1310 lines
39 KiB
C++
1310 lines
39 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 is_supported_protocol(string_view protocol) {
|
|
return (protocol == "HTTP/1.0") || (protocol == "HTTP/1.1");
|
|
}
|
|
|
|
bool is_supported_method(string_view method) {
|
|
return ((method == "GET") || (method == "HEAD") || (method == "POST"));
|
|
}
|
|
|
|
bool words_separated_by_dashes(string_view v) {
|
|
while (true) {
|
|
string_view word = sv::read_ascii_identifier(v);
|
|
if (word.empty()) return false;
|
|
if (v.empty()) return true;
|
|
char c = v.front();
|
|
if (c != '-') return false;
|
|
v.remove_prefix(1);
|
|
}
|
|
}
|
|
|
|
// This doesn't check whether the mime type is actually
|
|
// registered, obviously. It only checks that it's in
|
|
// the desired notation.
|
|
bool valid_mime_type(string_view method) {
|
|
string_view part1 = sv::read_ascii_identifier(method);
|
|
if (part1.empty()) return false;
|
|
if (sv::zfront(method) != '/') return false;
|
|
method.remove_prefix(1);
|
|
while (true) {
|
|
string_view word = sv::read_ascii_identifier(method);
|
|
if (word.empty()) return false;
|
|
if (method.empty()) return true;
|
|
char c = method.front();
|
|
if ((c != '-') && (c != '.') && (c != '+')) return false;
|
|
method.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();
|
|
|
|
if (!sv::has_prefix(url, "/")) {
|
|
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;
|
|
}
|
|
} else {
|
|
// Stick in some defaults for unspecified fields.
|
|
host = "host";
|
|
proto = "https";
|
|
}
|
|
|
|
// 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;
|
|
}
|
|
};
|
|
|
|
HttpClientRequest::HttpClientRequest() {
|
|
verify_certificate_ = false;
|
|
port_ = 0;
|
|
request_id_ = 0;
|
|
place_id_ = 0;
|
|
thread_id_ = 0;
|
|
}
|
|
|
|
void HttpClientRequest::fail(string_view s) {
|
|
if (error_.empty()) {
|
|
error_ = s;
|
|
}
|
|
}
|
|
|
|
eng::string HttpClientRequest::target() const {
|
|
assert(check().empty());
|
|
eng::ostringstream oss;
|
|
oss << (verify_certificate_ ? "cert" : "nocert");
|
|
oss << ':' << host_ << ':' << port_;
|
|
return oss.str();
|
|
}
|
|
|
|
void HttpClientRequest::set_verify_certificate(bool flag) {
|
|
verify_certificate_ = flag;
|
|
}
|
|
|
|
void HttpClientRequest::set_method(const eng::string &s) {
|
|
eng::string method = util::ascii_toupper(s);
|
|
if (!is_supported_method(method)) {
|
|
fail(util::ss("HTTP method not implemented: ", method, "."));
|
|
return;
|
|
}
|
|
if ((!method_.empty()) && (method_ != method)) {
|
|
fail(util::ss("HTTP method specified twice: ", method_, " and ", method));
|
|
return;
|
|
}
|
|
method_ = method;
|
|
}
|
|
|
|
void HttpClientRequest::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 HttpClientRequest::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 HttpClientRequest::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 HttpClientRequest::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 HttpClientRequest::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 HttpClientRequest::set_mime_type(const eng::string &mime_type) {
|
|
if (!valid_mime_type(mime_type)) {
|
|
fail(util::ss("Not a valid mime type: ", mime_type));
|
|
return;
|
|
}
|
|
if (!mime_type_.empty()) {
|
|
fail(util::ss("Mime type specified twice: ", mime_type_, " and ", mime_type));
|
|
return;
|
|
}
|
|
mime_type_ = mime_type;
|
|
}
|
|
|
|
void HttpClientRequest::set_content(const eng::string &content) {
|
|
if (!content_.empty()) {
|
|
fail(util::ss("Content specified twice"));
|
|
return;
|
|
}
|
|
content_ = content;
|
|
}
|
|
|
|
void HttpClientRequest::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 HttpClientRequest::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 HttpClientRequest::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 HttpClientRequest::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 HttpClientRequest::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 HttpClientRequest::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 HttpClientRequest::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 HttpClientRequest::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 HttpClientRequest::set_mime_type(LuaStack &LS, LuaSlot val) {
|
|
if (!LS.isstring(val)) {
|
|
fail(util::ss("HTTP mime type must be a string"));
|
|
return;
|
|
}
|
|
set_mime_type(LS.ckstring(val));
|
|
}
|
|
|
|
void HttpClientRequest::set_content(LuaStack &LS, LuaSlot val) {
|
|
if (!LS.isstring(val)) {
|
|
fail(util::ss("HTTP content must be a string"));
|
|
return;
|
|
}
|
|
set_content(LS.ckstring(val));
|
|
}
|
|
|
|
void HttpClientRequest::set_defaults() {
|
|
if (method_.empty()) {
|
|
method_ = "GET";
|
|
}
|
|
if (port_ == 0) {
|
|
port_ = 443;
|
|
}
|
|
}
|
|
|
|
void HttpClientRequest::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 == "mimetype") {
|
|
set_mime_type(LS, val);
|
|
} else if (kstr == "content") {
|
|
set_content(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 HttpClientRequest::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";
|
|
}
|
|
if (method_ == "POST") {
|
|
if (mime_type_.empty()) {
|
|
if (content_.empty()) {
|
|
return "HTTP mime type and content not set for POST request";
|
|
} else {
|
|
return "HTTP mime type has not been set for POST request";
|
|
}
|
|
}
|
|
} else {
|
|
if ((!mime_type_.empty()) || (!content_.empty())) {
|
|
return "HTTP mime type and content are only for POST requests";
|
|
}
|
|
}
|
|
return "";
|
|
}
|
|
|
|
|
|
void HttpClientRequest::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 = "\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);
|
|
}
|
|
|
|
// If it's a post request, send the content length and the content type.
|
|
if (method_ == "POST") {
|
|
sb->write_bytes("Content-length: ");
|
|
sb->ostream() << content_.size();
|
|
sb->write_bytes(linebreak);
|
|
sb->write_bytes("Content-type: ");
|
|
sb->write_bytes(mime_type_);
|
|
if (sv::has_prefix(mime_type_, "text/")) {
|
|
sb->write_bytes(" ; charset=utf-8");
|
|
}
|
|
sb->write_bytes(linebreak);
|
|
}
|
|
|
|
// Send the extra linebreak.
|
|
sb->write_bytes(linebreak);
|
|
|
|
// If it's a post request, send the content.
|
|
if (method_ == "POST") {
|
|
sb->write_bytes(content_);
|
|
}
|
|
}
|
|
|
|
void HttpClientRequest::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 HttpClientRequest::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 HttpClientRequest::DebugString() {
|
|
StreamBuffer sb;
|
|
send_internal(&sb, true);
|
|
return eng::string(sb.view());
|
|
}
|
|
|
|
void HttpClientRequestMap::serialize(StreamBuffer *sb) const {
|
|
sb->write_int32(size());
|
|
for (const auto &pair : *this) {
|
|
pair.second.serialize(sb);
|
|
}
|
|
}
|
|
|
|
void HttpClientRequestMap::deserialize(StreamBuffer *sb) {
|
|
int32_t count = sb->read_int32();
|
|
clear();
|
|
HttpClientRequest req;
|
|
for (int i = 0; i < count; i++) {
|
|
req.deserialize(sb);
|
|
(*this)[req.request_id()] = req;
|
|
}
|
|
}
|
|
|
|
HttpParser::HttpParser() {
|
|
request_id_ = 0;
|
|
is_request_ = false;
|
|
status_ = 0;
|
|
mime_type_ = "";
|
|
content_length_ = -1;
|
|
comm_length_ = 0;
|
|
}
|
|
|
|
void HttpParser::fail(int code, std::string_view message) {
|
|
status_ = code;
|
|
error_ = message;
|
|
mime_type_ = "";
|
|
charset_ = "";
|
|
content_ = "";
|
|
}
|
|
|
|
void HttpParser::syntax(std::string_view detail) {
|
|
if (is_request_) {
|
|
fail(400, util::ss("malformed request: ", detail));
|
|
} else {
|
|
fail(500, util::ss("malformed response: ", detail));
|
|
}
|
|
}
|
|
|
|
void HttpParser::incomplete(bool closed) {
|
|
if (closed) {
|
|
syntax("response truncated");
|
|
} else {
|
|
fail(0, "response not yet fully received");
|
|
}
|
|
}
|
|
|
|
void HttpParser::oversized() {
|
|
fail(413, util::ss("payload too large: Limit=", MAX_CONTENT_LENGTH));
|
|
}
|
|
|
|
bool HttpParser::parse_request_line(std::string_view &view, bool closed) {
|
|
// Extract the request line.
|
|
//
|
|
string_view request = sv::trim(sv::read_to_line(view));
|
|
if (sv::isnull(view)) {
|
|
incomplete(closed);
|
|
return false;
|
|
}
|
|
|
|
// Break down the request line.
|
|
//
|
|
eng::string method = util::ascii_toupper(sv::read_to_space(request));
|
|
string_view path = sv::read_to_space(request);
|
|
eng::string protocol = util::ascii_toupper(sv::read_to_space(request));
|
|
if ((!request.empty()) || (protocol.empty())) {
|
|
syntax("invalid request line");
|
|
return false;
|
|
}
|
|
if (!is_supported_method(method)) {
|
|
fail(405, util::ss("Method Not Allowed: ", method));
|
|
return false;
|
|
}
|
|
if (!is_supported_protocol(protocol)) {
|
|
syntax(util::ss("unsupported protocol: ", protocol));
|
|
return false;
|
|
}
|
|
|
|
// Parse the url.
|
|
//
|
|
ParsedURL url(path);
|
|
if (!url.valid) {
|
|
syntax(util::ss("Invalid URL path: ", path));
|
|
return false;
|
|
}
|
|
|
|
method_ = method;
|
|
path_ = url.path;
|
|
params_ = url.params;
|
|
return true;
|
|
}
|
|
|
|
bool HttpParser::parse_status_line(std::string_view &view, bool closed) {
|
|
// Extract the status line.
|
|
//
|
|
string_view status = sv::trim(sv::read_to_line(view));
|
|
if (sv::isnull(view)) {
|
|
incomplete(closed);
|
|
return false;
|
|
}
|
|
|
|
// Break down the status line.
|
|
//
|
|
string_view protoversion = sv::read_to_space(status);
|
|
if (!is_supported_protocol(protoversion)) {
|
|
syntax(util::ss("unsupported protocol: ", protoversion));
|
|
return false;
|
|
}
|
|
string_view scode = sv::read_to_space(status);
|
|
int64_t code = sv::to_int64(scode, 0);
|
|
if ((code < 100) || (code > 599)) {
|
|
syntax(util::ss("invalid response code: ", scode));
|
|
return false;
|
|
}
|
|
status_ = code;
|
|
|
|
// Responses outside the range 200-299 are errors,
|
|
// and therefore must store a nonempty error message.
|
|
//
|
|
if ((code < 200) || (code > 299)) {
|
|
if (status.empty()) {
|
|
error_ = util::ss("error code ", code);
|
|
} else {
|
|
error_ = status;
|
|
}
|
|
}
|
|
return true;
|
|
}
|
|
|
|
void HttpParser::parse_content_encoding(string_view value) {
|
|
content_encoding_ = util::ascii_tolower(value);
|
|
}
|
|
|
|
void HttpParser::parse_content_length(string_view value) {
|
|
int64_t code = sv::to_int64(value);
|
|
if ((code < 0) || (code > INT_MAX)) {
|
|
syntax(util::ss("unparseable content-length: ", value));
|
|
}
|
|
content_length_ = code;
|
|
}
|
|
|
|
void HttpParser::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()) {
|
|
syntax(util::ss("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 HttpParser::parse_location(string_view value) {
|
|
location_ = url_decode(value);
|
|
}
|
|
|
|
void HttpParser::parse_transfer_encoding(string_view value) {
|
|
transfer_encoding_ = util::ascii_tolower(value);
|
|
}
|
|
|
|
void HttpParser::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(416, util::ss("range not satisfiable: unsupported header: ", header));
|
|
}
|
|
}
|
|
|
|
bool HttpParser::parse_headers(std::string_view &view, bool closed) {
|
|
// Parse the headers.
|
|
while (true) {
|
|
string_view header = sv::read_to_line(view);
|
|
if (sv::isnull(view)) {
|
|
incomplete(closed);
|
|
return false;
|
|
}
|
|
if (header.empty()) {
|
|
return true;
|
|
}
|
|
eng::string command = util::ascii_tolower(sv::trim(sv::read_to_sep(header, ':')));
|
|
if (sv::isnull(header)) {
|
|
syntax(util::ss("no colon in header line: ", command));
|
|
return false;
|
|
}
|
|
if (!words_separated_by_dashes(command)) {
|
|
syntax(util::ss("invalid header: ", command));
|
|
return false;
|
|
}
|
|
parse_header(command, sv::trim(header));
|
|
}
|
|
}
|
|
|
|
bool HttpParser::parse_content_basic(std::string_view &view, bool closed) {
|
|
if (content_length_ >= 0) {
|
|
if (content_length_ > MAX_CONTENT_LENGTH) {
|
|
oversized();
|
|
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) {
|
|
oversized();
|
|
return false;
|
|
}
|
|
if (!closed) {
|
|
incomplete(closed);
|
|
return false;
|
|
}
|
|
content_ = sv::read_nbytes(view, view.size());
|
|
}
|
|
return true;
|
|
}
|
|
|
|
bool HttpParser::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) {
|
|
syntax("unparseable chunk header");
|
|
return false;
|
|
}
|
|
if (chunk_size > MAX_CONTENT_LENGTH) {
|
|
oversized();
|
|
return false;
|
|
}
|
|
if (chunk_size == 0) break;
|
|
total_size += chunk_size;
|
|
if (total_size > MAX_CONTENT_LENGTH) {
|
|
oversized();
|
|
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()) {
|
|
syntax("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;
|
|
}
|
|
|
|
bool HttpParser::parse_content(std::string_view &view, bool closed) {
|
|
// Parse the content.
|
|
if (transfer_encoding_ == "") {
|
|
if (!parse_content_basic(view, closed)) {
|
|
return false;
|
|
}
|
|
} else if (transfer_encoding_ == "chunked") {
|
|
if (!parse_content_chunked(view, closed)) {
|
|
return false;
|
|
}
|
|
} else {
|
|
syntax(util::ss("unsupported transfer-encoding: ", transfer_encoding_));
|
|
return false;
|
|
}
|
|
|
|
// Uncompress the content.
|
|
if ((content_encoding_ == "") || (content_encoding_ == "identity")) {
|
|
} else {
|
|
syntax(util::ss("content-encoding not supported: ", content_encoding_));
|
|
return true;
|
|
}
|
|
|
|
// If the sender didn't specify content-type, make a guess based on the content.
|
|
if (mime_type_.empty()) {
|
|
if (sv::valid_utf8(content_)) {
|
|
mime_type_ = "text/plain";
|
|
charset_ = "utf-8";
|
|
} else {
|
|
mime_type_ = "application/octet-stream";
|
|
charset_ = "";
|
|
}
|
|
}
|
|
|
|
// Switch the charset to utf-8, if it's text.
|
|
if (sv::has_prefix(mime_type_, "text/")) {
|
|
if (charset_.empty() || (charset_ == "ascii") || (charset_ == "utf-8")) {
|
|
// we're already good.
|
|
} else {
|
|
// We can't convert charsets yet.
|
|
syntax(util::ss("charset not supported: ", charset_));
|
|
return true;
|
|
}
|
|
} else {
|
|
// Not text. No need to specify charset.
|
|
charset_.clear();
|
|
}
|
|
return true;
|
|
}
|
|
|
|
void HttpParser::store(LuaStack &LS0, LuaSlot tab) const {
|
|
LuaVar ptab;
|
|
LuaStack LS(LS0.state(), ptab);
|
|
|
|
LS.newtable(tab);
|
|
LS.rawset(tab, "status", status_);
|
|
if (!error_.empty()) {
|
|
LS.rawset(tab, "error", error_);
|
|
}
|
|
if (!location_.empty()) {
|
|
LS.rawset(tab, "location", location_);
|
|
}
|
|
if (!mime_type_.empty() || !content_.empty()) {
|
|
LS.rawset(tab, "mimetype", mime_type_);
|
|
LS.rawset(tab, "content", content_);
|
|
}
|
|
if (!method_.empty()) {
|
|
LS.rawset(tab, "method", method_);
|
|
}
|
|
if (!path_.empty()) {
|
|
LS.rawset(tab, "path", path_);
|
|
LS.newtable(ptab);
|
|
LS.rawset(tab, "params", ptab);
|
|
for (const auto &pair : params_) {
|
|
LS.rawset(ptab, pair.first, pair.second);
|
|
}
|
|
}
|
|
|
|
// 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 (comm_length_ != 0) {
|
|
LS.rawset(tab, "dbg_commlength", comm_length_);
|
|
}
|
|
}
|
|
|
|
eng::string HttpParser::debug_string() const {
|
|
eng::ostringstream oss;
|
|
if (request_id_ != 0) {
|
|
oss << " request_id: " << request_id_ << std::endl;
|
|
}
|
|
oss << " status_code: " << status_ << std::endl;
|
|
oss << " error: " << error_ << std::endl;
|
|
if (content_length_ >= 0) {
|
|
oss << " content_length: " << content_length_ << std::endl;
|
|
}
|
|
if (!transfer_encoding_.empty()) {
|
|
oss << " transfer_encoding: " << transfer_encoding_ << std::endl;
|
|
}
|
|
if (!location_.empty()) {
|
|
oss << " location: " << location_ << std::endl;
|
|
}
|
|
if (!mime_type_.empty()) {
|
|
oss << " mime_type: " << mime_type_ << std::endl;
|
|
}
|
|
if (!charset_.empty()) {
|
|
oss << " charset: " << charset_ << std::endl;
|
|
}
|
|
if (!content_.empty()) {
|
|
oss << " content: " << content_ << std::endl;
|
|
}
|
|
if (!method_.empty()) {
|
|
oss << " method: " << method_ << std::endl;
|
|
}
|
|
if (!path_.empty()) {
|
|
oss << " path: " << path_ << std::endl;
|
|
}
|
|
for (const auto &pair : params_) {
|
|
oss << " param: " << pair.first << "=" << pair.second << std::endl;
|
|
}
|
|
if (comm_length_ > 0) {
|
|
oss << " comm_length: " << comm_length_ << std::endl;
|
|
}
|
|
return oss.str();
|
|
}
|
|
|
|
void HttpParser::clear_content_on_error() {
|
|
if ((status_ < 200) || (status_ > 299)) {
|
|
mime_type_.clear();
|
|
charset_.clear();
|
|
content_.clear();
|
|
}
|
|
}
|
|
|
|
void HttpParser::parse_response(std::string_view view, bool closed) {
|
|
std::string_view original_view = view;
|
|
is_request_ = false;
|
|
|
|
// Parse the status line.
|
|
if (!parse_status_line(view, closed)) {
|
|
return;
|
|
}
|
|
|
|
// Parse the headers.
|
|
if (!parse_headers(view, closed)) {
|
|
return;
|
|
}
|
|
|
|
// Process the content.
|
|
if (!parse_content(view, closed)) {
|
|
return;
|
|
}
|
|
|
|
// Calculate the response length.
|
|
comm_length_ = original_view.size() - view.size();
|
|
|
|
// If it's not a redirect, ignore location.
|
|
if ((status_ < 300) || (status_ > 399)) {
|
|
location_.clear();
|
|
}
|
|
|
|
// If it's multipart, reject it.
|
|
if (sv::has_prefix(mime_type_, "multipart/")) {
|
|
syntax("multipart messages not supported");
|
|
return;
|
|
}
|
|
|
|
// If there's an error code, throw out the content.
|
|
clear_content_on_error();
|
|
}
|
|
|
|
void HttpParser::parse_request(std::string_view view, bool closed) {
|
|
std::string_view original_view = view;
|
|
is_request_ = true;
|
|
|
|
// Parse the request line.
|
|
if (!parse_request_line(view, closed)) {
|
|
return;
|
|
}
|
|
|
|
// Parse the headers.
|
|
if (!parse_headers(view, closed)) {
|
|
return;
|
|
}
|
|
|
|
// Process the content, if any.
|
|
if (method_ == "POST") {
|
|
if (!parse_content(view, closed)) {
|
|
return;
|
|
}
|
|
}
|
|
|
|
// Calculate the comm length.
|
|
comm_length_ = original_view.size() - view.size();
|
|
|
|
// Always ignore location.
|
|
location_.clear();
|
|
|
|
// If it's multipart, reject it.
|
|
if (sv::has_prefix(mime_type_, "multipart/")) {
|
|
syntax("multipart messages not supported");
|
|
return;
|
|
}
|
|
|
|
// If we've made it this far, and there's no
|
|
// status code, set it to 200 OK.
|
|
if (status_ == 0) status_ = 200;
|
|
|
|
// If there's an error code, throw out the content.
|
|
clear_content_on_error();
|
|
}
|
|
|
|
void HttpParser::store_fail(LuaStack &LS, LuaSlot tab, int status_code, std::string_view error) {
|
|
HttpParser parser;
|
|
parser.fail(status_code, error);
|
|
parser.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_clientrequest, "request",
|
|
"|Takes an HTTP client 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.clientrequest, 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);
|
|
HttpClientRequest 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_clientresponse, "response",
|
|
"|Returns an HTTP client response in the form of a lua table."
|
|
"|The table will contain these important fields:"
|
|
"|"
|
|
"| status - 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_commlength - Total bytes in the communication."
|
|
"|"
|
|
"|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"
|
|
"| 413 (payload too large) - we refuse to download something so big"
|
|
"| 425 (can't resume) - reloaded a save game with a pending request"
|
|
"|"
|
|
"|Error messages that are generated locally consist of "
|
|
"|the standard message (eg, 'bad request') followed by more "
|
|
"|detailed information."
|
|
"|"
|
|
"|This routine, http.clientresponse, generates a response by parsing"
|
|
"|an actual HTTP response string. This is for debugging only.") {
|
|
LuaArg text;
|
|
LuaRet tab;
|
|
LuaStack LS(L, text, tab);
|
|
HttpParser parser;
|
|
parser.parse_response(LS.ckstring(text), true);
|
|
parser.store(LS, tab);
|
|
return LS.result();
|
|
}
|
|
|
|
LuaDefine(http_serverrequest, "request",
|
|
"|Returns an HTTP server request in the form of a lua table."
|
|
"|The table will contain these important fields:"
|
|
"|"
|
|
"| status - 3-digit HTTP response code."
|
|
"| error - an error message, or nil if no error."
|
|
"| method - GET, HEAD, or POST"
|
|
"| path - the url-decoded path, eg, '/index.html'"
|
|
"| params - a table of url-decoded URL parameters"
|
|
"| content - the content, as a string (POST only)"
|
|
"| mimetype - the mime type of the content (POST only)"
|
|
"|"
|
|
"|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_commlength - Total bytes in the communication."
|
|
"|"
|
|
"|None of the dbg fields is needed to understand the request."
|
|
"|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."
|
|
"|"
|
|
"|When the engine is functioning as a webserver, bad requests "
|
|
"|are never passed to lua. Therefore, a request that is passed "
|
|
"|to lua will always contain status=200 and error=nil. However, "
|
|
"|when debugging server requests using http.serverrequest, "
|
|
"|it is possible to see certain other errors:"
|
|
"|"
|
|
"| 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"
|
|
"| 413 (payload too large) - we refuse to download something so big"
|
|
"| 425 (can't resume) - reloaded a save game with a pending request"
|
|
"|"
|
|
"|Error messages that are generated locally consist of "
|
|
"|the standard message (eg, 'bad request') followed by more "
|
|
"|detailed information."
|
|
"|"
|
|
"|This routine, http.serverrequest, generates a request by parsing"
|
|
"|an actual HTTP request string. This is for debugging only.") {
|
|
LuaArg text;
|
|
LuaRet tab;
|
|
LuaStack LS(L, text, tab);
|
|
HttpParser parser;
|
|
parser.parse_request(LS.ckstring(text), true);
|
|
parser.store(LS, tab);
|
|
return LS.result();
|
|
}
|
|
|
|
LuaDefine(http_validmime, "(mt)", "") {
|
|
LuaArg str;
|
|
LuaRet ok;
|
|
LuaStack LS(L, str, ok);
|
|
LS.set(ok, valid_mime_type(LS.ckstring(str)));
|
|
return LS.result();
|
|
}
|