980 lines
30 KiB
C++
980 lines
30 KiB
C++
//
|
|
// Things to worry about:
|
|
// Expect: 100-Continue
|
|
|
|
|
|
|
|
#include "http.hpp"
|
|
#include "wrap-sstream.hpp"
|
|
#include "wrap-string.hpp"
|
|
#include "util.hpp"
|
|
#include "luastack.hpp"
|
|
|
|
#include <cstdint>
|
|
|
|
using string_view = std::string_view;
|
|
|
|
bool words_separated_by_dashes(string_view v) {
|
|
while (true) {
|
|
if (!sv::ascii_isalpha(sv::zfront(v))) return false;
|
|
v.remove_prefix(1);
|
|
while (sv::ascii_isalnum(sv::zfront(v))) v.remove_prefix(1);
|
|
if (v.empty()) return true;
|
|
if (sv::zfront(v) != '-') return false;
|
|
v.remove_prefix(1);
|
|
}
|
|
}
|
|
|
|
// Technically, this is a true, correct URL encode routine.
|
|
static eng::string url_encode_param(string_view value) {
|
|
eng::ostringstream result;
|
|
const char *hexdigits = "0123456789ABCDEF";
|
|
for (int i = 0; i < int(value.size()); i++) {
|
|
char c = value[i];
|
|
|
|
if (sv::ascii_isalnum(c) || c == '-' || c == '_' || c == '.' || c == '~') {
|
|
result << c;
|
|
} else if (c == ' ') {
|
|
result << '+';
|
|
} else {
|
|
result << '%' << hexdigits[c>>4] << hexdigits[c&15];
|
|
}
|
|
}
|
|
return result.str();
|
|
}
|
|
|
|
// This URL encode routine leaves slashes intact. That's not
|
|
// technically correct, but it's really what you want for paths.
|
|
static eng::string url_encode_path(string_view value) {
|
|
eng::ostringstream result;
|
|
const char *hexdigits = "0123456789ABCDEF";
|
|
for (int i = 0; i < int(value.size()); i++) {
|
|
char c = value[i];
|
|
|
|
if (sv::ascii_isalnum(c) || c == '-' || c == '_' || c == '.' || c == '~' || c == '/') {
|
|
result << c;
|
|
} else if (c == ' ') {
|
|
result << '+';
|
|
} else {
|
|
result << '%' << hexdigits[c>>4] << hexdigits[c&15];
|
|
}
|
|
}
|
|
return result.str();
|
|
}
|
|
|
|
static eng::string url_decode(string_view eurl) {
|
|
eng::ostringstream result;
|
|
int i = 0;
|
|
int len = eurl.size();
|
|
while (i < len) {
|
|
char c = eurl[i];
|
|
if (c == '+') {
|
|
result << ' ';
|
|
i += 1;
|
|
} else if ((c == '%') && (i + 2 < len)) {
|
|
std::string_view code = eurl.substr(i + 1, 2);
|
|
uint64_t value = sv::to_hex64(code);
|
|
if (value > 255) {
|
|
result << '?';
|
|
} else {
|
|
result << char(value);
|
|
}
|
|
i += 3;
|
|
} else {
|
|
result << c;
|
|
i += 1;
|
|
}
|
|
}
|
|
return result.str();
|
|
}
|
|
|
|
static void send_encoded_path(std::string_view path, const UrlParameters ¶ms, StreamBuffer *sb) {
|
|
sb->write_bytes(url_encode_path(path));
|
|
bool first_param = true;
|
|
for (const auto &pair : params) {
|
|
sb->write_char(first_param ? '?' : '&');
|
|
sb->write_bytes(url_encode_param(pair.first));
|
|
sb->write_char('=');
|
|
sb->write_bytes(url_encode_param(pair.second));
|
|
first_param = false;
|
|
}
|
|
}
|
|
|
|
static void send_host_and_port(std::string_view host, int port, StreamBuffer *sb) {
|
|
sb->write_bytes(host);
|
|
if (port != 0) {
|
|
sb->write_char(':');
|
|
sb->ostream() << port;
|
|
}
|
|
}
|
|
|
|
// In a properly-formed url, the hostname and path are url encoded.
|
|
// This parser expects an encoded URL.
|
|
|
|
struct ParsedURL {
|
|
public:
|
|
bool valid;
|
|
eng::string proto;
|
|
eng::string host;
|
|
int port;
|
|
eng::string path;
|
|
UrlParameters params;
|
|
|
|
public:
|
|
void clear() {
|
|
valid = false;
|
|
proto.clear();
|
|
host.clear();
|
|
port = 0;
|
|
path.clear();
|
|
params.clear();
|
|
}
|
|
|
|
eng::string str() {
|
|
StreamBuffer sb;
|
|
sb.write_bytes(proto);
|
|
sb.write_bytes("://");
|
|
send_host_and_port(host, port, &sb);
|
|
send_encoded_path(path, params, &sb);
|
|
return eng::string(sb.view());
|
|
}
|
|
|
|
ParsedURL(std::string_view url) {
|
|
clear();
|
|
|
|
proto = util::ascii_tolower(sv::read_to_sep(url, ':'));
|
|
if (!sv::has_prefix(url, "//")) { clear(); return; }
|
|
url.remove_prefix(2);
|
|
if (!words_separated_by_dashes(proto)) { clear(); return; }
|
|
|
|
// Extract the host and port as a single string.
|
|
string_view turl = url;
|
|
string_view hostport = sv::read_to_sep(turl, '/');
|
|
url.remove_prefix(hostport.size());
|
|
|
|
// Split the host and port from each other and parse them.
|
|
host = util::ascii_tolower(sv::read_to_sep(hostport, ':'));
|
|
if (host.empty()) { clear(); return; }
|
|
if (!hostport.empty()) {
|
|
int64_t iport = sv::to_int64(hostport);
|
|
if ((iport < 1) || (iport > 65535)) {
|
|
clear(); return;
|
|
}
|
|
port = iport;
|
|
}
|
|
|
|
// Split off the path.
|
|
path = url_decode(sv::read_to_sep(url, '?'));
|
|
if (path.empty()) {
|
|
path = "/";
|
|
}
|
|
|
|
// Process url parameters.
|
|
while (!sv::isnull(url)) {
|
|
std::string_view keyval = sv::read_to_sep(url, '&');
|
|
if (keyval.empty()) { clear(); return; }
|
|
std::string_view key = sv::read_to_sep(keyval, '=');
|
|
if (key.empty()) { clear(); return; }
|
|
if (sv::isnull(keyval)) { clear(); return; }
|
|
eng::string dkey = url_decode(key);
|
|
eng::string dval = url_decode(keyval);
|
|
params[dkey] = dval;
|
|
}
|
|
|
|
// If we made it here, we have a valid URL
|
|
valid = true;
|
|
}
|
|
};
|
|
|
|
HttpOutRequest::HttpOutRequest() {
|
|
verify_certificate_ = true;
|
|
port_ = 0;
|
|
request_id_ = 0;
|
|
place_id_ = 0;
|
|
thread_id_ = 0;
|
|
}
|
|
|
|
void HttpOutRequest::fail(string_view s) {
|
|
if (error_.empty()) {
|
|
error_ = s;
|
|
}
|
|
}
|
|
|
|
eng::string HttpOutRequest::target() const {
|
|
assert(check().empty());
|
|
eng::ostringstream oss;
|
|
oss << (verify_certificate_ ? "cert" : "nocert");
|
|
oss << ':' << host_ << ':' << port_;
|
|
return oss.str();
|
|
}
|
|
|
|
void HttpOutRequest::set_verify_certificate(bool flag) {
|
|
verify_certificate_ = flag;
|
|
}
|
|
|
|
void HttpOutRequest::set_method(const eng::string &s) {
|
|
eng::string method = util::ascii_toupper(s);
|
|
if ((method != "GET") && (method != "HEAD")) {
|
|
fail(util::ss("HTTP method not implemented: ", method, ".",
|
|
"Currently, only HEAD and GET are implemented."));
|
|
return;
|
|
}
|
|
if ((!method_.empty()) && (method_ != method)) {
|
|
fail(util::ss("HTTP method specified twice: ", method_, " and ", method));
|
|
return;
|
|
}
|
|
method_ = method;
|
|
}
|
|
|
|
void HttpOutRequest::set_host(const eng::string &s) {
|
|
eng::string host = util::ascii_tolower(s);
|
|
if (host.empty()) {
|
|
fail(util::ss("HTTP hostname cannot be empty string."));
|
|
return;
|
|
}
|
|
// This is not quite strict, but it's close. I believe
|
|
// the DNS lookup will fail for invalid hostnames anyway.
|
|
for (char c : host) {
|
|
if ((c != '-') && (c != '.') && (!sv::ascii_isalnum(c))) {
|
|
fail(util::ss("HTTP hostnames can only contain letters, digits, and hyphen: ", host));
|
|
return;
|
|
}
|
|
}
|
|
if (!host_.empty()) {
|
|
fail(util::ss("HTTP hostname specified twice: ", host_, " and ", host));
|
|
return;
|
|
}
|
|
host_ = host;
|
|
}
|
|
|
|
void HttpOutRequest::set_port(int port) {
|
|
if ((port < 1) || (port > 65535)) {
|
|
fail(util::ss("HTTP port must be between 1 and 65535: ", port));
|
|
return;
|
|
}
|
|
if (port_ != 0) {
|
|
fail(util::ss("HTTP port specified twice: ", port_, " and ", port));
|
|
return;
|
|
}
|
|
port_ = port;
|
|
}
|
|
|
|
void HttpOutRequest::set_path(string_view path) {
|
|
if (!sv::has_prefix(path, "/")) {
|
|
fail(util::ss("HTTP path must start with slash"));
|
|
return;
|
|
}
|
|
if (!path_.empty()) {
|
|
fail(util::ss("HTTP path specified twice: ", path_, " and ", path));
|
|
return;
|
|
}
|
|
path_ = path;
|
|
}
|
|
|
|
void HttpOutRequest::set_param(const eng::string &key, const eng::string &val) {
|
|
if (params_.find(key) != params_.end()) {
|
|
fail(util::ss("HTTP url parameter specified twice: ", key));
|
|
return;
|
|
}
|
|
if (key.empty()) {
|
|
fail(util::ss("HTTP parameter key cannot be empty"));
|
|
return;
|
|
}
|
|
params_[key] = val;
|
|
}
|
|
|
|
void HttpOutRequest::set_url(string_view url) {
|
|
ParsedURL parsed_url(url);
|
|
if (!parsed_url.valid) {
|
|
fail(util::ss("syntactically invalid URL: ", url));
|
|
return;
|
|
}
|
|
if (parsed_url.proto != "https") {
|
|
fail(util::ss("unsupported protocol: ", parsed_url.proto));
|
|
return;
|
|
}
|
|
set_host(parsed_url.host);
|
|
if (parsed_url.port) set_port(parsed_url.port);
|
|
set_path(parsed_url.path);
|
|
for (const auto &pair : parsed_url.params) {
|
|
set_param(pair.first, pair.second);
|
|
}
|
|
}
|
|
|
|
void HttpOutRequest::set_verify_certificate(LuaStack &LS, LuaSlot val) {
|
|
if (!LS.isboolean(val)) {
|
|
fail(util::ss("HTTP verify_certificate must be a boolean"));
|
|
return;
|
|
}
|
|
set_verify_certificate(LS.ckboolean(val));
|
|
}
|
|
|
|
void HttpOutRequest::set_method(LuaStack &LS, LuaSlot val) {
|
|
if (!LS.isstring(val)) {
|
|
fail(util::ss("HTTP method must be a string"));
|
|
return;
|
|
}
|
|
set_method(LS.ckstring(val));
|
|
}
|
|
|
|
void HttpOutRequest::set_host(LuaStack &LS, LuaSlot val) {
|
|
if (!LS.isstring(val)) {
|
|
fail(util::ss("HTTP host must be a string"));
|
|
return;
|
|
}
|
|
set_host(LS.ckstring(val));
|
|
}
|
|
|
|
void HttpOutRequest::set_port(LuaStack &LS, LuaSlot val) {
|
|
if (!LS.isint(val)) {
|
|
fail(util::ss("HTTP port must be an int"));
|
|
return;
|
|
}
|
|
set_port(LS.ckint(val));
|
|
}
|
|
|
|
void HttpOutRequest::set_path(LuaStack &LS, LuaSlot val) {
|
|
if (!LS.isstring(val)) {
|
|
fail(util::ss("HTTP path must be a string"));
|
|
return;
|
|
}
|
|
set_path(LS.ckstring(val));
|
|
}
|
|
|
|
void HttpOutRequest::set_param(LuaStack &LS, LuaSlot key, LuaSlot val) {
|
|
if (!LS.isstring(key)) {
|
|
fail(util::ss("HTTP url parameter key must be a string"));
|
|
return;
|
|
}
|
|
if (!LS.isstring(val)) {
|
|
fail(util::ss("HTTP url parameter val must be a string"));
|
|
return;
|
|
}
|
|
set_param(LS.ckstring(key), LS.ckstring(val));
|
|
}
|
|
|
|
void HttpOutRequest::set_params(LuaStack &LS0, LuaSlot tab) {
|
|
if (!LS0.istable(tab)) {
|
|
fail(util::ss("HTTP params must be a table"));
|
|
return;
|
|
}
|
|
LuaVar key, val;
|
|
LuaStack LS(LS0.state(), key, val);
|
|
LS.set(key, LuaNil);
|
|
while (LS.next(tab, key, val)) {
|
|
set_param(LS, key, val);
|
|
}
|
|
}
|
|
|
|
void HttpOutRequest::set_url(LuaStack &LS, LuaSlot val) {
|
|
if (!LS.isstring(val)) {
|
|
fail(util::ss("HTTP url must be a string"));
|
|
return;
|
|
}
|
|
set_url(LS.ckstring(val));
|
|
}
|
|
|
|
void HttpOutRequest::set_defaults() {
|
|
if (method_.empty()) {
|
|
method_ = "GET";
|
|
}
|
|
if (port_ == 0) {
|
|
port_ = 443;
|
|
}
|
|
}
|
|
|
|
void HttpOutRequest::set_config(LuaStack &LS0, LuaSlot tab) {
|
|
LuaVar key, val;
|
|
LuaStack LS(LS0.state(), key, val);
|
|
LS.set(key, LuaNil);
|
|
while (LS.next(tab, key, val)) {
|
|
eng::string kstr;
|
|
if (LS.isstring(key)) kstr = LS.ckstring(key);
|
|
if (kstr == "method") {
|
|
set_method(LS, val);
|
|
} else if (kstr == "host") {
|
|
set_host(LS, val);
|
|
} else if (kstr == "port") {
|
|
set_port(LS, val);
|
|
} else if (kstr == "path") {
|
|
set_path(LS, val);
|
|
} else if (kstr == "params") {
|
|
set_params(LS, val);
|
|
} else if (kstr == "url") {
|
|
set_url(LS, val);
|
|
} else if (kstr == "verifycertificate") {
|
|
set_verify_certificate(LS, val);
|
|
} else if (kstr == "") {
|
|
fail(util::ss("HTTP config parameter names must be strings."));
|
|
} else {
|
|
fail(util::ss("HTTP unrecognized config parameter: ", kstr));
|
|
}
|
|
}
|
|
}
|
|
|
|
eng::string HttpOutRequest::check() const {
|
|
if (!error_.empty()) {
|
|
return error_;
|
|
}
|
|
if (method_.empty()) {
|
|
return "HTTP method has not been set";
|
|
}
|
|
if (host_.empty()) {
|
|
return "HTTP host has not been set";
|
|
}
|
|
if (port_ == 0) {
|
|
return "HTTP port has not been set";
|
|
}
|
|
if (path_.empty()) {
|
|
return "HTTP url has not been set";
|
|
}
|
|
return "";
|
|
}
|
|
|
|
|
|
void HttpOutRequest::send_internal(StreamBuffer *sb, bool debug_string) const {
|
|
// If there's an error in the request, handle it. In debug string mode,
|
|
// we just put the error into the output. In production mode, we assert
|
|
// fail.
|
|
eng::string error = check();
|
|
if (debug_string) {
|
|
if (!error.empty()) {
|
|
sb->write_bytes(error);
|
|
return;
|
|
}
|
|
} else {
|
|
assert(error.empty());
|
|
}
|
|
|
|
// Choose a linebreak.
|
|
eng::string linebreak = (debug_string) ? "\n" : "\r\n";
|
|
|
|
// Send the command.
|
|
sb->write_bytes(method_);
|
|
sb->write_char(' ');
|
|
send_encoded_path(path_, params_, sb);
|
|
sb->write_bytes(" HTTP/1.1");
|
|
sb->write_bytes(linebreak);
|
|
|
|
// Send the host header.
|
|
sb->write_bytes("Host: ");
|
|
send_host_and_port(host_, port_, sb);
|
|
sb->write_bytes(linebreak);
|
|
|
|
// The empty accept-encoding header notifies the
|
|
// server that we don't support gzip, deflate, or
|
|
// other content compression.
|
|
sb->write_bytes("Accept-encoding:");
|
|
sb->write_bytes(linebreak);
|
|
|
|
// Add a user-agent header. Not sure why.
|
|
sb->write_bytes("User-agent: Mozilla 5.0 (luprex)");
|
|
sb->write_bytes(linebreak);
|
|
|
|
// Add the requester IDs (debug string only)
|
|
if (debug_string && ((request_id_ != 0) || (place_id_ != 0) || (thread_id_ != 0))) {
|
|
sb->write_bytes("X-Requester-Ids: ");
|
|
sb->ostream() << "rid=" << request_id_ << "; pid=" << place_id_ << "; tid=" << thread_id_;
|
|
sb->write_bytes(linebreak);
|
|
}
|
|
|
|
// Send the extra linebreak.
|
|
if (!debug_string) {
|
|
sb->write_bytes(linebreak);
|
|
}
|
|
}
|
|
|
|
void HttpOutRequest::serialize(StreamBuffer *sb) const {
|
|
sb->write_int64(request_id_);
|
|
sb->write_int64(place_id_);
|
|
sb->write_int64(thread_id_);
|
|
sb->write_string(error_);
|
|
sb->write_bool(verify_certificate_);
|
|
sb->write_string(method_);
|
|
sb->write_string(host_);
|
|
sb->write_int32(port_);
|
|
sb->write_string(path_);
|
|
sb->write_int32(params_.size());
|
|
for (const auto &pair : params_) {
|
|
sb->write_string(pair.first);
|
|
sb->write_string(pair.second);
|
|
}
|
|
}
|
|
|
|
void HttpOutRequest::deserialize(StreamBuffer *sb) {
|
|
request_id_ = sb->read_int64();
|
|
place_id_ = sb->read_int64();
|
|
thread_id_ = sb->read_int64();
|
|
error_ = sb->read_string();
|
|
verify_certificate_ = sb->read_bool();
|
|
method_ = sb->read_string();
|
|
host_ = sb->read_string();
|
|
port_ = sb->read_int32();
|
|
path_ = sb->read_string();
|
|
int32_t nparams = sb->read_int32();
|
|
params_.clear();
|
|
for (int i = 0; i < nparams; i++) {
|
|
eng::string k = sb->read_string();
|
|
eng::string v = sb->read_string();
|
|
params_[k] = v;
|
|
}
|
|
}
|
|
|
|
eng::string HttpOutRequest::DebugString() {
|
|
StreamBuffer sb;
|
|
send_internal(&sb, true);
|
|
return eng::string(sb.view());
|
|
}
|
|
|
|
void HttpOutRequestMap::serialize(StreamBuffer *sb) const {
|
|
sb->write_int32(size());
|
|
for (const auto &pair : *this) {
|
|
pair.second.serialize(sb);
|
|
}
|
|
}
|
|
|
|
void HttpOutRequestMap::deserialize(StreamBuffer *sb) {
|
|
int32_t count = sb->read_int32();
|
|
clear();
|
|
HttpOutRequest req;
|
|
for (int i = 0; i < count; i++) {
|
|
req.deserialize(sb);
|
|
(*this)[req.request_id()] = req;
|
|
}
|
|
}
|
|
|
|
HttpInResponse::HttpInResponse() {
|
|
status_code_ = 0;
|
|
response_length_ = 0;
|
|
mime_type_ = "";
|
|
content_length_ = -1;
|
|
}
|
|
|
|
eng::string HttpInResponse::DebugString() const {
|
|
eng::ostringstream oss;
|
|
oss << "HttpInResponse:" << std::endl;
|
|
oss << " status_code: " << status_code_ << std::endl;
|
|
oss << " error: " << error_ << std::endl;
|
|
oss << " content_length: " << content_length_ << std::endl;
|
|
oss << " transfer_encoding: " << transfer_encoding_ << std::endl;
|
|
oss << " location: " << location_ << std::endl;
|
|
oss << " mime_type: " << mime_type_ << std::endl;
|
|
oss << " charset: " << charset_ << std::endl;
|
|
oss << " content: " << content_ << std::endl;
|
|
oss << " response_length: " << response_length_ << std::endl;
|
|
return oss.str();
|
|
}
|
|
|
|
void HttpInResponse::fail(int code, string_view message) {
|
|
status_code_ = code;
|
|
error_ = message;
|
|
mime_type_ = "";
|
|
charset_ = "";
|
|
content_ = "";
|
|
}
|
|
|
|
void HttpInResponse::incomplete(bool closed) {
|
|
if (closed) {
|
|
fail(500, "internal server error: response truncated");
|
|
} else {
|
|
fail(0, "response not yet fully received");
|
|
}
|
|
}
|
|
|
|
void HttpInResponse::parse_content_encoding(string_view value) {
|
|
content_encoding_ = util::ascii_tolower(value);
|
|
}
|
|
|
|
void HttpInResponse::parse_content_length(string_view value) {
|
|
int64_t code = sv::to_int64(value);
|
|
if ((code < 0) || (code > INT_MAX)) {
|
|
fail(500, util::ss("internal server error: unparseable content-length: ", value));
|
|
}
|
|
content_length_ = code;
|
|
}
|
|
|
|
void HttpInResponse::parse_content_type(string_view value) {
|
|
eng::string ctype = util::ascii_tolower(value);
|
|
string_view ctview(ctype);
|
|
mime_type_ = sv::trim(sv::read_to_sep(ctview, ';'));
|
|
if (mime_type_.empty()) {
|
|
fail(500, util::ss("internal server error: unparseable content-type: ", value));
|
|
return;
|
|
}
|
|
while (true) {
|
|
string_view feature = sv::trim(sv::read_to_sep(ctview, ';'));
|
|
if (feature.empty()) {
|
|
return;
|
|
}
|
|
string_view ftype = sv::trim(sv::read_to_sep(feature, '='));
|
|
if (ftype == "charset") {
|
|
charset_ = sv::trim(feature);
|
|
}
|
|
}
|
|
}
|
|
|
|
void HttpInResponse::parse_location(string_view value) {
|
|
location_ = url_decode(value);
|
|
}
|
|
|
|
void HttpInResponse::parse_transfer_encoding(string_view value) {
|
|
transfer_encoding_ = util::ascii_tolower(value);
|
|
}
|
|
|
|
void HttpInResponse::parse_header(string_view header, string_view value) {
|
|
if (header == "content-encoding") {
|
|
parse_content_encoding(value);
|
|
} else if (header == "content-length") {
|
|
parse_content_length(value);
|
|
} else if (header == "content-type") {
|
|
parse_content_type(value);
|
|
} else if (header == "location") {
|
|
parse_location(value);
|
|
} else if (header == "transfer-encoding") {
|
|
parse_transfer_encoding(value);
|
|
} else if (header == "content-range") {
|
|
fail(406, util::ss("not acceptable: unsupported response header: ", header));
|
|
}
|
|
}
|
|
|
|
bool HttpInResponse::parse_content_basic(std::string_view &view, bool closed) {
|
|
if (content_length_ >= 0) {
|
|
if (content_length_ > MAX_CONTENT_LENGTH) {
|
|
fail(413, util::ss("payload too large: luprex limit=", MAX_CONTENT_LENGTH));
|
|
return false;
|
|
}
|
|
if (int(view.size()) < content_length_) {
|
|
incomplete(closed);
|
|
return false;
|
|
}
|
|
content_ = sv::read_nbytes(view, content_length_);
|
|
} else {
|
|
if (int64_t(view.size()) > MAX_CONTENT_LENGTH) {
|
|
fail(413, util::ss("payload too large: luprex limit=", MAX_CONTENT_LENGTH));
|
|
return false;
|
|
}
|
|
if (!closed) {
|
|
incomplete(closed);
|
|
return false;
|
|
}
|
|
content_ = sv::read_nbytes(view, view.size());
|
|
}
|
|
return true;
|
|
}
|
|
|
|
bool HttpInResponse::parse_content_chunked(std::string_view &view, bool closed) {
|
|
int64_t total_size = 0;
|
|
std::vector<string_view> chunks;
|
|
while (true) {
|
|
std::string_view chunk_header = sv::trim(sv::read_to_line(view));
|
|
if (sv::isnull(view)) {
|
|
incomplete(closed);
|
|
return false;
|
|
}
|
|
int64_t chunk_size = sv::to_hex64(chunk_header, -1);
|
|
if (chunk_size < 0) {
|
|
fail(500, "internal server error: unparseable chunk header");
|
|
return false;
|
|
}
|
|
if (chunk_size > MAX_CONTENT_LENGTH) {
|
|
fail(413, util::ss("payload too large: luprex limit=", MAX_CONTENT_LENGTH));
|
|
return false;
|
|
}
|
|
if (chunk_size == 0) break;
|
|
total_size += chunk_size;
|
|
if (total_size > MAX_CONTENT_LENGTH) {
|
|
fail(413, util::ss("payload too large: luprex limit=", MAX_CONTENT_LENGTH));
|
|
return false;
|
|
}
|
|
std::string_view chunk = sv::read_nbytes(view, chunk_size);
|
|
if (int64_t(chunk.size()) != chunk_size) {
|
|
incomplete(closed);
|
|
return false;
|
|
}
|
|
std::string_view newline = sv::read_to_line(view);
|
|
if (!newline.empty()) {
|
|
fail(500, "internal server error: corrupted chunk encoding");
|
|
return false;
|
|
}
|
|
if (sv::isnull(view)) {
|
|
incomplete(closed);
|
|
return false;
|
|
}
|
|
chunks.push_back(chunk);
|
|
}
|
|
content_.resize(total_size);
|
|
size_t offset = 0;
|
|
for (string_view chunk : chunks) {
|
|
content_.replace(offset, chunk.size(), chunk);
|
|
offset += chunk.size();
|
|
}
|
|
return true;
|
|
}
|
|
|
|
void HttpInResponse::parse(const StreamBuffer *sb, bool closed) {
|
|
// We're not going to modify the StreamBuffer at all.
|
|
// Instead, we work entirely on a view.
|
|
string_view view = sb->view();
|
|
|
|
// Get the status line.
|
|
string_view status = sv::trim(sv::read_to_line(view));
|
|
if (sv::isnull(view)) {
|
|
incomplete(closed);
|
|
return;
|
|
}
|
|
|
|
// Parse the status line.
|
|
string_view scode = sv::read_to_space(status);
|
|
int64_t code = sv::to_int64(scode, 0);
|
|
if ((code < 100) || (code > 599)) {
|
|
fail(500, util::ss("internal server error: invalid response code: ", scode));
|
|
}
|
|
status_code_ = code;
|
|
|
|
// Responses outside the range 200-299 are errors,
|
|
// and therefore must store an error message.
|
|
if ((code < 200) || (code > 299)) {
|
|
error_ = status;
|
|
if (error_.empty()) {
|
|
fail(code, util::ss("error code ", code));
|
|
}
|
|
}
|
|
|
|
// Parse the headers.
|
|
while (true) {
|
|
string_view header = sv::read_to_line(view);
|
|
if (sv::isnull(view)) {
|
|
incomplete(closed);
|
|
return;
|
|
}
|
|
if (header.empty()) {
|
|
break;
|
|
}
|
|
eng::string command = util::ascii_tolower(sv::trim(sv::read_to_sep(header, ':')));
|
|
if (sv::isnull(header)) {
|
|
fail(500, util::ss("internal server error: no colon in header line: ", command));
|
|
return;
|
|
}
|
|
if (!words_separated_by_dashes(command)) {
|
|
fail(500, util::ss("internal server error: invalid header: ", command));
|
|
return;
|
|
}
|
|
parse_header(command, sv::trim(header));
|
|
}
|
|
|
|
// Process the content using the transfer encoding.
|
|
if (transfer_encoding_ == "") {
|
|
if (!parse_content_basic(view, closed)) return;
|
|
} else if (transfer_encoding_ == "chunked") {
|
|
if (!parse_content_chunked(view, closed)) return;
|
|
} else {
|
|
fail(500, util::ss("unsupported transfer-encoding: ", transfer_encoding_));
|
|
return;
|
|
}
|
|
|
|
// Calculate the response length.
|
|
response_length_ = sb->fill() - view.size();
|
|
|
|
// If it's not a redirect, disallow 'location'.
|
|
if ((status_code_ < 300) || (status_code_ > 399)) {
|
|
if (!location_.empty()) {
|
|
fail(500, util::ss("internal server error: location specified, but result code not 300-399: ", code));
|
|
return;
|
|
}
|
|
}
|
|
|
|
// If the server didn't specify content-type, make a guess.
|
|
if (mime_type_.empty()) {
|
|
if (sv::valid_utf8(content_)) {
|
|
mime_type_ = "text/plain";
|
|
charset_ = "utf-8";
|
|
} else {
|
|
mime_type_ = "application/octet-stream";
|
|
charset_ = "";
|
|
}
|
|
}
|
|
|
|
// If it's multipart, reject it.
|
|
if (sv::has_prefix(mime_type_, "multipart/")) {
|
|
fail(406, "not acceptable: multipart messages not supported");
|
|
return;
|
|
}
|
|
|
|
// If it's text, demand a reasonable charset. Otherwise,
|
|
// ignore the charset.
|
|
if (sv::has_prefix(mime_type_, "text/")) {
|
|
if (charset_.empty() || (charset_ == "ascii")) {
|
|
charset_ = "utf-8";
|
|
}
|
|
if (charset_ != "utf-8") {
|
|
fail(406, util::ss("not acceptable: charset not supported: ", charset_));
|
|
return;
|
|
}
|
|
} else {
|
|
charset_.clear();
|
|
}
|
|
|
|
// Uncompress the content.
|
|
if ((content_encoding_ == "") || (content_encoding_ == "identity")) {
|
|
} else {
|
|
fail(406, util::ss("not acceptable: content-encoding not supported: ", content_encoding_));
|
|
return;
|
|
}
|
|
|
|
// If there's an error code, throw out the content.
|
|
if ((status_code_ < 200) || (status_code_ > 299)) {
|
|
mime_type_.clear();
|
|
charset_.clear();
|
|
content_.clear();
|
|
}
|
|
}
|
|
|
|
void HttpInResponse::store(LuaStack &LS0, LuaSlot tab) const {
|
|
LuaStack LS(LS0.state());
|
|
|
|
LS.newtable(tab);
|
|
LS.rawset(tab, "responsecode", status_code_);
|
|
if (!error_.empty()) {
|
|
LS.rawset(tab, "error", error_);
|
|
}
|
|
if (!location_.empty()) {
|
|
LS.rawset(tab, "location", location_);
|
|
}
|
|
if (!mime_type_.empty()) {
|
|
LS.rawset(tab, "mimetype", mime_type_);
|
|
LS.rawset(tab, "content", content_);
|
|
}
|
|
|
|
// Debugging fields. Do not use for lua programming.
|
|
if (content_length_ >= 0) {
|
|
LS.rawset(tab, "dbg_contentlength", content_length_);
|
|
}
|
|
if (!transfer_encoding_.empty()) {
|
|
LS.rawset(tab, "dbg_transferencoding", transfer_encoding_);
|
|
}
|
|
if (!charset_.empty()) {
|
|
LS.rawset(tab, "dbg_charset", charset_);
|
|
}
|
|
if (response_length_ != 0) {
|
|
LS.rawset(tab, "dbg_responselength", response_length_);
|
|
}
|
|
}
|
|
|
|
void HttpInResponse::store_fail(LuaStack &LS, LuaSlot tab, int status_code, std::string_view error) {
|
|
HttpInResponse response;
|
|
response.fail(status_code, error);
|
|
response.store(LS, tab);
|
|
}
|
|
|
|
LuaDefine(http_fixurl, "url", "validate URL and repair minor flaws in the URL syntax") {
|
|
LuaArg url;
|
|
LuaRet fixed;
|
|
LuaStack LS(L, url, fixed);
|
|
ParsedURL parsed(LS.ckstring(url));
|
|
if (!parsed.valid) {
|
|
luaL_error(L, "invalid URL, not fixable");
|
|
return LS.result();
|
|
}
|
|
LS.set(fixed, parsed.str());
|
|
return LS.result();
|
|
}
|
|
|
|
|
|
LuaDefine(http_request, "request",
|
|
"|Takes an HTTP request in the form of a lua table."
|
|
"|The table may contain these fields:"
|
|
"|"
|
|
"| method (ie, 'GET', 'POST', etc)"
|
|
"| host (ie, 'google.com')"
|
|
"| port (default: 443)"
|
|
"| path (ie, '/index.html')"
|
|
"| params (a table of url parameters)"
|
|
"| verifycertificate (default: true)"
|
|
"| url (ie, 'https://host:port/path.html?a=b&c=d')"
|
|
"|"
|
|
"|You can specify url components separately (host, port, path,"
|
|
"|and params), or you can specify the entire url as a unit. "
|
|
"|If you specify components, they must not be url-encoded. "
|
|
"|If you specify the url as a whole, it must already be url-encoded."
|
|
"|"
|
|
"|You can omit the port, in which case it defaults to the"
|
|
"|standard https port. You can omit verifycertificate, in which"
|
|
"|case it defaults to true. You can omit the method if the"
|
|
"|method is implied by the function you called (eg, 'http.get')."
|
|
"|"
|
|
"|Note that unencrypted http is not supported - we only allow https."
|
|
"|However, you can talk to a server that has a dummy certificate"
|
|
"|by specifying verifycertificate=false."
|
|
"|"
|
|
"|This routine, http.request, returns a debug string for the "
|
|
"|request. The debug string looks like the actual http headers"
|
|
"|that would be sent.") {
|
|
LuaArg tab;
|
|
LuaRet str;
|
|
LuaStack LS(L, tab, str);
|
|
HttpOutRequest req;
|
|
req.set_config(LS, tab);
|
|
req.set_defaults();
|
|
eng::string error = req.check();
|
|
if (!error.empty()) {
|
|
luaL_error(L, "%s", error.c_str());
|
|
return 0;
|
|
}
|
|
LS.set(str, req.DebugString());
|
|
return LS.result();
|
|
}
|
|
|
|
LuaDefine(http_response, "response",
|
|
"|Returns an HTTP response in the form of a lua table."
|
|
"|The table will contain these important fields:"
|
|
"|"
|
|
"| responsecode - 3-digit HTTP response code."
|
|
"| error - an error message, or nil if no error."
|
|
"| content - on success, the content, as a string."
|
|
"| mimetype - on success, the mime type of the content."
|
|
"| location - for HTTP redirects, the target url."
|
|
"|"
|
|
"|If the mimetype is a text mimetype, then the content"
|
|
"|is automatically converted to utf-8."
|
|
"|"
|
|
"|The table may also contain these debugging-only fields."
|
|
"|"
|
|
"| dbg_transferencoding - If there was a Transfer-Encoding header."
|
|
"| dbg_contentlength - If there was a Content-length header."
|
|
"| dbg_charset - Original character set for text mime types."
|
|
"| dbg_responselength - Total bytes in the response."
|
|
"|"
|
|
"|None of the dbg fields is needed to understand the response."
|
|
"|For example, consider dbg_charset. When text content is"
|
|
"|passed to lua, the content is automatically converted to utf-8."
|
|
"|So dbg_charset only tells you what the character set used"
|
|
"|to be, before it was converted to utf-8."
|
|
"|"
|
|
"|If an http routine generates an error, that error will be"
|
|
"|expressed as a status code. These locally-generated status"
|
|
"|codes can be:"
|
|
"|"
|
|
"| 400 (bad request) - the request was malformed"
|
|
"| 503 (service unavailable) - dns fail, connect fail, or ssl fail"
|
|
"| 500 (internal server error)- the response contains invalid HTTP"
|
|
"| 406 (not acceptable) - the response used a feature we don't support yet"
|
|
"| 413 (payload too large) - we refuse to download something so big"
|
|
"| 425 (can't resume) - reloaded a save game with a pending request"
|
|
"|"
|
|
"|In this case, the error message will be the stock"
|
|
"|error message, followed by more information."
|
|
"|"
|
|
"|This routine, http.response, generates a response by parsing"
|
|
"|an actual HTTP response string. This is for debugging only.") {
|
|
LuaArg text;
|
|
LuaRet tab;
|
|
LuaStack LS(L, text, tab);
|
|
HttpInResponse resp;
|
|
StreamBuffer sb;
|
|
sb.write_bytes(LS.ckstring(text));
|
|
resp.parse(&sb, true);
|
|
resp.store(LS, tab);
|
|
return LS.result();
|
|
}
|
|
|