Struct MessageParser

Source

pub struct MessageParser { /* private fields */ }

Expand description

RFC 5322 message parser and its configuration.

Build a parser once and reuse it; it is Send + Sync. The default configuration parses the header fields that 0.11 parsed into structured values the same way and keeps every other field raw.

use mail_parser::{HeaderForm, HeaderName, MessageParser};

let parser = MessageParser::new()
    .header(HeaderName::Received, HeaderForm::Raw)
    .header("X-Sender", HeaderForm::Addresses)
    .max_parts(1_000);
let message = parser
    .parse(b"X-Sender: <sender@example.com>\r\n\r\nbody")
    .expect("message");
let sender = message.headers().value("x-sender").and_then(|value| value.as_address());
let address = sender.and_then(|list| list.first()).and_then(|mailbox| mailbox.address());
assert_eq!(address, Some("sender@example.com"));

Implementations§

Source §

impl MessageParser

Source

pub fn new() -> Self

A parser with the default header table: every header 0.11 parsed into a structured value is parsed the same way, every other header is kept raw. Limits: nesting depth 64, 10,000 parts, 3 levels of transfer-encoded nested messages.

Parses the header name as form, in place of its default form. Content-Type, Content-Disposition and Content-Transfer-Encoding always keep their MIME form. Names compare ASCII case-insensitively, and a HeaderName::Other spelled like a known name configures that name; any other name is added to a short list checked for every unknown name.

Source

pub fn unknown_headers(self, form: HeaderForm) -> Self

The form used for header names that are neither known nor registered with MessageParser::header; HeaderForm::Raw by default.

It also applies to the vendor names that are known since 1.0 but were unknown to 0.11 (from HeaderName::XOriginalTo on: X-Mailer, User-Agent, X-Priority, the X-MS-Exchange, X-Google and X-GitHub fields, …), except those registered with header, whatever the order of the calls.

Source

pub fn max_depth(self, depth: usize) -> Self

Maximum number of nested containers (multiparts and messages), the root message included: with the default of 64, a part can be nested in at most 63 containers below the root message.

Source

pub fn max_parts(self, parts: usize) -> Self

Maximum number of parts; the content past the limit stays unparsed in the part flagged PartFlags::LIMIT_REACHED.

Source

pub fn max_encoded_nesting(self, levels: usize) -> Self

Maximum nesting of transfer-encoded message/rfc822 parts that are decoded and parsed; 0 leaves them as binary parts.

Source

pub fn parse<'x>( &self, raw: &'x (impl AsRef<[u8]> + ?Sized), ) -> Option<Message<'x>>

Parses a message that borrows raw. Returns None when the input has no header and no blank line, or is 4 GiB or larger.

Source

pub fn parse_owned(&self, raw: Vec<u8>) -> Option<Message<'static>>

Parses a message and takes ownership of its bytes, giving a Message<'static> that can be moved to another task.

Source

pub fn parse_headers<'x>( &self, raw: &'x (impl AsRef<[u8]> + ?Sized), ) -> Option<Message<'x>>

Parses only the root header block, for callers that do not need the MIME structure (DKIM verification, header indexing).

The message has a single part, the root part, which covers the whole body and keeps the kind its Content-Type gives (a multipart root has no children). No part is in a body list or in the attachments: crate::MessageRef::has_attachments is false, and a root that is not a multipart is reported by crate::MessageRef::other_parts.

Source

pub fn parse_with<'x>( &self, raw: &'x (impl AsRef<[u8]> + ?Sized), buffers: &mut MessageBuffers, ) -> Option<Message<'x>>

Parses a message into the storage held by buffers, which a previous message gave back with Message::into_buffers; once the storage fits the messages, parsing allocates nothing. See MessageBuffers for a loop.

The message takes the storage out of buffers, which stay usable: when the input does not parse, the storage is dropped and the next parse allocates.