/// <summary>
/// Extract the host and port to use from either the HTTP request
/// line, or the HTTP headers; update the request line to remove
/// the hostname and port
/// </summary>
/// <param name="hrl">
/// The HTTP request line; the URI will be updated to remove the
/// host name and port number</param>
/// <param name="hh_rq">The HTTP request headers</param>
/// <param name="port">
/// When this method returns, contains the request port
/// </param>
/// <remarks>
/// May modify the URI of <c>hrl</c>
/// </remarks>
protected string ParseDestinationHostAndPort(
HttpRequestLine hrl, HttpHeaders hh_rq, out int port)
{
string host = null;
port = DestinationPort;
if (DestinationPort == 0)
{
port = 80;
}
//bool bIsConnect = hrl.Method.Equals("CONNECT");
//port = bIsConnect ? 443 : 80;
bool bIsHTTP1_0 = hrl.ProtocolVersion.Equals("1.0");
if (hrl.URI.Equals("*"))
{
System.Diagnostics.Debug.Assert(!bIsHTTP1_0);
goto hostname_from_header;
}
// At this point, hrl.URI follows one of these forms:
// - scheme:(//authority)/abs_path
// - authority
// - /abs_path
int prefix = 0; // current parse position
if (hrl.URI.Contains("://"))
{
if (hrl.URI.StartsWith("http://"))
{
prefix = 7; // length of "http://"
}
else
if (hrl.URI.StartsWith("https://"))
{
prefix = 8; // length of "https://"
port = 443;
}
else
{
throw new HttpProtocolBroken(
"Expected scheme missing or unsupported");
}
}
// Starting from offset prefix, we now have either:
// 1) authority (only for CONNECT)
// 2) authority/abs_path
// 3) /abs_path
int slash = hrl.URI.IndexOf('/', prefix);
string authority = null;
if (slash == -1)
{
// case 1
authority = hrl.URI;
//System.Diagnostics.Debug.Assert(bIsConnect);
}
else
if (slash > 0) // Strict inequality
// case 2
{
authority = hrl.URI.Substring(prefix, slash - prefix);
}
if (authority != null)
{
// authority is either:
// a) hostname
// b) hostname:
// c) hostname:port
int c = authority.IndexOf(':');
if (c < 0)
{
// case a)
host = authority;
}
else
if (c == authority.Length - 1)
{
// case b)
host = authority.TrimEnd('/');
}
else
{
// case c)
host = authority.Substring(0, c);
port = int.Parse(authority.Substring(c + 1));
}
prefix += authority.Length;
}
if (host != null)
{
#if false
// XXX Not sure whether this can happen (without doing ad
// replacement) or if we want to prevent it
if (hh_rq.Host != null)
{
// Does hh_rq.Host and host match? (disregarding
// the potential ":port" prefix of hh_rq.Host)
int c2 = hh_rq.Host.IndexOf(':');
string rq_host = c2 < 0 ? hh_rq.Host :
hh_rq.Host.Substring(0, c2);
if (!rq_host.Equals(host))
{
// Host discrepancy: fix the 'Host' header
hh_rq.Host = host;
}
}
#endif
// Remove the host from the request URI, unless the "server"
// is actually a proxy, in which case the URI should remain
// unchanged. (RFC 2616, section 5.1.2)
//if (!UsingHttpProxy)
//{
// hrl.URI = hrl.URI.Substring(prefix);
// log.Debug("Rewriting request line as: " +
// hrl.RequestLine);
//}
return(host);
}
hostname_from_header:
host = hh_rq.Host;
if (host == null)
{
throw new HttpProtocolBroken("No host specified");
}
int cp = host.IndexOf(':');
if (cp < 0) /* nothing */ } {
/// <summary>
/// Extract the host and port to use from either the HTTP request
/// line, or the HTTP headers; update the request line to remove
/// the hostname and port
/// </summary>
/// <param name="hrl">
/// The HTTP request line; the URI will be updated to remove the
/// host name and port number</param>
/// <param name="hh_rq">The HTTP request headers</param>
/// <param name="port">
/// When this method returns, contains the request port
/// </param>
/// <remarks>
/// May modify the URI of <c>hrl</c>
/// </remarks>
protected string ParseDestinationHostAndPort(
HttpRequestLine hrl, HttpHeaders hh_rq, out int port)
{
string host = null;
bool bIsConnect = hrl.Method.Equals("CONNECT");
port = bIsConnect ? 443 : 80;
bool bIsHTTP1_0 = hrl.ProtocolVersion.Equals("1.0");
if (hrl.URI.Equals("*"))
{
System.Diagnostics.Debug.Assert(!bIsHTTP1_0);
goto hostname_from_header;
}
// At this point, hrl.URI follows one of these forms:
// - scheme:(//authority)/abs_path
// - authority
// - /abs_path
int prefix = 0; // current parse position
if (hrl.URI.Contains("://"))
{
if (hrl.URI.StartsWith("http://"))
{
prefix = 7; // length of "http://"
}
else
if (hrl.URI.StartsWith("https://"))
{
prefix = 8; // length of "https://"
port = 443;
}
else
{
throw new HttpProtocolBroken(
"Expected scheme missing or unsupported");
}
}
// Starting from offset prefix, we now have either:
// 1) authority (only for CONNECT)
// 2) authority/abs_path
// 3) /abs_path
int slash = hrl.URI.IndexOf('/', prefix);
string authority = null;
if (slash == -1)
{
// case 1
authority = hrl.URI;
System.Diagnostics.Debug.Assert(bIsConnect);
}
else
if (slash > 0) // Strict inequality
// case 2
{
authority = hrl.URI.Substring(prefix, slash - prefix);
}
if (authority != null)
{
// authority is either:
// a) hostname
// b) hostname:
// c) hostname:port
int c = authority.IndexOf(':');
if (c < 0)
{
// case a)
host = authority;
}
else
if (c == authority.Length - 1)
{
// case b)
host = authority.TrimEnd('/');
}
else
{
// case c)
host = authority.Substring(0, c);
port = int.Parse(authority.Substring(c + 1));
}
prefix += authority.Length;
}
if (host != null)
{
// Remove the host from the request URI, unless the "server"
// is actually a proxy, in which case the URI should remain
// unchanged. (RFC 2616, section 5.1.2)
if (RelayHttpProxyHost == null)
{
hrl.URI = hrl.URI.Substring(prefix);
log.Debug("Rewriting request line as: " +
hrl.RequestLine);
}
return(host);
}
hostname_from_header:
host = hh_rq.Host;
if (host == null)
{
throw new HttpProtocolBroken("No host specified");
}
int cp = host.IndexOf(':');
if (cp < 0) /* nothing */ } {
/// <summary>
/// Extract the host and port to use from either the HTTP request
/// line, or the HTTP headers; update the request line to remove
/// the hostname and port
/// </summary>
/// <param name="hrl">
/// The HTTP request line; the URI will be updated to remove the
/// host name and port number</param>
/// <param name="hh_rq">The HTTP request headers</param>
/// <param name="port">
/// When this method returns, contains the request port
/// </param>
/// <remarks>
/// May modify the URI of <c>hrl</c>
/// </remarks>
protected string ParseDestinationHostAndPort(
HttpRequestLine hrl, HttpHeaders hh_rq, out int port)
{
string host = null;
bool bIsConnect = hrl.Method.Equals("CONNECT");
port = bIsConnect ? 443 : 80;
bool bIsHTTP1_0 = hrl.ProtocolVersion.Equals("1.0");
if (hrl.URI.Equals("*"))
{
System.Diagnostics.Debug.Assert(!bIsHTTP1_0);
goto hostname_from_header;
}
// At this point, hrl.URI follows one of these forms:
// - scheme:(//authority)/abs_path
// - authority
// - /abs_path
int prefix = 0; // current parse position
if (hrl.URI.Contains("://"))
{
if (hrl.URI.StartsWith("http://"))
prefix = 7; // length of "http://"
else
if (hrl.URI.StartsWith("https://"))
{
prefix = 8; // length of "https://"
port = 443;
}
else
throw new HttpProtocolBroken(
"Expected scheme missing or unsupported");
}
// Starting from offset prefix, we now have either:
// 1) authority (only for CONNECT)
// 2) authority/abs_path
// 3) /abs_path
int slash = hrl.URI.IndexOf('/', prefix);
string authority = null;
if (slash == -1)
{
// case 1
authority = hrl.URI;
System.Diagnostics.Debug.Assert(bIsConnect);
}
else
if (slash > 0) // Strict inequality
// case 2
authority = hrl.URI.Substring(prefix, slash - prefix);
if (authority != null)
{
// authority is either:
// a) hostname
// b) hostname:
// c) hostname:port
int c = authority.IndexOf(':');
if (c < 0)
// case a)
host = authority;
else
if (c == authority.Length - 1)
// case b)
host = authority.TrimEnd('/');
else
{
// case c)
host = authority.Substring(0, c);
port = int.Parse(authority.Substring(c + 1));
}
prefix += authority.Length;
}
if (host != null)
{
#if false
// XXX Not sure whether this can happen (without doing ad
// replacement) or if we want to prevent it
if (hh_rq.Host != null)
{
// Does hh_rq.Host and host match? (disregarding
// the potential ":port" prefix of hh_rq.Host)
int c2 = hh_rq.Host.IndexOf(':');
string rq_host = c2 < 0 ? hh_rq.Host :
hh_rq.Host.Substring(0, c2);
if (!rq_host.Equals(host))
// Host discrepancy: fix the 'Host' header
hh_rq.Host = host;
}
#endif
// Remove the host from the request URI, unless the "server"
// is actually a proxy, in which case the URI should remain
// unchanged. (RFC 2616, section 5.1.2)
if (RelayHttpProxyHost == null)
{
hrl.URI = hrl.URI.Substring(prefix);
log.Debug("Rewriting request line as: " +
hrl.RequestLine);
}
return host;
}
hostname_from_header:
host = hh_rq.Host;
if (host == null)
throw new HttpProtocolBroken("No host specified");
int cp = host.IndexOf(':');
if (cp < 0) { /* nothing */ }
else
if (cp == host.Length - 1)
host = host.TrimEnd('/');
else
{
host = host.Substring(0, cp);
port = int.Parse(host.Substring(cp + 1));
}
return host;
}
/// <summary>
/// Pipeline step: read the HTTP request from the client, schedule
/// the next step to be <c>SendRequest</c>, and call
/// <c>OnReceiveRequest</c>
/// </summary>
protected virtual void ReadRequest()
{
try
{
RequestLine = new HttpRequestLine(SocketBP);
}
catch (TrotiNet.IoBroken)
{
// The request line is the first line of a HTTP request.
// If none comes in a timely fashion, then we eventually
// get a IoBroken exception, which is common enough
// not to be rethrown.
AbortRequest();
return;
}
catch (SocketException)
{
// Ditto
AbortRequest();
return;
}
RequestHeaders = new HttpHeaders(SocketBP);
if (RequestLine.Method.Equals("CONNECT"))
{
log.Debug("Method CONNECT not implemented");
SocketBP.Send501();
AbortRequest();
return;
}
log.Info("Got request " + RequestLine.RequestLine);
// We call OnReceiveRequest now because Connect() will
// modify the request URI.
State.NextStep = SendRequest;
OnReceiveRequest();
// Now we parse the request to:
// 1) find out where we should connect
// 2) find out whether there is a message body in the request
// 3) find out whether the BP connection should be kept-alive
if (State.NextStep != null)
{
// Step 1)
if (RelayHttpProxyHost == null)
{
int NewDestinationPort;
string NewDestinationHost = ParseDestinationHostAndPort(
RequestLine, RequestHeaders, out NewDestinationPort);
Connect(NewDestinationHost, NewDestinationPort);
}
else
Connect(RelayHttpProxyHost, RelayHttpProxyPort);
// Step 2)
// Find out whether the request has a message body
// (RFC 2616, section 4.3); if it has, get the message length
State.bRequestHasMessage = false;
State.RequestMessageLength = 0;
State.bRequestMessageChunked = false;
if (RequestHeaders.TransferEncoding != null)
{
State.bRequestHasMessage = true;
State.bRequestMessageChunked = Array.IndexOf<string>(
RequestHeaders.TransferEncoding, "chunked") >= 0;
System.Diagnostics.Debug.Assert(
State.bRequestMessageChunked);
}
else
if (RequestHeaders.ContentLength != null)
{
State.RequestMessageLength =
(uint)RequestHeaders.ContentLength;
// Note: HTTP 1.0 wants "Content-Length: 0" when there
// is no entity body. (RFC 1945, section 7.2)
if (State.RequestMessageLength > 0)
State.bRequestHasMessage = true;
}
}
// Step 3)
State.bUseDefaultPersistBP = true;
if (RequestHeaders.ProxyConnection != null)
{
// Note: This is not part of the HTTP 1.1 standard. See
// http://homepage.ntlworld.com./jonathan.deboynepollard/FGA/web-proxy-connection-header.html
foreach (string i in RequestHeaders.ProxyConnection)
{
if (i.Equals("close"))
{
State.bPersistConnectionBP = false;
State.bUseDefaultPersistBP = false;
break;
}
if (i.Equals("keep-alive"))
{
State.bPersistConnectionBP = true;
State.bUseDefaultPersistBP = false;
break;
}
}
if (RelayHttpProxyHost == null)
RequestHeaders.ProxyConnection = null;
}
// Note: we do not remove fields mentioned in the
// 'Connection' header (the specs say we should).
}
/// <summary>
/// Extract the host and port to use from either the HTTP request
/// line, or the HTTP headers; update the request line to remove
/// the hostname and port
/// </summary>
/// <param name="hrl">
/// The HTTP request line; the URI will be updated to remove the
/// host name and port number</param>
/// <param name="hh_rq">The HTTP request headers</param>
/// <param name="port">
/// When this method returns, contains the request port
/// </param>
/// <remarks>
/// May modify the URI of <c>hrl</c>
/// </remarks>
protected string ParseDestinationHostAndPort(
HttpRequestLine hrl, HttpHeaders hh_rq, out int port)
{
string host = null;
bool bIsConnect = hrl.Method.Equals("CONNECT");
port = bIsConnect ? 443 : 80;
bool bIsHTTP1_0 = hrl.ProtocolVersion.Equals("1.0");
if (hrl.URI.Equals("*"))
{
System.Diagnostics.Debug.Assert(!bIsHTTP1_0);
goto hostname_from_header;
}
// At this point, hrl.URI follows one of these forms:
// - scheme:(//authority)/abs_path
// - authority
// - /abs_path
int prefix = 0; // current parse position
if (hrl.URI.Contains("://"))
{
//当URI包含://，为普通网址时，需要规范化
//特别是如果是根目录，尾部不附加/，又是ip:port形式，则下面在解析port时会发生异常
//这个问题会导致部分网银客户端无法工作
//这里不能使用这种方法进行规范化，否则会导致在搜索页面时出现搜索结果都是？的错误
//因为在查询网址中搜索字符是url encode以后，通过这种方式进行规范化就会变成中文
//导致搜索结果出错
//log.Info("URI before normalization: " + hrl.URI);
//hrl.URI = new Uri(hrl.URI).ToString();
//log.Info("URI after normalization: " + hrl.URI);
if (hrl.URI.StartsWith("http://"))
prefix = 7; // length of "http://"
else
if (hrl.URI.StartsWith("https://"))
{
log.Error("Unexpected https request!");
prefix = 8; // length of "https://"
port = 443;
}
else
throw new HttpProtocolBroken(
"Expected scheme missing or unsupported");
}
// Starting from offset prefix, we now have either:
// 1) authority (only for CONNECT)
// 2) authority/abs_path
// 3) /abs_path
int slash = hrl.URI.IndexOf('/', prefix);
//这里如果找不到/，但是又确定包含http或是https，则表示为根目录，忘记添加/了
if (slash == -1 && prefix!= -1)
{
hrl.URI = hrl.URI + "/";
slash = hrl.URI.IndexOf('/', prefix);
log.Info("URI is not normal: " + hrl.URI);
}
string authority = null;
if (slash == -1)
{
// case 1
authority = hrl.URI;
System.Diagnostics.Debug.Assert(bIsConnect);
log.Error("URI only authority but not connect request!");
}
else
if (slash > 0) // Strict inequality
// case 2
authority = hrl.URI.Substring(prefix, slash - prefix);
if (authority != null)
{
// authority is either:
// a) hostname
// b) hostname:
// c) hostname:port
int c = authority.IndexOf(':');
if (c < 0)
// case a)
host = authority;
else
if (c == authority.Length - 1)
// case b)
host = authority.TrimEnd('/');
else
{
// case c)
host = authority.Substring(0, c);
port = int.Parse(authority.Substring(c + 1));
}
prefix += authority.Length;
}
if (host != null)
{
// Remove the host from the request URI, unless the "server"
// is actually a proxy, in which case the URI should remain
// unchanged. (RFC 2616, section 5.1.2)
if (RelayHttpProxyHost == null)
{
hrl.URI = hrl.URI.Substring(prefix);
log.Debug("Rewriting request line as: " +
hrl.RequestLine);
}
return host;
}
hostname_from_header:
host = hh_rq.Host;
if (host == null)
throw new HttpProtocolBroken("No host specified");
int cp = host.IndexOf(':');
if (cp < 0) { /* nothing */ }
else
if (cp == host.Length - 1)
host = host.TrimEnd('/');
else
{
host = host.Substring(0, cp);
port = int.Parse(host.Substring(cp + 1));
}
return host;
}
