给aria2打了个补丁，支持国内多数下载网站防盗链链接下载

这个补丁的作用，修改重定向时对引用及cookie的处理，使其能正确下载国内多数下载网站的防盗链链接。

补丁修改的机制说明：
在30x重定向时，RFC的HTTP标准规定不能发送Cookie, 不能使用引用。但国内多数下载站点这两者都同时使用，而且设置Cookie的域名比较复杂，这些都被作为防盗链的手段广泛使用，使其只能在浏览器中下载，遵守标准的普通下载工具却无法正常下载这些链接。
针对这种情况，在aria2中修正机制，在记录cookie的时候对域名进行处理，只保留有二级域名, 在查找cookie的时候也只是使用二级域名进行查询。
在做redirect的时候，仍旧按照服务器的响应记录cookie, 并在redirect请求中使用这些cookie。
在做redirect的时候，使用本次地址作为引用发送给服务器。
用户在--header参数中传递的Cookie会与本次服务器设置的Cookie合并，一起用于下一次转向链接的请求。

测试网站列表：
非凡软件下载
sina下载 down.tech.sina.com.cn
www.greendown.cn
...
如果您测试了其他站点，请email我。

补丁如下：
aria2-1.9.2_redirect_no_cookie_no_refer.patch
[code type="diff"]
diff --git a/src/HttpRequest.cc b/src/HttpRequest.cc
index c6c0afb..dfc981f 100644
--- a/src/HttpRequest.cc
+++ b/src/HttpRequest.cc
@@ -218,9 +218,10 @@ std::string HttpRequest::createRequest()
builtinHds.push_back(std::make_pair("Referer:", getPreviousURI()));
}
if(!_cookieStorage.isNull()) {
+ std::string baseDomain = util::getBaseDomainName(getHost());
std::string cookiesValue;
std::vector cookies =
- _cookieStorage->criteriaFind(getHost(),
+ _cookieStorage->criteriaFind(baseDomain, // getHost(),
getDir(),
Time().getTime(),
getProtocol() == Request::PROTO_HTTPS ?
@@ -235,15 +236,25 @@ std::string HttpRequest::createRequest()
}
for(std::vector >::const_iterator i =
builtinHds.begin(), eoi = builtinHds.end(); i != eoi; ++i) {
- std::vector::const_iterator j = _headers.begin();
- std::vector::const_iterator jend = _headers.end();
+ // std::vector::const_iterator j = _headers.begin();
+ // std::vector::const_iterator jend = _headers.end();
+ std::vector::iterator j = _headers.begin();
+ std::vector::iterator jend = _headers.end();
for(; j != jend; ++j) {
if(util::startsWith(*j, (*i).first)) {
break;
}
}
+ // hacked by liuguangzhao@users.sf.net
+ // if user header is not cookie, override it, or leave it there no touched.
+ // for this case, the best methord is combine to cookies part to one Cookie: header line
if(j == jend) {
strappend(requestLine, (*i).first, " ", (*i).second, A2STR::CRLF);
+ } else {
+ // combine the new cookie to header line
+ if (util::startsWith((*i).first, "Cookie") && !util::endsWith(*j, (*i).second)) {
+ (*j) += ";" + (*i).second;
+ }
}
}
// append additional headers given by user.
diff --git a/src/HttpResponse.cc b/src/HttpResponse.cc
index 6957fe9..79de4bd 100644
--- a/src/HttpResponse.cc
+++ b/src/HttpResponse.cc
@@ -116,11 +116,13 @@ std::string HttpResponse::determinFilename() const

void HttpResponse::retrieveCookie()
{
+ // hacked by liuguangzhao@users.sf.net
+ std::string baseDomain = util::getBaseDomainName(httpRequest->getHost());
std::vector v = httpHeader->get(HttpHeader::SET_COOKIE);
for(std::vector::const_iterator itr = v.begin(), eoi = v.end();
itr != eoi; ++itr) {
httpRequest->getCookieStorage()->parseAndStore(*itr,
- httpRequest->getHost(),
+ baseDomain, // httpRequest->getHost(),
httpRequest->getDir());
}
}
diff --git a/src/Request.cc b/src/Request.cc
index 9fbba5a..4204c0a 100644
--- a/src/Request.cc
+++ b/src/Request.cc
@@ -125,7 +125,8 @@ void Request::setReferer(const std::string& uri)
}

bool Request::redirectUri(const std::string& uri) {
- _previousUri = A2STR::NIL;
+ // _previousUri = A2STR::NIL;
+ _previousUri = _uri; // hacked by liuguangzhao@users.sf.net
_supportsPersistentConnection = true;
++_redirectCount;
std::string redirectedUri;
diff --git a/src/util.cc b/src/util.cc
index 58a37d4..bb10bdc 100644
--- a/src/util.cc
+++ b/src/util.cc
@@ -1305,6 +1305,48 @@ void removeMetalinkContentTypes(const SharedHandle& group)
}
}

+// hacked by liuguangzhao@users.sf.net
+/*
+ only leave the last 2 or three domain parts
+ */
+ std::string getBaseDomainName(const std::string &s)
+ {
+ std::string baseDomain;
+ std::string ts = s;
+ if (util::isNumericHost(s)) {
+ return s;
+ }
+ size_t colonPos = ts.find(':');
+ if (colonPos != std::string::npos) {
+ ts.resize(colonPos);
+ }
+
+ std::vector domParts;
+ std::string delims(".");
+ util::split(ts, std::back_inserter(domParts), delims);
+
+ int minDomParts = 2;
+ int domLevel = domParts.size();
+ if (domParts.at(domLevel - 1) == "cn"
+ || domParts.at(domLevel - 1) == "cc"
+ || domParts.at(domLevel - 1) == "us") {
+ if (domParts.at(domLevel - 2) == "org"
+ || domParts.at(domLevel - 2) == "com"
+ || domParts.at(domLevel - 2) == "net"
+ || domParts.at(domLevel - 2) == "info") {
+ minDomParts = 3;
+ }
+ }
+
+ baseDomain = domParts.at(domLevel - 2) + "." + domParts.at(domLevel - 1);
+ if (minDomParts == 3) {
+ baseDomain = domParts.at(domLevel - 3) + "." + baseDomain;
+ }
+
+ return baseDomain;
+ }
+
+
} // namespace util

} // namespace aria2
diff --git a/src/util.h b/src/util.h
index 9e1c2a5..01b6a1c 100644
--- a/src/util.h
+++ b/src/util.h
@@ -395,6 +395,9 @@ bool inSameCidrBlock(const std::string& ip1, const std::string& ip2, int bits);

void removeMetalinkContentTypes(const SharedHandle& group);

+ // hacked by liuguangzhao@users.sf.net
+ std::string getBaseDomainName(const std::string &s);
+
} // namespace util

} // namespace aria2

[/code]