index/mbox/istream-raw-mbox.c

	istream-raw-mbox.c revision 7f773564b94e6054a40d3785cb63c29f1e4d4dee
5e0ce63bb65db34d7f48b34bbb5545fa791781c4Timo Sirainen/* Copyright (C) 2003 Timo Sirainen */
ecc81625167ed96c04c02aa190a1ea5baa65b474Timo Sirainen
5254d77805cd35b9356d072ba325c356c43b0d51Timo Sirainen#include "lib.h"
5254d77805cd35b9356d072ba325c356c43b0d51Timo Sirainen#include "buffer.h"
ecc81625167ed96c04c02aa190a1ea5baa65b474Timo Sirainen#include "istream-internal.h"
07e4875d250e7a7157cd99132aafc773cf3cdf83Timo Sirainen#include "istream-raw-mbox.h"
ecc81625167ed96c04c02aa190a1ea5baa65b474Timo Sirainen#include "mbox-from.h"
ecc81625167ed96c04c02aa190a1ea5baa65b474Timo Sirainen
ecc81625167ed96c04c02aa190a1ea5baa65b474Timo Sirainenstruct raw_mbox_istream {
ecc81625167ed96c04c02aa190a1ea5baa65b474Timo Sirainen    struct _istream istream;
ecc81625167ed96c04c02aa190a1ea5baa65b474Timo Sirainen
ecc81625167ed96c04c02aa190a1ea5baa65b474Timo Sirainen    time_t received_time, next_received_time;
ecc81625167ed96c04c02aa190a1ea5baa65b474Timo Sirainen    char *sender, *next_sender;
ecc81625167ed96c04c02aa190a1ea5baa65b474Timo Sirainen
ecc81625167ed96c04c02aa190a1ea5baa65b474Timo Sirainen    uoff_t from_offset, hdr_offset, body_offset, mail_size;
c0435c854a0e7246373b9752d163095cc4fbe985Timo Sirainen    struct istream *input;
c0435c854a0e7246373b9752d163095cc4fbe985Timo Sirainen    uoff_t input_peak_offset;
ecc81625167ed96c04c02aa190a1ea5baa65b474Timo Sirainen
ecc81625167ed96c04c02aa190a1ea5baa65b474Timo Sirainen    unsigned int corrupted:1;
ecc81625167ed96c04c02aa190a1ea5baa65b474Timo Sirainen    unsigned int eom:1;
ecc81625167ed96c04c02aa190a1ea5baa65b474Timo Sirainen    unsigned int next_eof:1;
ecc81625167ed96c04c02aa190a1ea5baa65b474Timo Sirainen};
c0435c854a0e7246373b9752d163095cc4fbe985Timo Sirainen
ecc81625167ed96c04c02aa190a1ea5baa65b474Timo Sirainenstatic void _close(struct _iostream *stream __attr_unused__)
c0435c854a0e7246373b9752d163095cc4fbe985Timo Sirainen{
ecc81625167ed96c04c02aa190a1ea5baa65b474Timo Sirainen}
c0435c854a0e7246373b9752d163095cc4fbe985Timo Sirainen
c0435c854a0e7246373b9752d163095cc4fbe985Timo Sirainenstatic void _destroy(struct _iostream *stream)
ecc81625167ed96c04c02aa190a1ea5baa65b474Timo Sirainen{
ecc81625167ed96c04c02aa190a1ea5baa65b474Timo Sirainen    struct raw_mbox_istream *rstream = (struct raw_mbox_istream *)stream;
ecc81625167ed96c04c02aa190a1ea5baa65b474Timo Sirainen
c0435c854a0e7246373b9752d163095cc4fbe985Timo Sirainen    i_stream_seek(rstream->input, rstream->istream.istream.v_offset);
ecc81625167ed96c04c02aa190a1ea5baa65b474Timo Sirainen    i_stream_unref(rstream->input);
2c25e1360d4b5cc55eda969a3a7204d950de5a8fTimo Sirainen}
ecc81625167ed96c04c02aa190a1ea5baa65b474Timo Sirainen
ecc81625167ed96c04c02aa190a1ea5baa65b474Timo Sirainenstatic void _set_max_buffer_size(struct _iostream *stream, size_t max_size)
c0435c854a0e7246373b9752d163095cc4fbe985Timo Sirainen{
ecc81625167ed96c04c02aa190a1ea5baa65b474Timo Sirainen    struct raw_mbox_istream *rstream = (struct raw_mbox_istream *)stream;
c0435c854a0e7246373b9752d163095cc4fbe985Timo Sirainen
ecc81625167ed96c04c02aa190a1ea5baa65b474Timo Sirainen    i_stream_set_max_buffer_size(rstream->input, max_size);
ecc81625167ed96c04c02aa190a1ea5baa65b474Timo Sirainen}
ecc81625167ed96c04c02aa190a1ea5baa65b474Timo Sirainen
ecc81625167ed96c04c02aa190a1ea5baa65b474Timo Sirainenstatic void _set_blocking(struct _iostream *stream, int timeout_msecs,
c0435c854a0e7246373b9752d163095cc4fbe985Timo Sirainen              void (*timeout_cb)(void *), void *context)
ecc81625167ed96c04c02aa190a1ea5baa65b474Timo Sirainen{
ecc81625167ed96c04c02aa190a1ea5baa65b474Timo Sirainen    struct raw_mbox_istream *rstream = (struct raw_mbox_istream *)stream;
ecc81625167ed96c04c02aa190a1ea5baa65b474Timo Sirainen
ecc81625167ed96c04c02aa190a1ea5baa65b474Timo Sirainen    i_stream_set_blocking(rstream->input, timeout_msecs,
ecc81625167ed96c04c02aa190a1ea5baa65b474Timo Sirainen                  timeout_cb, context);
ecc81625167ed96c04c02aa190a1ea5baa65b474Timo Sirainen}
ecc81625167ed96c04c02aa190a1ea5baa65b474Timo Sirainen
ecc81625167ed96c04c02aa190a1ea5baa65b474Timo Sirainenstatic int mbox_read_from_line(struct raw_mbox_istream *rstream)
c0435c854a0e7246373b9752d163095cc4fbe985Timo Sirainen{
ecc81625167ed96c04c02aa190a1ea5baa65b474Timo Sirainen    const unsigned char *buf, *p;
c0435c854a0e7246373b9752d163095cc4fbe985Timo Sirainen    char *sender;
ecc81625167ed96c04c02aa190a1ea5baa65b474Timo Sirainen    time_t received_time;
ecc81625167ed96c04c02aa190a1ea5baa65b474Timo Sirainen    size_t pos, line_pos;
ecc81625167ed96c04c02aa190a1ea5baa65b474Timo Sirainen    int skip;
ecc81625167ed96c04c02aa190a1ea5baa65b474Timo Sirainen
ecc81625167ed96c04c02aa190a1ea5baa65b474Timo Sirainen    buf = i_stream_get_data(rstream->input, &pos);
ecc81625167ed96c04c02aa190a1ea5baa65b474Timo Sirainen    i_assert(pos > 0);
ecc81625167ed96c04c02aa190a1ea5baa65b474Timo Sirainen
ecc81625167ed96c04c02aa190a1ea5baa65b474Timo Sirainen    /* from_offset points to "\nFrom ", so unless we're at the beginning
ecc81625167ed96c04c02aa190a1ea5baa65b474Timo Sirainen       of the file, skip the initial \n */
ecc81625167ed96c04c02aa190a1ea5baa65b474Timo Sirainen    skip = rstream->from_offset != 0;
ecc81625167ed96c04c02aa190a1ea5baa65b474Timo Sirainen
ecc81625167ed96c04c02aa190a1ea5baa65b474Timo Sirainen    while ((p = memchr(buf+skip, '\n', pos-skip)) == NULL) {
7ef5ca6fb59a318c821a852ae48a2edbb671d7ddTimo Sirainen        if (i_stream_read(rstream->input) < 0) {
7ef5ca6fb59a318c821a852ae48a2edbb671d7ddTimo Sirainen            /* EOF - shouldn't happen */
ecc81625167ed96c04c02aa190a1ea5baa65b474Timo Sirainen            return -1;
7ef5ca6fb59a318c821a852ae48a2edbb671d7ddTimo Sirainen        }
ecc81625167ed96c04c02aa190a1ea5baa65b474Timo Sirainen        buf = i_stream_get_data(rstream->input, &pos);
ecc81625167ed96c04c02aa190a1ea5baa65b474Timo Sirainen    }
ecc81625167ed96c04c02aa190a1ea5baa65b474Timo Sirainen    line_pos = (size_t)(p - buf);
ecc81625167ed96c04c02aa190a1ea5baa65b474Timo Sirainen
ecc81625167ed96c04c02aa190a1ea5baa65b474Timo Sirainen    if (rstream->from_offset != 0) {
ecc81625167ed96c04c02aa190a1ea5baa65b474Timo Sirainen        buf++;
ecc81625167ed96c04c02aa190a1ea5baa65b474Timo Sirainen        pos--;
ecc81625167ed96c04c02aa190a1ea5baa65b474Timo Sirainen    }
ecc81625167ed96c04c02aa190a1ea5baa65b474Timo Sirainen
2767104d81e97a109f0aa9758792bfa1da325a97Timo Sirainen    /* beginning of mbox */
c06f4017027263cf3a08becc551f5126409e2a83Timo Sirainen    if (memcmp(buf, "From ", 5) != 0 ||
c06f4017027263cf3a08becc551f5126409e2a83Timo Sirainen        mbox_from_parse(buf+5, pos-5, &received_time, &sender) < 0) {
1060afdc2fcdf647dbb3bc11647401f1b44a3a8aTimo Sirainen        /* broken From - should happen only at beginning of
1060afdc2fcdf647dbb3bc11647401f1b44a3a8aTimo Sirainen           file if this isn't a mbox.. */
1060afdc2fcdf647dbb3bc11647401f1b44a3a8aTimo Sirainen        return -1;
c06f4017027263cf3a08becc551f5126409e2a83Timo Sirainen    }
c06f4017027263cf3a08becc551f5126409e2a83Timo Sirainen
c06f4017027263cf3a08becc551f5126409e2a83Timo Sirainen    if (rstream->istream.istream.v_offset == rstream->from_offset) {
c06f4017027263cf3a08becc551f5126409e2a83Timo Sirainen        rstream->received_time = received_time;
c06f4017027263cf3a08becc551f5126409e2a83Timo Sirainen        i_free(rstream->sender);
c06f4017027263cf3a08becc551f5126409e2a83Timo Sirainen        rstream->sender = sender;
2767104d81e97a109f0aa9758792bfa1da325a97Timo Sirainen    } else {
2767104d81e97a109f0aa9758792bfa1da325a97Timo Sirainen        rstream->next_received_time = received_time;
9e095dd6a77097356aca8216356d4d71ef1bea45Timo Sirainen        i_free(rstream->next_sender);
2767104d81e97a109f0aa9758792bfa1da325a97Timo Sirainen        rstream->next_sender = sender;
2767104d81e97a109f0aa9758792bfa1da325a97Timo Sirainen    }
2767104d81e97a109f0aa9758792bfa1da325a97Timo Sirainen
2767104d81e97a109f0aa9758792bfa1da325a97Timo Sirainen    /* we'll skip over From-line */
c06f4017027263cf3a08becc551f5126409e2a83Timo Sirainen    rstream->istream.istream.v_offset += line_pos+1;
c06f4017027263cf3a08becc551f5126409e2a83Timo Sirainen    i_stream_skip(rstream->input, line_pos+1);
2767104d81e97a109f0aa9758792bfa1da325a97Timo Sirainen    rstream->hdr_offset = rstream->istream.istream.v_offset;
c06f4017027263cf3a08becc551f5126409e2a83Timo Sirainen    return 0;
9e095dd6a77097356aca8216356d4d71ef1bea45Timo Sirainen}
2767104d81e97a109f0aa9758792bfa1da325a97Timo Sirainen
ecc81625167ed96c04c02aa190a1ea5baa65b474Timo Sirainenstatic ssize_t _read(struct _istream *stream)
ecc81625167ed96c04c02aa190a1ea5baa65b474Timo Sirainen{
2767104d81e97a109f0aa9758792bfa1da325a97Timo Sirainen    static const char *mbox_from = "\nFrom ";
ecc81625167ed96c04c02aa190a1ea5baa65b474Timo Sirainen    struct raw_mbox_istream *rstream = (struct raw_mbox_istream *)stream;
2767104d81e97a109f0aa9758792bfa1da325a97Timo Sirainen    const unsigned char *buf;
2767104d81e97a109f0aa9758792bfa1da325a97Timo Sirainen    const char *fromp;
2767104d81e97a109f0aa9758792bfa1da325a97Timo Sirainen    char *sender, eoh_char;
ecc81625167ed96c04c02aa190a1ea5baa65b474Timo Sirainen    time_t received_time;
2767104d81e97a109f0aa9758792bfa1da325a97Timo Sirainen    size_t i, pos, new_pos, from_start_pos;
2767104d81e97a109f0aa9758792bfa1da325a97Timo Sirainen    ssize_t ret = 0;
2767104d81e97a109f0aa9758792bfa1da325a97Timo Sirainen
2767104d81e97a109f0aa9758792bfa1da325a97Timo Sirainen    if (rstream->eom) {
2767104d81e97a109f0aa9758792bfa1da325a97Timo Sirainen        if (rstream->body_offset == (uoff_t)-1) {
2767104d81e97a109f0aa9758792bfa1da325a97Timo Sirainen            /* missing \n from headers */
2767104d81e97a109f0aa9758792bfa1da325a97Timo Sirainen            rstream->body_offset =
2767104d81e97a109f0aa9758792bfa1da325a97Timo Sirainen                stream->istream.v_offset +
ecc81625167ed96c04c02aa190a1ea5baa65b474Timo Sirainen                (stream->pos - stream->skip);
2767104d81e97a109f0aa9758792bfa1da325a97Timo Sirainen        }
ecc81625167ed96c04c02aa190a1ea5baa65b474Timo Sirainen        stream->istream.eof = rstream->next_eof;
ecc81625167ed96c04c02aa190a1ea5baa65b474Timo Sirainen        return -1;
1060afdc2fcdf647dbb3bc11647401f1b44a3a8aTimo Sirainen    }
ecc81625167ed96c04c02aa190a1ea5baa65b474Timo Sirainen
ecc81625167ed96c04c02aa190a1ea5baa65b474Timo Sirainen    i_stream_seek(rstream->input, stream->istream.v_offset);
ecc81625167ed96c04c02aa190a1ea5baa65b474Timo Sirainen
602a0434db30d8e3292d1c161a803d96a879a74fTimo Sirainen    stream->pos -= stream->skip;
6ef7e31619edfaa17ed044b45861d106a86191efTimo Sirainen    stream->skip = 0;
ecc81625167ed96c04c02aa190a1ea5baa65b474Timo Sirainen    stream->buffer = NULL;
c0435c854a0e7246373b9752d163095cc4fbe985Timo Sirainen
ecc81625167ed96c04c02aa190a1ea5baa65b474Timo Sirainen    do {
411d6baa37f31d90730e90c4a28c43e1974bbe58Timo Sirainen        ret = i_stream_read(rstream->input);
dd62b77c932d1b518f2a3e4bf80e36542becc256Timo Sirainen        buf = i_stream_get_data(rstream->input, &pos);
dd62b77c932d1b518f2a3e4bf80e36542becc256Timo Sirainen    } while (ret > 0 &&
dd62b77c932d1b518f2a3e4bf80e36542becc256Timo Sirainen         stream->istream.v_offset + pos <= rstream->input_peak_offset);
dd62b77c932d1b518f2a3e4bf80e36542becc256Timo Sirainen
dd62b77c932d1b518f2a3e4bf80e36542becc256Timo Sirainen    if (ret < 0) {
ecc81625167ed96c04c02aa190a1ea5baa65b474Timo Sirainen        if (ret == -2)
143cb2e0744e647f8fc637bbdea1106c1587a4bfTimo Sirainen            return -2;
143cb2e0744e647f8fc637bbdea1106c1587a4bfTimo Sirainen
143cb2e0744e647f8fc637bbdea1106c1587a4bfTimo Sirainen        /* we've read the whole file, final byte should be
143cb2e0744e647f8fc637bbdea1106c1587a4bfTimo Sirainen           the \n trailer */
ecc81625167ed96c04c02aa190a1ea5baa65b474Timo Sirainen        if (pos > 0 && buf[pos-1] == '\n')
ecc81625167ed96c04c02aa190a1ea5baa65b474Timo Sirainen            pos--;
07e4875d250e7a7157cd99132aafc773cf3cdf83Timo Sirainen
dd62b77c932d1b518f2a3e4bf80e36542becc256Timo Sirainen        i_assert(pos >= stream->pos);
07e4875d250e7a7157cd99132aafc773cf3cdf83Timo Sirainen        ret = pos == stream->pos ? -1 :
07e4875d250e7a7157cd99132aafc773cf3cdf83Timo Sirainen            (ssize_t)(pos - stream->pos);
07e4875d250e7a7157cd99132aafc773cf3cdf83Timo Sirainen
07e4875d250e7a7157cd99132aafc773cf3cdf83Timo Sirainen        stream->buffer = buf;
07e4875d250e7a7157cd99132aafc773cf3cdf83Timo Sirainen        stream->pos = pos;
07e4875d250e7a7157cd99132aafc773cf3cdf83Timo Sirainen
07e4875d250e7a7157cd99132aafc773cf3cdf83Timo Sirainen        rstream->eom = TRUE;
07e4875d250e7a7157cd99132aafc773cf3cdf83Timo Sirainen        rstream->next_eof = TRUE;
7d207b1e77a7b5e3fda640e353acfc86d261fedfTimo Sirainen        rstream->mail_size = stream->istream.v_offset + pos -
6ef7e31619edfaa17ed044b45861d106a86191efTimo Sirainen            rstream->hdr_offset;
07e4875d250e7a7157cd99132aafc773cf3cdf83Timo Sirainen        return ret < 0 ? _read(stream) : ret;
07e4875d250e7a7157cd99132aafc773cf3cdf83Timo Sirainen    }
07e4875d250e7a7157cd99132aafc773cf3cdf83Timo Sirainen
07e4875d250e7a7157cd99132aafc773cf3cdf83Timo Sirainen    if (stream->istream.v_offset == rstream->from_offset) {
cd1eef2109b4476842b7757f1d69b104196d5941Timo Sirainen        /* beginning of message, we haven't yet read our From-line */
cd1eef2109b4476842b7757f1d69b104196d5941Timo Sirainen        if (mbox_read_from_line(rstream) < 0) {
07e4875d250e7a7157cd99132aafc773cf3cdf83Timo Sirainen            stream->pos = 0;
cd1eef2109b4476842b7757f1d69b104196d5941Timo Sirainen            stream->istream.eof = TRUE;
07e4875d250e7a7157cd99132aafc773cf3cdf83Timo Sirainen            rstream->corrupted = TRUE;
07e4875d250e7a7157cd99132aafc773cf3cdf83Timo Sirainen            return -1;
07e4875d250e7a7157cd99132aafc773cf3cdf83Timo Sirainen        }
dd62b77c932d1b518f2a3e4bf80e36542becc256Timo Sirainen
dd62b77c932d1b518f2a3e4bf80e36542becc256Timo Sirainen        /* got it. we don't want to return it however,
c0435c854a0e7246373b9752d163095cc4fbe985Timo Sirainen           so start again from headers */
6ef7e31619edfaa17ed044b45861d106a86191efTimo Sirainen        buf = i_stream_get_data(rstream->input, &pos);
ecc81625167ed96c04c02aa190a1ea5baa65b474Timo Sirainen        if (pos == 0)
c0435c854a0e7246373b9752d163095cc4fbe985Timo Sirainen            return _read(stream);
ecc81625167ed96c04c02aa190a1ea5baa65b474Timo Sirainen    }
ecc81625167ed96c04c02aa190a1ea5baa65b474Timo Sirainen
c0435c854a0e7246373b9752d163095cc4fbe985Timo Sirainen    /* See if we have From-line here - note that it works right only
ecc81625167ed96c04c02aa190a1ea5baa65b474Timo Sirainen       because all characters are different in mbox_from. */
ecc81625167ed96c04c02aa190a1ea5baa65b474Timo Sirainen        fromp = mbox_from; from_start_pos = 0;
ecc81625167ed96c04c02aa190a1ea5baa65b474Timo Sirainen    eoh_char = rstream->body_offset == (uoff_t)-1 ? '\n' : '\0';
ecc81625167ed96c04c02aa190a1ea5baa65b474Timo Sirainen    for (i = 0; i < pos; i++) {
ecc81625167ed96c04c02aa190a1ea5baa65b474Timo Sirainen        if (buf[i] == eoh_char && i > 0 && buf[i-1] == '\n') {
7ef5ca6fb59a318c821a852ae48a2edbb671d7ddTimo Sirainen            rstream->body_offset = stream->istream.v_offset + i + 1;
ecc81625167ed96c04c02aa190a1ea5baa65b474Timo Sirainen            eoh_char = '\0';
ecc81625167ed96c04c02aa190a1ea5baa65b474Timo Sirainen        }
07e4875d250e7a7157cd99132aafc773cf3cdf83Timo Sirainen        if (buf[i] == *fromp) {
07e4875d250e7a7157cd99132aafc773cf3cdf83Timo Sirainen            if (*++fromp == '\0') {
ecc81625167ed96c04c02aa190a1ea5baa65b474Timo Sirainen                /* potential From-line, see if we have the
c06f4017027263cf3a08becc551f5126409e2a83Timo Sirainen                   rest of the line buffered.
411d6baa37f31d90730e90c4a28c43e1974bbe58Timo Sirainen                   FIXME: if From-line is longer than input
ecc81625167ed96c04c02aa190a1ea5baa65b474Timo Sirainen                   buffer, we break. probably irrelevant.. */
c06f4017027263cf3a08becc551f5126409e2a83Timo Sirainen                i++;
411d6baa37f31d90730e90c4a28c43e1974bbe58Timo Sirainen                from_start_pos = i;
411d6baa37f31d90730e90c4a28c43e1974bbe58Timo Sirainen                fromp = mbox_from;
ecc81625167ed96c04c02aa190a1ea5baa65b474Timo Sirainen            } else if (from_start_pos != 0) {
dd62b77c932d1b518f2a3e4bf80e36542becc256Timo Sirainen                /* we have the whole From-line here now.
ecc81625167ed96c04c02aa190a1ea5baa65b474Timo Sirainen                   See if it's a valid one. */
                if (mbox_from_parse(buf + from_start_pos,
                            pos - from_start_pos,
                            &received_time,
                            &sender) == 0) {
                    /* yep, we stop here. */
                    rstream->next_received_time =
                        received_time;
                    i_free(rstream->next_sender);
                    rstream->next_sender = sender;
                    rstream->eom = TRUE;

                                        /* rewind "\nFrom " */
                    from_start_pos -= 6;

                    rstream->mail_size =
                        stream->istream.v_offset +
                        from_start_pos -
                        rstream->hdr_offset;
                    break;
                }
                from_start_pos = 0;
            }
        } else {
            fromp = mbox_from;
            if (buf[i] == *fromp)
                fromp++;
        }
    }

    /* we want to go at least one byte further next time */
    rstream->input_peak_offset = stream->istream.v_offset + i;

    if (from_start_pos != 0) {
        /* we're waiting for the \n at the end of From-line */
        new_pos = from_start_pos;
    } else {
        /* leave out the beginnings of potential From-line */
        new_pos = i - (fromp - mbox_from);
    }
    i_assert(new_pos > stream->pos);
    ret = new_pos - stream->pos;

    stream->buffer = buf;
    stream->pos = new_pos;
    return ret;
}

static void _seek(struct _istream *stream, uoff_t v_offset)
{
    struct raw_mbox_istream *rstream = (struct raw_mbox_istream *)stream;

    stream->istream.v_offset = v_offset;
    stream->skip = stream->pos = 0;
    stream->buffer = NULL;

        rstream->input_peak_offset = 0;
    rstream->eom = FALSE;
    rstream->next_eof = FALSE;
}

struct istream *i_stream_create_raw_mbox(pool_t pool, struct istream *input)
{
    struct raw_mbox_istream *rstream;

    i_stream_ref(input);

    rstream = p_new(pool, struct raw_mbox_istream, 1);

    rstream->input = input;
    rstream->body_offset = (uoff_t)-1;
    rstream->mail_size = (uoff_t)-1;
    rstream->received_time = (time_t)-1;
    rstream->next_received_time = (time_t)-1;

    rstream->istream.iostream.close = _close;
    rstream->istream.iostream.destroy = _destroy;
    rstream->istream.iostream.set_max_buffer_size = _set_max_buffer_size;
    rstream->istream.iostream.set_blocking = _set_blocking;

    rstream->istream.read = _read;
    rstream->istream.seek = _seek;

    return _i_stream_create(&rstream->istream, pool, -1,
                input->real_stream->abs_start_offset);
}

static int istream_raw_mbox_is_valid_from(struct raw_mbox_istream *rstream)
{
    const unsigned char *data;
    size_t size;
    time_t received_time;
    char *sender;

    /* minimal: "From x Thu Nov 29 22:33:52 2001" = 31 chars */
    if (i_stream_read_data(rstream->input, &data, &size, 30) == -1)
        return -1;

    if (size == 1 && data[0] == '\n') {
        /* EOF */
        return TRUE;
    }

    if (size < 31 || memcmp(data, "\nFrom ", 6) != 0)
        return FALSE;

    while (memchr(data+1, '\n', size-1) == NULL) {
        if (i_stream_read_data(rstream->input, &data, &size, size) < 0)
            break;
    }

    if (mbox_from_parse(data+6, size-6, &received_time, &sender) < 0)
        return FALSE;

    rstream->next_received_time = received_time;
    i_free(rstream->next_sender);
    rstream->next_sender = sender;
    return TRUE;
}

uoff_t istream_raw_mbox_get_start_offset(struct istream *stream)
{
    struct raw_mbox_istream *rstream =
        (struct raw_mbox_istream *)stream->real_stream;

    return rstream->from_offset;
}

uoff_t istream_raw_mbox_get_header_offset(struct istream *stream)
{
    struct raw_mbox_istream *rstream =
        (struct raw_mbox_istream *)stream->real_stream;

    if (rstream->hdr_offset == rstream->from_offset)
        (void)_read(&rstream->istream);

    return rstream->hdr_offset;
}

uoff_t istream_raw_mbox_get_body_offset(struct istream *stream)
{
    struct raw_mbox_istream *rstream =
        (struct raw_mbox_istream *)stream->real_stream;
    uoff_t offset;
    size_t pos;

    if (rstream->body_offset != (uoff_t)-1)
        return rstream->body_offset;

    offset = stream->v_offset;
    i_stream_seek(stream, rstream->hdr_offset);
    while (rstream->body_offset == (uoff_t)-1) {
        i_stream_get_data(stream, &pos);
        i_stream_skip(stream, pos);

        if (_read(&rstream->istream) < 0)
            break;
    }

    i_stream_seek(stream, offset);
    return rstream->body_offset;
}

uoff_t istream_raw_mbox_get_body_size(struct istream *stream, uoff_t body_size)
{
    struct raw_mbox_istream *rstream =
        (struct raw_mbox_istream *)stream->real_stream;
    const unsigned char *data;
    size_t size;

    i_assert(rstream->hdr_offset != (uoff_t)-1);
    i_assert(rstream->body_offset != (uoff_t)-1);

    if (rstream->mail_size != (uoff_t)-1) {
        return rstream->mail_size -
            (rstream->body_offset - rstream->hdr_offset);
    }

    if (body_size != (uoff_t)-1) {
        i_stream_seek(rstream->input, rstream->body_offset + body_size);
        if (istream_raw_mbox_is_valid_from(rstream) > 0) {
            rstream->mail_size = body_size +
                (rstream->body_offset - rstream->hdr_offset);
            return body_size;
        }
    }

    /* have to read through the message body */
    while (i_stream_read_data(stream, &data, &size, 0) > 0)
        i_stream_skip(stream, size);

    i_assert(rstream->mail_size != (uoff_t)-1);
    return rstream->mail_size -
        (rstream->body_offset - rstream->hdr_offset);
}

time_t istream_raw_mbox_get_received_time(struct istream *stream)
{
    struct raw_mbox_istream *rstream =
        (struct raw_mbox_istream *)stream->real_stream;

    if (rstream->received_time == (time_t)-1)
        (void)_read(&rstream->istream);
    return rstream->received_time;
}

const char *istream_raw_mbox_get_sender(struct istream *stream)
{
    struct raw_mbox_istream *rstream =
        (struct raw_mbox_istream *)stream->real_stream;

    if (rstream->sender == NULL)
        (void)_read(&rstream->istream);
    return rstream->sender == NULL ? "" : rstream->sender;
}

void istream_raw_mbox_next(struct istream *stream, uoff_t body_size)
{
    struct raw_mbox_istream *rstream =
        (struct raw_mbox_istream *)stream->real_stream;

    body_size = istream_raw_mbox_get_body_size(stream, body_size);
    rstream->mail_size = (uoff_t)-1;

    rstream->received_time = rstream->next_received_time;
    rstream->next_received_time = (time_t)-1;

    i_free(rstream->sender);
    rstream->sender = rstream->next_sender;
    rstream->next_sender = NULL;

    rstream->from_offset = rstream->body_offset + body_size;
    rstream->hdr_offset = rstream->from_offset;
    rstream->body_offset = (uoff_t)-1;

    /* don't clear stream->eof if we don't have to */
    if (stream->v_offset != rstream->from_offset)
        i_stream_seek(stream, rstream->from_offset);
    i_stream_seek(rstream->input, rstream->from_offset);

        rstream->input_peak_offset = 0;
    rstream->eom = FALSE;
    rstream->next_eof = FALSE;
}

int istream_raw_mbox_seek(struct istream *stream, uoff_t offset)
{
    struct raw_mbox_istream *rstream =
        (struct raw_mbox_istream *)stream->real_stream;
    int check;

    rstream->corrupted = FALSE;
    rstream->eom = FALSE;
    rstream->next_eof = FALSE;
        rstream->input_peak_offset = 0;

    if (rstream->mail_size != (uoff_t)-1 &&
        rstream->hdr_offset + rstream->mail_size == offset) {
        istream_raw_mbox_next(stream, (uoff_t)-1);
        return 0;
    }

    if (offset == rstream->from_offset) {
        /* back to beginning of current message */
        offset = rstream->hdr_offset;
        check = offset == 0;
    } else {
        rstream->body_offset = (uoff_t)-1;
        rstream->mail_size = (uoff_t)-1;
        rstream->received_time = (time_t)-1;
        rstream->next_received_time = (time_t)-1;

        i_free(rstream->sender);
        rstream->sender = NULL;
        i_free(rstream->next_sender);
        rstream->next_sender = NULL;

                rstream->from_offset = offset;
        rstream->hdr_offset = offset;
        check = TRUE;
    }

    i_stream_seek(stream, offset);
    i_stream_seek(rstream->input, offset);

    if (check)
        (void)_read(&rstream->istream);
    return rstream->corrupted ? -1 : 0;
}

void istream_raw_mbox_flush(struct istream *stream)
{
    struct raw_mbox_istream *rstream =
        (struct raw_mbox_istream *)stream->real_stream;

    /* kludgy */
    rstream->input->real_stream->skip = 0;
    rstream->input->real_stream->pos = 0;

    rstream->istream.skip = 0;
    rstream->istream.pos = 0;
}