG,"Parse: "+iurl);
+	irc::sepstream tokenizer(iurl, '/');
 	
-	// Tokenize by slashes (protocol:, blank, domain, request..)
-	int pos = 0, pstart = 0, pend = 0;
-	
-	for (;;)
+	for (int p = 0;; p++)
 	{
-		pend = url.url.find('/', pstart);
-		string part = url.url.substr(pstart, pend);
-
-		switch (pos)
+		std::string part = tokenizer.GetToken();
+		if (part.empty() && tokenizer.StreamEnd())
+			break;
+		
+		if ((p == 0) && (part[part.length() - 1] == ':'))
 		{
-			case 0:
-				Instance->Log(DEBUG,"PART 0: "+part);
-				// Protocol
-				if (part[part.length()-1] != ':')
-					return false;
-				url.protocol = part.substr(0, part.length() - 1);
-				break;
-			case 1:
-				// Empty, skip
-				break;
-			case 2:
-				Instance->Log(DEBUG,"PART 2: "+part);
-				// User and password (user:pass@)
-				string::size_type aend = part.find('@', 0);
-				if (aend != string::npos)
+			// Protocol ('http:')
+			url.protocol = part.substr(0, part.length() - 1);
+		}
+		else if ((p == 1) && (part.empty()))
+		{
+			continue;
+		}
+		else if (url.domain.empty())
+		{
+			// Domain part: [user[:pass]@]domain[:port]
+			std::string::size_type usrpos = part.find('@');
+			if (usrpos != std::string::npos)
+			{
+				// Have a user (and possibly password) part
+				std::string::size_type ppos = part.find(':');
+				if ((ppos != std::string::npos) && (ppos < usrpos))
 				{
-					// Technically, it is valid to not have a password (username@domain)
-					string::size_type usrend = part.find(':', 0);
-					
-					if ((usrend != string::npos) && (usrend < aend))
-						url.password = part.substr(usrend + 1, aend);
-					else
-						usrend = aend;
-					
-					url.username = part.substr(0, usrend);
+					// Have password too
+					url.password = part.substr(ppos + 1, usrpos - ppos - 1);
+					url.username = part.substr(0, ppos);
 				}
 				else
-					aend = 0;
-				
-				// Port (:port)
-				string::size_type dend = part.find(':', aend);
-				if (dend != string::npos)
-					url.port = atoi(part.substr(dend + 1).c_str());
-
-				// Domain
-				url.domain = part.substr(aend + 1, dend);
+				{
+					url.username = part.substr(0, usrpos);
+				}
 				
-				// The rest of the string is the request
-				url.request = url.url.substr(pend);
-				break;
+				part = part.substr(usrpos + 1);
+			}
+			
+			std::string::size_type popos = part.rfind(':');
+			if (popos != std::string::npos)
+			{
+				url.port = atoi(part.substr(popos + 1).c_str());
+				url.domain = part.substr(0, popos);
+			}
+			else
+			{
+				url.domain = part;
+			}
 		}
-		
-		if (pos++ == 2)
-			break;
+		else
+		{
+			// Request (part of it)..
+			url.request.append("/");
+			url.request.append(part);
+		}
+	}
+	
+	if (url.request.empty())
+		url.request = "/";
 
-		pstart = pend + 1;
+	if ((url.domain.empty()) || (!url.port) || (url.protocol.empty()))
+	{
+		Instance->Log(DEFAULT, "Invalid URL (%s): Missing required value", iurl.c_str());
+		return false;
+	}
+	
+	if (url.protocol != "http")
+	{
+		Instance->Log(DEFAULT, "Invalid URL (%s): Unsupported protocol '%s'", iurl.c_str(), url.protocol.c_str());
+		return false;
 	}
 	
 	return true;
-- 
cgit v1.2.3