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

	message-header-parser.c revision 15362cdf9df29fef8795e865957e17ec027a9ebf
2454dfa32c93c20a8522c6ed42fe057baaac9f9aStephan Bosch/* Copyright (c) 2002-2008 Dovecot authors, see the included COPYING file */
f8740ac53310cd28ba4ec6dc9e9ce6e9a3688f39Timo Sirainen
f8740ac53310cd28ba4ec6dc9e9ce6e9a3688f39Timo Sirainen#include "lib.h"
f8740ac53310cd28ba4ec6dc9e9ce6e9a3688f39Timo Sirainen#include "buffer.h"
f8740ac53310cd28ba4ec6dc9e9ce6e9a3688f39Timo Sirainen#include "istream.h"
f8740ac53310cd28ba4ec6dc9e9ce6e9a3688f39Timo Sirainen#include "str.h"
f8740ac53310cd28ba4ec6dc9e9ce6e9a3688f39Timo Sirainen#include "message-size.h"
f8740ac53310cd28ba4ec6dc9e9ce6e9a3688f39Timo Sirainen#include "message-header-parser.h"
f8740ac53310cd28ba4ec6dc9e9ce6e9a3688f39Timo Sirainen
b7324e421e2132cbbf753e6fdbe675bbaecdf929Timo Sirainenstruct message_header_parser_ctx {
f8740ac53310cd28ba4ec6dc9e9ce6e9a3688f39Timo Sirainen    struct message_header_line line;
f8740ac53310cd28ba4ec6dc9e9ce6e9a3688f39Timo Sirainen
f8740ac53310cd28ba4ec6dc9e9ce6e9a3688f39Timo Sirainen    struct istream *input;
c9099b08b3cae8a849098ca776b4363c6d5f5f36Timo Sirainen    struct message_size *hdr_size;
c9099b08b3cae8a849098ca776b4363c6d5f5f36Timo Sirainen
c9099b08b3cae8a849098ca776b4363c6d5f5f36Timo Sirainen    string_t *name;
c9099b08b3cae8a849098ca776b4363c6d5f5f36Timo Sirainen    buffer_t *value_buf;
c9099b08b3cae8a849098ca776b4363c6d5f5f36Timo Sirainen    size_t skip;
c9099b08b3cae8a849098ca776b4363c6d5f5f36Timo Sirainen
c9099b08b3cae8a849098ca776b4363c6d5f5f36Timo Sirainen    enum message_header_parser_flags flags;
c9099b08b3cae8a849098ca776b4363c6d5f5f36Timo Sirainen    unsigned int skip_line:1;
c9099b08b3cae8a849098ca776b4363c6d5f5f36Timo Sirainen    unsigned int has_nuls:1;
c9099b08b3cae8a849098ca776b4363c6d5f5f36Timo Sirainen};
f8740ac53310cd28ba4ec6dc9e9ce6e9a3688f39Timo Sirainen
f8740ac53310cd28ba4ec6dc9e9ce6e9a3688f39Timo Sirainenstruct message_header_parser_ctx *
f8740ac53310cd28ba4ec6dc9e9ce6e9a3688f39Timo Sirainenmessage_parse_header_init(struct istream *input, struct message_size *hdr_size,
f8740ac53310cd28ba4ec6dc9e9ce6e9a3688f39Timo Sirainen              enum message_header_parser_flags flags)
f8740ac53310cd28ba4ec6dc9e9ce6e9a3688f39Timo Sirainen{
f8740ac53310cd28ba4ec6dc9e9ce6e9a3688f39Timo Sirainen    struct message_header_parser_ctx *ctx;
f8740ac53310cd28ba4ec6dc9e9ce6e9a3688f39Timo Sirainen
f8740ac53310cd28ba4ec6dc9e9ce6e9a3688f39Timo Sirainen    ctx = i_new(struct message_header_parser_ctx, 1);
f8740ac53310cd28ba4ec6dc9e9ce6e9a3688f39Timo Sirainen    ctx->input = input;
f8740ac53310cd28ba4ec6dc9e9ce6e9a3688f39Timo Sirainen    ctx->hdr_size = hdr_size;
f8740ac53310cd28ba4ec6dc9e9ce6e9a3688f39Timo Sirainen    ctx->name = str_new(default_pool, 128);
f8740ac53310cd28ba4ec6dc9e9ce6e9a3688f39Timo Sirainen    ctx->flags = flags;
f8740ac53310cd28ba4ec6dc9e9ce6e9a3688f39Timo Sirainen
f8740ac53310cd28ba4ec6dc9e9ce6e9a3688f39Timo Sirainen    if (hdr_size != NULL)
f8740ac53310cd28ba4ec6dc9e9ce6e9a3688f39Timo Sirainen        memset(hdr_size, 0, sizeof(*hdr_size));
f8740ac53310cd28ba4ec6dc9e9ce6e9a3688f39Timo Sirainen    return ctx;
f8740ac53310cd28ba4ec6dc9e9ce6e9a3688f39Timo Sirainen}
f8740ac53310cd28ba4ec6dc9e9ce6e9a3688f39Timo Sirainen
92f5ea24e989266539e97c6fe59ede0565aec6fdTimo Sirainenvoid message_parse_header_deinit(struct message_header_parser_ctx **_ctx)
f8740ac53310cd28ba4ec6dc9e9ce6e9a3688f39Timo Sirainen{
fb79b36eb34532dbe67caf99eefe3660b8c841e0Timo Sirainen    struct message_header_parser_ctx *ctx = *_ctx;
7000810786f2959f02cd6d2f4151a9eb61ff5db8Timo Sirainen
92f5ea24e989266539e97c6fe59ede0565aec6fdTimo Sirainen    i_stream_skip(ctx->input, ctx->skip);
7000810786f2959f02cd6d2f4151a9eb61ff5db8Timo Sirainen    if (ctx->value_buf != NULL)
7000810786f2959f02cd6d2f4151a9eb61ff5db8Timo Sirainen        buffer_free(&ctx->value_buf);
1c0020171b04d14adc4966ed963361abc9a86787Timo Sirainen    str_free(&ctx->name);
1c0020171b04d14adc4966ed963361abc9a86787Timo Sirainen    i_free(ctx);
1c0020171b04d14adc4966ed963361abc9a86787Timo Sirainen
1c0020171b04d14adc4966ed963361abc9a86787Timo Sirainen    *_ctx = NULL;
1c0020171b04d14adc4966ed963361abc9a86787Timo Sirainen}
1c0020171b04d14adc4966ed963361abc9a86787Timo Sirainen
1c0020171b04d14adc4966ed963361abc9a86787Timo Sirainenint message_parse_header_next(struct message_header_parser_ctx *ctx,
1c0020171b04d14adc4966ed963361abc9a86787Timo Sirainen                  struct message_header_line **hdr_r)
1c0020171b04d14adc4966ed963361abc9a86787Timo Sirainen{
1c0020171b04d14adc4966ed963361abc9a86787Timo Sirainen        struct message_header_line *line = &ctx->line;
1c0020171b04d14adc4966ed963361abc9a86787Timo Sirainen    const unsigned char *msg;
df8046c9a4f6bc2a478ad1e74504d50f3110c906Timo Sirainen    size_t i, size, startpos, colon_pos, parse_size;
df8046c9a4f6bc2a478ad1e74504d50f3110c906Timo Sirainen    int ret;
7000810786f2959f02cd6d2f4151a9eb61ff5db8Timo Sirainen    bool continued, continues, last_no_newline, last_crlf;
abb5d20d3155db02a1afec4066d52707ba9d4e52Timo Sirainen    bool no_newline, crlf_newline;
abb5d20d3155db02a1afec4066d52707ba9d4e52Timo Sirainen
abb5d20d3155db02a1afec4066d52707ba9d4e52Timo Sirainen    *hdr_r = NULL;
df8046c9a4f6bc2a478ad1e74504d50f3110c906Timo Sirainen    if (line->eoh)
df8046c9a4f6bc2a478ad1e74504d50f3110c906Timo Sirainen        return -1;
df8046c9a4f6bc2a478ad1e74504d50f3110c906Timo Sirainen
abb5d20d3155db02a1afec4066d52707ba9d4e52Timo Sirainen    if (ctx->skip > 0) {
abb5d20d3155db02a1afec4066d52707ba9d4e52Timo Sirainen        i_stream_skip(ctx->input, ctx->skip);
abb5d20d3155db02a1afec4066d52707ba9d4e52Timo Sirainen        ctx->skip = 0;
df8046c9a4f6bc2a478ad1e74504d50f3110c906Timo Sirainen    }
df8046c9a4f6bc2a478ad1e74504d50f3110c906Timo Sirainen
df8046c9a4f6bc2a478ad1e74504d50f3110c906Timo Sirainen    if (line->continues) {
df8046c9a4f6bc2a478ad1e74504d50f3110c906Timo Sirainen        if (line->use_full_value && !line->continued) {
df8046c9a4f6bc2a478ad1e74504d50f3110c906Timo Sirainen            /* save the first line */
abb5d20d3155db02a1afec4066d52707ba9d4e52Timo Sirainen            if (ctx->value_buf != NULL)
abb5d20d3155db02a1afec4066d52707ba9d4e52Timo Sirainen                buffer_set_used_size(ctx->value_buf, 0);
abb5d20d3155db02a1afec4066d52707ba9d4e52Timo Sirainen            else {
df8046c9a4f6bc2a478ad1e74504d50f3110c906Timo Sirainen                ctx->value_buf =
df8046c9a4f6bc2a478ad1e74504d50f3110c906Timo Sirainen                    buffer_create_dynamic(default_pool,
df8046c9a4f6bc2a478ad1e74504d50f3110c906Timo Sirainen                                  4096);
df8046c9a4f6bc2a478ad1e74504d50f3110c906Timo Sirainen            }
abb5d20d3155db02a1afec4066d52707ba9d4e52Timo Sirainen            buffer_append(ctx->value_buf,
7000810786f2959f02cd6d2f4151a9eb61ff5db8Timo Sirainen                      line->value, line->value_len);
7000810786f2959f02cd6d2f4151a9eb61ff5db8Timo Sirainen        }
abb5d20d3155db02a1afec4066d52707ba9d4e52Timo Sirainen
d6a7cb184cc882a90aa3d9312082e0029f354ff6Timo Sirainen        colon_pos = 0;
abb5d20d3155db02a1afec4066d52707ba9d4e52Timo Sirainen    } else {
abb5d20d3155db02a1afec4066d52707ba9d4e52Timo Sirainen        /* new header line */
2aee623fcad4b931c27435ceaa495c3d3edd69b6Aki Tuomi        line->name_offset = ctx->input->v_offset;
d6a7cb184cc882a90aa3d9312082e0029f354ff6Timo Sirainen        colon_pos = UINT_MAX;
d6a7cb184cc882a90aa3d9312082e0029f354ff6Timo Sirainen    }
d6a7cb184cc882a90aa3d9312082e0029f354ff6Timo Sirainen
d6a7cb184cc882a90aa3d9312082e0029f354ff6Timo Sirainen    no_newline = FALSE;
d6a7cb184cc882a90aa3d9312082e0029f354ff6Timo Sirainen    crlf_newline = FALSE;
d6a7cb184cc882a90aa3d9312082e0029f354ff6Timo Sirainen    continued = line->continues;
d6a7cb184cc882a90aa3d9312082e0029f354ff6Timo Sirainen    continues = FALSE;
d6a7cb184cc882a90aa3d9312082e0029f354ff6Timo Sirainen
d6a7cb184cc882a90aa3d9312082e0029f354ff6Timo Sirainen    for (startpos = 0;;) {
d6a7cb184cc882a90aa3d9312082e0029f354ff6Timo Sirainen        ret = i_stream_read_data(ctx->input, &msg, &size, startpos+1);
d6a7cb184cc882a90aa3d9312082e0029f354ff6Timo Sirainen
d6a7cb184cc882a90aa3d9312082e0029f354ff6Timo Sirainen        if (ret >= 0) {
d6a7cb184cc882a90aa3d9312082e0029f354ff6Timo Sirainen            /* we want to know one byte in advance to find out
d6a7cb184cc882a90aa3d9312082e0029f354ff6Timo Sirainen               if it's multiline header */
d6a7cb184cc882a90aa3d9312082e0029f354ff6Timo Sirainen            parse_size = size == 0 ? 0 : size-1;
d6a7cb184cc882a90aa3d9312082e0029f354ff6Timo Sirainen        } else {
d6a7cb184cc882a90aa3d9312082e0029f354ff6Timo Sirainen            parse_size = size;
f5e8a76a128d4e92f0641135183c164fd5c5ce5eTimo Sirainen        }
f5e8a76a128d4e92f0641135183c164fd5c5ce5eTimo Sirainen
f5e8a76a128d4e92f0641135183c164fd5c5ce5eTimo Sirainen        if (ret <= 0 && startpos == parse_size) {
f5e8a76a128d4e92f0641135183c164fd5c5ce5eTimo Sirainen            if (ret == -1) {
f5e8a76a128d4e92f0641135183c164fd5c5ce5eTimo Sirainen                if (startpos > 0) {
f5e8a76a128d4e92f0641135183c164fd5c5ce5eTimo Sirainen                    /* header ended unexpectedly. */
f5e8a76a128d4e92f0641135183c164fd5c5ce5eTimo Sirainen                    no_newline = TRUE;
f5e8a76a128d4e92f0641135183c164fd5c5ce5eTimo Sirainen                    ctx->skip = startpos;
f5e8a76a128d4e92f0641135183c164fd5c5ce5eTimo Sirainen                    break;
f5e8a76a128d4e92f0641135183c164fd5c5ce5eTimo Sirainen                }
f5e8a76a128d4e92f0641135183c164fd5c5ce5eTimo Sirainen                /* error / EOF with no bytes */
cc935aff970ed6c24d136cc560c7e705a49d536cTimo Sirainen                return -1;
cc935aff970ed6c24d136cc560c7e705a49d536cTimo Sirainen            }
cc935aff970ed6c24d136cc560c7e705a49d536cTimo Sirainen            if (ret == 0 && !ctx->input->eof) {
cc935aff970ed6c24d136cc560c7e705a49d536cTimo Sirainen                /* stream is nonblocking - need more data */
cc935aff970ed6c24d136cc560c7e705a49d536cTimo Sirainen                return 0;
cc935aff970ed6c24d136cc560c7e705a49d536cTimo Sirainen            }
cc935aff970ed6c24d136cc560c7e705a49d536cTimo Sirainen
cc935aff970ed6c24d136cc560c7e705a49d536cTimo Sirainen            if (msg[0] == '\n' ||
cc935aff970ed6c24d136cc560c7e705a49d536cTimo Sirainen                (msg[0] == '\r' && size > 1 && msg[1] == '\n')) {
cc935aff970ed6c24d136cc560c7e705a49d536cTimo Sirainen                /* end of headers - this mostly happens just
cc935aff970ed6c24d136cc560c7e705a49d536cTimo Sirainen                   with mbox where headers are read separately
cc935aff970ed6c24d136cc560c7e705a49d536cTimo Sirainen                   from body */
cc935aff970ed6c24d136cc560c7e705a49d536cTimo Sirainen                size = 0;
cc935aff970ed6c24d136cc560c7e705a49d536cTimo Sirainen                if (ctx->hdr_size != NULL)
cc935aff970ed6c24d136cc560c7e705a49d536cTimo Sirainen                    ctx->hdr_size->lines++;
cc935aff970ed6c24d136cc560c7e705a49d536cTimo Sirainen                if (msg[0] == '\r') {
cc935aff970ed6c24d136cc560c7e705a49d536cTimo Sirainen                    ctx->skip = 2;
27e859cee42654bff801ba96677cfc4e4e0108c7Timo Sirainen                    crlf_newline = TRUE;
279c6b6d0b0a159c8533102e7e914db21dadcb03Timo Sirainen                } else {
279c6b6d0b0a159c8533102e7e914db21dadcb03Timo Sirainen                    ctx->skip = 1;
279c6b6d0b0a159c8533102e7e914db21dadcb03Timo Sirainen                    if (ctx->hdr_size != NULL)
279c6b6d0b0a159c8533102e7e914db21dadcb03Timo Sirainen                        ctx->hdr_size->virtual_size++;
279c6b6d0b0a159c8533102e7e914db21dadcb03Timo Sirainen                }
279c6b6d0b0a159c8533102e7e914db21dadcb03Timo Sirainen                break;
279c6b6d0b0a159c8533102e7e914db21dadcb03Timo Sirainen            }
279c6b6d0b0a159c8533102e7e914db21dadcb03Timo Sirainen
279c6b6d0b0a159c8533102e7e914db21dadcb03Timo Sirainen            /* a) line is larger than input buffer
279c6b6d0b0a159c8533102e7e914db21dadcb03Timo Sirainen               b) header ended unexpectedly */
387f9e3b4120273ad0213206a0e9cc2dc0e62ccaTimo Sirainen            if (colon_pos == UINT_MAX && ret == -2 && !continued) {
27e859cee42654bff801ba96677cfc4e4e0108c7Timo Sirainen                /* header name is huge. just skip it. */
27e859cee42654bff801ba96677cfc4e4e0108c7Timo Sirainen                if (msg[size-1] == '\r')
27e859cee42654bff801ba96677cfc4e4e0108c7Timo Sirainen                    size--;
27e859cee42654bff801ba96677cfc4e4e0108c7Timo Sirainen
071543cc13df9600d2e97aa35f28907be5a79477Timo Sirainen                if (ctx->hdr_size != NULL) {
27e859cee42654bff801ba96677cfc4e4e0108c7Timo Sirainen                    ctx->hdr_size->physical_size += size;
632018810af689442569cbb0139c55868923ccfeTimo Sirainen                    ctx->hdr_size->virtual_size += size;
387f9e3b4120273ad0213206a0e9cc2dc0e62ccaTimo Sirainen                }
632018810af689442569cbb0139c55868923ccfeTimo Sirainen                i_stream_skip(ctx->input, size);
632018810af689442569cbb0139c55868923ccfeTimo Sirainen                ctx->skip_line = TRUE;
632018810af689442569cbb0139c55868923ccfeTimo Sirainen                startpos = 0;
632018810af689442569cbb0139c55868923ccfeTimo Sirainen                continue;
632018810af689442569cbb0139c55868923ccfeTimo Sirainen            }
632018810af689442569cbb0139c55868923ccfeTimo Sirainen
632018810af689442569cbb0139c55868923ccfeTimo Sirainen            if (ret == -2) {
632018810af689442569cbb0139c55868923ccfeTimo Sirainen                /* go back to last LWSP if found. */
632018810af689442569cbb0139c55868923ccfeTimo Sirainen                size_t min_pos = !continued ? colon_pos : 0;
387f9e3b4120273ad0213206a0e9cc2dc0e62ccaTimo Sirainen                for (i = size-1; i > min_pos; i--) {
632018810af689442569cbb0139c55868923ccfeTimo Sirainen                    if (IS_LWSP(msg[i])) {
632018810af689442569cbb0139c55868923ccfeTimo Sirainen                        size = i;
632018810af689442569cbb0139c55868923ccfeTimo Sirainen                        break;
632018810af689442569cbb0139c55868923ccfeTimo Sirainen                    }
8d3af185ae454653fad60e41c5f36edb1d45c868Timo Sirainen                }
8d3af185ae454653fad60e41c5f36edb1d45c868Timo Sirainen
8d3af185ae454653fad60e41c5f36edb1d45c868Timo Sirainen                continues = TRUE;
8d3af185ae454653fad60e41c5f36edb1d45c868Timo Sirainen            }
8d3af185ae454653fad60e41c5f36edb1d45c868Timo Sirainen            no_newline = TRUE;
8d3af185ae454653fad60e41c5f36edb1d45c868Timo Sirainen            ctx->skip = size;
8d3af185ae454653fad60e41c5f36edb1d45c868Timo Sirainen            break;
8d3af185ae454653fad60e41c5f36edb1d45c868Timo Sirainen        }
8d3af185ae454653fad60e41c5f36edb1d45c868Timo Sirainen
dea8bfa31729dbdde1b12718f1ef98fac4e99db9Timo Sirainen        /* find ':' */
dea8bfa31729dbdde1b12718f1ef98fac4e99db9Timo Sirainen        if (colon_pos == UINT_MAX) {
dea8bfa31729dbdde1b12718f1ef98fac4e99db9Timo Sirainen            for (i = startpos; i < parse_size; i++) {
dea8bfa31729dbdde1b12718f1ef98fac4e99db9Timo Sirainen                if (msg[i] > ':')
dea8bfa31729dbdde1b12718f1ef98fac4e99db9Timo Sirainen                    continue;
dea8bfa31729dbdde1b12718f1ef98fac4e99db9Timo Sirainen
dea8bfa31729dbdde1b12718f1ef98fac4e99db9Timo Sirainen                if (msg[i] == ':') {
dea8bfa31729dbdde1b12718f1ef98fac4e99db9Timo Sirainen                    colon_pos = i;
dea8bfa31729dbdde1b12718f1ef98fac4e99db9Timo Sirainen                    line->full_value_offset =
dea8bfa31729dbdde1b12718f1ef98fac4e99db9Timo Sirainen                        ctx->input->v_offset + i + 1;
dea8bfa31729dbdde1b12718f1ef98fac4e99db9Timo Sirainen                    break;
dea8bfa31729dbdde1b12718f1ef98fac4e99db9Timo Sirainen                }
dea8bfa31729dbdde1b12718f1ef98fac4e99db9Timo Sirainen                if (msg[i] == '\n') {
dea8bfa31729dbdde1b12718f1ef98fac4e99db9Timo Sirainen                    /* end of headers, or error */
dea8bfa31729dbdde1b12718f1ef98fac4e99db9Timo Sirainen                    break;
dea8bfa31729dbdde1b12718f1ef98fac4e99db9Timo Sirainen                }
f8740ac53310cd28ba4ec6dc9e9ce6e9a3688f39Timo Sirainen
f8740ac53310cd28ba4ec6dc9e9ce6e9a3688f39Timo Sirainen                if (msg[i] == '\0')
f8740ac53310cd28ba4ec6dc9e9ce6e9a3688f39Timo Sirainen                    ctx->has_nuls = TRUE;
f8740ac53310cd28ba4ec6dc9e9ce6e9a3688f39Timo Sirainen            }
f8740ac53310cd28ba4ec6dc9e9ce6e9a3688f39Timo Sirainen        } else {
f8740ac53310cd28ba4ec6dc9e9ce6e9a3688f39Timo Sirainen            i = startpos;
f8740ac53310cd28ba4ec6dc9e9ce6e9a3688f39Timo Sirainen        }
f8740ac53310cd28ba4ec6dc9e9ce6e9a3688f39Timo Sirainen
f8740ac53310cd28ba4ec6dc9e9ce6e9a3688f39Timo Sirainen        /* find '\n' */
f8740ac53310cd28ba4ec6dc9e9ce6e9a3688f39Timo Sirainen        for (; i < parse_size; i++) {
f8740ac53310cd28ba4ec6dc9e9ce6e9a3688f39Timo Sirainen            if (msg[i] <= '\n') {
f8740ac53310cd28ba4ec6dc9e9ce6e9a3688f39Timo Sirainen                if (msg[i] == '\n')
f8740ac53310cd28ba4ec6dc9e9ce6e9a3688f39Timo Sirainen                    break;
f8740ac53310cd28ba4ec6dc9e9ce6e9a3688f39Timo Sirainen                if (msg[i] == '\0')
f8740ac53310cd28ba4ec6dc9e9ce6e9a3688f39Timo Sirainen                    ctx->has_nuls = TRUE;
f8740ac53310cd28ba4ec6dc9e9ce6e9a3688f39Timo Sirainen            }
f8740ac53310cd28ba4ec6dc9e9ce6e9a3688f39Timo Sirainen        }
f8740ac53310cd28ba4ec6dc9e9ce6e9a3688f39Timo Sirainen
f8740ac53310cd28ba4ec6dc9e9ce6e9a3688f39Timo Sirainen        if (i < parse_size) {
f8740ac53310cd28ba4ec6dc9e9ce6e9a3688f39Timo Sirainen            /* got a line */
f8740ac53310cd28ba4ec6dc9e9ce6e9a3688f39Timo Sirainen            if (ctx->skip_line) {
f8740ac53310cd28ba4ec6dc9e9ce6e9a3688f39Timo Sirainen                /* skipping a huge line */
f8740ac53310cd28ba4ec6dc9e9ce6e9a3688f39Timo Sirainen                if (ctx->hdr_size != NULL) {
f8740ac53310cd28ba4ec6dc9e9ce6e9a3688f39Timo Sirainen                    ctx->hdr_size->physical_size += i;
f8740ac53310cd28ba4ec6dc9e9ce6e9a3688f39Timo Sirainen                    ctx->hdr_size->virtual_size += i;
f8740ac53310cd28ba4ec6dc9e9ce6e9a3688f39Timo Sirainen                }
f8740ac53310cd28ba4ec6dc9e9ce6e9a3688f39Timo Sirainen
f8740ac53310cd28ba4ec6dc9e9ce6e9a3688f39Timo Sirainen                if (i == 0 || msg[i-1] != '\r') {
f8740ac53310cd28ba4ec6dc9e9ce6e9a3688f39Timo Sirainen                    /* missing CR */
f8740ac53310cd28ba4ec6dc9e9ce6e9a3688f39Timo Sirainen                    if (ctx->hdr_size != NULL)
f8740ac53310cd28ba4ec6dc9e9ce6e9a3688f39Timo Sirainen                        ctx->hdr_size->virtual_size++;
f8740ac53310cd28ba4ec6dc9e9ce6e9a3688f39Timo Sirainen                } else {
f8740ac53310cd28ba4ec6dc9e9ce6e9a3688f39Timo Sirainen                    crlf_newline = TRUE;
f8740ac53310cd28ba4ec6dc9e9ce6e9a3688f39Timo Sirainen                }
f8740ac53310cd28ba4ec6dc9e9ce6e9a3688f39Timo Sirainen                i_stream_skip(ctx->input, i);
f8740ac53310cd28ba4ec6dc9e9ce6e9a3688f39Timo Sirainen                startpos = 0;
f8740ac53310cd28ba4ec6dc9e9ce6e9a3688f39Timo Sirainen                ctx->skip_line = FALSE;
6de0cbb4a498ce0519b0f28e221164ce8d39736aTimo Sirainen                continue;
6de0cbb4a498ce0519b0f28e221164ce8d39736aTimo Sirainen            }
6de0cbb4a498ce0519b0f28e221164ce8d39736aTimo Sirainen            continues = i+1 < size && IS_LWSP(msg[i+1]);
6de0cbb4a498ce0519b0f28e221164ce8d39736aTimo Sirainen
6de0cbb4a498ce0519b0f28e221164ce8d39736aTimo Sirainen            if (ctx->hdr_size != NULL)
6de0cbb4a498ce0519b0f28e221164ce8d39736aTimo Sirainen                ctx->hdr_size->lines++;
6de0cbb4a498ce0519b0f28e221164ce8d39736aTimo Sirainen            if (i == 0 || msg[i-1] != '\r') {
6de0cbb4a498ce0519b0f28e221164ce8d39736aTimo Sirainen                /* missing CR */
6de0cbb4a498ce0519b0f28e221164ce8d39736aTimo Sirainen                if (ctx->hdr_size != NULL)
6de0cbb4a498ce0519b0f28e221164ce8d39736aTimo Sirainen                    ctx->hdr_size->virtual_size++;
6de0cbb4a498ce0519b0f28e221164ce8d39736aTimo Sirainen                size = i;
6de0cbb4a498ce0519b0f28e221164ce8d39736aTimo Sirainen            } else {
6de0cbb4a498ce0519b0f28e221164ce8d39736aTimo Sirainen                size = i-1;
f8740ac53310cd28ba4ec6dc9e9ce6e9a3688f39Timo Sirainen                crlf_newline = TRUE;
f8740ac53310cd28ba4ec6dc9e9ce6e9a3688f39Timo Sirainen            }
baf3e87e186453fda13bd21f7cbcb2efc8492e8bTimo Sirainen
f8740ac53310cd28ba4ec6dc9e9ce6e9a3688f39Timo Sirainen            ctx->skip = i+1;
6de0cbb4a498ce0519b0f28e221164ce8d39736aTimo Sirainen            break;
f8740ac53310cd28ba4ec6dc9e9ce6e9a3688f39Timo Sirainen        }
f8740ac53310cd28ba4ec6dc9e9ce6e9a3688f39Timo Sirainen
f8740ac53310cd28ba4ec6dc9e9ce6e9a3688f39Timo Sirainen        startpos = i;
f8740ac53310cd28ba4ec6dc9e9ce6e9a3688f39Timo Sirainen    }
f8740ac53310cd28ba4ec6dc9e9ce6e9a3688f39Timo Sirainen
    last_crlf = line->crlf_newline &&
        (ctx->flags & MESSAGE_HEADER_PARSER_FLAG_DROP_CR) == 0;
    last_no_newline = line->no_newline ||
        (ctx->flags & MESSAGE_HEADER_PARSER_FLAG_CLEAN_ONELINE) != 0;

    line->continues = continues;
    line->continued = continued;
    line->crlf_newline = crlf_newline;
    line->no_newline = no_newline;
    if (size == 0) {
        /* end of headers */
        line->eoh = TRUE;
        line->name_len = line->value_len = line->full_value_len = 0;
        line->name = ""; line->value = line->full_value = NULL;
        line->middle = NULL; line->middle_len = 0;
    } else if (line->continued) {
        line->value = msg;
        line->value_len = size;
    } else if (colon_pos == UINT_MAX) {
        /* missing ':', assume the whole line is name */
        line->value = NULL;
        line->value_len = 0;

        str_truncate(ctx->name, 0);
        str_append_n(ctx->name, msg, size);
        line->name = str_c(ctx->name);
        line->name_len = str_len(ctx->name);

        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[0]))
                    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);
        str_append_n(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, set full_value = value */
        line->full_value = line->value;
        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(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);
}