integration/luprex/core/cpp/http.hpp

/////////////////////////////////////////////////////////
//
// HTTP Implementation.
//
// This is a fairly limited implementation of HTTP.
//
// It only supports HTTP 1.1
//
// It only supports GET, POST, and HEAD.
//
/////////////////////////////////////////////////////////

#ifndef HTTP_HPP
#define HTTP_HPP

#include "eng-malloc.hpp"
#include "wrap-string.hpp"
#include "wrap-vector.hpp"
#include "wrap-map.hpp"
#include "luastack.hpp"
#include "streambuffer.hpp"
#include "drivenengine.hpp"
#include <ostream>

using UrlParameters = eng::map<eng::string, eng::string>;

class HttpClientRequest : public eng::nevernew {    
private:
    // Request IDs.
    int64_t request_id_;
    int64_t place_id_;
    int64_t thread_id_;
    
    // If the request contains an error, the error
    // message is stored here.
    eng::string error_;

    // If true, verify the server's certificate.
    // True is the default.
    bool verify_certificate_;

    // Method: GET, HEAD, or POST.  Must be all-caps.
    eng::string method_;

    // The hostname. Not yet url-encoded.
    eng::string host_;

    // Port number.
    int port_;

    // The path.  Not yet url-encoded.  Can not include URL parameters.
    eng::string path_;
    
    // URL parameters to append to the path.  Not yet url-encoded.
    UrlParameters params_;

    // The mime type of the content, only for POST requests.
    eng::string mime_type_;

    // The content as a string, only for POST requests.
    eng::string content_;

private:
    void fail(std::string_view error);
    void send_internal(StreamBuffer *target, bool debug_string) const;

public:
    // Construct an empty HTTP request.
    // All of the fields have empty values.
    HttpClientRequest();
    
    // Get request-related fields.
    //
    const eng::string &error() const { return error_; }
    bool verify_certificate() const { return verify_certificate_; }
    const eng::string &method() const { return method_; }
    const eng::string &host() const { return host_; }
    int port() const { return port_; }
    const eng::string &path() const { return path_; }
    const eng::string &mime_type() const { return mime_type_; }
    const eng::string &content() const { return content_; }
    
    // Populate an request-related fields one piece at a time.
    // If you pass an invalid value, or if the field is
    // already set, the routine will generate an error message
    // and store it in the error field.  In that case, the set
    // will not happen.  If there's already an error in the error
    // field, it will not be overwritten.
    //
    void set_verify_certificate(bool flag);
    void set_method(const eng::string &method);
    void set_host(const eng::string &host);
    void set_port(int port);
    void set_path(std::string_view path);
    void set_param(const eng::string &key, const eng::string &value);
    void set_url(std::string_view url);
    void set_mime_type(const eng::string &mime_type);
    void set_content(const eng::string &content);
 
    void set_verify_certificate(LuaStack &LS, LuaSlot val);
    void set_method(LuaStack &LS, LuaSlot val);
    void set_host(LuaStack &LS, LuaSlot val);
    void set_port(LuaStack &LS, LuaSlot val);
    void set_path(LuaStack &LS, LuaSlot path);
    void set_param(LuaStack &LS, LuaSlot key, LuaSlot val);
    void set_params(LuaStack &LS, LuaSlot tab);
    void set_url(LuaStack &LS, LuaSlot val);
    void set_mime_type(LuaStack &LS, LuaSlot val);
    void set_content(LuaStack &LS, LuaSlot val);

    // Set default values for method and port.
    // This must be done after setting regular values.
    void set_defaults();

    // Populate request-related fields from a Lua table.
    void set_config(LuaStack &LS0, LuaSlot tab);

    // Get or Set the request IDs.
    // This class does just stores these.
    int64_t request_id() const { return request_id_; }
    int64_t place_id() const { return place_id_; }
    int64_t thread_id() const { return thread_id_; }
    void set_request_id(int64_t request_id) { request_id_ = request_id; }
    void set_place_id(int64_t place_id) { place_id_ = place_id; }
    void set_thread_id(int64_t thread_id) { thread_id_ = thread_id; }
    
    // Get the network target, eg, "cert:host:port"
    eng::string target() const;

    // Verify that the request is error free and that
    // defaults have been set.
    eng::string check() const;

    // Put the request into the stream, assuming HTTP/1.1
    void send(StreamBuffer *target) const { send_internal(target, false); }

    // Serialize and deserialize.
    void serialize(StreamBuffer *sb) const;
    void deserialize(StreamBuffer *sb);

    // Get the request as a debug string.
    eng::string DebugString();
};

// HttpParser is used for parsing both requests and responses.
// Stores a status code, an error message, headers, and content.
//
class HttpParser : public eng::nevernew {
protected:
    // True if this is for parsing a request.
    // This is only used for generating error messages.
    bool is_request_;

    // The status code, a 3-digit number.
    int status_;

    // If the communication contains an error, the error can
    // be stored here.  In the event of a successful communication,
    // this should always be empty string.
    eng::string error_;

    // Only if content-length header present, otherwise, -1.
    int64_t content_length_;

    // If empty, it means there was no transfer-encoding header.
    eng::string transfer_encoding_;

    // If empty, it means there was no content-encoding header.
    eng::string content_encoding_;

    // Only if location header present.
    eng::string location_;

    // MIME type of the content.
    eng::string mime_type_;

    // Charset of the content before the content was translated to utf-8.
    eng::string charset_;

    // The method, GET, HEAD, or POST. (only when parsing requests)
    eng::string method_;

    // The URL path, not url-encoded. (only when parsing requests)
    eng::string path_;

    // The URL parameters, not url-encoded (only when parsing requests)
    UrlParameters params_;

    // The content as string.  If it's text, it's been translated to utf-8.
    eng::string content_;

    // The length in bytes of the entire communication.
    // If the response is complete, but the comm_length_
    // is zero, it means we couldn't find the end of 
    // the request.
    int comm_length_;

protected:
    // Store a message indicating that we haven't received enough
    // bytes yet.  If the connection is closed and we still haven't
    // received enough bytes, that's a fatal error.
    //
    void incomplete(bool closed);

    // Store a message indicating that we couldn't parse because
    // something was syntactically malformed, ie, not valid HTTP.
    //
    void syntax(std::string_view detail);

    // Store a message indicating that the content was too large
    // and we refused to download it.
    //
    void oversized();

    // If the status is not in the range 200-299 (OK),
    // then discard the content.
    void clear_content_on_error();

    // Parse a response status line, such as "HTTP/1.1 200 OK"
    // returns false if we couldn't parse the whole line.
    //
    bool parse_status_line(std::string_view &view, bool closed);

    // Parse a request line, such as "GET /index.html HTTP/1.1"
    // returns false if we couldn't parse the whole line.
    //
    bool parse_request_line(std::string_view &view, bool closed);

    // Parse specific headers.
    //
    void parse_content_encoding(std::string_view value);
    void parse_content_length(std::string_view value);
    void parse_content_type(std::string_view value);
    void parse_location(std::string_view value);
    void parse_transfer_encoding(std::string_view value);

    // Parse a single response header.  Most headers are ignored.
    // If the header contains an error, the error is stored.
    //
    void parse_header(std::string_view header, std::string_view value);

    // Parse all of the headers, including the blank line after
    // the headers.  Return true if the view is at the end of the headers.
    //
    bool parse_headers(std::string_view &view, bool closed);
    
    // parse the content, for different transfer encodings.
    // Returns true if the view is at the end of the content.
    //
    bool parse_content_basic(std::string_view &view, bool closed);
    bool parse_content_chunked(std::string_view &view, bool closed);

    // Parse the content.  Return true if the view is at the end of the
    // content.
    //
    // - Uses the appropriate transfer-encoding specified in the headers.
    // - Decompresses the content using the appropriate content-encoding.
    // - Guesses a mimetype and charset if one was not specified in headers.
    // - If it's text, converts the charset to utf-8.
    //
    bool parse_content(std::string_view &view, bool closed);

    // Get or set the communication length.
    //
    int comm_length() const { return comm_length_; }
    void set_comm_length(int len) { comm_length_ = len; }

    // Construct a blank parser.
    //
    HttpParser();

    // Store the parsed fields into a lua table.
    //
    void store_parsed(LuaStack &LS, LuaSlot tab) const;

    // Emit all the parser fields as a debug string.
    //
    void parser_generate_debug_string(std::ostream &oss) const;

public:
    // The parser will not try to parse content longer than this.
    //
    const int64_t MAX_CONTENT_LENGTH = 1000000;

    // Store a status code and an error message, and clear the content.
    // This is generally used when the client detects an error,
    // such as a DNS lookup fail, a connection failed, an SSL negotiation
    // failed, or the like.
    //
    void fail(int status, std::string_view error);

    // Return true if the communication was complete.
    //
    bool complete() const { return status_ != 0; }
};

class HttpClientResponse : public HttpParser {
private:
    // Most of the data is stored in the HttpParser.
    // That includes the status code, error, headers, and content.
    
    // The request ID.
    int64_t request_id_;

    // The length in bytes of the entire communication.
    // If the response is complete, but the comm_length_
    // is zero, it means we couldn't find the end of 
    // the request.
    //
    int comm_length_;

public:
    // Construct a blank response.
    HttpClientResponse();

    // Parse the HTTP response.  The closed flag is to be set to true if the
    // remote has closed the connection.
    //
    // If the request is incomplete, generates a status code of zero. In that
    // case, loading more data from the server might improve the situation.
    //
    // If successful, indicates how much of the text was consumed by
    // setting comm_length_.
    //
    void parse(std::string_view text, bool closed);

    // Get or Set the request ID.
    // This class does nothing with the request ID, it just stores it.
    //
    int64_t request_id() const { return request_id_; }
    void set_request_id(int64_t v) { request_id_ = v; }
    
    // Convert the HTTP response to a lua table.
    //
    void store(LuaStack &LS, LuaSlot tab) const;

    // Convert to a debug string.
    //
    eng::string DebugString() const;

    // Synthesize an error response and store it in lua.
    //
    static void store_fail(LuaStack &LS, LuaSlot tab, int status, std::string_view error);
};

class HttpServerRequest : public HttpParser {
private:
    // The length in bytes of the entire communication.
    // If the request is complete, but the comm_length_
    // is zero, it means we couldn't find the end of 
    // the request.
    //
    int comm_length_;

public:
    // Construct a blank request.
    //
    HttpServerRequest();

    // Parse the HTTP request.  The closed flag is to be set to true if the
    // remote has closed the connection.
    //
    // If the request is incomplete, generates a status code of zero. In that
    // case, loading more data from the server might improve the situation.
    //
    // If successful, indicates how much of the text was consumed by
    // setting comm_length.
    //
    void parse(std::string_view text, bool closed);
    
    // Convert the HTTP request to a lua table.
    //
    void store(LuaStack &LS, LuaSlot tab) const;

    // Convert to a debug string.
    //
    eng::string DebugString() const;
};

class HttpClientRequestMap : public eng::map<int64_t, HttpClientRequest> {
public:
    void serialize(StreamBuffer *sb) const;
    void deserialize(StreamBuffer *sb);
};

using HttpClientResponseVec = eng::vector<HttpClientResponse>;

// This class associates an HTTP request to an actual communication
// channel that is executing that request.

class HttpClientChannel {
public:
    SharedChannel channel_;
    int64_t parsed_bytes_;

    HttpClientChannel() : parsed_bytes_(0) {}
};

using HttpClientChannelMap = eng::map<int64_t, HttpClientChannel>;

#endif // HTTP_HPP