src/lib/istream-crlf.c

	istream-crlf.c revision 6006aa91665cec400dcec3c40176361a84305aea
5f5870385cff47efd2f58e7892f251cf13761528Timo Sirainen/* Copyright (c) 2007-2008 Dovecot authors, see the included COPYING file */
596f582a422eddef951ee445f9a0ed3d4bf14ff8Timo Sirainen
60bdf7c0422caeb08cc615c60b659d4d1ff30614Timo Sirainen#include "lib.h"
596f582a422eddef951ee445f9a0ed3d4bf14ff8Timo Sirainen#include "istream-internal.h"
596f582a422eddef951ee445f9a0ed3d4bf14ff8Timo Sirainen#include "istream-crlf.h"
596f582a422eddef951ee445f9a0ed3d4bf14ff8Timo Sirainen
596f582a422eddef951ee445f9a0ed3d4bf14ff8Timo Sirainenstruct crlf_istream {
596f582a422eddef951ee445f9a0ed3d4bf14ff8Timo Sirainen    struct istream_private istream;
596f582a422eddef951ee445f9a0ed3d4bf14ff8Timo Sirainen
596f582a422eddef951ee445f9a0ed3d4bf14ff8Timo Sirainen    unsigned int pending_cr:1;
596f582a422eddef951ee445f9a0ed3d4bf14ff8Timo Sirainen    unsigned int last_cr:1;
596f582a422eddef951ee445f9a0ed3d4bf14ff8Timo Sirainen};
596f582a422eddef951ee445f9a0ed3d4bf14ff8Timo Sirainen
596f582a422eddef951ee445f9a0ed3d4bf14ff8Timo Sirainenstatic void i_stream_crlf_destroy(struct iostream_private *stream)
596f582a422eddef951ee445f9a0ed3d4bf14ff8Timo Sirainen{
596f582a422eddef951ee445f9a0ed3d4bf14ff8Timo Sirainen    struct crlf_istream *cstream = (struct crlf_istream *)stream;
596f582a422eddef951ee445f9a0ed3d4bf14ff8Timo Sirainen
596f582a422eddef951ee445f9a0ed3d4bf14ff8Timo Sirainen    i_free(cstream->istream.w_buffer);
596f582a422eddef951ee445f9a0ed3d4bf14ff8Timo Sirainen    i_stream_unref(&cstream->istream.parent);
596f582a422eddef951ee445f9a0ed3d4bf14ff8Timo Sirainen}
596f582a422eddef951ee445f9a0ed3d4bf14ff8Timo Sirainen
596f582a422eddef951ee445f9a0ed3d4bf14ff8Timo Sirainenstatic void
596f582a422eddef951ee445f9a0ed3d4bf14ff8Timo Siraineni_stream_crlf_set_max_buffer_size(struct iostream_private *stream,
596f582a422eddef951ee445f9a0ed3d4bf14ff8Timo Sirainen                   size_t max_size)
596f582a422eddef951ee445f9a0ed3d4bf14ff8Timo Sirainen{
596f582a422eddef951ee445f9a0ed3d4bf14ff8Timo Sirainen    struct crlf_istream *cstream = (struct crlf_istream *)stream;
596f582a422eddef951ee445f9a0ed3d4bf14ff8Timo Sirainen
596f582a422eddef951ee445f9a0ed3d4bf14ff8Timo Sirainen    cstream->istream.max_buffer_size = max_size;
596f582a422eddef951ee445f9a0ed3d4bf14ff8Timo Sirainen    i_stream_set_max_buffer_size(cstream->istream.parent, max_size);
596f582a422eddef951ee445f9a0ed3d4bf14ff8Timo Sirainen}
596f582a422eddef951ee445f9a0ed3d4bf14ff8Timo Sirainen
596f582a422eddef951ee445f9a0ed3d4bf14ff8Timo Sirainenstatic int i_stream_crlf_read_common(struct crlf_istream *cstream)
596f582a422eddef951ee445f9a0ed3d4bf14ff8Timo Sirainen{
596f582a422eddef951ee445f9a0ed3d4bf14ff8Timo Sirainen    struct istream_private *stream = &cstream->istream;
5cc9289a900f5d40d126ca663790dca2a7456416Timo Sirainen    const unsigned char *data;
5cc9289a900f5d40d126ca663790dca2a7456416Timo Sirainen    size_t size;
5cc9289a900f5d40d126ca663790dca2a7456416Timo Sirainen    ssize_t ret;
5cc9289a900f5d40d126ca663790dca2a7456416Timo Sirainen
5cc9289a900f5d40d126ca663790dca2a7456416Timo Sirainen    data = i_stream_get_data(stream->parent, &size);
596f582a422eddef951ee445f9a0ed3d4bf14ff8Timo Sirainen    if (size == 0) {
596f582a422eddef951ee445f9a0ed3d4bf14ff8Timo Sirainen        ret = i_stream_read(stream->parent);
596f582a422eddef951ee445f9a0ed3d4bf14ff8Timo Sirainen        if (ret <= 0 && (ret != -2 || stream->skip == 0)) {
596f582a422eddef951ee445f9a0ed3d4bf14ff8Timo Sirainen            stream->istream.stream_errno =
596f582a422eddef951ee445f9a0ed3d4bf14ff8Timo Sirainen                stream->parent->stream_errno;
596f582a422eddef951ee445f9a0ed3d4bf14ff8Timo Sirainen            stream->istream.eof = stream->parent->eof;
596f582a422eddef951ee445f9a0ed3d4bf14ff8Timo Sirainen            return ret;
596f582a422eddef951ee445f9a0ed3d4bf14ff8Timo Sirainen        }
596f582a422eddef951ee445f9a0ed3d4bf14ff8Timo Sirainen        data = i_stream_get_data(stream->parent, &size);
596f582a422eddef951ee445f9a0ed3d4bf14ff8Timo Sirainen        i_assert(size != 0);
596f582a422eddef951ee445f9a0ed3d4bf14ff8Timo Sirainen    }
596f582a422eddef951ee445f9a0ed3d4bf14ff8Timo Sirainen
596f582a422eddef951ee445f9a0ed3d4bf14ff8Timo Sirainen    if (!i_stream_get_buffer_space(stream, size, NULL))
596f582a422eddef951ee445f9a0ed3d4bf14ff8Timo Sirainen        return -2;
596f582a422eddef951ee445f9a0ed3d4bf14ff8Timo Sirainen    return 1;
596f582a422eddef951ee445f9a0ed3d4bf14ff8Timo Sirainen}
596f582a422eddef951ee445f9a0ed3d4bf14ff8Timo Sirainen
596f582a422eddef951ee445f9a0ed3d4bf14ff8Timo Sirainenstatic ssize_t i_stream_crlf_read_crlf(struct istream_private *stream)
596f582a422eddef951ee445f9a0ed3d4bf14ff8Timo Sirainen{
596f582a422eddef951ee445f9a0ed3d4bf14ff8Timo Sirainen    struct crlf_istream *cstream = (struct crlf_istream *)stream;
596f582a422eddef951ee445f9a0ed3d4bf14ff8Timo Sirainen    const unsigned char *data;
9a1a4c2ca1b9a8bbadd38224b15a364c78cb9938Timo Sirainen    size_t i, dest, size;
9a1a4c2ca1b9a8bbadd38224b15a364c78cb9938Timo Sirainen    ssize_t ret;
9a1a4c2ca1b9a8bbadd38224b15a364c78cb9938Timo Sirainen
9a1a4c2ca1b9a8bbadd38224b15a364c78cb9938Timo Sirainen    ret = i_stream_crlf_read_common(cstream);
596f582a422eddef951ee445f9a0ed3d4bf14ff8Timo Sirainen    if (ret <= 0)
596f582a422eddef951ee445f9a0ed3d4bf14ff8Timo Sirainen        return ret;
596f582a422eddef951ee445f9a0ed3d4bf14ff8Timo Sirainen
596f582a422eddef951ee445f9a0ed3d4bf14ff8Timo Sirainen    data = i_stream_get_data(stream->parent, &size);
596f582a422eddef951ee445f9a0ed3d4bf14ff8Timo Sirainen
596f582a422eddef951ee445f9a0ed3d4bf14ff8Timo Sirainen    /* @UNSAFE: add missing CRs */
596f582a422eddef951ee445f9a0ed3d4bf14ff8Timo Sirainen    dest = stream->pos;
596f582a422eddef951ee445f9a0ed3d4bf14ff8Timo Sirainen    for (i = 0; i < size && dest < stream->buffer_size; i++) {
596f582a422eddef951ee445f9a0ed3d4bf14ff8Timo Sirainen        if (data[i] == '\n') {
596f582a422eddef951ee445f9a0ed3d4bf14ff8Timo Sirainen            if (i == 0) {
596f582a422eddef951ee445f9a0ed3d4bf14ff8Timo Sirainen                if (!cstream->last_cr)
596f582a422eddef951ee445f9a0ed3d4bf14ff8Timo Sirainen                    stream->w_buffer[dest++] = '\r';
596f582a422eddef951ee445f9a0ed3d4bf14ff8Timo Sirainen            } else {
596f582a422eddef951ee445f9a0ed3d4bf14ff8Timo Sirainen                if (data[i-1] != '\r')
596f582a422eddef951ee445f9a0ed3d4bf14ff8Timo Sirainen                    stream->w_buffer[dest++] = '\r';
596f582a422eddef951ee445f9a0ed3d4bf14ff8Timo Sirainen            }
596f582a422eddef951ee445f9a0ed3d4bf14ff8Timo Sirainen            if (dest == stream->buffer_size)
596f582a422eddef951ee445f9a0ed3d4bf14ff8Timo Sirainen                break;
596f582a422eddef951ee445f9a0ed3d4bf14ff8Timo Sirainen        }
596f582a422eddef951ee445f9a0ed3d4bf14ff8Timo Sirainen        stream->w_buffer[dest++] = data[i];
596f582a422eddef951ee445f9a0ed3d4bf14ff8Timo Sirainen    }
596f582a422eddef951ee445f9a0ed3d4bf14ff8Timo Sirainen    cstream->last_cr = stream->w_buffer[dest-1] == '\r';
596f582a422eddef951ee445f9a0ed3d4bf14ff8Timo Sirainen    i_stream_skip(stream->parent, i);
596f582a422eddef951ee445f9a0ed3d4bf14ff8Timo Sirainen
596f582a422eddef951ee445f9a0ed3d4bf14ff8Timo Sirainen    ret = dest - stream->pos;
596f582a422eddef951ee445f9a0ed3d4bf14ff8Timo Sirainen    i_assert(ret != 0);
596f582a422eddef951ee445f9a0ed3d4bf14ff8Timo Sirainen    stream->pos = dest;
596f582a422eddef951ee445f9a0ed3d4bf14ff8Timo Sirainen    return ret;
596f582a422eddef951ee445f9a0ed3d4bf14ff8Timo Sirainen}
596f582a422eddef951ee445f9a0ed3d4bf14ff8Timo Sirainen
596f582a422eddef951ee445f9a0ed3d4bf14ff8Timo Sirainenstatic ssize_t i_stream_crlf_read_lf(struct istream_private *stream)
596f582a422eddef951ee445f9a0ed3d4bf14ff8Timo Sirainen{
596f582a422eddef951ee445f9a0ed3d4bf14ff8Timo Sirainen    struct crlf_istream *cstream = (struct crlf_istream *)stream;
596f582a422eddef951ee445f9a0ed3d4bf14ff8Timo Sirainen    const unsigned char *data;
596f582a422eddef951ee445f9a0ed3d4bf14ff8Timo Sirainen    size_t i, dest, size;
    ssize_t ret;
    int diff;

    ret = i_stream_crlf_read_common(cstream);
    if (ret <= 0)
        return ret;

    data = i_stream_get_data(stream->parent, &size);

    /* @UNSAFE */
    dest = stream->pos;
    if (data[0] == '\n') {
        stream->w_buffer[dest++] = '\n';
        cstream->pending_cr = FALSE;
    } else {
        if (cstream->pending_cr) {
            /* CR without LF */
            stream->w_buffer[dest++] = '\r';
            if (dest == stream->buffer_size) {
                cstream->pending_cr = FALSE;
                return 1;
            }
        }
        if (data[0] != '\r')
            stream->w_buffer[dest++] = data[0];
    }

    diff = -1;
    for (i = 1; i < size && dest < stream->buffer_size; i++) {
        if (data[i] == '\r') {
            if (data[i-1] != '\r')
                continue;
        } else if (data[i-1] == '\r' && data[i] != '\n') {
            stream->w_buffer[dest++] = '\r';
            if (dest == stream->buffer_size) {
                diff = 0;
                break;
            }
        }

        stream->w_buffer[dest++] = data[i];
    }
    cstream->pending_cr = data[i+diff] == '\r';
    i_stream_skip(stream->parent, i);

    ret = dest - stream->pos;
    if (ret == 0) {
        i_assert(cstream->pending_cr && size == 1);
        return i_stream_crlf_read_lf(stream);
    }
    stream->pos = dest;
    return ret;
}

static void
i_stream_crlf_seek(struct istream_private *stream,
           uoff_t v_offset, bool mark ATTR_UNUSED)
{
    size_t available;

    if (stream->istream.v_offset > v_offset)
        i_panic("crlf-istream: seeking unsupported currently");

    while (stream->istream.v_offset < v_offset) {
        (void)i_stream_crlf_read_crlf(stream);

        available = stream->pos - stream->skip;
        if (available == 0) {
            stream->istream.stream_errno = ESPIPE;
            return;
        }
        if (available <= v_offset - stream->istream.v_offset)
            i_stream_skip(&stream->istream, available);
        else {
            i_stream_skip(&stream->istream,
                      v_offset - stream->istream.v_offset);
        }
    }
}

static const struct stat *
i_stream_crlf_stat(struct istream_private *stream, bool exact)
{
    return i_stream_stat(stream->parent, exact);
}

static struct istream *
i_stream_create_crlf_full(struct istream *input, bool crlf)
{
    struct crlf_istream *cstream;

    i_stream_ref(input);

    cstream = i_new(struct crlf_istream, 1);
    cstream->istream.max_buffer_size = input->real_stream->max_buffer_size;

    cstream->istream.iostream.destroy = i_stream_crlf_destroy;
    cstream->istream.iostream.set_max_buffer_size =
        i_stream_crlf_set_max_buffer_size;

    cstream->istream.read = crlf ? i_stream_crlf_read_crlf :
        i_stream_crlf_read_lf;
    cstream->istream.seek = i_stream_crlf_seek;
    cstream->istream.stat = i_stream_crlf_stat;

    cstream->istream.istream.blocking = input->blocking;
    cstream->istream.istream.seekable = FALSE;
    return i_stream_create(&cstream->istream, input,
                   i_stream_get_fd(input));
}

struct istream *i_stream_create_crlf(struct istream *input)
{
    return i_stream_create_crlf_full(input, TRUE);
}

struct istream *i_stream_create_lf(struct istream *input)
{
    return i_stream_create_crlf_full(input, FALSE);
}