src/lib/AbstractParser.h
author František Kučera <franta-hg@frantovo.cz>
Sun, 25 Jul 2021 11:53:55 +0200
branchv_0
changeset 2 7128fabeede0
parent 1 68a281aefa76
permissions -rw-r--r--
FreeformASN1ContentHandler: basic ASN.1 tree in a relation

/**
 * Relational pipes
 * Copyright © 2021 František Kučera (Frantovo.cz, GlobalCode.info)
 *
 * This program is free software: you can redistribute it and/or modify
 * it under the terms of the GNU General Public License as published by
 * the Free Software Foundation, version 3 of the License.
 *
 * This program is distributed in the hope that it will be useful,
 * but WITHOUT ANY WARRANTY; without even the implied warranty of
 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
 * GNU General Public License for more details.
 *
 * You should have received a copy of the GNU General Public License
 * along with this program. If not, see <http://www.gnu.org/licenses/>.
 */
#pragma once

#include <cstring>
#include <cstdint>

namespace relpipe {
namespace in {
namespace asn1 {
namespace lib {

class AbstractParserImpl;

/**
 * Event-driven parser that consumes sequence of byte buffers (chunked stream)
 * and produces some output (implementation specific).
 * 
 * The flow is controlled from outside: incomming data pieces are passed to the parser through the write() method as they come.
 * 
 * From the child class perspective (implementation of particular parser)
 * data pieces are obtained through the read() method from the internal buffer maintained by AbstractParser.
 */
class AbstractParser {
private:
	friend AbstractParserImpl;
	AbstractParserImpl* implementation;
	void rollback();
public:
	virtual ~AbstractParser();

	/**
	 * Submit a part of incomming data to the parser.
	 * When possible, such data are usually processed synchronously during this method call.
	 * If not (incomplete TLV/AVP/PDU/token/etc.) this part is appended to the internal buffer maintained by AbstractParser
	 * and processed in some next cycle or – in worst case – during the close() call.
	 * 
	 * @param buffer readable memory
	 * @param length size of the buffer
	 */
	void write(const uint8_t* buffer, const size_t length);

	/**
	 * Finalize the parsing process.
	 * After calling this method, all data from AbstractParser buffers should be consumed, parsed and results published.
	 * No write() call is expected after close() and would fail.
	 * However the parser object remains valid and may be used to get some auxiliary information (if supperted by given implementation).
	 */
	virtual void close();
protected:
	AbstractParser();

	/**
	 * May be thrown from the update() method in order to cancel currenty cycle and do explicit rollback.
	 * Same data will be processed in the next cycle.
	 */
	class ExplicitRollbackException {
		// TODO: common super-class for exceptions, hidden implementation
	};

	/**
	 * May be called from the update() method in order to explicitly confirm that read data was successfully processed.
	 * Such data will not be processed again during the next cycle (even if ReadBufferUnderflowException occurs later in this cycle).
	 
	 * Explicit commit() call is useful when we did some demanding work (so we are not willing to do it again in case of ReadBufferUnderflowException);
	 * and is necessary when we already published some outputs or did other non-idempotent operation or caused some other significant side effects.
	 
	 * If there is no commit() called and update() just finishes, commit() is called implicitly.
	 * 
	 * Note: There is no accessible rollback() method – throw ExplicitRollbackException instead.
	 */
	void commit();

	/**
	 * Fill the buffer with incomming data of given length (exactly).
	 * 
	 * If there are not enough data available, ReadBufferUnderflowException is thrown.
	 * This exception should not be caught in the child class – it should propagate back to the AbstractParser
	 * where it causes rollback(). In such case, the same data will be available during next update() cycle.
	 * 
	 * @param buffer writable memory
	 * @param length size of the buffer
	 */
	void read(uint8_t* buffer, const size_t length);

	/**
	 * Like read(), but does not update the marks (buffer positions), so it can be called again and again with same result.
	 * 
	 * @param buffer writable memory
	 * @param length size of the buffer
	 */
	void peek(uint8_t* buffer, const size_t length);

	/**
	 * @return total number of bytes that were successfully read
	 */
	size_t getBytesRead();

	/**
	 * Reads input from buffers using read(), parses data and usually emits the result (e.g. to a handler/listener).
	 * Is specific for particular format/parser.
	 * 
	 * This method is called at least once at the end of the stream (with whole stream content in read buffers).
	 * Usually it is called more often, by default: after each write (with just already received data part in read buffers).
	 */
	virtual void update() = 0;
};

}
}
}
}