| Author: | Marc Schiereck |
|---|---|
| Last-Modified: | 2003-03-31 |
| Revision: | 1.1 |
| Date-Created: | 2002-10-28 |
| Status: | incomplete |
Description of a concept for storing mails in Storm.
How are emails stored in a Storm-Pool.
Separate header and body?
RESOLVED: Yes: We store messages in Storm blocks, where we have to consider some points:
The way header and body of e-mails get separated.
RESOLVED: The Header is saved in one block and the body of the mail in another.
The same for multipart-message.
RESOLVED: The individual body parts get stored in a block for each part while the headers get stored in a single block.
How are header-blocks and body-blocks connected.
RESOLVED: Introducing a new Acces-Type x-storm with a parameter block for the ID of the email body's block.
We need a system for saving emails in Storm (especially for the planned email-client.) so we can transclude and link. A way to do this is to put all headers of the mail (In the case of a multipart-message there could be sub-headers for each part) in one block and to put all bodies in different blocks, as said in Issues and described further later on.
The reason for putting the the bodies in separate blocks is that their contents so have xanalogical IDs, which don't depend on the mail-header. So if one recives mails with the same body but different headers the body will only be stored once and the headers point to the body's content using only one xanalogical ID.
And it should be possible to fetch emails on different computers and to synchronize the Storm-Pool.
When saving non-multipart-messages, header and body get separated and saved in a block respectively. In order to preserve the connection between header and body the content-type of the e-mail needs to be changed to message/external-body. In addition a new access-type, called x-storm is introduced. It has a parameter block which holds the ID of the block containing the body.
The Content-Type of the header-block is message/rfc822. After the block's header the following is inserted:
Content-Type: message/external-body;
access-type="x-storm";
block="<ID>"
where <ID> is the ID of the body's block. Then the original mail header follows. There Content-ID: xxx and Content-Transfer-Encoding: binary are added.
The body-block's header has the following fields, and only these:
Content-Type: <content-type of the mail> Message-ID: <message-ID of the mail, if existent> Content-Transfer-Encoding: <...>
An example:
E-Mail:
From: Marc Schiereck <kanonensau@gmx.net> To: gzz developers list <gzz-dev@mail.freesoftware.fsf.org> Subject: Sample message Content-Type: text/plain Sample message.
Header block:
Content-Type: message/rfc822
Content-Type: message/external-body;
access-type="x-storm";
block="ID"
From: Marc Schiereck <kanonensau@gmx.net>
To: gzz developers list <gzz-dev@mail.freesoftware.fsf.org>
Subject: Sample message
Message-ID: <id@foo.org>
Content-ID: storm:block:01A1F452AE2B3441AB2234B1A2378B24DFA12B3212
Content-Type: text/plain
Content-Transfer-Encoding: binary
Body Block:
Content-Type: text/plain Message-ID: <id@foo.org> Sample message.
Storing multipart messages works similar to storing non-multipart messages. There is one block for the headers and a block for each part's body.
Content-Type of the header-block is message/rfc822. The header of the message stays untouched. Just the headers of the individual parts are changed. They are replaced by:
Content-Type: message/external-body;
access-type="x-storm";
block="<block-ID>"
<replaced header>
Content-ID: xxx
Content-Transfer-Encoding: binary
where <block-ID> is the ID of the body's block.
The body blocks have the same format as the body blocks of non-multipart messages.
An example:
E-Mail:
From: Marc Schiereck <kanonensau@gmx.net>
To: gzz developers list <gzz-dev@mail.freesoftware.fsf.org>
Subject: Sample message
Content-type: multipart/mixed;
boundary="boundary"
Message-ID: <id1@foo.org>
--boundary
Content-Type: text/plain
part1
--boundary
Content-Type: text/plain
part2
--boundary--
Header block:
Content-Type: message/rfc822
From: Marc Schiereck <kanonensau@gmx.net>
To: gzz developers list <gzz-dev@mail.freesoftware.fsf.org>
Subject: Sample message
Content-type: multipart/mixed;
boundary="boundary"
Message-ID: <id1@foo.org>
--boundary
Content-Type: message/external-body;
access-type="x-storm";
block="<ID-part1>"
Content-Type: text/plain
Content-ID: storm:block:01A1F452AE2B3441AB2234B1A2378B24DFA12B3213
Content-Transfer-Encoding: binary
--boundary
Content-Type: message/external-body;
access-type="x-storm";
block="<ID-part2>"
Content-Type: text/plain
Content-ID: storm:block:01A1F452AE2B3441AB2234B1A2378B24DFA12B3214
Content-Transfer-Encoding: binary
--boundary--
Body block 1:
Content-Type: text/plain Message-ID: <id1@foo.org> part1
Body block 2:
Content-Type: text/plain Message-ID: <id1@foo.org> part2
This functionality will be implemented in a Jython module, gzz.modules.email.converter.
To store a single mail the following function will be defined:
storeMail(mail, mediaserver)
mail is a string holding the complete mail (header and body. They get separated inside the function).