URL parser fixes/improvements

This commit is contained in:
Thulinma 2017-10-24 13:58:56 +02:00
parent 7d420d884e
commit 047aebdb27
3 changed files with 144 additions and 106 deletions

View file

@ -8,7 +8,7 @@
/// Helper function to check if the given c-string is numeric or not
static bool is_numeric(const char * str){
while (str != 0){
while (str[0] != 0){
if (str[0] < 48 || str[0] > 57){return false;}
++str;
}
@ -17,6 +17,7 @@ static bool is_numeric(const char * str){
///Constructor that does the actual parsing
HTTP::URL::URL(const std::string & url){
IPv6Addr = false;
//first detect protocol at the start, if any
size_t proto_sep = url.find("://");
if (proto_sep != std::string::npos){
@ -24,6 +25,9 @@ HTTP::URL::URL(const std::string & url){
proto_sep += 3;
}else{
proto_sep = 0;
if (url.substr(0, 2) == "//"){
proto_sep = 2;
}
}
//proto_sep now points to the start of the host, guaranteed
//continue by finding the path, if any
@ -36,7 +40,7 @@ HTTP::URL::URL(const std::string & url){
}
size_t hmark = path.find('#');
if (hmark != std::string::npos){
frag = path.substr(hmark+1);
frag = Encodings::URL::decode(path.substr(hmark+1));
path.erase(hmark);
}
size_t qmark = path.find('?');
@ -45,15 +49,28 @@ HTTP::URL::URL(const std::string & url){
path.erase(qmark);
}
if (path.size()){
if (path[0] == '/'){
path.erase(0, 1);
}
size_t dots = path.find("/./");
while (dots != std::string::npos){
DONTEVEN_MSG("%s (/./ -> /)", path.c_str());
path.erase(dots, 2);
dots = path.find("/./");
}
dots = path.find("//");
while (dots != std::string::npos){
DONTEVEN_MSG("%s (// -> /)", path.c_str());
path.erase(dots, 1);
dots = path.find("//");
}
if (path[0] == '/'){
path.erase(0, 1);
}
dots = path.find("/../");
while (dots != std::string::npos){
size_t prevslash = path.rfind('/', dots-1);
if (prevslash == std::string::npos){
if (prevslash == std::string::npos || dots == 0){
path.erase(0, dots+4);
}else{
path.erase(prevslash+1, dots-prevslash+3);
@ -66,39 +83,59 @@ HTTP::URL::URL(const std::string & url){
if (path.substr(0, 3) == "../"){
path.erase(0, 3);
}
path = Encodings::URL::decode(path);
}
}
//host and port are now definitely between proto_sep and first_slash
//we check for [ at the start because we may have an IPv6 address as host
if (url[proto_sep] == '['){
//IPv6 address - find matching brace
size_t closing_brace = url.find(']', proto_sep);
//check if it exists at all
if (closing_brace == std::string::npos || closing_brace > first_slash){
//assume host ends at first slash if there is no closing brace before it
closing_brace = first_slash;
//user, pass, host and port are now definitely between proto_sep and first_slash
std::string uphp = url.substr(proto_sep, first_slash-proto_sep);//user+pass+host+port
//Check if we have a user/pass before the host
size_t at_sign = uphp.find('@');
if (at_sign != std::string::npos){
std::string creds = uphp.substr(0, at_sign);
uphp.erase(0, at_sign+1);
size_t colon = creds.find(':');
if (colon != std::string::npos){
user = Encodings::URL::decode(creds.substr(0, colon));
pass = Encodings::URL::decode(creds.substr(colon+1));
}else{
user = Encodings::URL::decode(creds);
}
host = url.substr(proto_sep+1, closing_brace-(proto_sep+1));
}
//we check for [ at the start because we may have an IPv6 address as host
if (uphp[0] == '['){
//IPv6 address - find matching brace
IPv6Addr = true;
size_t closing_brace = uphp.find(']');
host = uphp.substr(1, closing_brace-1);
//continue by finding port, if any
size_t colon = url.rfind(':', first_slash);
if (colon == std::string::npos || colon <= closing_brace){
//no port. Assume 80
port = "80";
size_t colon = uphp.find(':', closing_brace);
if (colon == std::string::npos){
//no port. Assume default
port = "";
}else{
//we have a port number, read it
port = url.substr(colon+1, first_slash-(colon+1));
port = uphp.substr(colon+1);
if (!is_numeric(port.c_str())){
host += ":" + port;
port = "";
}
}
}else{
//"normal" host - first find port, if any
size_t colon = url.rfind(':', first_slash);
if (colon == std::string::npos || colon < proto_sep){
size_t colon = uphp.rfind(':');
if (colon == std::string::npos){
//no port. Assume default
port = "";
host = url.substr(proto_sep, first_slash-proto_sep);
host = uphp;
}else{
//we have a port number, read it
port = url.substr(colon+1, first_slash-(colon+1));
host = url.substr(proto_sep, colon-proto_sep);
port = uphp.substr(colon+1);
host = uphp.substr(0, colon);
if (!is_numeric(port.c_str())){
IPv6Addr = true;
host += ":" + port;
port = "";
}
}
}
//if the host is numeric, assume it is a port, instead
@ -121,25 +158,35 @@ uint32_t HTTP::URL::getPort() const{
///Returns the default port for the protocol in numeric format
uint32_t HTTP::URL::getDefaultPort() const{
if (protocol == "http"){return 80;}
if (protocol == "https"){return 443;}
if (protocol == "rtmp"){return 1935;}
if (protocol == "dtsc"){return 4200;}
return 80;
if (protocol == "rtsp"){return 554;}
return 0;
}
///Returns the full URL in string format
std::string HTTP::URL::getUrl() const{
std::string ret;
if (protocol.size()){
ret = protocol + "://" + host;
ret = protocol + "://";
}else{
ret = "//" + host;
ret = "//";
}
if (user.size() || pass.size()){
ret += Encodings::URL::encode(user) + ":" + Encodings::URL::encode(pass) + "@";
}
if (IPv6Addr){
ret += "[" + host + "]";
}else{
ret += host;
}
if (port.size() && getPort() != getDefaultPort()){ret += ":" + port;}
ret += "/";
if (path.size()){ret += path;}
if (path.size()){ret += Encodings::URL::encode(path);}
if (args.size()){ret += "?" + args;}
if (frag.size()){ret += "#" + frag;}
if (frag.size()){ret += "#" + Encodings::URL::encode(frag);}
return ret;
}
@ -147,13 +194,21 @@ std::string HTTP::URL::getUrl() const{
std::string HTTP::URL::getBareUrl() const{
std::string ret;
if (protocol.size()){
ret = protocol + "://" + host;
ret = protocol + "://";
}else{
ret = "//" + host;
ret = "//";
}
if (user.size() || pass.size()){
ret += Encodings::URL::encode(user) + ":" + Encodings::URL::encode(pass) + "@";
}
if (IPv6Addr){
ret += "[" + host + "]";
}else{
ret += host;
}
if (port.size() && getPort() != getDefaultPort()){ret += ":" + port;}
ret += "/";
if (path.size()){ret += path;}
if (path.size()){ret += Encodings::URL::encode(path);}
return ret;
}