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

	message-header-parser.c revision 6cb3c4f4276531258be706821e034f1f0a8cd276
2e37d45867d081db150ab78dad303b9077aea24fTimo Sirainen/* Copyright (c) 2002-2012 Dovecot authors, see the included COPYING file */
f89cb43088c8b46d12d66ac924724b53ab14ce66Timo Sirainen
f89cb43088c8b46d12d66ac924724b53ab14ce66Timo Sirainen#include "lib.h"
f89cb43088c8b46d12d66ac924724b53ab14ce66Timo Sirainen#include "buffer.h"
0536ccb51d41e3078c3a9fa33e509fb4b2420f95Timo Sirainen#include "istream.h"
f89cb43088c8b46d12d66ac924724b53ab14ce66Timo Sirainen#include "str.h"
f89cb43088c8b46d12d66ac924724b53ab14ce66Timo Sirainen#include "message-size.h"
f89cb43088c8b46d12d66ac924724b53ab14ce66Timo Sirainen#include "message-header-parser.h"
f89cb43088c8b46d12d66ac924724b53ab14ce66Timo Sirainen
f89cb43088c8b46d12d66ac924724b53ab14ce66Timo Sirainenstruct message_header_parser_ctx {
f89cb43088c8b46d12d66ac924724b53ab14ce66Timo Sirainen    struct message_header_line line;
f89cb43088c8b46d12d66ac924724b53ab14ce66Timo Sirainen
f89cb43088c8b46d12d66ac924724b53ab14ce66Timo Sirainen    struct istream *input;
f89cb43088c8b46d12d66ac924724b53ab14ce66Timo Sirainen    struct message_size *hdr_size;
beffc30d933c5e134c45cc871852a8427eba7e70Timo Sirainen
f89cb43088c8b46d12d66ac924724b53ab14ce66Timo Sirainen    string_t *name;
f89cb43088c8b46d12d66ac924724b53ab14ce66Timo Sirainen    buffer_t *value_buf;
f89cb43088c8b46d12d66ac924724b53ab14ce66Timo Sirainen    size_t skip;
f89cb43088c8b46d12d66ac924724b53ab14ce66Timo Sirainen
f89cb43088c8b46d12d66ac924724b53ab14ce66Timo Sirainen    enum message_header_parser_flags flags;
f89cb43088c8b46d12d66ac924724b53ab14ce66Timo Sirainen    unsigned int skip_line:1;
f89cb43088c8b46d12d66ac924724b53ab14ce66Timo Sirainen    unsigned int has_nuls:1;
f89cb43088c8b46d12d66ac924724b53ab14ce66Timo Sirainen};
f89cb43088c8b46d12d66ac924724b53ab14ce66Timo Sirainen
f89cb43088c8b46d12d66ac924724b53ab14ce66Timo Sirainenstruct message_header_parser_ctx *
f89cb43088c8b46d12d66ac924724b53ab14ce66Timo Sirainenmessage_parse_header_init(struct istream *input, struct message_size *hdr_size,
f89cb43088c8b46d12d66ac924724b53ab14ce66Timo Sirainen              enum message_header_parser_flags flags)
f89cb43088c8b46d12d66ac924724b53ab14ce66Timo Sirainen{
f89cb43088c8b46d12d66ac924724b53ab14ce66Timo Sirainen    struct message_header_parser_ctx *ctx;
f89cb43088c8b46d12d66ac924724b53ab14ce66Timo Sirainen
f89cb43088c8b46d12d66ac924724b53ab14ce66Timo Sirainen    ctx = i_new(struct message_header_parser_ctx, 1);
f89cb43088c8b46d12d66ac924724b53ab14ce66Timo Sirainen    ctx->input = input;
f89cb43088c8b46d12d66ac924724b53ab14ce66Timo Sirainen    ctx->hdr_size = hdr_size;
5a0ac2e5ef482016e00575a7dce83f52c1704732Timo Sirainen    ctx->name = str_new(default_pool, 128);
5a0ac2e5ef482016e00575a7dce83f52c1704732Timo Sirainen    ctx->flags = flags;
7823ef73e51bb81a17dcb306aff89016d4ce258fTimo Sirainen    ctx->value_buf = buffer_create_dynamic(default_pool, 4096);
f89cb43088c8b46d12d66ac924724b53ab14ce66Timo Sirainen
f89cb43088c8b46d12d66ac924724b53ab14ce66Timo Sirainen    if (hdr_size != NULL)
f89cb43088c8b46d12d66ac924724b53ab14ce66Timo Sirainen        memset(hdr_size, 0, sizeof(*hdr_size));
f89cb43088c8b46d12d66ac924724b53ab14ce66Timo Sirainen    return ctx;
f89cb43088c8b46d12d66ac924724b53ab14ce66Timo Sirainen}
f89cb43088c8b46d12d66ac924724b53ab14ce66Timo Sirainen
f89cb43088c8b46d12d66ac924724b53ab14ce66Timo Sirainenvoid message_parse_header_deinit(struct message_header_parser_ctx **_ctx)
f89cb43088c8b46d12d66ac924724b53ab14ce66Timo Sirainen{
f89cb43088c8b46d12d66ac924724b53ab14ce66Timo Sirainen    struct message_header_parser_ctx *ctx = *_ctx;
f89cb43088c8b46d12d66ac924724b53ab14ce66Timo Sirainen
f89cb43088c8b46d12d66ac924724b53ab14ce66Timo Sirainen    i_stream_skip(ctx->input, ctx->skip);
f89cb43088c8b46d12d66ac924724b53ab14ce66Timo Sirainen    buffer_free(&ctx->value_buf);
f89cb43088c8b46d12d66ac924724b53ab14ce66Timo Sirainen    str_free(&ctx->name);
f89cb43088c8b46d12d66ac924724b53ab14ce66Timo Sirainen    i_free(ctx);
f89cb43088c8b46d12d66ac924724b53ab14ce66Timo Sirainen
f89cb43088c8b46d12d66ac924724b53ab14ce66Timo Sirainen    *_ctx = NULL;
f89cb43088c8b46d12d66ac924724b53ab14ce66Timo Sirainen}
f89cb43088c8b46d12d66ac924724b53ab14ce66Timo Sirainen
f89cb43088c8b46d12d66ac924724b53ab14ce66Timo Sirainenint message_parse_header_next(struct message_header_parser_ctx *ctx,
f89cb43088c8b46d12d66ac924724b53ab14ce66Timo Sirainen                  struct message_header_line **hdr_r)
f89cb43088c8b46d12d66ac924724b53ab14ce66Timo Sirainen{
beffc30d933c5e134c45cc871852a8427eba7e70Timo Sirainen        struct message_header_line *line = &ctx->line;
f89cb43088c8b46d12d66ac924724b53ab14ce66Timo Sirainen    const unsigned char *msg;
f89cb43088c8b46d12d66ac924724b53ab14ce66Timo Sirainen    size_t i, size, startpos, colon_pos, parse_size;
4ce6338bf945cccfff9e4ce7cc6aa2246851b84aTimo Sirainen    int ret;
f89cb43088c8b46d12d66ac924724b53ab14ce66Timo Sirainen    bool continued, continues, last_no_newline, last_crlf;
beffc30d933c5e134c45cc871852a8427eba7e70Timo Sirainen    bool no_newline, crlf_newline;
beffc30d933c5e134c45cc871852a8427eba7e70Timo Sirainen
beffc30d933c5e134c45cc871852a8427eba7e70Timo Sirainen    *hdr_r = NULL;
beffc30d933c5e134c45cc871852a8427eba7e70Timo Sirainen    if (line->eoh)
f89cb43088c8b46d12d66ac924724b53ab14ce66Timo Sirainen        return -1;
4ce6338bf945cccfff9e4ce7cc6aa2246851b84aTimo Sirainen
f89cb43088c8b46d12d66ac924724b53ab14ce66Timo Sirainen    if (ctx->skip > 0) {
f89cb43088c8b46d12d66ac924724b53ab14ce66Timo Sirainen        i_stream_skip(ctx->input, ctx->skip);
f89cb43088c8b46d12d66ac924724b53ab14ce66Timo Sirainen        ctx->skip = 0;
beffc30d933c5e134c45cc871852a8427eba7e70Timo Sirainen    }
64055bc6d2ed9e25b3b1db3b5b90d0bdb77cd715Timo Sirainen
64055bc6d2ed9e25b3b1db3b5b90d0bdb77cd715Timo Sirainen    if (line->continues)
64055bc6d2ed9e25b3b1db3b5b90d0bdb77cd715Timo Sirainen        colon_pos = 0;
64055bc6d2ed9e25b3b1db3b5b90d0bdb77cd715Timo Sirainen    else {
64055bc6d2ed9e25b3b1db3b5b90d0bdb77cd715Timo Sirainen        /* new header line */
64055bc6d2ed9e25b3b1db3b5b90d0bdb77cd715Timo Sirainen        line->name_offset = ctx->input->v_offset;
f89cb43088c8b46d12d66ac924724b53ab14ce66Timo Sirainen        colon_pos = UINT_MAX;
beffc30d933c5e134c45cc871852a8427eba7e70Timo Sirainen        buffer_set_used_size(ctx->value_buf, 0);
f89cb43088c8b46d12d66ac924724b53ab14ce66Timo Sirainen    }
beffc30d933c5e134c45cc871852a8427eba7e70Timo Sirainen
4ce6338bf945cccfff9e4ce7cc6aa2246851b84aTimo Sirainen    no_newline = FALSE;
f89cb43088c8b46d12d66ac924724b53ab14ce66Timo Sirainen    crlf_newline = FALSE;
f89cb43088c8b46d12d66ac924724b53ab14ce66Timo Sirainen    continued = line->continues;
f89cb43088c8b46d12d66ac924724b53ab14ce66Timo Sirainen    continues = FALSE;
f89cb43088c8b46d12d66ac924724b53ab14ce66Timo Sirainen
f89cb43088c8b46d12d66ac924724b53ab14ce66Timo Sirainen    for (startpos = 0;;) {
f89cb43088c8b46d12d66ac924724b53ab14ce66Timo Sirainen        ret = i_stream_read_data(ctx->input, &msg, &size, startpos+1);
beffc30d933c5e134c45cc871852a8427eba7e70Timo Sirainen        if (ret >= 0) {
f89cb43088c8b46d12d66ac924724b53ab14ce66Timo Sirainen            /* we want to know one byte in advance to find out
f89cb43088c8b46d12d66ac924724b53ab14ce66Timo Sirainen               if it's multiline header */
f89cb43088c8b46d12d66ac924724b53ab14ce66Timo Sirainen            parse_size = size == 0 ? 0 : size-1;
7026c16186f543e11af12b8b87f396006db93297Timo Sirainen        } else {
7026c16186f543e11af12b8b87f396006db93297Timo Sirainen            parse_size = size;
f89cb43088c8b46d12d66ac924724b53ab14ce66Timo Sirainen        }
7026c16186f543e11af12b8b87f396006db93297Timo Sirainen
f89cb43088c8b46d12d66ac924724b53ab14ce66Timo Sirainen        if (ret <= 0 && startpos == parse_size) {
beffc30d933c5e134c45cc871852a8427eba7e70Timo Sirainen            if (ret == -1) {
beffc30d933c5e134c45cc871852a8427eba7e70Timo Sirainen                if (startpos > 0) {
beffc30d933c5e134c45cc871852a8427eba7e70Timo Sirainen                    /* header ended unexpectedly. */
beffc30d933c5e134c45cc871852a8427eba7e70Timo Sirainen                    no_newline = TRUE;
beffc30d933c5e134c45cc871852a8427eba7e70Timo Sirainen                    ctx->skip = startpos;
beffc30d933c5e134c45cc871852a8427eba7e70Timo Sirainen                    break;
beffc30d933c5e134c45cc871852a8427eba7e70Timo Sirainen                }
beffc30d933c5e134c45cc871852a8427eba7e70Timo Sirainen                /* error / EOF with no bytes */
beffc30d933c5e134c45cc871852a8427eba7e70Timo Sirainen                return -1;
beffc30d933c5e134c45cc871852a8427eba7e70Timo Sirainen            }
beffc30d933c5e134c45cc871852a8427eba7e70Timo Sirainen
beffc30d933c5e134c45cc871852a8427eba7e70Timo Sirainen            if (size > 0 && !ctx->skip_line &&
beffc30d933c5e134c45cc871852a8427eba7e70Timo Sirainen                (msg[0] == '\n' ||
beffc30d933c5e134c45cc871852a8427eba7e70Timo Sirainen                 (msg[0] == '\r' && size > 1 && msg[1] == '\n'))) {
beffc30d933c5e134c45cc871852a8427eba7e70Timo Sirainen                /* end of headers - this mostly happens just
beffc30d933c5e134c45cc871852a8427eba7e70Timo Sirainen                   with mbox where headers are read separately
beffc30d933c5e134c45cc871852a8427eba7e70Timo Sirainen                   from body */
beffc30d933c5e134c45cc871852a8427eba7e70Timo Sirainen                size = 0;
f89cb43088c8b46d12d66ac924724b53ab14ce66Timo Sirainen                if (ctx->hdr_size != NULL)
beffc30d933c5e134c45cc871852a8427eba7e70Timo Sirainen                    ctx->hdr_size->lines++;
f89cb43088c8b46d12d66ac924724b53ab14ce66Timo Sirainen                if (msg[0] == '\r') {
beffc30d933c5e134c45cc871852a8427eba7e70Timo Sirainen                    ctx->skip = 2;
f89cb43088c8b46d12d66ac924724b53ab14ce66Timo Sirainen                    crlf_newline = TRUE;
f89cb43088c8b46d12d66ac924724b53ab14ce66Timo Sirainen                } else {
f89cb43088c8b46d12d66ac924724b53ab14ce66Timo Sirainen                    ctx->skip = 1;
beffc30d933c5e134c45cc871852a8427eba7e70Timo Sirainen                    if (ctx->hdr_size != NULL)
f89cb43088c8b46d12d66ac924724b53ab14ce66Timo Sirainen                        ctx->hdr_size->virtual_size++;
f89cb43088c8b46d12d66ac924724b53ab14ce66Timo Sirainen                }
f89cb43088c8b46d12d66ac924724b53ab14ce66Timo Sirainen                break;
f89cb43088c8b46d12d66ac924724b53ab14ce66Timo Sirainen            }
64c48ffb71f1cf99acf375768fde4cff9b512648Timo Sirainen            if (ret == 0 && !ctx->input->eof) {
f89cb43088c8b46d12d66ac924724b53ab14ce66Timo Sirainen                /* stream is nonblocking - need more data */
f89cb43088c8b46d12d66ac924724b53ab14ce66Timo Sirainen                return 0;
f89cb43088c8b46d12d66ac924724b53ab14ce66Timo Sirainen            }
f89cb43088c8b46d12d66ac924724b53ab14ce66Timo Sirainen            i_assert(size > 0);
f89cb43088c8b46d12d66ac924724b53ab14ce66Timo Sirainen
f89cb43088c8b46d12d66ac924724b53ab14ce66Timo Sirainen            /* a) line is larger than input buffer
f89cb43088c8b46d12d66ac924724b53ab14ce66Timo Sirainen               b) header ended unexpectedly */
f89cb43088c8b46d12d66ac924724b53ab14ce66Timo Sirainen            if (colon_pos == UINT_MAX && ret == -2 && !continued) {
beffc30d933c5e134c45cc871852a8427eba7e70Timo Sirainen                /* header name is huge. just skip it. */
f89cb43088c8b46d12d66ac924724b53ab14ce66Timo Sirainen                i_assert(size > 1);
f89cb43088c8b46d12d66ac924724b53ab14ce66Timo Sirainen                if (msg[size-1] == '\r')
f89cb43088c8b46d12d66ac924724b53ab14ce66Timo Sirainen                    size--;
beffc30d933c5e134c45cc871852a8427eba7e70Timo Sirainen
4ce6338bf945cccfff9e4ce7cc6aa2246851b84aTimo Sirainen                if (ctx->hdr_size != NULL) {
f89cb43088c8b46d12d66ac924724b53ab14ce66Timo Sirainen                    ctx->hdr_size->physical_size += size;
f89cb43088c8b46d12d66ac924724b53ab14ce66Timo Sirainen                    ctx->hdr_size->virtual_size += size;
f89cb43088c8b46d12d66ac924724b53ab14ce66Timo Sirainen                }
c09f9f95db314e7482c95e502e1c56ed6c555797Timo Sirainen                i_stream_skip(ctx->input, size);
f89cb43088c8b46d12d66ac924724b53ab14ce66Timo Sirainen                ctx->skip_line = TRUE;
f89cb43088c8b46d12d66ac924724b53ab14ce66Timo Sirainen                startpos = 0;
f89cb43088c8b46d12d66ac924724b53ab14ce66Timo Sirainen                continue;
beffc30d933c5e134c45cc871852a8427eba7e70Timo Sirainen            }
f89cb43088c8b46d12d66ac924724b53ab14ce66Timo Sirainen
f89cb43088c8b46d12d66ac924724b53ab14ce66Timo Sirainen            if (ret == -2) {
f89cb43088c8b46d12d66ac924724b53ab14ce66Timo Sirainen                /* go back to last LWSP if found. */
beffc30d933c5e134c45cc871852a8427eba7e70Timo Sirainen                size_t min_pos = !continued ? colon_pos : 0;
beffc30d933c5e134c45cc871852a8427eba7e70Timo Sirainen                for (i = size-1; i > min_pos; i--) {
f89cb43088c8b46d12d66ac924724b53ab14ce66Timo Sirainen                    if (IS_LWSP(msg[i])) {
beffc30d933c5e134c45cc871852a8427eba7e70Timo Sirainen                        size = i;
beffc30d933c5e134c45cc871852a8427eba7e70Timo Sirainen                        break;
f89cb43088c8b46d12d66ac924724b53ab14ce66Timo Sirainen                    }
4ce6338bf945cccfff9e4ce7cc6aa2246851b84aTimo Sirainen                }
4ce6338bf945cccfff9e4ce7cc6aa2246851b84aTimo Sirainen                if (i == min_pos && (msg[size-1] == '\r' ||
4ce6338bf945cccfff9e4ce7cc6aa2246851b84aTimo Sirainen                             msg[size-1] == '\n')) {
beffc30d933c5e134c45cc871852a8427eba7e70Timo Sirainen                    /* we may or may not have a full header,
beffc30d933c5e134c45cc871852a8427eba7e70Timo Sirainen                       but we don't know until we get the
f89cb43088c8b46d12d66ac924724b53ab14ce66Timo Sirainen                       next character. leave out the
beffc30d933c5e134c45cc871852a8427eba7e70Timo Sirainen                       linefeed and finish the header on
f89cb43088c8b46d12d66ac924724b53ab14ce66Timo Sirainen                       the next run. */
f89cb43088c8b46d12d66ac924724b53ab14ce66Timo Sirainen                    size--;
f89cb43088c8b46d12d66ac924724b53ab14ce66Timo Sirainen                    if (size > 0 && msg[size-1] == '\r')
f89cb43088c8b46d12d66ac924724b53ab14ce66Timo Sirainen                        size--;
beffc30d933c5e134c45cc871852a8427eba7e70Timo Sirainen                }
f89cb43088c8b46d12d66ac924724b53ab14ce66Timo Sirainen
f89cb43088c8b46d12d66ac924724b53ab14ce66Timo Sirainen                continues = TRUE;
f89cb43088c8b46d12d66ac924724b53ab14ce66Timo Sirainen            }
f89cb43088c8b46d12d66ac924724b53ab14ce66Timo Sirainen            no_newline = TRUE;
f89cb43088c8b46d12d66ac924724b53ab14ce66Timo Sirainen            ctx->skip = size;
f89cb43088c8b46d12d66ac924724b53ab14ce66Timo Sirainen            break;
f89cb43088c8b46d12d66ac924724b53ab14ce66Timo Sirainen        }
f89cb43088c8b46d12d66ac924724b53ab14ce66Timo Sirainen
f89cb43088c8b46d12d66ac924724b53ab14ce66Timo Sirainen        /* find ':' */
f89cb43088c8b46d12d66ac924724b53ab14ce66Timo Sirainen        if (colon_pos == UINT_MAX) {
f89cb43088c8b46d12d66ac924724b53ab14ce66Timo Sirainen            for (i = startpos; i < parse_size; i++) {
f89cb43088c8b46d12d66ac924724b53ab14ce66Timo Sirainen                if (msg[i] > ':')
f89cb43088c8b46d12d66ac924724b53ab14ce66Timo Sirainen                    continue;
f89cb43088c8b46d12d66ac924724b53ab14ce66Timo Sirainen
f89cb43088c8b46d12d66ac924724b53ab14ce66Timo Sirainen                if (msg[i] == ':' && !ctx->skip_line) {
f89cb43088c8b46d12d66ac924724b53ab14ce66Timo Sirainen                    colon_pos = i;
f89cb43088c8b46d12d66ac924724b53ab14ce66Timo Sirainen                    line->full_value_offset =
f89cb43088c8b46d12d66ac924724b53ab14ce66Timo Sirainen                        ctx->input->v_offset + i + 1;
f89cb43088c8b46d12d66ac924724b53ab14ce66Timo Sirainen                    break;
f89cb43088c8b46d12d66ac924724b53ab14ce66Timo Sirainen                }
f89cb43088c8b46d12d66ac924724b53ab14ce66Timo Sirainen                if (msg[i] == '\n') {
f89cb43088c8b46d12d66ac924724b53ab14ce66Timo Sirainen                    /* end of headers, or error */
f89cb43088c8b46d12d66ac924724b53ab14ce66Timo Sirainen                    break;
f89cb43088c8b46d12d66ac924724b53ab14ce66Timo Sirainen                }
f89cb43088c8b46d12d66ac924724b53ab14ce66Timo Sirainen
f89cb43088c8b46d12d66ac924724b53ab14ce66Timo Sirainen                if (msg[i] == '\0')
f89cb43088c8b46d12d66ac924724b53ab14ce66Timo Sirainen                    ctx->has_nuls = TRUE;
f89cb43088c8b46d12d66ac924724b53ab14ce66Timo Sirainen            }
f89cb43088c8b46d12d66ac924724b53ab14ce66Timo Sirainen        } else {
f89cb43088c8b46d12d66ac924724b53ab14ce66Timo Sirainen            i = startpos;
f89cb43088c8b46d12d66ac924724b53ab14ce66Timo Sirainen        }
f89cb43088c8b46d12d66ac924724b53ab14ce66Timo Sirainen
f89cb43088c8b46d12d66ac924724b53ab14ce66Timo Sirainen        /* find '\n' */
f89cb43088c8b46d12d66ac924724b53ab14ce66Timo Sirainen        for (; i < parse_size; i++) {
f89cb43088c8b46d12d66ac924724b53ab14ce66Timo Sirainen            if (msg[i] <= '\n') {
f89cb43088c8b46d12d66ac924724b53ab14ce66Timo Sirainen                if (msg[i] == '\n')
f89cb43088c8b46d12d66ac924724b53ab14ce66Timo Sirainen                    break;
f89cb43088c8b46d12d66ac924724b53ab14ce66Timo Sirainen                if (msg[i] == '\0')
f89cb43088c8b46d12d66ac924724b53ab14ce66Timo Sirainen                    ctx->has_nuls = TRUE;
f89cb43088c8b46d12d66ac924724b53ab14ce66Timo Sirainen            }
f89cb43088c8b46d12d66ac924724b53ab14ce66Timo Sirainen        }
f89cb43088c8b46d12d66ac924724b53ab14ce66Timo Sirainen
beffc30d933c5e134c45cc871852a8427eba7e70Timo Sirainen        if (i < parse_size && i+1 == size && ret == -2) {
beffc30d933c5e134c45cc871852a8427eba7e70Timo Sirainen            /* we don't know if the line continues. */
f89cb43088c8b46d12d66ac924724b53ab14ce66Timo Sirainen            i++;
f89cb43088c8b46d12d66ac924724b53ab14ce66Timo Sirainen        } else if (i < parse_size) {
f89cb43088c8b46d12d66ac924724b53ab14ce66Timo Sirainen            /* got a line */
f89cb43088c8b46d12d66ac924724b53ab14ce66Timo Sirainen            if (ctx->skip_line) {
f89cb43088c8b46d12d66ac924724b53ab14ce66Timo Sirainen                /* skipping a line with a huge header name */
f89cb43088c8b46d12d66ac924724b53ab14ce66Timo Sirainen                if (ctx->hdr_size != NULL) {
f89cb43088c8b46d12d66ac924724b53ab14ce66Timo Sirainen                    ctx->hdr_size->lines++;
7026c16186f543e11af12b8b87f396006db93297Timo Sirainen                    ctx->hdr_size->physical_size += i + 1;
7026c16186f543e11af12b8b87f396006db93297Timo Sirainen                    ctx->hdr_size->virtual_size += i + 1;
f89cb43088c8b46d12d66ac924724b53ab14ce66Timo Sirainen                }
f89cb43088c8b46d12d66ac924724b53ab14ce66Timo Sirainen                if (i == 0 || msg[i-1] != '\r') {
f89cb43088c8b46d12d66ac924724b53ab14ce66Timo Sirainen                    /* missing CR */
f89cb43088c8b46d12d66ac924724b53ab14ce66Timo Sirainen                    if (ctx->hdr_size != NULL)
f89cb43088c8b46d12d66ac924724b53ab14ce66Timo Sirainen                        ctx->hdr_size->virtual_size++;
f89cb43088c8b46d12d66ac924724b53ab14ce66Timo Sirainen                }
f89cb43088c8b46d12d66ac924724b53ab14ce66Timo Sirainen
f89cb43088c8b46d12d66ac924724b53ab14ce66Timo Sirainen                i_stream_skip(ctx->input, i + 1);
f89cb43088c8b46d12d66ac924724b53ab14ce66Timo Sirainen                startpos = 0;
f89cb43088c8b46d12d66ac924724b53ab14ce66Timo Sirainen                ctx->skip_line = FALSE;
f89cb43088c8b46d12d66ac924724b53ab14ce66Timo Sirainen                continue;
f89cb43088c8b46d12d66ac924724b53ab14ce66Timo Sirainen            }
f89cb43088c8b46d12d66ac924724b53ab14ce66Timo Sirainen            continues = i+1 < size && IS_LWSP(msg[i+1]);
f89cb43088c8b46d12d66ac924724b53ab14ce66Timo Sirainen
f89cb43088c8b46d12d66ac924724b53ab14ce66Timo Sirainen            if (ctx->hdr_size != NULL)
f89cb43088c8b46d12d66ac924724b53ab14ce66Timo Sirainen                ctx->hdr_size->lines++;
f89cb43088c8b46d12d66ac924724b53ab14ce66Timo Sirainen            if (i == 0 || msg[i-1] != '\r') {
f89cb43088c8b46d12d66ac924724b53ab14ce66Timo Sirainen                /* missing CR */
f89cb43088c8b46d12d66ac924724b53ab14ce66Timo Sirainen                if (ctx->hdr_size != NULL)
f89cb43088c8b46d12d66ac924724b53ab14ce66Timo Sirainen                    ctx->hdr_size->virtual_size++;
f89cb43088c8b46d12d66ac924724b53ab14ce66Timo Sirainen                size = i;
f89cb43088c8b46d12d66ac924724b53ab14ce66Timo Sirainen            } else {
f89cb43088c8b46d12d66ac924724b53ab14ce66Timo Sirainen                size = i-1;
ab7b5b9286104974c2a572a499ccf8b56c5d2955Timo Sirainen                crlf_newline = TRUE;
f89cb43088c8b46d12d66ac924724b53ab14ce66Timo Sirainen            }
ab7b5b9286104974c2a572a499ccf8b56c5d2955Timo Sirainen
ab7b5b9286104974c2a572a499ccf8b56c5d2955Timo Sirainen            ctx->skip = i+1;
f89cb43088c8b46d12d66ac924724b53ab14ce66Timo Sirainen            break;
f89cb43088c8b46d12d66ac924724b53ab14ce66Timo Sirainen        }
ab7b5b9286104974c2a572a499ccf8b56c5d2955Timo Sirainen
f89cb43088c8b46d12d66ac924724b53ab14ce66Timo Sirainen        startpos = i;
f89cb43088c8b46d12d66ac924724b53ab14ce66Timo Sirainen    }
ab7b5b9286104974c2a572a499ccf8b56c5d2955Timo Sirainen
ab7b5b9286104974c2a572a499ccf8b56c5d2955Timo Sirainen    last_crlf = line->crlf_newline &&
ab7b5b9286104974c2a572a499ccf8b56c5d2955Timo Sirainen        (ctx->flags & MESSAGE_HEADER_PARSER_FLAG_DROP_CR) == 0;
ab7b5b9286104974c2a572a499ccf8b56c5d2955Timo Sirainen    last_no_newline = line->no_newline ||
f89cb43088c8b46d12d66ac924724b53ab14ce66Timo Sirainen        (ctx->flags & MESSAGE_HEADER_PARSER_FLAG_CLEAN_ONELINE) != 0;
f89cb43088c8b46d12d66ac924724b53ab14ce66Timo Sirainen
f89cb43088c8b46d12d66ac924724b53ab14ce66Timo Sirainen    line->continues = continues;
f89cb43088c8b46d12d66ac924724b53ab14ce66Timo Sirainen    line->continued = continued;
f89cb43088c8b46d12d66ac924724b53ab14ce66Timo Sirainen    line->crlf_newline = crlf_newline;
f89cb43088c8b46d12d66ac924724b53ab14ce66Timo Sirainen    line->no_newline = no_newline;
f89cb43088c8b46d12d66ac924724b53ab14ce66Timo Sirainen    if (size == 0) {
f89cb43088c8b46d12d66ac924724b53ab14ce66Timo Sirainen        /* end of headers */
f89cb43088c8b46d12d66ac924724b53ab14ce66Timo Sirainen        line->eoh = TRUE;
f89cb43088c8b46d12d66ac924724b53ab14ce66Timo Sirainen        line->name_len = line->value_len = line->full_value_len = 0;
4ce6338bf945cccfff9e4ce7cc6aa2246851b84aTimo Sirainen        line->name = ""; line->value = line->full_value = NULL;
f89cb43088c8b46d12d66ac924724b53ab14ce66Timo Sirainen        line->middle = NULL; line->middle_len = 0;
f89cb43088c8b46d12d66ac924724b53ab14ce66Timo Sirainen        line->full_value_offset = line->name_offset;
f89cb43088c8b46d12d66ac924724b53ab14ce66Timo Sirainen        line->continues = FALSE;
f89cb43088c8b46d12d66ac924724b53ab14ce66Timo Sirainen    } else if (line->continued) {
f89cb43088c8b46d12d66ac924724b53ab14ce66Timo Sirainen        line->value = msg;
f89cb43088c8b46d12d66ac924724b53ab14ce66Timo Sirainen        line->value_len = size;
f89cb43088c8b46d12d66ac924724b53ab14ce66Timo Sirainen    } else if (colon_pos == UINT_MAX) {
f89cb43088c8b46d12d66ac924724b53ab14ce66Timo Sirainen        /* missing ':', assume the whole line is name */
f89cb43088c8b46d12d66ac924724b53ab14ce66Timo Sirainen        line->value = NULL;
f89cb43088c8b46d12d66ac924724b53ab14ce66Timo Sirainen        line->value_len = 0;
f89cb43088c8b46d12d66ac924724b53ab14ce66Timo Sirainen
da2aa032ccfa8e7e4a4380ef738014549f4d2c2dTimo Sirainen        str_truncate(ctx->name, 0);
ab7b5b9286104974c2a572a499ccf8b56c5d2955Timo Sirainen        buffer_append(ctx->name, msg, size);
ab7b5b9286104974c2a572a499ccf8b56c5d2955Timo Sirainen        line->name = str_c(ctx->name);
9511a40d933181045343110c8101b75887062aaeTimo Sirainen        line->name_len = str_len(ctx->name);
f89cb43088c8b46d12d66ac924724b53ab14ce66Timo Sirainen
        line->middle = NULL;
        line->middle_len = 0;
    } else {
        size_t pos;

        line->value = msg + colon_pos+1;
        line->value_len = size - colon_pos - 1;
        if (ctx->flags & MESSAGE_HEADER_PARSER_FLAG_SKIP_INITIAL_LWSP) {
            /* get value. skip all LWSP after ':'. Note that
               RFC2822 doesn't say we should, but history behind
               it..

               Exception to this is if the value consists only of
               LWSP, then skip only the one LWSP after ':'. */
            for (pos = 0; pos < line->value_len; pos++) {
                if (!IS_LWSP(line->value[pos]))
                    break;
            }

            if (pos == line->value_len) {
                /* everything was LWSP */
                if (line->value_len > 0 &&
                    IS_LWSP(line->value[0]))
                    pos = 1;
            }
        } else {
            pos = line->value_len > 0 &&
                IS_LWSP(line->value[0]) ? 1 : 0;
        }

        line->value += pos;
        line->value_len -= pos;
        line->full_value_offset += pos;

        /* get name, skip LWSP before ':' */
        while (colon_pos > 0 && IS_LWSP(msg[colon_pos-1]))
            colon_pos--;

        str_truncate(ctx->name, 0);
        /* use buffer_append() so the name won't be truncated if there
           are NULs. */
        buffer_append(ctx->name, msg, colon_pos);
        str_append_c(ctx->name, '\0');

        /* keep middle stored also in ctx->name so it's available
           with use_full_value */
        line->middle = msg + colon_pos;
        line->middle_len = (size_t)(line->value - line->middle);
        str_append_n(ctx->name, line->middle, line->middle_len);

        line->name = str_c(ctx->name);
        line->name_len = colon_pos;
        line->middle = str_data(ctx->name) + line->name_len + 1;
    }

    if (!line->continued) {
        /* first header line. make a copy of the line since we can't
           really trust input stream not to lose it. */
        buffer_append(ctx->value_buf, line->value, line->value_len);
        line->value = line->full_value = ctx->value_buf->data;
        line->full_value_len = line->value_len;
    } else if (line->use_full_value) {
        /* continue saving the full value. */
        if (last_no_newline) {
            /* line is longer than fit into our buffer, so we
               were forced to break it into multiple
               message_header_lines */
        } else {
            if (last_crlf)
                buffer_append_c(ctx->value_buf, '\r');
            buffer_append_c(ctx->value_buf, '\n');
        }
        if ((ctx->flags & MESSAGE_HEADER_PARSER_FLAG_CLEAN_ONELINE) &&
            line->value_len > 0 && line->value[0] != ' ' &&
            IS_LWSP(line->value[0])) {
            buffer_append_c(ctx->value_buf, ' ');
            buffer_append(ctx->value_buf,
                      line->value + 1, line->value_len - 1);
        } else {
            buffer_append(ctx->value_buf,
                      line->value, line->value_len);
        }
        line->full_value = buffer_get_data(ctx->value_buf,
                           &line->full_value_len);
    } else {
        /* we didn't want full_value, and this is a continued line. */
        line->full_value = NULL;
        line->full_value_len = 0;
    }

    /* always reset it */
    line->use_full_value = FALSE;

    if (ctx->hdr_size != NULL) {
        ctx->hdr_size->physical_size += ctx->skip;
        ctx->hdr_size->virtual_size += ctx->skip;
    }

    *hdr_r = line;
    return 1;
}

bool message_parse_header_has_nuls(const struct message_header_parser_ctx *ctx)
{
    return ctx->has_nuls;
}

#undef message_parse_header
void message_parse_header(struct istream *input, struct message_size *hdr_size,
              enum message_header_parser_flags flags,
              message_header_callback_t *callback, void *context)
{
    struct message_header_parser_ctx *hdr_ctx;
    struct message_header_line *hdr;
    int ret;

    hdr_ctx = message_parse_header_init(input, hdr_size, flags);
    while ((ret = message_parse_header_next(hdr_ctx, &hdr)) > 0)
        callback(hdr, context);
    i_assert(ret != 0);
    message_parse_header_deinit(&hdr_ctx);

    /* call after the final skipping */
    callback(NULL, context);
}