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

	message-header-parser.c revision 51c331377beb4a2acb81aee4d12bc8ef6c496625
bcb4e51a409d94ae670de96afb8483a4f7855294Stephan Bosch/* Copyright (C) 2002-2006 Timo Sirainen */
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch#include "lib.h"
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch#include "buffer.h"
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch#include "istream.h"
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch#include "str.h"
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch#include "message-size.h"
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch#include "message-header-parser.h"
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Boschstruct message_header_parser_ctx {
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch    struct message_header_line line;
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch    struct istream *input;
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch    struct message_size *hdr_size;
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch    string_t *name;
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch    buffer_t *value_buf;
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch    size_t skip;
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch    unsigned int skip_initial_lwsp:1;
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch    unsigned int skip_line:1;
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch    unsigned int has_nuls:1;
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch};
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Boschstruct message_header_parser_ctx *
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Boschmessage_parse_header_init(struct istream *input, struct message_size *hdr_size,
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch              bool skip_initial_lwsp)
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch{
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch    struct message_header_parser_ctx *ctx;
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch    ctx = i_new(struct message_header_parser_ctx, 1);
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch    ctx->input = input;
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch    ctx->hdr_size = hdr_size;
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch    ctx->name = str_new(default_pool, 128);
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch    ctx->skip_initial_lwsp = skip_initial_lwsp;
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch    if (hdr_size != NULL)
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch        memset(hdr_size, 0, sizeof(*hdr_size));
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch    return ctx;
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch}
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Boschvoid message_parse_header_deinit(struct message_header_parser_ctx **_ctx)
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch{
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch    struct message_header_parser_ctx *ctx = *_ctx;
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch    i_stream_skip(ctx->input, ctx->skip);
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch    if (ctx->value_buf != NULL)
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch        buffer_free(ctx->value_buf);
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch    str_free(&ctx->name);
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch    i_free(ctx);
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch    *_ctx = NULL;
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch}
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Boschint message_parse_header_next(struct message_header_parser_ctx *ctx,
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch                  struct message_header_line **hdr_r)
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch{
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch        struct message_header_line *line = &ctx->line;
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch    const unsigned char *msg;
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch    size_t i, size, startpos, colon_pos, parse_size;
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch    int ret;
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch    bool last_no_newline;
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch    *hdr_r = NULL;
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch    if (line->eoh)
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch        return -1;
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch    if (ctx->skip > 0) {
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch        i_stream_skip(ctx->input, ctx->skip);
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch        ctx->skip = 0;
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch    }
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch    startpos = 0; colon_pos = UINT_MAX;
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch    last_no_newline = line->no_newline;
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch    line->no_newline = FALSE;
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch    line->crlf_newline = FALSE;
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch    if (line->continues) {
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch        if (line->use_full_value && !line->continued) {
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch            /* save the first line */
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch            if (ctx->value_buf != NULL)
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch                buffer_set_used_size(ctx->value_buf, 0);
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch            else {
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch                ctx->value_buf =
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch                    buffer_create_dynamic(default_pool,
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch                                  4096);
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch            }
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch            buffer_append(ctx->value_buf,
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch                      line->value, line->value_len);
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch        }
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch        line->continued = TRUE;
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch        line->continues = FALSE;
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch        colon_pos = 0;
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch    } else {
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch        /* new header line */
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch        line->continued = FALSE;
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch                line->name_offset = ctx->input->v_offset;
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch    }
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch    for (;;) {
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch        ret = i_stream_read_data(ctx->input, &msg, &size, startpos+1);
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch        if (ret > 0) {
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch            /* we want to know one byte in advance to find out
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch               if it's multiline header */
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch            parse_size = size-1;
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch        } else {
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch            parse_size = size;
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch        }
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch        if (ret <= 0 && startpos == size) {
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch            if (ret == -1) {
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch                if (startpos > 0) {
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch                    /* header ended unexpectedly. */
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch                    line->no_newline = TRUE;
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch                    ctx->skip = startpos;
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch                    break;
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch                }
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch                /* error / EOF with no bytes */
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch                return -1;
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch            }
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch            if (ret == 0 && !ctx->input->eof) {
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch                /* stream is nonblocking - need more data */
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch                return 0;
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch            }
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch            if (msg[0] == '\n' ||
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch                (msg[0] == '\r' && size > 1 && msg[1] == '\n')) {
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch                /* end of headers - this mostly happens just
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch                   with mbox where headers are read separately
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch                   from body */
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch                size = 0;
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch                if (ctx->hdr_size != NULL)
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch                    ctx->hdr_size->lines++;
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch                if (msg[0] == '\r') {
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch                    ctx->skip = 2;
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch                    line->crlf_newline = TRUE;
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch                } else {
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch                    ctx->skip = 1;
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch                    if (ctx->hdr_size != NULL)
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch                        ctx->hdr_size->virtual_size++;
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch                }
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch                break;
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch            }
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch            /* a) line is larger than input buffer
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch               b) header ended unexpectedly */
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch            if (colon_pos == UINT_MAX && ret == -2 &&
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch                !line->continued) {
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch                /* header name is huge. just skip it. */
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch                if (msg[size-1] == '\r')
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch                    size--;
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch                if (ctx->hdr_size != NULL) {
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch                    ctx->hdr_size->physical_size += size;
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch                    ctx->hdr_size->virtual_size += size;
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch                }
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch                i_stream_skip(ctx->input, size);
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch                ctx->skip_line = TRUE;
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch                startpos = 0;
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch                continue;
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch            }
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch            if (ret == -2) {
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch                /* go back to last LWSP if found. */
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch                size_t min_pos =
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch                    !line->continued ? colon_pos : 0;
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch                for (i = size-1; i > min_pos; i--) {
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch                    if (IS_LWSP(msg[i])) {
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch                        size = i;
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch                        break;
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch                    }
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch                }
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch                line->continues = TRUE;
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch            }
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch            line->no_newline = TRUE;
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch            ctx->skip = size;
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch            break;
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch        }
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch        /* find ':' */
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch        if (colon_pos == UINT_MAX) {
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch            for (i = startpos; i < parse_size; i++) {
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch                if (msg[i] <= ':') {
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch                    if (msg[i] == ':') {
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch                        colon_pos = i;
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch                        line->full_value_offset =
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch                            ctx->input->v_offset +
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch                            i + 1;
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch                        break;
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch                    }
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch                    if (msg[i] == '\n') {
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch                        /* end of headers, or error */
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch                        break;
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch                    }
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch                    if (msg[i] == '\0')
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch                        ctx->has_nuls = TRUE;
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch                }
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch            }
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch        }
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch        /* find '\n' */
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch        for (i = startpos; i < parse_size; i++) {
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch            if (msg[i] <= '\n') {
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch                if (msg[i] == '\n')
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch                    break;
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch                if (msg[i] == '\0')
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch                    ctx->has_nuls = TRUE;
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch            }
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch        }
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch        if (i < parse_size) {
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch            /* got a line */
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch            if (ctx->skip_line) {
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch                /* skipping a huge line */
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch                if (ctx->hdr_size != NULL) {
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch                    ctx->hdr_size->physical_size += i;
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch                    ctx->hdr_size->virtual_size += i;
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch                }
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch                if (i == 0 || msg[i-1] != '\r') {
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch                    /* missing CR */
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch                    if (ctx->hdr_size != NULL)
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch                        ctx->hdr_size->virtual_size++;
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch                } else {
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch                    line->crlf_newline = TRUE;
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch                }
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch                i_stream_skip(ctx->input, i);
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch                startpos = 0;
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch                ctx->skip_line = FALSE;
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch                continue;
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch            }
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch            line->continues = i+1 < size && IS_LWSP(msg[i+1]);
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch            if (ctx->hdr_size != NULL)
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch                ctx->hdr_size->lines++;
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch            if (i == 0 || msg[i-1] != '\r') {
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch                /* missing CR */
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch                if (ctx->hdr_size != NULL)
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch                    ctx->hdr_size->virtual_size++;
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch                size = i;
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch            } else {
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch                size = i-1;
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch                line->crlf_newline = TRUE;
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch            }
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch            ctx->skip = i+1;
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch            break;
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch        }
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch        startpos = i;
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch    }
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch    if (size == 0) {
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch        /* end of headers */
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch        line->eoh = TRUE;
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch        line->name_len = line->value_len = line->full_value_len = 0;
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch        line->name = ""; line->value = line->full_value = NULL;
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch        line->middle = NULL; line->middle_len = 0;
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch    } else if (line->continued) {
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch        line->value = msg;
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch        line->value_len = size;
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch        line->middle = NULL;
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch        line->middle_len = 0;
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch    } else if (colon_pos == UINT_MAX) {
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch        /* missing ':', assume the whole line is name */
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch        line->value = NULL;
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch        line->value_len = 0;
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch        str_truncate(ctx->name, 0);
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch        str_append_n(ctx->name, msg, size);
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch        line->name = str_c(ctx->name);
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch        line->name_len = str_len(ctx->name);
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch        line->middle = NULL;
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch        line->middle_len = 0;
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch    } else {
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch        size_t pos;
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch        line->value = msg + colon_pos+1;
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch        line->value_len = size - colon_pos - 1;
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch        if (ctx->skip_initial_lwsp) {
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch            /* get value. skip all LWSP after ':'. Note that
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch               RFC2822 doesn't say we should, but history behind
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch               it..
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch               Exception to this is if the value consists only of
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch               LWSP, then skip only the one LWSP after ':'. */
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch            for (pos = 0; pos < line->value_len; pos++) {
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch                if (!IS_LWSP(line->value[0]))
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch                    break;
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch            }
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch            if (pos == line->value_len) {
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch                /* everything was LWSP */
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch                if (line->value_len > 0 &&
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch                    IS_LWSP(line->value[0]))
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch                    pos = 1;
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch            }
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch        } else {
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch            pos = line->value_len > 0 &&
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch                IS_LWSP(line->value[0]) ? 1 : 0;
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch        }
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch        line->value += pos;
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch        line->value_len -= pos;
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch        line->full_value_offset += pos;
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch        /* get name, skip LWSP before ':' */
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch        while (colon_pos > 0 && IS_LWSP(msg[colon_pos-1]))
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch            colon_pos--;
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch        line->middle = msg + colon_pos;
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch        line->middle_len = (size_t)(line->value - line->middle);
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch        str_truncate(ctx->name, 0);
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch        str_append_n(ctx->name, msg, colon_pos);
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch        line->name = str_c(ctx->name);
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch        line->name_len = str_len(ctx->name);
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch    }
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch    if (!line->continued) {
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch        /* first header line, set full_value = value */
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch        line->full_value = line->value;
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch        line->full_value_len = line->value_len;
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch    } else if (line->use_full_value) {
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch        /* continue saving the full value */
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch        if (!last_no_newline)
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch            buffer_append_c(ctx->value_buf, '\n');
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch        buffer_append(ctx->value_buf, line->value, line->value_len);
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch        line->full_value = buffer_get_data(ctx->value_buf,
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch                           &line->full_value_len);
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch    } else {
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch        /* we didn't want full_value, and this is a continued line. */
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch        line->full_value = NULL;
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch        line->full_value_len = 0;
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch    }
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch    /* always reset it */
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch    line->use_full_value = FALSE;
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch    if (ctx->hdr_size != NULL) {
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch        ctx->hdr_size->physical_size += ctx->skip;
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch        ctx->hdr_size->virtual_size += ctx->skip;
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch    }
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch    *hdr_r = line;
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch    return 1;
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch}
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Boschbool message_parse_header_has_nuls(struct message_header_parser_ctx *ctx)
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch{
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch    return ctx->has_nuls;
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch}
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Boschvoid message_parse_header(struct istream *input, struct message_size *hdr_size,
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch              message_header_callback_t *callback, void *context)
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch{
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch    struct message_header_parser_ctx *hdr_ctx;
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch    struct message_header_line *hdr;
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch    int ret;
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch    hdr_ctx = message_parse_header_init(input, hdr_size, TRUE);
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch    while ((ret = message_parse_header_next(hdr_ctx, &hdr)) > 0)
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch        callback(hdr, context);
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch    i_assert(ret != 0);
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch    message_parse_header_deinit(&hdr_ctx);
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch    /* call after the final skipping */
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch    callback(NULL, context);
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch}
03f08e4abef2afc782069bdfef305f84e98f0922Stephan Bosch