stream.c

/*= -*- c-basic-offset: 4; indent-tabs-mode: nil; -*-
 *
 * librsync -- dynamic caching and delta update in HTTP
 * 
 * Copyright (C) 2000, 2001 by Martin Pool <mbp@sourcefrog.net>
 * 
 * This program is free software; you can redistribute it and/or
 * modify it under the terms of the GNU Lesser General Public License
 * as published by the Free Software Foundation; either version 2.1 of
 * the License, or (at your option) any later version.
 * 
 * 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
 * Lesser General Public License for more details.
 * 
 * You should have received a copy of the GNU Lesser General Public
 * License along with this program; if not, write to the Free Software
 * Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
 */

                     /*
                      * Programming languages should be designed not
                      * by piling feature on top of feature, but by
                      * removing the weaknesses and restrictions that
                      * make additional features appear necessary.
                      *    -- Revised^5 Report on Scheme
                      */


/*
 * OK, so I'll admit IO here is a little complex.  The most important
 * player here is the stream, which is an object for managing filter
 * operations.  It has both input and output sides, both of which is
 * just a (pointer,len) pair into a buffer provided by the client.
 * The code controlling the stream handles however much data it wants,
 * and the client provides or accepts however much is convenient.
 *
 * At the same time as being friendly to the client, we also try to be
 * very friendly to the internal code.  It wants to be able to ask for
 * arbitrary amounts of input or output and get it without having to
 * keep track of partial completion.  So there are functions which
 * either complete, or queue whatever was not sent and return
 * RS_BLOCKED.
 *
 * The output buffer is a little more clever than simply a data
 * buffer.  Instead it knows that we can send either literal data, or
 * data copied through from the input of the stream.
 *
 * In buf.c you will find functions that then map buffers onto stdio
 * files.
 *
 * So on return from an encoding function, either the input or the
 * output or possibly both will have no more bytes available.
 */

/*
 * Manage librsync streams of IO.  See scoop.c and tube.c for related
 * code for input and output respectively.
 *
 * librsync never does IO or memory allocation, but relies on the
 * caller.  This is very nice for integration, but means that we have
 * to be fairly flexible as to when we can `read' or `write' stuff
 * internally.
 *
 * librsync basically does two types of IO.  It reads network integers
 * of various lengths which encode command and control information
 * such as versions and signatures.  It also does bulk data transfer.
 *
 * IO of network integers is internally buffered, because higher
 * levels of the code need to see them transmitted atomically: it's no
 * good to read half of a uint32.  So there is a small and fixed
 * length internal buffer which accumulates these.  Unlike previous
 * versions of the library, we don't require that the caller hold the
 * start until the whole thing has arrived, which guarantees that we
 * can always make progress.
 *
 * On each call into a stream iterator, it should begin by trying to
 * flush output.  This may well use up all the remaining stream space,
 * in which case nothing else can be done.
 */

/* TODO: Return errors rather than aborting if something goes wrong.  */


#include <config.h>

#include <assert.h>
#include <stdlib.h>
#include <string.h>
#include <stdio.h>

#include "librsync.h"
#include "stream.h"
#include "util.h"
#include "trace.h"


/**
 * \brief Copy up to \p max_len bytes from input of \b stream to its output.
 *
 * Return the number of bytes actually copied, which may be less than
 * LEN if there is not enough space in one or the other stream.
 *
 * This always does the copy immediately.  Most functions should call
 * rs_tube_copy() to cause the copy to happen gradually as space
 * becomes available.
 */
int rs_buffers_copy(rs_buffers_t *stream, int max_len)
{
    int len = max_len;
    
    assert(len > 0);

    if ((unsigned) len > stream->avail_in) {
        rs_trace("copy limited to %ld available input bytes",
                 (long) stream->avail_in);
        len = stream->avail_in;
    }


    if ((unsigned) len > stream->avail_out) {
        rs_trace("copy limited to %ld available output bytes",
                 (long) stream->avail_out);
        len = stream->avail_out;
    }

    if (!len)
        return 0;
/*     rs_trace("stream copied chunk of %d bytes", len); */

    memcpy(stream->next_out, stream->next_in, len);
    
    stream->next_out += len;
    stream->avail_out -= len;

    stream->next_in += len;
    stream->avail_in -= len;

    return len;
}


/**
 * Whenever a stream processing function exits, it should have done so
 * because it has either consumed all the input or has filled the
 * output buffer.  This function checks that simple postcondition.
 */
void rs_buffers_check_exit(rs_buffers_t const *stream)
{
    assert(stream->avail_in == 0  ||  stream->avail_out == 0);
}