Fastmail US LLC

1429 Walnut Street - Suite 1201 Philadelphia PA 19102 USA murch@fastmailteam.com

ART JMAP JMAP JSON Sieve This document specifies a data model for managing Sieve scripts on a server using the JSON Meta Application Protocol (JMAP).

JMAP (JSON Meta Application Protocol) is a generic protocol for synchronizing data, such as mail, calendars or contacts, between a client and a server. It is optimized for mobile and web environments, and aims to provide a consistent interface to different data types. This specification defines a data model for managing Sieve scripts on a server using JMAP. The data model is designed to allow a server to provide consistent access to the same scripts via ManageSieve as well as JMAP, however the functionality offered over the two protocols may differ.

The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT", "SHOULD", "SHOULD NOT", "RECOMMENDED", "NOT RECOMMENDED", "MAY", and "OPTIONAL" in this document are to be interpreted as described in BCP 14 when, and only when, they appear in all capitals, as shown here. Type signatures, examples, and property descriptions in this document follow the conventions established in Section 1.1 of . Data types defined in the core specification are also used in this document. Servers MUST support all properties specified for the new data type defined in this document. For compatibility with publishing requirements, line breaks have been inserted inside long JSON strings, with the following continuation lines indented. To form the valid JSON example, any line breaks inside a string must be replaced with a space and any other white space after the line break removed.

The same terminology is used in this document as in the core JMAP specification, see , Section 1.6. The term SieveScript (with this specific capitalization) is used to refer to the data type defined in this document and instances of those data types.

The capabilities object is returned as part of the JMAP Session object; see , Section 2. This document defines one additional capability URI.

This represents support for the SieveScript data type and associated API methods. The value of this property in the JMAP Session capabilities property is an empty object. The value of this property in an account’s accountCapabilities property is an object that MUST contain the following information on server capabilities: supportsTest: Boolean If true, the server supports the SieveScript/test method. maxSizeScriptName: UnsignedInt The maximum length, in (UTF-8) octets, allowed for the name of a SieveScript. For compatibility with ManageSieve, this MUST be at least 512 (up to 128 Unicode characters). maxSizeScript: UnsignedInt|null The maximum size (in octets) of a Sieve script the server is willing to store for the user, or null for no limit. maxNumberScripts: UnsignedInt|null The maximum number of Sieve scripts the server is willing to store for the user, or null for no limit. maxNumberRedirects: UnsignedInt|null The maximum number of Sieve "redirect" actions a script can perform during a single evaluation or null for no limit. Note that this is different from the total number of "redirect" actions a script can contain. sieveExtensions: String[] A list of case-sensitive Sieve capability strings (as listed in Sieve "require" action; see , Section 3.2) indicating the extensions supported by the Sieve engine. notificationMethods: String[]|null A list of URI schema parts for notification methods supported by the Sieve "enotify" extension, or null if the extension is not supported by the Sieve engine. externalLists: String[]|null A list of URI schema parts for externally stored list types supported by the Sieve "extlists" extension, or null if the extension is not supported by the Sieve engine.

A SieveScript object represents a single Sieve script for filtering email messages at time of final delivery. A SieveScript object has the following properties: id: Id (immutable; server-set) The id of the script. name: String|null (optional; default is server-dependent) User-visible name for the SieveScript. If non-null, this MUST be a Net-Unicode string of at least 1 character in length, subject to the maximum size given in the capability object. For compatibility with ManageSieve, servers MUST reject names that contain control characters. Servers MAY reject names that violate server policy (e.g., names containing slash (/)). The name MUST be unique among all SieveScripts within an account. blobId: Id The id of the blob containing the raw octets of the script. The script MUST be UTF-8 content of at least 1 character in length, subject to the syntax of Sieve. The script MUST NOT contain any "require" statement(s) mentioning Sieve capabiltity strings not present in the capability object. Note that if the Sieve "ihave" capability string is present in the capability object, the script MAY mention unrecognized/unsupported extensions in the "ihave" test. isActive: Boolean (server-set; default: false) A user may have multiple SieveScripts on the server, yet only one script may be used for filtering of incoming messages. This is the active script. Users may have zero or one active script. The SieveScript/set method is used for changing the active script or disabling Sieve processing.

This is a standard "/get" method as described in , Section 5.1. The ids argument may be null to fetch all at once. This method provides similar functionality to the GETSCRIPT and LISTSCRIPTS commands in .

This is a standard "/set" method as described in , Section 5.3 but with the following additional request argument, which may be omitted: onSuccessActivateScript: Id|null (optional) If null, the currently active SieveScript (if any) will be deactivated if and only if all of the creations, modifications, and destructions (if any) succeed. Otherwise, the id of the SieveScript to activate if and only if all of the creations, modifications, and destructions (if any) succeed. (For references to SieveScript creations, this is equivalent to a creation-reference, so the id will be the creation id prefixed with a "#".) If this argument is not present in the request, the currently active SieveScript (if any) will remain as such. The id of any activated SieveScript MUST be reported in either the "created" or "updated" argument in the response as appropriate. The id of any deactivated SieveScript MUST be reported in the "updated" argument in the response. This method provides similar functionality to the PUTSCRIPT, DELETESCRIPT, RENAMESCRIPT, and SETACTIVE commands in . Script content must first be uploaded as a blob using either the standard upload mechanism (see Section 6.1) or the JMAP Blob management extension (see Section 3.1). If the SieveScript can not be created or updated because it would result in two SieveScripts with the same name, the server MUST reject the request with an "alreadyExists" SetError. An "existingId" property of type "Id" MUST be included on the SetError object with the id of the existing SieveScript. If the SieveScript can not be created or updated because its size exceeds the "maxSizeScript" limit, the server MUST reject the request with a "tooLarge" SetError. If the Sieve Script can not be created because it would exceed the "maxNumberScripts" limit, the server MUST reject the request with an "overQuota" SetError. The active SieveScript MUST NOT be destroyed unless it is first deactivated in a separate SieveScript/set method call. The following extra SetError types are defined: For "create" and "update": invalidScript: The SieveScript content violates the Sieve grammar and/or one or more extensions mentioned in the script's "require" statement(s) are not supported by the Sieve interpreter. The description property on the SetError object SHOULD contain a specific error message giving at least the line number of the first error. For "destroy": scriptIsActive: The SieveScript is active.

Request (and response) to upload a script requiring the Imap4Flags Extension (assuming that the JMAP Upload URL has been advertised in the JMAP Session object as having a path of "/jmap/upload/{accountId}/"):

Request (and response) to create and activate a script using the uploaded blob:

Request (and response) to update script content using the JMAP Blob management extension:

Request (and response) to update script name and deactivate:

Request (and response) to activate a script:

Requests (and responses) to deactivate and destroy the active script:

This is a standard "/query" method as described in , Section 5.5. A FilterCondition object has the following properties, either of which may be omitted: name: String The SieveScript "name" property contains the given string. isActive: Boolean The "isActive" property of the SieveScript must be identical to the value given to match the condition. The following SieveScript properties MUST be supported for sorting: name isActive

This method is used by the client to verify Sieve script validity without storing the script on the server, providing similar functionality to the CHECKSCRIPT command in . The method takes the following arguments: accountId: Id The id of the account to use. blobId: Id The id of the blob containing the raw octets of the script to validate, subject to the same requirements in . The response has the following arguments: accountId: Id The id of the account used for this call. error: SetError|null A "invalidScript" SetError object if the script content is invalid (see ), or null if the script content is valid. As with the SieveScript/set method, script content must first be uploaded as a blob using either the standard upload mechanism (see Section 6.1) or the JMAP Blob management extension (see Section 3.1).

This method is used by the client to ask the Sieve interpreter to evaluate a Sieve script against a set of emails and report the actions that would be performed for each. When calling this method the "using" property of the Request object MUST contain the capabilities "urn:ietf:params:jmap:sieve" and "urn:ietf:params:jmap:mail". The latter is required due to the use of blob ids which may reference Email objects and the use of the Envelope object, as described below. The SieveScript/test method takes the following arguments: accountId: Id The id of the account to use. scriptBlobId: String The id of the blob containing the raw octets of the script to validate, subject to the same requirements in . emailBlobIds: Id[] The ids representing the raw octets of the messages to test against. envelope: Envelope|null Information that the Sieve interpreter should assume was present in the SMTP transaction that delivered the message when evaluating "envelope" tests. If null, all "envelope" tests MUST evaluate to false. See Section 7 of for the contents of the Envelope object. lastVacationResponse: UTCDate|null The UTC date-time at which the Sieve interpreter should assume that it last auto-replied to the sender of the message, or null if the Sieve interpreter should assume that it has not auto-replied to the sender. The response has the following arguments: accountId: Id The id of the account used for this call. completed: Id[Action[]]|null A map of the blob id to a set of Action types for each message successfully processed by the script, or null if none. The Action data type is a tuple, represented as a JSON array containing three elements: A String name of the Sieve action (e.g., "keep"). A String[*] object containing any named (tagged) arguments for the action. The name MUST be the tag for the argument as given in the specification of the action (e.g., ":flags"). This may be an empty object if the action does not have any tagged arguments, or none were specified in the Sieve script (e.g., discard or ereject action). An *[] array containing any positional arguments for the action in the order as given in the specification of the action. This may be an empty array if the action does not have any positional arguments (e.g., discard or keep action). notCompleted: Id[SetError]|null A map of the blob id to a SetError object for each message that was not successfully processed by the script, or null if none. A "serverFail" SetError (see Section 3.6.2 of ) MUST be used to indicate a Sieve interpreter run-time error. The following additional errors may be returned instead of the "SieveScript/test" response: "invalidScript": The script content is invalid (see ). "notFound": The script referenced by the id could not be found. "rateLimit": The number of recent test method calls has reached a server-defined limit. "requestTooLarge": The total number of emailBlobIds exceeds the maximum number the server is willing to process in a single test method call. "serverFail": The script failed preparation to be executed for some other reason. The JSON data type to use for each argument value is a direct mapping from its Sieve data type, per the following table: Sieve Type JSON Type Number Number String String String List String[] tag with no value Boolean (true) Recommendations for constructing the list of arguments are as follows: Optional arguments in which the value is supplied by the Sieve interpreter SHOULD be included (e.g., ":from" and ":subject" arguments to the "vacation" action). Optional arguments in which the value is implicitly supplied by a Sieve variable SHOULD be included (e.g., "keep" or "fileinto" actions without an explicit ":flags" argument, but "imap4flags" have been set on the internal variable). Optional arguments in which the value is the specfied default MAY be omitted. Tagged arguments that are only used to determine whether the action will be executed and have no impact on the result of the action MAY be omitted (e.g., ":days" and ":addresses" arguments to the vacation action).

Assume that the following script has been created and has blob id "S123".

Assume that the following email has been uploaded and assigned blob id "B456". To: ken@example.com Subject: test email Date: Wed, 23 Sep 2020 12:11:11 -0500 Content-Type: text/plain; charset="UTF-8" MIME-Version: 1.0 This is a test email. ]]>

The following request executes the script against the email and provides envelope information for use by the "vacation" action.

The following response lists the actions that would be performed by the script.

Section 8 of defines a VacationResponse object to represent an autoresponder to incoming email messages. Servers that implement the VacationResponse as a Sieve script that resides amongst other user scripts are subject to the following requirements: MUST allow the VacationResponse Sieve script to be fetched by the SieveScript/get method. MUST allow the VacationResponse Sieve script to be [de]activated via the "onSuccessActivateScript" argument to the SieveScript/set method. MUST NOT allow the VacationResponse Sieve script to be destroyed or have its content updated by the SieveScript/set method. Any such request MUST be rejected with a "forbidden" SetError. A "description" property MAY be present with an explanation that the script can only be modified by a VacationResponse/set method.

All security considerations of JMAP and Sieve apply to this specification.

IANA will register the "sieve" JMAP Capability as follows: Capability Name: urn:ietf:params:jmap:sieve Specification document: this document Intended use: common Change Controller: IETF Security and privacy considerations: this document,

The following sub-sections register two new error codes in the JMAP Error Codes registry, as defined in .

JMAP Error Code: invalidScript Intended use: common Change controller: IETF Reference: This document, Description: The SieveScript violates the Sieve grammar and/or one or more extensions mentioned in the script's "require" statement(s) are not supported by the Sieve interpreter.

JMAP Error Code: scriptIsActive Intended use: common Change controller: IETF Reference: This document, Description: The client tried to destroy the active SieveScript.

The concepts in this document are based largely on those in . The author would like to thank the authors of that document for providing both inspiration and some borrowed text for this document. The author would also like to thank the following individuals for contributing their ideas and support for writing this specification: Bron Gondwana, Neil Jenkins, Alexey Melnikov, and Ricardo Signes.

Changes since ietf-03: SieveScript/test: Moved positional arguments into their own array (because the specfications don't use a consistent method for defining the action syntax or naming of positional arguments). Changes since ietf-02: Removed open issues. Reverted back to using only blob ids for script content. Added "rateLimit" and "requestTooLarge" to the list of possible error codes for /set method. Added Compatibility with JMAP Vacation Response section. Added RFC5228 to Security Considerations. Miscellaneous editorial changes. Changes since ietf-01: Removed normative references to ManageSieve (RFC 5804). Added the 'maxSizeScriptName' capability. Made the 'name' property in the SieveScript object optional. Added requirements for the 'name' property in the SieveScript object. Removed the 'blobId' property from the SieveScript object. Removed the 'replaceOnCreate' argument from the /set method. Removed the 'blobId' argument from the /validate method. Removed the 'scriptBlobId' argument from, and added the 'scriptContent' argument to, the /test method. Editorial fixes from Neil Jenkins and Ricardo Signes. Other miscellaneous text reorganization and editorial fixes. Changes since ietf-00: Specified that changes made by onSuccessActivateScript MUST be reported in the /set response as created and/or updated as appropriate. Reworked and specified more of the /test response based on implementation experience. Changes since murchison-01: Explicitly stated that Sieve capability strings are case-sensitive. errorDescription is now String|null. Added /query method. Added /test method. Changes since murchison-00: Added IANA registration for "scriptIsActive" JMAP error code. Added open issue about /set{create} with an existing script name.