src/lib-mail/message-header-parser.c

bcb4e51a409d94ae670de96afb8483a4f7855294Stephan Bosch/* Copyright (c) 2002-2018 Dovecot authors, see the included COPYING file */
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen#include "lib.h"
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen#include "buffer.h"
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen#include "istream.h"
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen#include "str.h"
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen#include "message-size.h"
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen#include "message-header-parser.h"
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainenstruct message_header_parser_ctx {
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen    struct message_header_line line;
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen    struct istream *input;
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen    struct message_size *hdr_size;
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen    string_t *name;
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen    buffer_t *value_buf;
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen
043c8a96a035379bcba04f487d58457beefdfcaaTimo Sirainen    enum message_header_parser_flags flags;
0dffa25d211be541ee3c953b23566a1a990789dfTimo Sirainen    bool skip_line:1;
0dffa25d211be541ee3c953b23566a1a990789dfTimo Sirainen    bool has_nuls:1;
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen};
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainenstruct message_header_parser_ctx *
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainenmessage_parse_header_init(struct istream *input, struct message_size *hdr_size,
043c8a96a035379bcba04f487d58457beefdfcaaTimo Sirainen              enum message_header_parser_flags flags)
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen{
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen    struct message_header_parser_ctx *ctx;
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen    ctx = i_new(struct message_header_parser_ctx, 1);
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen    ctx->input = input;
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen    ctx->hdr_size = hdr_size;
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen    ctx->name = str_new(default_pool, 128);
043c8a96a035379bcba04f487d58457beefdfcaaTimo Sirainen    ctx->flags = flags;
1ff487015234b23c32cf8bb4c9f8c02922535b8eTimo Sirainen    ctx->value_buf = buffer_create_dynamic(default_pool, 4096);
20faa69d801460e89aa0b1214f3db4b026999b1eTimo Sirainen    i_stream_ref(input);
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen    if (hdr_size != NULL)
efe78d3ba24fc866af1c79b9223dc0809ba26cadStephan Bosch        i_zero(hdr_size);
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen    return ctx;
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen}
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainenvoid message_parse_header_deinit(struct message_header_parser_ctx **_ctx)
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen{
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen    struct message_header_parser_ctx *ctx = *_ctx;
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen
20faa69d801460e89aa0b1214f3db4b026999b1eTimo Sirainen    i_stream_unref(&ctx->input);
1ff487015234b23c32cf8bb4c9f8c02922535b8eTimo Sirainen    buffer_free(&ctx->value_buf);
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen    str_free(&ctx->name);
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen    i_free(ctx);
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen    *_ctx = NULL;
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen}
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainenint message_parse_header_next(struct message_header_parser_ctx *ctx,
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen                  struct message_header_line **hdr_r)
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen{
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen        struct message_header_line *line = &ctx->line;
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen    const unsigned char *msg;
a556e29bb9e48968b88c783148f3308bf90c7912Timo Sirainen    size_t i, size, startpos, colon_pos, parse_size, skip = 0;
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen    int ret;
c2f2a6cdf7dd55a4aabe72495cabb4deaa59cffcTimo Sirainen    bool continued, continues, last_no_newline, last_crlf;
675b2b1c4587a79478062b05725da61afa5d8045Timo Sirainen    bool no_newline, crlf_newline;
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen    *hdr_r = NULL;
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen    if (line->eoh)
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen        return -1;
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen
1ff487015234b23c32cf8bb4c9f8c02922535b8eTimo Sirainen    if (line->continues)
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen        colon_pos = 0;
1ff487015234b23c32cf8bb4c9f8c02922535b8eTimo Sirainen    else {
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen        /* new header line */
675b2b1c4587a79478062b05725da61afa5d8045Timo Sirainen        line->name_offset = ctx->input->v_offset;
675b2b1c4587a79478062b05725da61afa5d8045Timo Sirainen        colon_pos = UINT_MAX;
1ff487015234b23c32cf8bb4c9f8c02922535b8eTimo Sirainen        buffer_set_used_size(ctx->value_buf, 0);
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen    }
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen
675b2b1c4587a79478062b05725da61afa5d8045Timo Sirainen    no_newline = FALSE;
675b2b1c4587a79478062b05725da61afa5d8045Timo Sirainen    crlf_newline = FALSE;
c2f2a6cdf7dd55a4aabe72495cabb4deaa59cffcTimo Sirainen    continued = line->continues;
c2f2a6cdf7dd55a4aabe72495cabb4deaa59cffcTimo Sirainen    continues = FALSE;
675b2b1c4587a79478062b05725da61afa5d8045Timo Sirainen
675b2b1c4587a79478062b05725da61afa5d8045Timo Sirainen    for (startpos = 0;;) {
651f981ca065d2365bf2ea07b04318a5402f047aTimo Sirainen        ret = i_stream_read_bytes(ctx->input, &msg, &size, startpos+2);
c2f2a6cdf7dd55a4aabe72495cabb4deaa59cffcTimo Sirainen        if (ret >= 0) {
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen            /* we want to know one byte in advance to find out
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen               if it's multiline header */
675b2b1c4587a79478062b05725da61afa5d8045Timo Sirainen            parse_size = size == 0 ? 0 : size-1;
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen        } else {
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen            parse_size = size;
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen        }
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen
c2f2a6cdf7dd55a4aabe72495cabb4deaa59cffcTimo Sirainen        if (ret <= 0 && startpos == parse_size) {
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen            if (ret == -1) {
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen                if (startpos > 0) {
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen                    /* header ended unexpectedly. */
675b2b1c4587a79478062b05725da61afa5d8045Timo Sirainen                    no_newline = TRUE;
a556e29bb9e48968b88c783148f3308bf90c7912Timo Sirainen                    skip = startpos;
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen                    break;
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen                }
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen                /* error / EOF with no bytes */
a556e29bb9e48968b88c783148f3308bf90c7912Timo Sirainen                i_assert(skip == 0);
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen                return -1;
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen            }
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen
01ca85e6e763cfa0d146166fdd60654f04f51199Timo Sirainen            if (size > 0 && !ctx->skip_line && !continued &&
cc3da53f3a49304a251bfae88f814505326ac210Timo Sirainen                (msg[0] == '\n' ||
cc3da53f3a49304a251bfae88f814505326ac210Timo Sirainen                 (msg[0] == '\r' && size > 1 && msg[1] == '\n'))) {
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen                /* end of headers - this mostly happens just
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen                   with mbox where headers are read separately
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen                   from body */
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen                size = 0;
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen                if (ctx->hdr_size != NULL)
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen                    ctx->hdr_size->lines++;
51c331377beb4a2acb81aee4d12bc8ef6c496625Timo Sirainen                if (msg[0] == '\r') {
a556e29bb9e48968b88c783148f3308bf90c7912Timo Sirainen                    skip = 2;
675b2b1c4587a79478062b05725da61afa5d8045Timo Sirainen                    crlf_newline = TRUE;
51c331377beb4a2acb81aee4d12bc8ef6c496625Timo Sirainen                } else {
a556e29bb9e48968b88c783148f3308bf90c7912Timo Sirainen                    skip = 1;
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen                    if (ctx->hdr_size != NULL)
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen                        ctx->hdr_size->virtual_size++;
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen                }
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen                break;
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen            }
cc3da53f3a49304a251bfae88f814505326ac210Timo Sirainen            if (ret == 0 && !ctx->input->eof) {
cc3da53f3a49304a251bfae88f814505326ac210Timo Sirainen                /* stream is nonblocking - need more data */
a556e29bb9e48968b88c783148f3308bf90c7912Timo Sirainen                i_assert(skip == 0);
cc3da53f3a49304a251bfae88f814505326ac210Timo Sirainen                return 0;
cc3da53f3a49304a251bfae88f814505326ac210Timo Sirainen            }
cc3da53f3a49304a251bfae88f814505326ac210Timo Sirainen            i_assert(size > 0);
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen            /* a) line is larger than input buffer
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen               b) header ended unexpectedly */
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen            if (ret == -2) {
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen                /* go back to last LWSP if found. */
c2f2a6cdf7dd55a4aabe72495cabb4deaa59cffcTimo Sirainen                size_t min_pos = !continued ? colon_pos : 0;
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen                for (i = size-1; i > min_pos; i--) {
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen                    if (IS_LWSP(msg[i])) {
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen                        size = i;
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen                        break;
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen                    }
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen                }
6cb3c4f4276531258be706821e034f1f0a8cd276Timo Sirainen                if (i == min_pos && (msg[size-1] == '\r' ||
6cb3c4f4276531258be706821e034f1f0a8cd276Timo Sirainen                             msg[size-1] == '\n')) {
6cb3c4f4276531258be706821e034f1f0a8cd276Timo Sirainen                    /* we may or may not have a full header,
6cb3c4f4276531258be706821e034f1f0a8cd276Timo Sirainen                       but we don't know until we get the
6cb3c4f4276531258be706821e034f1f0a8cd276Timo Sirainen                       next character. leave out the
6cb3c4f4276531258be706821e034f1f0a8cd276Timo Sirainen                       linefeed and finish the header on
6cb3c4f4276531258be706821e034f1f0a8cd276Timo Sirainen                       the next run. */
6cb3c4f4276531258be706821e034f1f0a8cd276Timo Sirainen                    size--;
6cb3c4f4276531258be706821e034f1f0a8cd276Timo Sirainen                    if (size > 0 && msg[size-1] == '\r')
6cb3c4f4276531258be706821e034f1f0a8cd276Timo Sirainen                        size--;
6cb3c4f4276531258be706821e034f1f0a8cd276Timo Sirainen                }
01ca85e6e763cfa0d146166fdd60654f04f51199Timo Sirainen                /* the buffer really has to be more than 2 to
01ca85e6e763cfa0d146166fdd60654f04f51199Timo Sirainen                   avoid CRLF looping forever */
01ca85e6e763cfa0d146166fdd60654f04f51199Timo Sirainen                i_assert(size > 0);
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen
c2f2a6cdf7dd55a4aabe72495cabb4deaa59cffcTimo Sirainen                continues = TRUE;
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen            }
675b2b1c4587a79478062b05725da61afa5d8045Timo Sirainen            no_newline = TRUE;
a556e29bb9e48968b88c783148f3308bf90c7912Timo Sirainen            skip = size;
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen            break;
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen        }
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen        /* find ':' */
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen        if (colon_pos == UINT_MAX) {
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen            for (i = startpos; i < parse_size; i++) {
064ccad9cea2a75a0d6caa2c679b6cd22be58174Timo Sirainen                if (msg[i] > ':')
064ccad9cea2a75a0d6caa2c679b6cd22be58174Timo Sirainen                    continue;
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen
f16712057c1b82c6d2a3a4267c4521d357cd4b4cTimo Sirainen                if (msg[i] == ':' && !ctx->skip_line) {
064ccad9cea2a75a0d6caa2c679b6cd22be58174Timo Sirainen                    colon_pos = i;
064ccad9cea2a75a0d6caa2c679b6cd22be58174Timo Sirainen                    line->full_value_offset =
064ccad9cea2a75a0d6caa2c679b6cd22be58174Timo Sirainen                        ctx->input->v_offset + i + 1;
064ccad9cea2a75a0d6caa2c679b6cd22be58174Timo Sirainen                    break;
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen                }
064ccad9cea2a75a0d6caa2c679b6cd22be58174Timo Sirainen                if (msg[i] == '\n') {
064ccad9cea2a75a0d6caa2c679b6cd22be58174Timo Sirainen                    /* end of headers, or error */
064ccad9cea2a75a0d6caa2c679b6cd22be58174Timo Sirainen                    break;
064ccad9cea2a75a0d6caa2c679b6cd22be58174Timo Sirainen                }
064ccad9cea2a75a0d6caa2c679b6cd22be58174Timo Sirainen
064ccad9cea2a75a0d6caa2c679b6cd22be58174Timo Sirainen                if (msg[i] == '\0')
064ccad9cea2a75a0d6caa2c679b6cd22be58174Timo Sirainen                    ctx->has_nuls = TRUE;
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen            }
064ccad9cea2a75a0d6caa2c679b6cd22be58174Timo Sirainen        } else {
064ccad9cea2a75a0d6caa2c679b6cd22be58174Timo Sirainen            i = startpos;
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen        }
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen        /* find '\n' */
064ccad9cea2a75a0d6caa2c679b6cd22be58174Timo Sirainen        for (; i < parse_size; i++) {
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen            if (msg[i] <= '\n') {
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen                if (msg[i] == '\n')
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen                    break;
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen                if (msg[i] == '\0')
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen                    ctx->has_nuls = TRUE;
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen            }
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen        }
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen
6cb3c4f4276531258be706821e034f1f0a8cd276Timo Sirainen        if (i < parse_size && i+1 == size && ret == -2) {
6cb3c4f4276531258be706821e034f1f0a8cd276Timo Sirainen            /* we don't know if the line continues. */
6cb3c4f4276531258be706821e034f1f0a8cd276Timo Sirainen            i++;
6cb3c4f4276531258be706821e034f1f0a8cd276Timo Sirainen        } else if (i < parse_size) {
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen            /* got a line */
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen            if (ctx->skip_line) {
f16712057c1b82c6d2a3a4267c4521d357cd4b4cTimo Sirainen                /* skipping a line with a huge header name */
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen                if (ctx->hdr_size != NULL) {
f16712057c1b82c6d2a3a4267c4521d357cd4b4cTimo Sirainen                    ctx->hdr_size->lines++;
f16712057c1b82c6d2a3a4267c4521d357cd4b4cTimo Sirainen                    ctx->hdr_size->physical_size += i + 1;
f16712057c1b82c6d2a3a4267c4521d357cd4b4cTimo Sirainen                    ctx->hdr_size->virtual_size += i + 1;
51c331377beb4a2acb81aee4d12bc8ef6c496625Timo Sirainen                }
51c331377beb4a2acb81aee4d12bc8ef6c496625Timo Sirainen                if (i == 0 || msg[i-1] != '\r') {
51c331377beb4a2acb81aee4d12bc8ef6c496625Timo Sirainen                    /* missing CR */
51c331377beb4a2acb81aee4d12bc8ef6c496625Timo Sirainen                    if (ctx->hdr_size != NULL)
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen                        ctx->hdr_size->virtual_size++;
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen                }
f16712057c1b82c6d2a3a4267c4521d357cd4b4cTimo Sirainen
f16712057c1b82c6d2a3a4267c4521d357cd4b4cTimo Sirainen                i_stream_skip(ctx->input, i + 1);
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen                startpos = 0;
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen                ctx->skip_line = FALSE;
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen                continue;
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen            }
c2f2a6cdf7dd55a4aabe72495cabb4deaa59cffcTimo Sirainen            continues = i+1 < size && IS_LWSP(msg[i+1]);
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen            if (ctx->hdr_size != NULL)
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen                ctx->hdr_size->lines++;
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen            if (i == 0 || msg[i-1] != '\r') {
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen                /* missing CR */
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen                if (ctx->hdr_size != NULL)
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen                    ctx->hdr_size->virtual_size++;
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen                size = i;
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen            } else {
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen                size = i-1;
675b2b1c4587a79478062b05725da61afa5d8045Timo Sirainen                crlf_newline = TRUE;
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen            }
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen
a556e29bb9e48968b88c783148f3308bf90c7912Timo Sirainen            skip = i+1;
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen            break;
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen        }
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen        startpos = i;
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen    }
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen
675b2b1c4587a79478062b05725da61afa5d8045Timo Sirainen    last_crlf = line->crlf_newline &&
675b2b1c4587a79478062b05725da61afa5d8045Timo Sirainen        (ctx->flags & MESSAGE_HEADER_PARSER_FLAG_DROP_CR) == 0;
675b2b1c4587a79478062b05725da61afa5d8045Timo Sirainen    last_no_newline = line->no_newline ||
675b2b1c4587a79478062b05725da61afa5d8045Timo Sirainen        (ctx->flags & MESSAGE_HEADER_PARSER_FLAG_CLEAN_ONELINE) != 0;
675b2b1c4587a79478062b05725da61afa5d8045Timo Sirainen
c2f2a6cdf7dd55a4aabe72495cabb4deaa59cffcTimo Sirainen    line->continues = continues;
675b2b1c4587a79478062b05725da61afa5d8045Timo Sirainen    line->continued = continued;
675b2b1c4587a79478062b05725da61afa5d8045Timo Sirainen    line->crlf_newline = crlf_newline;
675b2b1c4587a79478062b05725da61afa5d8045Timo Sirainen    line->no_newline = no_newline;
01ca85e6e763cfa0d146166fdd60654f04f51199Timo Sirainen    if (size == 0 && !continued) {
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen        /* end of headers */
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen        line->eoh = TRUE;
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen        line->name_len = line->value_len = line->full_value_len = 0;
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen        line->name = ""; line->value = line->full_value = NULL;
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen        line->middle = NULL; line->middle_len = 0;
cc3da53f3a49304a251bfae88f814505326ac210Timo Sirainen        line->full_value_offset = line->name_offset;
cc3da53f3a49304a251bfae88f814505326ac210Timo Sirainen        line->continues = FALSE;
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen    } else if (line->continued) {
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen        line->value = msg;
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen        line->value_len = size;
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen    } else if (colon_pos == UINT_MAX) {
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen        /* missing ':', assume the whole line is name */
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen        line->value = NULL;
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen        line->value_len = 0;
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen        str_truncate(ctx->name, 0);
4bd6702bfcb3568959acdb3cb68450aa76ed703bTimo Sirainen        buffer_append(ctx->name, msg, size);
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen        line->name = str_c(ctx->name);
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen        line->name_len = str_len(ctx->name);
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen        line->middle = NULL;
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen        line->middle_len = 0;
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen    } else {
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen        size_t pos;
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen        line->value = msg + colon_pos+1;
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen        line->value_len = size - colon_pos - 1;
23bdbb7b1831785c6ba6df190f6369da882d2b9dTimo Sirainen        if ((ctx->flags & MESSAGE_HEADER_PARSER_FLAG_SKIP_INITIAL_LWSP) != 0) {
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen            /* get value. skip all LWSP after ':'. Note that
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen               RFC2822 doesn't say we should, but history behind
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen               it..
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen               Exception to this is if the value consists only of
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen               LWSP, then skip only the one LWSP after ':'. */
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen            for (pos = 0; pos < line->value_len; pos++) {
c52eba0224a0ff239f4778a7f6ed5ce38d92a5ddTimo Sirainen                if (!IS_LWSP(line->value[pos]))
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen                    break;
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen            }
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen            if (pos == line->value_len) {
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen                /* everything was LWSP */
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen                if (line->value_len > 0 &&
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen                    IS_LWSP(line->value[0]))
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen                    pos = 1;
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen            }
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen        } else {
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen            pos = line->value_len > 0 &&
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen                IS_LWSP(line->value[0]) ? 1 : 0;
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen        }
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen        line->value += pos;
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen        line->value_len -= pos;
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen        line->full_value_offset += pos;
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen        /* get name, skip LWSP before ':' */
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen        while (colon_pos > 0 && IS_LWSP(msg[colon_pos-1]))
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen            colon_pos--;
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen
f3aff3c6bf32d5bb0a61993b76a3fbe5ad798c09Timo Sirainen        str_truncate(ctx->name, 0);
200500ebd3c93a5771e7ae2f442659b9acb16eb6Timo Sirainen        /* use buffer_append() so the name won't be truncated if there
200500ebd3c93a5771e7ae2f442659b9acb16eb6Timo Sirainen           are NULs. */
200500ebd3c93a5771e7ae2f442659b9acb16eb6Timo Sirainen        buffer_append(ctx->name, msg, colon_pos);
f3aff3c6bf32d5bb0a61993b76a3fbe5ad798c09Timo Sirainen        str_append_c(ctx->name, '\0');
f3aff3c6bf32d5bb0a61993b76a3fbe5ad798c09Timo Sirainen
f3aff3c6bf32d5bb0a61993b76a3fbe5ad798c09Timo Sirainen        /* keep middle stored also in ctx->name so it's available
f3aff3c6bf32d5bb0a61993b76a3fbe5ad798c09Timo Sirainen           with use_full_value */
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen        line->middle = msg + colon_pos;
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen        line->middle_len = (size_t)(line->value - line->middle);
f3aff3c6bf32d5bb0a61993b76a3fbe5ad798c09Timo Sirainen        str_append_n(ctx->name, line->middle, line->middle_len);
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen        line->name = str_c(ctx->name);
f3aff3c6bf32d5bb0a61993b76a3fbe5ad798c09Timo Sirainen        line->name_len = colon_pos;
f3aff3c6bf32d5bb0a61993b76a3fbe5ad798c09Timo Sirainen        line->middle = str_data(ctx->name) + line->name_len + 1;
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen    }
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen    if (!line->continued) {
1ff487015234b23c32cf8bb4c9f8c02922535b8eTimo Sirainen        /* first header line. make a copy of the line since we can't
1ff487015234b23c32cf8bb4c9f8c02922535b8eTimo Sirainen           really trust input stream not to lose it. */
1ff487015234b23c32cf8bb4c9f8c02922535b8eTimo Sirainen        buffer_append(ctx->value_buf, line->value, line->value_len);
1ff487015234b23c32cf8bb4c9f8c02922535b8eTimo Sirainen        line->value = line->full_value = ctx->value_buf->data;
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen        line->full_value_len = line->value_len;
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen    } else if (line->use_full_value) {
15362cdf9df29fef8795e865957e17ec027a9ebfTimo Sirainen        /* continue saving the full value. */
15362cdf9df29fef8795e865957e17ec027a9ebfTimo Sirainen        if (last_no_newline) {
15362cdf9df29fef8795e865957e17ec027a9ebfTimo Sirainen            /* line is longer than fit into our buffer, so we
15362cdf9df29fef8795e865957e17ec027a9ebfTimo Sirainen               were forced to break it into multiple
15362cdf9df29fef8795e865957e17ec027a9ebfTimo Sirainen               message_header_lines */
15362cdf9df29fef8795e865957e17ec027a9ebfTimo Sirainen        } else {
e161a5225abda0837b5deb8746ef808ba5e98d94Timo Sirainen            if (last_crlf)
e161a5225abda0837b5deb8746ef808ba5e98d94Timo Sirainen                buffer_append_c(ctx->value_buf, '\r');
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen            buffer_append_c(ctx->value_buf, '\n');
e161a5225abda0837b5deb8746ef808ba5e98d94Timo Sirainen        }
23bdbb7b1831785c6ba6df190f6369da882d2b9dTimo Sirainen        if ((ctx->flags & MESSAGE_HEADER_PARSER_FLAG_CLEAN_ONELINE) != 0 &&
15362cdf9df29fef8795e865957e17ec027a9ebfTimo Sirainen            line->value_len > 0 && line->value[0] != ' ' &&
15362cdf9df29fef8795e865957e17ec027a9ebfTimo Sirainen            IS_LWSP(line->value[0])) {
043c8a96a035379bcba04f487d58457beefdfcaaTimo Sirainen            buffer_append_c(ctx->value_buf, ' ');
043c8a96a035379bcba04f487d58457beefdfcaaTimo Sirainen            buffer_append(ctx->value_buf,
043c8a96a035379bcba04f487d58457beefdfcaaTimo Sirainen                      line->value + 1, line->value_len - 1);
043c8a96a035379bcba04f487d58457beefdfcaaTimo Sirainen        } else {
043c8a96a035379bcba04f487d58457beefdfcaaTimo Sirainen            buffer_append(ctx->value_buf,
043c8a96a035379bcba04f487d58457beefdfcaaTimo Sirainen                      line->value, line->value_len);
043c8a96a035379bcba04f487d58457beefdfcaaTimo Sirainen        }
b88c43d09a288e99d439c78de4cc613212ea924cTimo Sirainen        line->full_value = ctx->value_buf->data;
b88c43d09a288e99d439c78de4cc613212ea924cTimo Sirainen        line->full_value_len = ctx->value_buf->used;
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen    } else {
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen        /* we didn't want full_value, and this is a continued line. */
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen        line->full_value = NULL;
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen        line->full_value_len = 0;
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen    }
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen    /* always reset it */
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen    line->use_full_value = FALSE;
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen    if (ctx->hdr_size != NULL) {
a556e29bb9e48968b88c783148f3308bf90c7912Timo Sirainen        ctx->hdr_size->physical_size += skip;
a556e29bb9e48968b88c783148f3308bf90c7912Timo Sirainen        ctx->hdr_size->virtual_size += skip;
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen    }
a556e29bb9e48968b88c783148f3308bf90c7912Timo Sirainen    i_stream_skip(ctx->input, skip);
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen    *hdr_r = line;
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen    return 1;
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen}
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen
68a4946b12583b88fa802e52ebee45cd96056772Timo Sirainenbool message_parse_header_has_nuls(const struct message_header_parser_ctx *ctx)
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen{
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen    return ctx->has_nuls;
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen}
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen
59151b71059df1190acd75d8717ed04a7920c862Timo Sirainen#undef message_parse_header
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainenvoid message_parse_header(struct istream *input, struct message_size *hdr_size,
043c8a96a035379bcba04f487d58457beefdfcaaTimo Sirainen              enum message_header_parser_flags flags,
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen              message_header_callback_t *callback, void *context)
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen{
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen    struct message_header_parser_ctx *hdr_ctx;
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen    struct message_header_line *hdr;
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen    int ret;
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen
043c8a96a035379bcba04f487d58457beefdfcaaTimo Sirainen    hdr_ctx = message_parse_header_init(input, hdr_size, flags);
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen    while ((ret = message_parse_header_next(hdr_ctx, &hdr)) > 0)
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen        callback(hdr, context);
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen    i_assert(ret != 0);
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen    message_parse_header_deinit(&hdr_ctx);
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen    /* call after the final skipping */
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen    callback(NULL, context);
d96f86fb881c5b106649e8994ead1052acf24030Timo Sirainen}
d45a4c7d1df04dd93d57aa8a29b76b55d4905341Timo Sirainen
d45a4c7d1df04dd93d57aa8a29b76b55d4905341Timo Sirainenvoid message_header_line_write(buffer_t *output,
d45a4c7d1df04dd93d57aa8a29b76b55d4905341Timo Sirainen                   const struct message_header_line *hdr)
d45a4c7d1df04dd93d57aa8a29b76b55d4905341Timo Sirainen{
d45a4c7d1df04dd93d57aa8a29b76b55d4905341Timo Sirainen    if (!hdr->continued) {
d45a4c7d1df04dd93d57aa8a29b76b55d4905341Timo Sirainen        buffer_append(output, hdr->name, strlen(hdr->name));
d45a4c7d1df04dd93d57aa8a29b76b55d4905341Timo Sirainen        buffer_append(output, hdr->middle, hdr->middle_len);
d45a4c7d1df04dd93d57aa8a29b76b55d4905341Timo Sirainen    }
d45a4c7d1df04dd93d57aa8a29b76b55d4905341Timo Sirainen    buffer_append(output, hdr->value, hdr->value_len);
d45a4c7d1df04dd93d57aa8a29b76b55d4905341Timo Sirainen    if (!hdr->no_newline) {
d45a4c7d1df04dd93d57aa8a29b76b55d4905341Timo Sirainen        if (hdr->crlf_newline)
d45a4c7d1df04dd93d57aa8a29b76b55d4905341Timo Sirainen            buffer_append_c(output, '\r');
d45a4c7d1df04dd93d57aa8a29b76b55d4905341Timo Sirainen        buffer_append_c(output, '\n');
d45a4c7d1df04dd93d57aa8a29b76b55d4905341Timo Sirainen    }
d45a4c7d1df04dd93d57aa8a29b76b55d4905341Timo Sirainen}