src/lib-mail/message-parser.c

	message-parser.c revision f755df78a17a7c003a182594ab78b0325d929ad4
a8c5a86d183db25a57bf193c06b41e092ec2e151Timo Sirainen/* Copyright (C) 2002 Timo Sirainen */
ecc81625167ed96c04c02aa190a1ea5baa65b474Timo Sirainen
ecc81625167ed96c04c02aa190a1ea5baa65b474Timo Sirainen#include "lib.h"
07e4875d250e7a7157cd99132aafc773cf3cdf83Timo Sirainen#include "iobuffer.h"
1a0ece3e873e3864269ed7eaed957dc10c56d25fTimo Sirainen#include "rfc822-tokenize.h"
e0c3d5460d1cc0c440cb7723c8c2eef8d0afe9b9Timo Sirainen#include "message-content-parser.h"
0536ccb51d41e3078c3a9fa33e509fb4b2420f95Timo Sirainen#include "message-parser.h"
ecc81625167ed96c04c02aa190a1ea5baa65b474Timo Sirainen#include "message-size.h"
8cb72c59d5ea4e9e5f638d7ec840bb853f5a188eTimo Sirainen
8cb72c59d5ea4e9e5f638d7ec840bb853f5a188eTimo Sirainentypedef struct _MessageBoundary {
8cb72c59d5ea4e9e5f638d7ec840bb853f5a188eTimo Sirainen    struct _MessageBoundary *next;
8cb72c59d5ea4e9e5f638d7ec840bb853f5a188eTimo Sirainen
8cb72c59d5ea4e9e5f638d7ec840bb853f5a188eTimo Sirainen    MessagePart *part;
8cb72c59d5ea4e9e5f638d7ec840bb853f5a188eTimo Sirainen    const char *boundary;
8cb72c59d5ea4e9e5f638d7ec840bb853f5a188eTimo Sirainen    unsigned int len;
8cb72c59d5ea4e9e5f638d7ec840bb853f5a188eTimo Sirainen} MessageBoundary;
8cb72c59d5ea4e9e5f638d7ec840bb853f5a188eTimo Sirainen
8cb72c59d5ea4e9e5f638d7ec840bb853f5a188eTimo Sirainentypedef struct {
8cb72c59d5ea4e9e5f638d7ec840bb853f5a188eTimo Sirainen    Pool pool;
8cb72c59d5ea4e9e5f638d7ec840bb853f5a188eTimo Sirainen    MessagePart *part;
8cb72c59d5ea4e9e5f638d7ec840bb853f5a188eTimo Sirainen
8cb72c59d5ea4e9e5f638d7ec840bb853f5a188eTimo Sirainen    char *last_boundary;
8cb72c59d5ea4e9e5f638d7ec840bb853f5a188eTimo Sirainen    char *last_content_type;
8cb72c59d5ea4e9e5f638d7ec840bb853f5a188eTimo Sirainen    MessageBoundary *boundaries;
e2ce8d4a6ac5d82a906178148453e7613fab9ba0Timo Sirainen
e2ce8d4a6ac5d82a906178148453e7613fab9ba0Timo Sirainen    MessageHeaderFunc func;
e2ce8d4a6ac5d82a906178148453e7613fab9ba0Timo Sirainen    void *context;
e2ce8d4a6ac5d82a906178148453e7613fab9ba0Timo Sirainen} MessageParseContext;
e2ce8d4a6ac5d82a906178148453e7613fab9ba0Timo Sirainen
e2ce8d4a6ac5d82a906178148453e7613fab9ba0Timo Sirainenstatic MessagePart *message_parse_part(IOBuffer *inbuf,
e2ce8d4a6ac5d82a906178148453e7613fab9ba0Timo Sirainen                       MessageParseContext *parse_ctx);
e2ce8d4a6ac5d82a906178148453e7613fab9ba0Timo Sirainenstatic MessagePart *message_parse_body(IOBuffer *inbuf,
e2ce8d4a6ac5d82a906178148453e7613fab9ba0Timo Sirainen                       MessageBoundary *boundaries,
cd56a23e21f1df3f79648cf07e2f4385e2fadebbTimo Sirainen                       MessageSize *body_size);
cd56a23e21f1df3f79648cf07e2f4385e2fadebbTimo Sirainenstatic MessagePart *message_skip_boundary(IOBuffer *inbuf,
e2ce8d4a6ac5d82a906178148453e7613fab9ba0Timo Sirainen                      MessageBoundary *boundaries,
cd56a23e21f1df3f79648cf07e2f4385e2fadebbTimo Sirainen                      MessageSize *boundary_size);
cd56a23e21f1df3f79648cf07e2f4385e2fadebbTimo Sirainen
cd56a23e21f1df3f79648cf07e2f4385e2fadebbTimo Sirainenstatic void message_size_add_part(MessageSize *dest, MessagePart *part)
c0435c854a0e7246373b9752d163095cc4fbe985Timo Sirainen{
ecc81625167ed96c04c02aa190a1ea5baa65b474Timo Sirainen    dest->physical_size +=
252db51b6c0a605163326b3ea5d09e9936ca3b29Timo Sirainen        part->header_size.physical_size +
ecc81625167ed96c04c02aa190a1ea5baa65b474Timo Sirainen        part->body_size.physical_size;
ecc81625167ed96c04c02aa190a1ea5baa65b474Timo Sirainen    dest->virtual_size +=
d5cebe7f98e63d4e2822863ef2faa4971e8b3a5dTimo Sirainen        part->header_size.virtual_size +
ecc81625167ed96c04c02aa190a1ea5baa65b474Timo Sirainen        part->body_size.virtual_size;
252db51b6c0a605163326b3ea5d09e9936ca3b29Timo Sirainen    dest->lines += part->header_size.lines + part->body_size.lines;
e0c3d5460d1cc0c440cb7723c8c2eef8d0afe9b9Timo Sirainen}
e0c3d5460d1cc0c440cb7723c8c2eef8d0afe9b9Timo Sirainen
e0c3d5460d1cc0c440cb7723c8c2eef8d0afe9b9Timo Sirainenstatic MessagePart *message_part_append(Pool pool, MessagePart *parent)
d5cebe7f98e63d4e2822863ef2faa4971e8b3a5dTimo Sirainen{
e0c3d5460d1cc0c440cb7723c8c2eef8d0afe9b9Timo Sirainen    MessagePart *part, **list;
252db51b6c0a605163326b3ea5d09e9936ca3b29Timo Sirainen
d5cebe7f98e63d4e2822863ef2faa4971e8b3a5dTimo Sirainen    part = p_new(pool, MessagePart, 1);
ecc81625167ed96c04c02aa190a1ea5baa65b474Timo Sirainen    part->parent = parent;
ecc81625167ed96c04c02aa190a1ea5baa65b474Timo Sirainen
1a0ece3e873e3864269ed7eaed957dc10c56d25fTimo Sirainen    list = &part->parent->children;
1a0ece3e873e3864269ed7eaed957dc10c56d25fTimo Sirainen    while (*list != NULL)
5ac0b0bf32898c63da086ae169674ecac151a31eTimo Sirainen        list = &(*list)->next;
5ac0b0bf32898c63da086ae169674ecac151a31eTimo Sirainen
5ac0b0bf32898c63da086ae169674ecac151a31eTimo Sirainen    *list = part;
1a0ece3e873e3864269ed7eaed957dc10c56d25fTimo Sirainen    return part;
5ac0b0bf32898c63da086ae169674ecac151a31eTimo Sirainen}
1a0ece3e873e3864269ed7eaed957dc10c56d25fTimo Sirainen
1a0ece3e873e3864269ed7eaed957dc10c56d25fTimo Sirainenstatic void parse_content_type(const Rfc822Token *tokens, int count,
1a0ece3e873e3864269ed7eaed957dc10c56d25fTimo Sirainen                   void *context)
1a0ece3e873e3864269ed7eaed957dc10c56d25fTimo Sirainen{
1a0ece3e873e3864269ed7eaed957dc10c56d25fTimo Sirainen    MessageParseContext *parse_ctx = context;
5ac0b0bf32898c63da086ae169674ecac151a31eTimo Sirainen    const char *str;
5ac0b0bf32898c63da086ae169674ecac151a31eTimo Sirainen
1a0ece3e873e3864269ed7eaed957dc10c56d25fTimo Sirainen    if (tokens[0].token != 'A')
1a0ece3e873e3864269ed7eaed957dc10c56d25fTimo Sirainen        return;
c28f6aa0b70af4811c9ace9114fe827c2f503455Timo Sirainen
c28f6aa0b70af4811c9ace9114fe827c2f503455Timo Sirainen    if (parse_ctx->last_content_type != NULL)
1a0ece3e873e3864269ed7eaed957dc10c56d25fTimo Sirainen        return;
1a0ece3e873e3864269ed7eaed957dc10c56d25fTimo Sirainen
c28f6aa0b70af4811c9ace9114fe827c2f503455Timo Sirainen    str = rfc822_tokens_get_value(tokens, count);
1a0ece3e873e3864269ed7eaed957dc10c56d25fTimo Sirainen    parse_ctx->last_content_type = p_strdup(parse_ctx->pool, str);
1a0ece3e873e3864269ed7eaed957dc10c56d25fTimo Sirainen
df831edaa3b3aa22e03bc5fd416a0553c5600a69Phil Carmody    if (strcasecmp(str, "message/rfc822") == 0)
1a0ece3e873e3864269ed7eaed957dc10c56d25fTimo Sirainen        parse_ctx->part->flags |= MESSAGE_PART_FLAG_MESSAGE_RFC822;
1a0ece3e873e3864269ed7eaed957dc10c56d25fTimo Sirainen    else if (strncasecmp(str, "text/", 5) == 0)
1a0ece3e873e3864269ed7eaed957dc10c56d25fTimo Sirainen        parse_ctx->part->flags |= MESSAGE_PART_FLAG_TEXT;
1a0ece3e873e3864269ed7eaed957dc10c56d25fTimo Sirainen    else if (strncasecmp(str, "multipart/", 10) == 0) {
1a0ece3e873e3864269ed7eaed957dc10c56d25fTimo Sirainen        parse_ctx->part->flags |= MESSAGE_PART_FLAG_MULTIPART;
c28f6aa0b70af4811c9ace9114fe827c2f503455Timo Sirainen
c28f6aa0b70af4811c9ace9114fe827c2f503455Timo Sirainen        if (strcasecmp(str+10, "digest") == 0) {
c0435c854a0e7246373b9752d163095cc4fbe985Timo Sirainen            parse_ctx->part->flags |=
ecc81625167ed96c04c02aa190a1ea5baa65b474Timo Sirainen                MESSAGE_PART_FLAG_MULTIPART_DIGEST;
252db51b6c0a605163326b3ea5d09e9936ca3b29Timo Sirainen        }
ecc81625167ed96c04c02aa190a1ea5baa65b474Timo Sirainen    }
ecc81625167ed96c04c02aa190a1ea5baa65b474Timo Sirainen}
ecc81625167ed96c04c02aa190a1ea5baa65b474Timo Sirainen
ecc81625167ed96c04c02aa190a1ea5baa65b474Timo Sirainenstatic void parse_content_type_param(const Rfc822Token *name,
862ec874f9373e3e499e237d3b9f71fdf1413feeTimo Sirainen                     const Rfc822Token *value,
862ec874f9373e3e499e237d3b9f71fdf1413feeTimo Sirainen                     int value_count, void *context)
862ec874f9373e3e499e237d3b9f71fdf1413feeTimo Sirainen{
862ec874f9373e3e499e237d3b9f71fdf1413feeTimo Sirainen    MessageParseContext *parse_ctx = context;
862ec874f9373e3e499e237d3b9f71fdf1413feeTimo Sirainen    const char *str;
862ec874f9373e3e499e237d3b9f71fdf1413feeTimo Sirainen
862ec874f9373e3e499e237d3b9f71fdf1413feeTimo Sirainen    if ((parse_ctx->part->flags & MESSAGE_PART_FLAG_MULTIPART) == 0 ||
862ec874f9373e3e499e237d3b9f71fdf1413feeTimo Sirainen        name->len != 8 || strncasecmp(name->ptr, "boundary", 8) != 0)
862ec874f9373e3e499e237d3b9f71fdf1413feeTimo Sirainen        return;
862ec874f9373e3e499e237d3b9f71fdf1413feeTimo Sirainen
862ec874f9373e3e499e237d3b9f71fdf1413feeTimo Sirainen    if (parse_ctx->last_boundary == NULL) {
862ec874f9373e3e499e237d3b9f71fdf1413feeTimo Sirainen        str = rfc822_tokens_get_value(value, value_count);
862ec874f9373e3e499e237d3b9f71fdf1413feeTimo Sirainen        parse_ctx->last_boundary = p_strdup(parse_ctx->pool, str);
862ec874f9373e3e499e237d3b9f71fdf1413feeTimo Sirainen    }
862ec874f9373e3e499e237d3b9f71fdf1413feeTimo Sirainen}
862ec874f9373e3e499e237d3b9f71fdf1413feeTimo Sirainen
862ec874f9373e3e499e237d3b9f71fdf1413feeTimo Sirainenstatic void parse_header_field(MessagePart *part,
862ec874f9373e3e499e237d3b9f71fdf1413feeTimo Sirainen                   const char *name, unsigned int name_len,
c0435c854a0e7246373b9752d163095cc4fbe985Timo Sirainen                   const char *value, unsigned int value_len,
ecc81625167ed96c04c02aa190a1ea5baa65b474Timo Sirainen                   void *context)
e2ce8d4a6ac5d82a906178148453e7613fab9ba0Timo Sirainen{
ecc81625167ed96c04c02aa190a1ea5baa65b474Timo Sirainen    MessageParseContext *parse_ctx = context;
ecc81625167ed96c04c02aa190a1ea5baa65b474Timo Sirainen
7662010b03ffe5f2a6ecf4b4eb220d1c65efea76Timo Sirainen    /* call the user-defined header parser */
7662010b03ffe5f2a6ecf4b4eb220d1c65efea76Timo Sirainen    if (parse_ctx->func != NULL) {
7662010b03ffe5f2a6ecf4b4eb220d1c65efea76Timo Sirainen        parse_ctx->func(part, name, name_len, value, value_len,
7662010b03ffe5f2a6ecf4b4eb220d1c65efea76Timo Sirainen                parse_ctx->context);
7662010b03ffe5f2a6ecf4b4eb220d1c65efea76Timo Sirainen    }
c0435c854a0e7246373b9752d163095cc4fbe985Timo Sirainen
ecc81625167ed96c04c02aa190a1ea5baa65b474Timo Sirainen    if (name_len == 12 && strncasecmp(name, "Content-Type", 12) == 0) {
252db51b6c0a605163326b3ea5d09e9936ca3b29Timo Sirainen        /* we need to know the boundary */
ecc81625167ed96c04c02aa190a1ea5baa65b474Timo Sirainen        (void)message_content_parse_header(t_strndup(value, value_len),
ecc81625167ed96c04c02aa190a1ea5baa65b474Timo Sirainen                           parse_content_type,
0a49b316fc729e5d57268ffa63c7122ac73f994cTimo Sirainen                           parse_content_type_param,
0a49b316fc729e5d57268ffa63c7122ac73f994cTimo Sirainen                           parse_ctx);
0a49b316fc729e5d57268ffa63c7122ac73f994cTimo Sirainen    }
0a49b316fc729e5d57268ffa63c7122ac73f994cTimo Sirainen}
0a49b316fc729e5d57268ffa63c7122ac73f994cTimo Sirainen
51e1a1c280ccb461a15827f7987d09cb9708b6e3Timo Sirainenstatic MessagePart *message_parse_multipart(IOBuffer *inbuf,
51e1a1c280ccb461a15827f7987d09cb9708b6e3Timo Sirainen                        MessageParseContext *parse_ctx)
51e1a1c280ccb461a15827f7987d09cb9708b6e3Timo Sirainen{
51e1a1c280ccb461a15827f7987d09cb9708b6e3Timo Sirainen    MessagePart *parent_part, *next_part, *part;
51e1a1c280ccb461a15827f7987d09cb9708b6e3Timo Sirainen    MessageBoundary *b;
4334b9b032298defd4d3906f5357698ff016ead0Timo Sirainen
4334b9b032298defd4d3906f5357698ff016ead0Timo Sirainen    /* multipart message. add new boundary */
4334b9b032298defd4d3906f5357698ff016ead0Timo Sirainen    b = t_new(MessageBoundary, 1);
4334b9b032298defd4d3906f5357698ff016ead0Timo Sirainen    b->part = parse_ctx->part;
4334b9b032298defd4d3906f5357698ff016ead0Timo Sirainen    b->boundary = parse_ctx->last_boundary;
4334b9b032298defd4d3906f5357698ff016ead0Timo Sirainen    b->len = strlen(b->boundary);
4334b9b032298defd4d3906f5357698ff016ead0Timo Sirainen
4334b9b032298defd4d3906f5357698ff016ead0Timo Sirainen    b->next = parse_ctx->boundaries;
4334b9b032298defd4d3906f5357698ff016ead0Timo Sirainen    parse_ctx->boundaries = b;
4334b9b032298defd4d3906f5357698ff016ead0Timo Sirainen
4334b9b032298defd4d3906f5357698ff016ead0Timo Sirainen    /* reset fields */
c0435c854a0e7246373b9752d163095cc4fbe985Timo Sirainen    parse_ctx->last_boundary = NULL;
ecc81625167ed96c04c02aa190a1ea5baa65b474Timo Sirainen    parse_ctx->last_content_type = NULL;
252db51b6c0a605163326b3ea5d09e9936ca3b29Timo Sirainen
0892446b45c195461bb7be6599f02d97e1e2c9b2Timo Sirainen    /* skip the data before the first boundary */
c09f9f95db314e7482c95e502e1c56ed6c555797Timo Sirainen    parent_part = parse_ctx->part;
ecc81625167ed96c04c02aa190a1ea5baa65b474Timo Sirainen    next_part = message_skip_boundary(inbuf, parse_ctx->boundaries,
a2738cdb6d2733fb3e186331d68009421a19ea00Timo Sirainen                      &parent_part->body_size);
d66ef20c30fee728899ee168c75fcc5ff8fbdac1Timo Sirainen
ecc81625167ed96c04c02aa190a1ea5baa65b474Timo Sirainen    /* now, parse the parts */
d66ef20c30fee728899ee168c75fcc5ff8fbdac1Timo Sirainen    while (next_part == parent_part) {
ecc81625167ed96c04c02aa190a1ea5baa65b474Timo Sirainen        /* new child */
9e095dd6a77097356aca8216356d4d71ef1bea45Timo Sirainen        part = message_part_append(parse_ctx->pool, parent_part);
c09f9f95db314e7482c95e502e1c56ed6c555797Timo Sirainen
4334b9b032298defd4d3906f5357698ff016ead0Timo Sirainen        /* set child position */
4334b9b032298defd4d3906f5357698ff016ead0Timo Sirainen        part->physical_pos =
4334b9b032298defd4d3906f5357698ff016ead0Timo Sirainen            parent_part->physical_pos +
0892446b45c195461bb7be6599f02d97e1e2c9b2Timo Sirainen            parent_part->body_size.physical_size +
c09f9f95db314e7482c95e502e1c56ed6c555797Timo Sirainen            parent_part->header_size.physical_size;
7af4788b402346c94496095dd819f95ce03fe431Timo Sirainen
0a0cd45a633112a2ae6aad801c1e6afe53ab95deTimo Sirainen                parse_ctx->part = part;
0a0cd45a633112a2ae6aad801c1e6afe53ab95deTimo Sirainen        next_part = message_parse_part(inbuf, parse_ctx);
0a0cd45a633112a2ae6aad801c1e6afe53ab95deTimo Sirainen
0a0cd45a633112a2ae6aad801c1e6afe53ab95deTimo Sirainen        /* update our size */
0a0cd45a633112a2ae6aad801c1e6afe53ab95deTimo Sirainen        message_size_add_part(&parent_part->body_size, part);
c09f9f95db314e7482c95e502e1c56ed6c555797Timo Sirainen
c09f9f95db314e7482c95e502e1c56ed6c555797Timo Sirainen        if (next_part != parent_part)
c09f9f95db314e7482c95e502e1c56ed6c555797Timo Sirainen            break;
c09f9f95db314e7482c95e502e1c56ed6c555797Timo Sirainen
d66ef20c30fee728899ee168c75fcc5ff8fbdac1Timo Sirainen        /* skip the boundary */
c09f9f95db314e7482c95e502e1c56ed6c555797Timo Sirainen        next_part = message_skip_boundary(inbuf, parse_ctx->boundaries,
c09f9f95db314e7482c95e502e1c56ed6c555797Timo Sirainen                          &parent_part->body_size);
c24ef531ca58abad996482f5c2e8992be9ae8981Timo Sirainen    }
c09f9f95db314e7482c95e502e1c56ed6c555797Timo Sirainen
0a0cd45a633112a2ae6aad801c1e6afe53ab95deTimo Sirainen    /* remove boundary */
0a0cd45a633112a2ae6aad801c1e6afe53ab95deTimo Sirainen    i_assert(parse_ctx->boundaries == b);
0a0cd45a633112a2ae6aad801c1e6afe53ab95deTimo Sirainen    parse_ctx->boundaries = b->next;
0a0cd45a633112a2ae6aad801c1e6afe53ab95deTimo Sirainen    return next_part;
e4c81823af1fc43ca3f2ce9eb4af7fc8f57b13a5Timo Sirainen}
e4c81823af1fc43ca3f2ce9eb4af7fc8f57b13a5Timo Sirainen
2524ef7b34965a1b1895d6140fd8296bf57c78d2Timo Sirainenstatic MessagePart *message_parse_part(IOBuffer *inbuf,
0892446b45c195461bb7be6599f02d97e1e2c9b2Timo Sirainen                       MessageParseContext *parse_ctx)
e4c81823af1fc43ca3f2ce9eb4af7fc8f57b13a5Timo Sirainen{
c09f9f95db314e7482c95e502e1c56ed6c555797Timo Sirainen    MessagePart *next_part, *part;
4334b9b032298defd4d3906f5357698ff016ead0Timo Sirainen    uoff_t hdr_size;
4334b9b032298defd4d3906f5357698ff016ead0Timo Sirainen
c09f9f95db314e7482c95e502e1c56ed6c555797Timo Sirainen    message_parse_header(parse_ctx->part, inbuf,
ecc81625167ed96c04c02aa190a1ea5baa65b474Timo Sirainen                 &parse_ctx->part->header_size,
ecc81625167ed96c04c02aa190a1ea5baa65b474Timo Sirainen                 parse_header_field, parse_ctx);
61e6367a259e2473f33df42fda8ceeb3b8b48416Timo Sirainen
61e6367a259e2473f33df42fda8ceeb3b8b48416Timo Sirainen    /* update message position/size */
61e6367a259e2473f33df42fda8ceeb3b8b48416Timo Sirainen    hdr_size = parse_ctx->part->header_size.physical_size;
61e6367a259e2473f33df42fda8ceeb3b8b48416Timo Sirainen
61e6367a259e2473f33df42fda8ceeb3b8b48416Timo Sirainen    if (parse_ctx->last_boundary != NULL)
61e6367a259e2473f33df42fda8ceeb3b8b48416Timo Sirainen        return message_parse_multipart(inbuf, parse_ctx);
61e6367a259e2473f33df42fda8ceeb3b8b48416Timo Sirainen
61e6367a259e2473f33df42fda8ceeb3b8b48416Timo Sirainen    if (parse_ctx->last_content_type == NULL) {
61e6367a259e2473f33df42fda8ceeb3b8b48416Timo Sirainen        if (parse_ctx->part->parent != NULL &&
61e6367a259e2473f33df42fda8ceeb3b8b48416Timo Sirainen            (parse_ctx->part->parent->flags &
61e6367a259e2473f33df42fda8ceeb3b8b48416Timo Sirainen             MESSAGE_PART_FLAG_MULTIPART_DIGEST)) {
61e6367a259e2473f33df42fda8ceeb3b8b48416Timo Sirainen            /* when there's no content-type specified and we're
61e6367a259e2473f33df42fda8ceeb3b8b48416Timo Sirainen               below multipart/digest, the assume message/rfc822
66dc739bb67d678770e1b7a7bc75f4f6f9523d2aTimo Sirainen               content-type */
66dc739bb67d678770e1b7a7bc75f4f6f9523d2aTimo Sirainen            parse_ctx->part->flags |=
61e6367a259e2473f33df42fda8ceeb3b8b48416Timo Sirainen                MESSAGE_PART_FLAG_MESSAGE_RFC822;
66dc739bb67d678770e1b7a7bc75f4f6f9523d2aTimo Sirainen        } else {
61e6367a259e2473f33df42fda8ceeb3b8b48416Timo Sirainen            /* otherwise we default to text/plain */
61e6367a259e2473f33df42fda8ceeb3b8b48416Timo Sirainen            parse_ctx->part->flags |= MESSAGE_PART_FLAG_TEXT;
61e6367a259e2473f33df42fda8ceeb3b8b48416Timo Sirainen        }
61e6367a259e2473f33df42fda8ceeb3b8b48416Timo Sirainen    }
61e6367a259e2473f33df42fda8ceeb3b8b48416Timo Sirainen
61e6367a259e2473f33df42fda8ceeb3b8b48416Timo Sirainen    parse_ctx->last_boundary = NULL;
61e6367a259e2473f33df42fda8ceeb3b8b48416Timo Sirainen        parse_ctx->last_content_type = NULL;
61e6367a259e2473f33df42fda8ceeb3b8b48416Timo Sirainen
61e6367a259e2473f33df42fda8ceeb3b8b48416Timo Sirainen    if (parse_ctx->part->flags & MESSAGE_PART_FLAG_MESSAGE_RFC822) {
61e6367a259e2473f33df42fda8ceeb3b8b48416Timo Sirainen        /* message/rfc822 part - the message body begins with
61e6367a259e2473f33df42fda8ceeb3b8b48416Timo Sirainen           headers again, this works pretty much the same as
61e6367a259e2473f33df42fda8ceeb3b8b48416Timo Sirainen           a single multipart/mixed item */
61e6367a259e2473f33df42fda8ceeb3b8b48416Timo Sirainen        part = message_part_append(parse_ctx->pool, parse_ctx->part);
61e6367a259e2473f33df42fda8ceeb3b8b48416Timo Sirainen
c6b6ac7819931dfa92c0182ffaa7db07ac6ab0daTimo Sirainen        parse_ctx->part = part;
61e6367a259e2473f33df42fda8ceeb3b8b48416Timo Sirainen        next_part = message_parse_part(inbuf, parse_ctx);
61e6367a259e2473f33df42fda8ceeb3b8b48416Timo Sirainen        parse_ctx->part = part->parent;
61e6367a259e2473f33df42fda8ceeb3b8b48416Timo Sirainen
c0435c854a0e7246373b9752d163095cc4fbe985Timo Sirainen        /* our body size is the size of header+body in message/rfc822 */
ecc81625167ed96c04c02aa190a1ea5baa65b474Timo Sirainen        message_size_add_part(&part->parent->body_size, part);
252db51b6c0a605163326b3ea5d09e9936ca3b29Timo Sirainen    } else {
ecc81625167ed96c04c02aa190a1ea5baa65b474Timo Sirainen        /* normal message, read until the next boundary */
ecc81625167ed96c04c02aa190a1ea5baa65b474Timo Sirainen        part = parse_ctx->part;
20344c0e814139e3c365fbb9287478f91512089eTimo Sirainen        next_part = message_parse_body(inbuf, parse_ctx->boundaries,
20344c0e814139e3c365fbb9287478f91512089eTimo Sirainen                           &part->body_size);
dd62b77c932d1b518f2a3e4bf80e36542becc256Timo Sirainen    }
ecc81625167ed96c04c02aa190a1ea5baa65b474Timo Sirainen
ecc81625167ed96c04c02aa190a1ea5baa65b474Timo Sirainen    return next_part;
ecc81625167ed96c04c02aa190a1ea5baa65b474Timo Sirainen}
ecc81625167ed96c04c02aa190a1ea5baa65b474Timo Sirainen
ecc81625167ed96c04c02aa190a1ea5baa65b474Timo SirainenMessagePart *message_parse(Pool pool, IOBuffer *inbuf,
dd62b77c932d1b518f2a3e4bf80e36542becc256Timo Sirainen               MessageHeaderFunc func, void *context)
ecc81625167ed96c04c02aa190a1ea5baa65b474Timo Sirainen{
20344c0e814139e3c365fbb9287478f91512089eTimo Sirainen    MessagePart *part;
ecc81625167ed96c04c02aa190a1ea5baa65b474Timo Sirainen    MessageParseContext parse_ctx;
ecc81625167ed96c04c02aa190a1ea5baa65b474Timo Sirainen
c9bf63e9094761767a63ac6b189bcf60bcffdc44Timo Sirainen    memset(&parse_ctx, 0, sizeof(parse_ctx));
dd62b77c932d1b518f2a3e4bf80e36542becc256Timo Sirainen    parse_ctx.pool = pool;
dd62b77c932d1b518f2a3e4bf80e36542becc256Timo Sirainen    parse_ctx.func = func;
c09f9f95db314e7482c95e502e1c56ed6c555797Timo Sirainen    parse_ctx.context = context;
602a0434db30d8e3292d1c161a803d96a879a74fTimo Sirainen    parse_ctx.part = part = p_new(pool, MessagePart, 1);
ecc81625167ed96c04c02aa190a1ea5baa65b474Timo Sirainen
ecc81625167ed96c04c02aa190a1ea5baa65b474Timo Sirainen    t_push();
4334b9b032298defd4d3906f5357698ff016ead0Timo Sirainen    message_parse_part(inbuf, &parse_ctx);
e3fc1874694a8ddba9552ec23f9952f74f33d1d5Timo Sirainen    t_pop();
4334b9b032298defd4d3906f5357698ff016ead0Timo Sirainen    return part;
e3fc1874694a8ddba9552ec23f9952f74f33d1d5Timo Sirainen}
e3fc1874694a8ddba9552ec23f9952f74f33d1d5Timo Sirainen
4334b9b032298defd4d3906f5357698ff016ead0Timo Sirainen/* skip over to next line increasing message size */
4334b9b032298defd4d3906f5357698ff016ead0Timo Sirainenstatic void message_skip_line(IOBuffer *inbuf, MessageSize *msg_size)
4334b9b032298defd4d3906f5357698ff016ead0Timo Sirainen{
e3fc1874694a8ddba9552ec23f9952f74f33d1d5Timo Sirainen    unsigned char *msg;
e3fc1874694a8ddba9552ec23f9952f74f33d1d5Timo Sirainen    unsigned int i, size, startpos;
4334b9b032298defd4d3906f5357698ff016ead0Timo Sirainen
e3fc1874694a8ddba9552ec23f9952f74f33d1d5Timo Sirainen    startpos = 0;
e3fc1874694a8ddba9552ec23f9952f74f33d1d5Timo Sirainen
c0435c854a0e7246373b9752d163095cc4fbe985Timo Sirainen    while (io_buffer_read_data(inbuf, &msg, &size, startpos) >= 0) {
ecc81625167ed96c04c02aa190a1ea5baa65b474Timo Sirainen        for (i = startpos; i < size; i++) {
252db51b6c0a605163326b3ea5d09e9936ca3b29Timo Sirainen            if (msg[i] == '\n') {
ecc81625167ed96c04c02aa190a1ea5baa65b474Timo Sirainen                if (msg_size != NULL) {
e3fc1874694a8ddba9552ec23f9952f74f33d1d5Timo Sirainen                    if (i == 0 || msg[i-1] != '\r')
4334b9b032298defd4d3906f5357698ff016ead0Timo Sirainen                        msg_size->virtual_size++;
601f5f14c6cde28f0e0c6ca7c5d735315d3d48dfTimo Sirainen                    msg_size->lines++;
4334b9b032298defd4d3906f5357698ff016ead0Timo Sirainen                }
4334b9b032298defd4d3906f5357698ff016ead0Timo Sirainen                break;
4334b9b032298defd4d3906f5357698ff016ead0Timo Sirainen            }
ecc81625167ed96c04c02aa190a1ea5baa65b474Timo Sirainen        }
4334b9b032298defd4d3906f5357698ff016ead0Timo Sirainen
4334b9b032298defd4d3906f5357698ff016ead0Timo Sirainen        if (i < size) {
4334b9b032298defd4d3906f5357698ff016ead0Timo Sirainen            startpos = i+1;
4334b9b032298defd4d3906f5357698ff016ead0Timo Sirainen            break;
602a0434db30d8e3292d1c161a803d96a879a74fTimo Sirainen        }
602a0434db30d8e3292d1c161a803d96a879a74fTimo Sirainen
602a0434db30d8e3292d1c161a803d96a879a74fTimo Sirainen        if (i > 0) {
602a0434db30d8e3292d1c161a803d96a879a74fTimo Sirainen            /* leave the last character, it may be \r */
252db51b6c0a605163326b3ea5d09e9936ca3b29Timo Sirainen            io_buffer_skip(inbuf, i - 1);
602a0434db30d8e3292d1c161a803d96a879a74fTimo Sirainen            startpos = 1;
c9bf63e9094761767a63ac6b189bcf60bcffdc44Timo Sirainen
602a0434db30d8e3292d1c161a803d96a879a74fTimo Sirainen            if (msg_size != NULL) {
602a0434db30d8e3292d1c161a803d96a879a74fTimo Sirainen                msg_size->physical_size += i - 1;
602a0434db30d8e3292d1c161a803d96a879a74fTimo Sirainen                msg_size->virtual_size += i - 1;
602a0434db30d8e3292d1c161a803d96a879a74fTimo Sirainen            }
4334b9b032298defd4d3906f5357698ff016ead0Timo Sirainen        }
ecc81625167ed96c04c02aa190a1ea5baa65b474Timo Sirainen    }
ecc81625167ed96c04c02aa190a1ea5baa65b474Timo Sirainen
07e4875d250e7a7157cd99132aafc773cf3cdf83Timo Sirainen    io_buffer_skip(inbuf, startpos);
dd62b77c932d1b518f2a3e4bf80e36542becc256Timo Sirainen
252db51b6c0a605163326b3ea5d09e9936ca3b29Timo Sirainen    if (msg_size != NULL) {
dd62b77c932d1b518f2a3e4bf80e36542becc256Timo Sirainen        msg_size->physical_size += startpos;
c9bf63e9094761767a63ac6b189bcf60bcffdc44Timo Sirainen        msg_size->virtual_size += startpos;
c9bf63e9094761767a63ac6b189bcf60bcffdc44Timo Sirainen    }
c9bf63e9094761767a63ac6b189bcf60bcffdc44Timo Sirainen}
4334b9b032298defd4d3906f5357698ff016ead0Timo Sirainen
07e4875d250e7a7157cd99132aafc773cf3cdf83Timo Sirainenvoid message_parse_header(MessagePart *part, IOBuffer *inbuf,
4334b9b032298defd4d3906f5357698ff016ead0Timo Sirainen              MessageSize *hdr_size,
4334b9b032298defd4d3906f5357698ff016ead0Timo Sirainen              MessageHeaderFunc func, void *context)
07e4875d250e7a7157cd99132aafc773cf3cdf83Timo Sirainen{
07e4875d250e7a7157cd99132aafc773cf3cdf83Timo Sirainen    unsigned char *msg;
01f4ee4a0243f3fe9af763e1a540cd5cff0d63f5Timo Sirainen    unsigned int i, size, startpos, missing_cr_count;
07e4875d250e7a7157cd99132aafc773cf3cdf83Timo Sirainen    unsigned int line_start, colon_pos, end_pos, name_len, value_len;
252db51b6c0a605163326b3ea5d09e9936ca3b29Timo Sirainen    int ret;
07e4875d250e7a7157cd99132aafc773cf3cdf83Timo Sirainen
c9bf63e9094761767a63ac6b189bcf60bcffdc44Timo Sirainen    if (hdr_size != NULL)
01f4ee4a0243f3fe9af763e1a540cd5cff0d63f5Timo Sirainen        memset(hdr_size, 0, sizeof(MessageSize));
548e394330621952db0f03dd667b70184c4a37b6Timo Sirainen
01f4ee4a0243f3fe9af763e1a540cd5cff0d63f5Timo Sirainen    missing_cr_count = startpos = line_start = 0;
01f4ee4a0243f3fe9af763e1a540cd5cff0d63f5Timo Sirainen    colon_pos = UINT_MAX;
01f4ee4a0243f3fe9af763e1a540cd5cff0d63f5Timo Sirainen    while ((ret = io_buffer_read_data(inbuf, &msg,
01f4ee4a0243f3fe9af763e1a540cd5cff0d63f5Timo Sirainen                      &size, startpos+1)) != -1) {
dd62b77c932d1b518f2a3e4bf80e36542becc256Timo Sirainen        if (ret == -2) {
dd62b77c932d1b518f2a3e4bf80e36542becc256Timo Sirainen            /* overflow, line is too long. just skip it. */
4b9f99761df5014c659cd87fddaf6854af428cfcTimo Sirainen            i_assert(size > 2);
4b9f99761df5014c659cd87fddaf6854af428cfcTimo Sirainen
4b9f99761df5014c659cd87fddaf6854af428cfcTimo Sirainen                        message_skip_line(inbuf, hdr_size);
4b9f99761df5014c659cd87fddaf6854af428cfcTimo Sirainen            startpos = line_start = 0;
4b9f99761df5014c659cd87fddaf6854af428cfcTimo Sirainen            colon_pos = UINT_MAX;
4b9f99761df5014c659cd87fddaf6854af428cfcTimo Sirainen            continue;
4b9f99761df5014c659cd87fddaf6854af428cfcTimo Sirainen        }
4b9f99761df5014c659cd87fddaf6854af428cfcTimo Sirainen
4b9f99761df5014c659cd87fddaf6854af428cfcTimo Sirainen        if (size == 0) {
4b9f99761df5014c659cd87fddaf6854af428cfcTimo Sirainen            /* no, we never want empty buffer */
4106a25399703eb6cbb166dcbd5bb932cb2f7ad2Timo Sirainen            continue;
7e1f68ad71d3485f1882142837b01f7a98ca8467Timo Sirainen        }
4106a25399703eb6cbb166dcbd5bb932cb2f7ad2Timo Sirainen
7e1f68ad71d3485f1882142837b01f7a98ca8467Timo Sirainen        /* don't parse the last character, so we can always have
7e1f68ad71d3485f1882142837b01f7a98ca8467Timo Sirainen           one character read-ahead. we never care about the last
a3c197999dfe2b0c8ea38cb77cfa5e95026005c0Timo Sirainen           character anyway, it's either the first character in
a3c197999dfe2b0c8ea38cb77cfa5e95026005c0Timo Sirainen           message body, or if there's no body for any reason, it's
4106a25399703eb6cbb166dcbd5bb932cb2f7ad2Timo Sirainen           the \n ending the header. */
a3c197999dfe2b0c8ea38cb77cfa5e95026005c0Timo Sirainen        size--;
a3c197999dfe2b0c8ea38cb77cfa5e95026005c0Timo Sirainen        for (i = startpos; i < size; i++) {
a3c197999dfe2b0c8ea38cb77cfa5e95026005c0Timo Sirainen            if (msg[i] == ':' && colon_pos == UINT_MAX) {
a3c197999dfe2b0c8ea38cb77cfa5e95026005c0Timo Sirainen                colon_pos = i;
923115fd382904fa13bb09bf307bf2835b52df60Timo Sirainen                continue;
923115fd382904fa13bb09bf307bf2835b52df60Timo Sirainen            }
923115fd382904fa13bb09bf307bf2835b52df60Timo Sirainen
923115fd382904fa13bb09bf307bf2835b52df60Timo Sirainen            if (msg[i] != '\n')
923115fd382904fa13bb09bf307bf2835b52df60Timo Sirainen                continue;
252db51b6c0a605163326b3ea5d09e9936ca3b29Timo Sirainen
e0c3d5460d1cc0c440cb7723c8c2eef8d0afe9b9Timo Sirainen            if (hdr_size != NULL)
e0c3d5460d1cc0c440cb7723c8c2eef8d0afe9b9Timo Sirainen                hdr_size->lines++;
e0c3d5460d1cc0c440cb7723c8c2eef8d0afe9b9Timo Sirainen
e0c3d5460d1cc0c440cb7723c8c2eef8d0afe9b9Timo Sirainen            if (i == 0 || msg[i-1] != '\r') {
6f08b98ac63c25b747120d0c8f8e319b4e26ab0fTimo Sirainen                /* missing CR */
e0c3d5460d1cc0c440cb7723c8c2eef8d0afe9b9Timo Sirainen                missing_cr_count++;
6f08b98ac63c25b747120d0c8f8e319b4e26ab0fTimo Sirainen            }
6f08b98ac63c25b747120d0c8f8e319b4e26ab0fTimo Sirainen
e0c3d5460d1cc0c440cb7723c8c2eef8d0afe9b9Timo Sirainen            if (i == 0 || (i == 1 && msg[i-1] == '\r')) {
6f08b98ac63c25b747120d0c8f8e319b4e26ab0fTimo Sirainen                /* no headers at all */
6f08b98ac63c25b747120d0c8f8e319b4e26ab0fTimo Sirainen                break;
e0c3d5460d1cc0c440cb7723c8c2eef8d0afe9b9Timo Sirainen            }
e0c3d5460d1cc0c440cb7723c8c2eef8d0afe9b9Timo Sirainen
e0c3d5460d1cc0c440cb7723c8c2eef8d0afe9b9Timo Sirainen            if ((i > 0 && msg[i-1] == '\n') ||
e0c3d5460d1cc0c440cb7723c8c2eef8d0afe9b9Timo Sirainen                (i > 1 && msg[i-2] == '\n' && msg[i-1] == '\r')) {
e0c3d5460d1cc0c440cb7723c8c2eef8d0afe9b9Timo Sirainen                /* \n\n or \n\r\n - end of headers */
e0c3d5460d1cc0c440cb7723c8c2eef8d0afe9b9Timo Sirainen                break;
e0c3d5460d1cc0c440cb7723c8c2eef8d0afe9b9Timo Sirainen            }
e0c3d5460d1cc0c440cb7723c8c2eef8d0afe9b9Timo Sirainen
e0c3d5460d1cc0c440cb7723c8c2eef8d0afe9b9Timo Sirainen            /* make sure the header doesn't continue to next line */
e0c3d5460d1cc0c440cb7723c8c2eef8d0afe9b9Timo Sirainen            if (!IS_LWSP(msg[i+1])) {
e0c3d5460d1cc0c440cb7723c8c2eef8d0afe9b9Timo Sirainen                if (colon_pos != UINT_MAX &&
e0c3d5460d1cc0c440cb7723c8c2eef8d0afe9b9Timo Sirainen                    colon_pos != line_start && func != NULL &&
8d80659e504ffb34bb0c6a633184fece35751b18Timo Sirainen                    !IS_LWSP(msg[line_start])) {
e0c3d5460d1cc0c440cb7723c8c2eef8d0afe9b9Timo Sirainen                    /* we have a valid header line */
e0c3d5460d1cc0c440cb7723c8c2eef8d0afe9b9Timo Sirainen
51e1a1c280ccb461a15827f7987d09cb9708b6e3Timo Sirainen                    /* get length of name-field */
51e1a1c280ccb461a15827f7987d09cb9708b6e3Timo Sirainen                    end_pos = colon_pos-1;
e0c3d5460d1cc0c440cb7723c8c2eef8d0afe9b9Timo Sirainen                    while (end_pos > line_start &&
e0c3d5460d1cc0c440cb7723c8c2eef8d0afe9b9Timo Sirainen                           IS_LWSP(msg[end_pos]))
e0c3d5460d1cc0c440cb7723c8c2eef8d0afe9b9Timo Sirainen                        end_pos--;
e0c3d5460d1cc0c440cb7723c8c2eef8d0afe9b9Timo Sirainen                    name_len = end_pos - line_start + 1;
e0c3d5460d1cc0c440cb7723c8c2eef8d0afe9b9Timo Sirainen
51e1a1c280ccb461a15827f7987d09cb9708b6e3Timo Sirainen                    /* get length of value field. skip
51e1a1c280ccb461a15827f7987d09cb9708b6e3Timo Sirainen                       only the initial LWSP after ':'.
51e1a1c280ccb461a15827f7987d09cb9708b6e3Timo Sirainen                       some fields may want to keep
51e1a1c280ccb461a15827f7987d09cb9708b6e3Timo Sirainen                       the extra spaces.. */
51e1a1c280ccb461a15827f7987d09cb9708b6e3Timo Sirainen                    colon_pos++;
51e1a1c280ccb461a15827f7987d09cb9708b6e3Timo Sirainen                    if (colon_pos < i &&
51e1a1c280ccb461a15827f7987d09cb9708b6e3Timo Sirainen                        IS_LWSP(msg[colon_pos]))
51e1a1c280ccb461a15827f7987d09cb9708b6e3Timo Sirainen                        colon_pos++;
51e1a1c280ccb461a15827f7987d09cb9708b6e3Timo Sirainen                    value_len = i - colon_pos;
51e1a1c280ccb461a15827f7987d09cb9708b6e3Timo Sirainen                    if (msg[i-1] == '\r') value_len--;
c0435c854a0e7246373b9752d163095cc4fbe985Timo Sirainen
ecc81625167ed96c04c02aa190a1ea5baa65b474Timo Sirainen                    /* and finally call the function */
252db51b6c0a605163326b3ea5d09e9936ca3b29Timo Sirainen                    func(part, msg + line_start, name_len,
7b42d6cbee8186195d8c5e66078043a0fa1f25c1Timo Sirainen                         msg + colon_pos, value_len,
ecc81625167ed96c04c02aa190a1ea5baa65b474Timo Sirainen                         context);
0611067f385a37773800225256dcd5cf6aa34212Timo Sirainen                }
df16c7e87511fed827e6890a2a47d13ca48716deTimo Sirainen
df16c7e87511fed827e6890a2a47d13ca48716deTimo Sirainen                colon_pos = UINT_MAX;
ecc81625167ed96c04c02aa190a1ea5baa65b474Timo Sirainen                line_start = i+1;
0611067f385a37773800225256dcd5cf6aa34212Timo Sirainen            }
ecc81625167ed96c04c02aa190a1ea5baa65b474Timo Sirainen        }
7b42d6cbee8186195d8c5e66078043a0fa1f25c1Timo Sirainen
7b42d6cbee8186195d8c5e66078043a0fa1f25c1Timo Sirainen        if (i < size) {
b7651d283ca261015ef3c445f1f27f340f0864e2Timo Sirainen            /* end of header */
7b42d6cbee8186195d8c5e66078043a0fa1f25c1Timo Sirainen            startpos = i+1;
7b42d6cbee8186195d8c5e66078043a0fa1f25c1Timo Sirainen            break;
7b42d6cbee8186195d8c5e66078043a0fa1f25c1Timo Sirainen        }
51e1a1c280ccb461a15827f7987d09cb9708b6e3Timo Sirainen
7b42d6cbee8186195d8c5e66078043a0fa1f25c1Timo Sirainen        if (i > 0) {
ecc81625167ed96c04c02aa190a1ea5baa65b474Timo Sirainen            /* leave the last line to buffer */
ecc81625167ed96c04c02aa190a1ea5baa65b474Timo Sirainen            if (colon_pos != UINT_MAX)
89e195dfb5c4b0efd9b9f459771a4467674e5b1fTimo Sirainen                colon_pos -= line_start;
89e195dfb5c4b0efd9b9f459771a4467674e5b1fTimo Sirainen            if (hdr_size != NULL)
89e195dfb5c4b0efd9b9f459771a4467674e5b1fTimo Sirainen                hdr_size->physical_size += line_start;
89e195dfb5c4b0efd9b9f459771a4467674e5b1fTimo Sirainen            io_buffer_skip(inbuf, line_start);
1036ad17ac837a451f6b045cac504d3efa2edb8eTimo Sirainen
89e195dfb5c4b0efd9b9f459771a4467674e5b1fTimo Sirainen            startpos = i-line_start;
1036ad17ac837a451f6b045cac504d3efa2edb8eTimo Sirainen            line_start = 0;
1036ad17ac837a451f6b045cac504d3efa2edb8eTimo Sirainen        }
1036ad17ac837a451f6b045cac504d3efa2edb8eTimo Sirainen    }
a0b6b441fc679e562e79be0fb2819ffc24ab5b74Timo Sirainen    io_buffer_skip(inbuf, startpos);
a0b6b441fc679e562e79be0fb2819ffc24ab5b74Timo Sirainen
89c8d5f336e44cca091a1f588d51ba26e5145ad2Timo Sirainen    if (hdr_size != NULL) {
89c8d5f336e44cca091a1f588d51ba26e5145ad2Timo Sirainen        hdr_size->physical_size += startpos;
89c8d5f336e44cca091a1f588d51ba26e5145ad2Timo Sirainen        hdr_size->virtual_size +=
89c8d5f336e44cca091a1f588d51ba26e5145ad2Timo Sirainen            hdr_size->physical_size + missing_cr_count;
f20e7fbdc9bdbe8fecb9c661c9b8175f3bb78c69Timo Sirainen        i_assert(hdr_size->virtual_size >= hdr_size->physical_size);
c0b1543512bc3e0a3a9f526056a3678a07ce32f5Timo Sirainen    }
a0b6b441fc679e562e79be0fb2819ffc24ab5b74Timo Sirainen}
a0b6b441fc679e562e79be0fb2819ffc24ab5b74Timo Sirainen
51e1a1c280ccb461a15827f7987d09cb9708b6e3Timo Sirainenstatic MessageBoundary *boundary_find(MessageBoundary *boundaries,
a0b6b441fc679e562e79be0fb2819ffc24ab5b74Timo Sirainen                      const char *msg, unsigned int len)
a0b6b441fc679e562e79be0fb2819ffc24ab5b74Timo Sirainen{
a0b6b441fc679e562e79be0fb2819ffc24ab5b74Timo Sirainen    while (boundaries != NULL) {
1036ad17ac837a451f6b045cac504d3efa2edb8eTimo Sirainen        if (boundaries->len <= len &&
89e195dfb5c4b0efd9b9f459771a4467674e5b1fTimo Sirainen            strncmp(boundaries->boundary, msg, boundaries->len) == 0)
89e195dfb5c4b0efd9b9f459771a4467674e5b1fTimo Sirainen            return boundaries;
89e195dfb5c4b0efd9b9f459771a4467674e5b1fTimo Sirainen
6f08b98ac63c25b747120d0c8f8e319b4e26ab0fTimo Sirainen        boundaries = boundaries->next;
6f08b98ac63c25b747120d0c8f8e319b4e26ab0fTimo Sirainen    }
6f08b98ac63c25b747120d0c8f8e319b4e26ab0fTimo Sirainen
6f08b98ac63c25b747120d0c8f8e319b4e26ab0fTimo Sirainen    return NULL;
6f08b98ac63c25b747120d0c8f8e319b4e26ab0fTimo Sirainen}
f5982bb5b0a704e88fa2b44b0b74e365d13103b9Timo Sirainen
f5982bb5b0a704e88fa2b44b0b74e365d13103b9Timo Sirainen/* read until next boundary is found. if skip_over = FALSE, stop at the
f5982bb5b0a704e88fa2b44b0b74e365d13103b9Timo Sirainen   [\r]\n before the boundary, otherwise leave it right after the known
f5982bb5b0a704e88fa2b44b0b74e365d13103b9Timo Sirainen   boundary so the ending "--" can be checked. */
f5982bb5b0a704e88fa2b44b0b74e365d13103b9Timo Sirainenstatic MessageBoundary *
f5982bb5b0a704e88fa2b44b0b74e365d13103b9Timo Sirainenmessage_find_boundary(IOBuffer *inbuf, MessageBoundary *boundaries,
f5982bb5b0a704e88fa2b44b0b74e365d13103b9Timo Sirainen              MessageSize *msg_size, int skip_over)
f5982bb5b0a704e88fa2b44b0b74e365d13103b9Timo Sirainen{
f5982bb5b0a704e88fa2b44b0b74e365d13103b9Timo Sirainen    MessageBoundary *boundary;
f5982bb5b0a704e88fa2b44b0b74e365d13103b9Timo Sirainen    unsigned char *msg;
f5982bb5b0a704e88fa2b44b0b74e365d13103b9Timo Sirainen    unsigned int i, size, startpos, line_start, missing_cr_count;
f5982bb5b0a704e88fa2b44b0b74e365d13103b9Timo Sirainen
f5982bb5b0a704e88fa2b44b0b74e365d13103b9Timo Sirainen    boundary = NULL;
f5982bb5b0a704e88fa2b44b0b74e365d13103b9Timo Sirainen    missing_cr_count = startpos = line_start = 0;
f5982bb5b0a704e88fa2b44b0b74e365d13103b9Timo Sirainen
f5982bb5b0a704e88fa2b44b0b74e365d13103b9Timo Sirainen    while (io_buffer_read_data(inbuf, &msg, &size, startpos) >= 0) {
f5982bb5b0a704e88fa2b44b0b74e365d13103b9Timo Sirainen        for (i = startpos; i < size; i++) {
f5982bb5b0a704e88fa2b44b0b74e365d13103b9Timo Sirainen            if (msg[i] != '\n')
f5982bb5b0a704e88fa2b44b0b74e365d13103b9Timo Sirainen                continue;
f5982bb5b0a704e88fa2b44b0b74e365d13103b9Timo Sirainen
68a4946b12583b88fa802e52ebee45cd96056772Timo Sirainen            if (i > line_start+2 && msg[line_start] == '-' &&
4106a25399703eb6cbb166dcbd5bb932cb2f7ad2Timo Sirainen                msg[line_start+1] == '-') {
ecc81625167ed96c04c02aa190a1ea5baa65b474Timo Sirainen                /* possible boundary */
4106a25399703eb6cbb166dcbd5bb932cb2f7ad2Timo Sirainen                boundary = boundary_find(boundaries,
ecc81625167ed96c04c02aa190a1ea5baa65b474Timo Sirainen                             msg + line_start + 2,
ecc81625167ed96c04c02aa190a1ea5baa65b474Timo Sirainen                             i - line_start - 2);
c06f4017027263cf3a08becc551f5126409e2a83Timo Sirainen                if (boundary != NULL)
ecc81625167ed96c04c02aa190a1ea5baa65b474Timo Sirainen                    break;
ecc81625167ed96c04c02aa190a1ea5baa65b474Timo Sirainen            }
ecc81625167ed96c04c02aa190a1ea5baa65b474Timo Sirainen
f5982bb5b0a704e88fa2b44b0b74e365d13103b9Timo Sirainen            if (i == 0 || msg[i-1] != '\r') {
f5982bb5b0a704e88fa2b44b0b74e365d13103b9Timo Sirainen                /* missing CR */
f5982bb5b0a704e88fa2b44b0b74e365d13103b9Timo Sirainen                missing_cr_count++;
f5982bb5b0a704e88fa2b44b0b74e365d13103b9Timo Sirainen            }
f5982bb5b0a704e88fa2b44b0b74e365d13103b9Timo Sirainen
f5982bb5b0a704e88fa2b44b0b74e365d13103b9Timo Sirainen            msg_size->lines++;
f5982bb5b0a704e88fa2b44b0b74e365d13103b9Timo Sirainen            line_start = i+1;
f5982bb5b0a704e88fa2b44b0b74e365d13103b9Timo Sirainen        }
f5982bb5b0a704e88fa2b44b0b74e365d13103b9Timo Sirainen
f5982bb5b0a704e88fa2b44b0b74e365d13103b9Timo Sirainen        if (boundary != NULL) {
f5982bb5b0a704e88fa2b44b0b74e365d13103b9Timo Sirainen            /* boundary found */
f5982bb5b0a704e88fa2b44b0b74e365d13103b9Timo Sirainen            break;
f5982bb5b0a704e88fa2b44b0b74e365d13103b9Timo Sirainen        }
f5982bb5b0a704e88fa2b44b0b74e365d13103b9Timo Sirainen
f5982bb5b0a704e88fa2b44b0b74e365d13103b9Timo Sirainen        if (i > 0) {
f5982bb5b0a704e88fa2b44b0b74e365d13103b9Timo Sirainen            if (i - line_start > 128 &&
4106a25399703eb6cbb166dcbd5bb932cb2f7ad2Timo Sirainen                msg[line_start] == '-' && msg[line_start+1] == '-') {
4106a25399703eb6cbb166dcbd5bb932cb2f7ad2Timo Sirainen                /* long partial line, see if it's a boundary.
4106a25399703eb6cbb166dcbd5bb932cb2f7ad2Timo Sirainen                   RFC-2046 says that the boundaries must be
4106a25399703eb6cbb166dcbd5bb932cb2f7ad2Timo Sirainen                   70 chars without "--" or less. We allow
4106a25399703eb6cbb166dcbd5bb932cb2f7ad2Timo Sirainen                   a bit larger.. */
4106a25399703eb6cbb166dcbd5bb932cb2f7ad2Timo Sirainen                boundary = boundary_find(boundaries,
f5982bb5b0a704e88fa2b44b0b74e365d13103b9Timo Sirainen                             msg + line_start + 2,
f5982bb5b0a704e88fa2b44b0b74e365d13103b9Timo Sirainen                             i - line_start - 2);
f5982bb5b0a704e88fa2b44b0b74e365d13103b9Timo Sirainen                if (boundary != NULL)
c06f4017027263cf3a08becc551f5126409e2a83Timo Sirainen                    break;
ecc81625167ed96c04c02aa190a1ea5baa65b474Timo Sirainen
ecc81625167ed96c04c02aa190a1ea5baa65b474Timo Sirainen                /* nope, we can skip over the line, just
ecc81625167ed96c04c02aa190a1ea5baa65b474Timo Sirainen                   leave the last char since it may be \r */
4106a25399703eb6cbb166dcbd5bb932cb2f7ad2Timo Sirainen                i--;
3785910c303507db5f629684e6dde2cc7f83668eTimo Sirainen            } else {
f5982bb5b0a704e88fa2b44b0b74e365d13103b9Timo Sirainen                /* leave the last line to buffer, it may be
3785910c303507db5f629684e6dde2cc7f83668eTimo Sirainen                   boundary */
f5982bb5b0a704e88fa2b44b0b74e365d13103b9Timo Sirainen                i = line_start;
f5982bb5b0a704e88fa2b44b0b74e365d13103b9Timo Sirainen                if (i > 2) i -= 2; /* leave the \r\n too */
3785910c303507db5f629684e6dde2cc7f83668eTimo Sirainen                line_start -= i;
3785910c303507db5f629684e6dde2cc7f83668eTimo Sirainen            }
4106a25399703eb6cbb166dcbd5bb932cb2f7ad2Timo Sirainen
c06f4017027263cf3a08becc551f5126409e2a83Timo Sirainen            io_buffer_skip(inbuf, i);
ecc81625167ed96c04c02aa190a1ea5baa65b474Timo Sirainen            msg_size->physical_size += i;
4106a25399703eb6cbb166dcbd5bb932cb2f7ad2Timo Sirainen            msg_size->virtual_size += i;
ecc81625167ed96c04c02aa190a1ea5baa65b474Timo Sirainen
ecc81625167ed96c04c02aa190a1ea5baa65b474Timo Sirainen            startpos = size - i;
c06f4017027263cf3a08becc551f5126409e2a83Timo Sirainen        }
ecc81625167ed96c04c02aa190a1ea5baa65b474Timo Sirainen    }
ecc81625167ed96c04c02aa190a1ea5baa65b474Timo Sirainen
ecc81625167ed96c04c02aa190a1ea5baa65b474Timo Sirainen    if (boundary != NULL) {
c06f4017027263cf3a08becc551f5126409e2a83Timo Sirainen        if (skip_over) {
ecc81625167ed96c04c02aa190a1ea5baa65b474Timo Sirainen            /* leave the pointer right after the boundary */
ecc81625167ed96c04c02aa190a1ea5baa65b474Timo Sirainen            line_start += 2 + boundary->len;
ecc81625167ed96c04c02aa190a1ea5baa65b474Timo Sirainen        } else if (line_start > 0 && msg[line_start-1] == '\n') {
c06f4017027263cf3a08becc551f5126409e2a83Timo Sirainen            /* leave the \r\n before the boundary */
c06f4017027263cf3a08becc551f5126409e2a83Timo Sirainen            line_start--;
ecc81625167ed96c04c02aa190a1ea5baa65b474Timo Sirainen            msg_size->lines--;
ecc81625167ed96c04c02aa190a1ea5baa65b474Timo Sirainen
6ef7e31619edfaa17ed044b45861d106a86191efTimo Sirainen            if (line_start > 0 && msg[line_start-1] == '\r')
ecc81625167ed96c04c02aa190a1ea5baa65b474Timo Sirainen                line_start--;
37ab3cde96bfa4bc5304c0c348fc420aec79572dTimo Sirainen            else
c06f4017027263cf3a08becc551f5126409e2a83Timo Sirainen                missing_cr_count--;
c06f4017027263cf3a08becc551f5126409e2a83Timo Sirainen        }
37ab3cde96bfa4bc5304c0c348fc420aec79572dTimo Sirainen        startpos = line_start;
37ab3cde96bfa4bc5304c0c348fc420aec79572dTimo Sirainen    }
ecc81625167ed96c04c02aa190a1ea5baa65b474Timo Sirainen
f4bbeadda12fbd7c219063db68f3e78646d83c2cTimo Sirainen    io_buffer_skip(inbuf, startpos);
0b47e9f5e0181053b4d9ca7b426b0e5c185e820eTimo Sirainen    msg_size->physical_size += startpos;
0b47e9f5e0181053b4d9ca7b426b0e5c185e820eTimo Sirainen    msg_size->virtual_size += startpos + missing_cr_count;
abe8754852e70763e92f74caabbcc13d0917714cTimo Sirainen
ecc81625167ed96c04c02aa190a1ea5baa65b474Timo Sirainen    i_assert(msg_size->virtual_size >= msg_size->physical_size);
c06f4017027263cf3a08becc551f5126409e2a83Timo Sirainen
c06f4017027263cf3a08becc551f5126409e2a83Timo Sirainen    return boundary;
c06f4017027263cf3a08becc551f5126409e2a83Timo Sirainen}
c06f4017027263cf3a08becc551f5126409e2a83Timo Sirainen
90b8f131849540fa374aede95edd86d47d35c09dTimo Sirainenstatic MessagePart *message_parse_body(IOBuffer *inbuf,
90b8f131849540fa374aede95edd86d47d35c09dTimo Sirainen                       MessageBoundary *boundaries,
90b8f131849540fa374aede95edd86d47d35c09dTimo Sirainen                       MessageSize *body_size)
c06f4017027263cf3a08becc551f5126409e2a83Timo Sirainen{
c06f4017027263cf3a08becc551f5126409e2a83Timo Sirainen    MessageBoundary *boundary;
c09f9f95db314e7482c95e502e1c56ed6c555797Timo Sirainen
c09f9f95db314e7482c95e502e1c56ed6c555797Timo Sirainen    if (boundaries == NULL) {
c09f9f95db314e7482c95e502e1c56ed6c555797Timo Sirainen        message_get_body_size(inbuf, body_size, (uoff_t)-1);
c09f9f95db314e7482c95e502e1c56ed6c555797Timo Sirainen        return NULL;
c09f9f95db314e7482c95e502e1c56ed6c555797Timo Sirainen    } else {
c09f9f95db314e7482c95e502e1c56ed6c555797Timo Sirainen        boundary = message_find_boundary(inbuf, boundaries,
c09f9f95db314e7482c95e502e1c56ed6c555797Timo Sirainen                         body_size, FALSE);
90b8f131849540fa374aede95edd86d47d35c09dTimo Sirainen        return boundary == NULL ? NULL : boundary->part;
c06f4017027263cf3a08becc551f5126409e2a83Timo Sirainen    }
ecc81625167ed96c04c02aa190a1ea5baa65b474Timo Sirainen}
ecc81625167ed96c04c02aa190a1ea5baa65b474Timo Sirainen
252db51b6c0a605163326b3ea5d09e9936ca3b29Timo Sirainen/* skip data until next boundary is found. if it's end boundary,
904f9d5654b9c39edcdf32883e5e88771faf4d69Timo Sirainen   skip the footer as well. */
904f9d5654b9c39edcdf32883e5e88771faf4d69Timo Sirainenstatic MessagePart *message_skip_boundary(IOBuffer *inbuf,
904f9d5654b9c39edcdf32883e5e88771faf4d69Timo Sirainen                      MessageBoundary *boundaries,
904f9d5654b9c39edcdf32883e5e88771faf4d69Timo Sirainen                      MessageSize *boundary_size)
904f9d5654b9c39edcdf32883e5e88771faf4d69Timo Sirainen{
904f9d5654b9c39edcdf32883e5e88771faf4d69Timo Sirainen    MessageBoundary *boundary;
904f9d5654b9c39edcdf32883e5e88771faf4d69Timo Sirainen    unsigned char *msg;
904f9d5654b9c39edcdf32883e5e88771faf4d69Timo Sirainen    unsigned int size;
252db51b6c0a605163326b3ea5d09e9936ca3b29Timo Sirainen    int end_boundary;
7ef5ca6fb59a318c821a852ae48a2edbb671d7ddTimo Sirainen
7ef5ca6fb59a318c821a852ae48a2edbb671d7ddTimo Sirainen    boundary = message_find_boundary(inbuf, boundaries,
7ef5ca6fb59a318c821a852ae48a2edbb671d7ddTimo Sirainen                     boundary_size, TRUE);
97e62b2b36dda0acb3215667042f5c80cdee8155Timo Sirainen    if (boundary == NULL)
97e62b2b36dda0acb3215667042f5c80cdee8155Timo Sirainen        return NULL;
7ef5ca6fb59a318c821a852ae48a2edbb671d7ddTimo Sirainen
7ef5ca6fb59a318c821a852ae48a2edbb671d7ddTimo Sirainen    /* now, see if it's end boundary */
7ef5ca6fb59a318c821a852ae48a2edbb671d7ddTimo Sirainen    end_boundary = FALSE;
7662010b03ffe5f2a6ecf4b4eb220d1c65efea76Timo Sirainen    while (io_buffer_read_data(inbuf, &msg, &size, 1) >= 0) {
7662010b03ffe5f2a6ecf4b4eb220d1c65efea76Timo Sirainen        if (size >= 2) {
fe363b433b8038a69b55169da9dca27892ad7d18Timo Sirainen            end_boundary = msg[0] == '-' && msg[1] == '-';
fe363b433b8038a69b55169da9dca27892ad7d18Timo Sirainen            break;
7ef5ca6fb59a318c821a852ae48a2edbb671d7ddTimo Sirainen        }
97e62b2b36dda0acb3215667042f5c80cdee8155Timo Sirainen    }
7ef5ca6fb59a318c821a852ae48a2edbb671d7ddTimo Sirainen
7ef5ca6fb59a318c821a852ae48a2edbb671d7ddTimo Sirainen    /* skip the rest of the line */
7d5d50dd9a8c2539d7025a69e39d34fca56daeafTimo Sirainen    message_skip_line(inbuf, boundary_size);
7d5d50dd9a8c2539d7025a69e39d34fca56daeafTimo Sirainen
7d5d50dd9a8c2539d7025a69e39d34fca56daeafTimo Sirainen    if (end_boundary) {
7d5d50dd9a8c2539d7025a69e39d34fca56daeafTimo Sirainen        /* skip the footer */
7d5d50dd9a8c2539d7025a69e39d34fca56daeafTimo Sirainen        return message_parse_body(inbuf, boundaries, boundary_size);
7d5d50dd9a8c2539d7025a69e39d34fca56daeafTimo Sirainen    }
7d5d50dd9a8c2539d7025a69e39d34fca56daeafTimo Sirainen
7ef5ca6fb59a318c821a852ae48a2edbb671d7ddTimo Sirainen    return boundary == NULL ? NULL : boundary->part;
7ef5ca6fb59a318c821a852ae48a2edbb671d7ddTimo Sirainen}
dd2df6a67f10792ce31a3666197c0b6885893a3aTimo Sirainen