Search MIME headers

This commit is contained in:
Quentin 2024-01-06 22:53:41 +01:00
parent 73fc5e77df
commit 5622a71cd1
Signed by: quentin
GPG key ID: E9602264D639FF68
4 changed files with 77 additions and 29 deletions

View file

@ -1,3 +1,6 @@
use anyhow::{anyhow, Result};
use chrono::naive::NaiveDate;
use imap_codec::imap_types::core::{IString, NString};
use imap_codec::imap_types::envelope::{Address, Envelope};
@ -6,6 +9,10 @@ use eml_codec::imf;
pub struct ImfView<'a>(pub &'a imf::Imf<'a>);
impl<'a> ImfView<'a> {
pub fn naive_date(&self) -> Result<NaiveDate> {
Ok(self.0.date.ok_or(anyhow!("date is not set"))?.date_naive())
}
/// Envelope rules are defined in RFC 3501, section 7.4.2
/// https://datatracker.ietf.org/doc/html/rfc3501#section-7.4.2
///

View file

@ -40,12 +40,12 @@ impl<'a> MailView<'a> {
QueryResult::FullResult { content, .. } => {
let (_, parsed) =
eml_codec::parse_message(&content).or(Err(anyhow!("Invalid mail body")))?;
FetchedMail::new_from_message(parsed)
FetchedMail::full_from_message(parsed)
}
QueryResult::PartialResult { metadata, .. } => {
let (_, parsed) = eml_codec::parse_imf(&metadata.headers)
let (_, parsed) = eml_codec::parse_message(&metadata.headers)
.or(Err(anyhow!("unable to parse email headers")))?;
FetchedMail::Partial(parsed)
FetchedMail::partial_from_message(parsed)
}
QueryResult::IndexResult { .. } => FetchedMail::IndexOnly,
},
@ -53,7 +53,11 @@ impl<'a> MailView<'a> {
}
pub fn imf(&self) -> Option<ImfView> {
self.content.imf().map(ImfView)
self.content.as_imf().map(ImfView)
}
pub fn selected_mime(&'a self) -> Option<mime_view::SelectedMime<'a>> {
self.content.as_anypart().ok().map(mime_view::SelectedMime)
}
pub fn filter(&self, ap: &AttributesProxy) -> Result<(Body<'static>, SeenFlag)> {
@ -103,6 +107,20 @@ impl<'a> MailView<'a> {
Ok(msg_date.date_naive())
}
pub fn is_header_contains_pattern(&self, hdr: &[u8], pattern: &[u8]) -> bool {
let mime = match self.selected_mime() {
None => return false,
Some(x) => x,
};
let val = match mime.header_value(hdr) {
None => return false,
Some(x) => x,
};
val.windows(pattern.len()).any(|win| win == pattern)
}
// Private function, mainly for filter!
fn uid(&self) -> MessageDataItem<'static> {
MessageDataItem::Uid(self.in_idx.uid.clone())
@ -139,12 +157,12 @@ impl<'a> MailView<'a> {
}
fn rfc_822_text(&self) -> Result<MessageDataItem<'static>> {
let txt: NString = self.content.as_full()?.raw_body.to_vec().try_into()?;
let txt: NString = self.content.as_msg()?.raw_body.to_vec().try_into()?;
Ok(MessageDataItem::Rfc822Text(txt))
}
fn rfc822(&self) -> Result<MessageDataItem<'static>> {
let full: NString = self.content.as_full()?.raw_part.to_vec().try_into()?;
let full: NString = self.content.as_msg()?.raw_part.to_vec().try_into()?;
Ok(MessageDataItem::Rfc822(full))
}
@ -154,13 +172,13 @@ impl<'a> MailView<'a> {
fn body(&self) -> Result<MessageDataItem<'static>> {
Ok(MessageDataItem::Body(mime_view::bodystructure(
self.content.as_full()?.child.as_ref(),
self.content.as_msg()?.child.as_ref(),
)?))
}
fn body_structure(&self) -> Result<MessageDataItem<'static>> {
Ok(MessageDataItem::Body(mime_view::bodystructure(
self.content.as_full()?.child.as_ref(),
self.content.as_msg()?.child.as_ref(),
)?))
}
@ -231,32 +249,38 @@ pub enum SeenFlag {
pub enum FetchedMail<'a> {
IndexOnly,
Partial(imf::Imf<'a>),
Partial(AnyPart<'a>),
Full(AnyPart<'a>),
}
impl<'a> FetchedMail<'a> {
pub fn new_from_message(msg: Message<'a>) -> Self {
pub fn full_from_message(msg: Message<'a>) -> Self {
Self::Full(AnyPart::Msg(msg))
}
pub fn partial_from_message(msg: Message<'a>) -> Self {
Self::Partial(AnyPart::Msg(msg))
}
fn as_anypart(&self) -> Result<&AnyPart<'a>> {
match self {
FetchedMail::Full(x) => Ok(&x),
FetchedMail::Partial(x) => Ok(&x),
_ => bail!("The full message must be fetched, not only its headers"),
}
}
fn as_full(&self) -> Result<&Message<'a>> {
fn as_msg(&self) -> Result<&Message<'a>> {
match self {
FetchedMail::Full(AnyPart::Msg(x)) => Ok(&x),
FetchedMail::Partial(AnyPart::Msg(x)) => Ok(&x),
_ => bail!("The full message must be fetched, not only its headers AND it must be an AnyPart::Msg."),
}
}
fn imf(&self) -> Option<&imf::Imf<'a>> {
fn as_imf(&self) -> Option<&imf::Imf<'a>> {
match self {
FetchedMail::Full(AnyPart::Msg(x)) => Some(&x.imf),
FetchedMail::Partial(x) => Some(&x),
FetchedMail::Partial(AnyPart::Msg(x)) => Some(&x.imf),
_ => None,
}
}

View file

@ -164,8 +164,23 @@ impl<'a> SubsettedSection<'a> {
/// Used for current MIME inspection
///
/// See NodeMime for recursive logic
struct SelectedMime<'a>(&'a AnyPart<'a>);
pub struct SelectedMime<'a>(pub &'a AnyPart<'a>);
impl<'a> SelectedMime<'a> {
pub fn header_value(&'a self, to_match_ext: &[u8]) -> Option<&'a [u8]> {
let to_match = to_match_ext.to_ascii_lowercase();
self.eml_mime()
.kv
.iter()
.filter_map(|field| match field {
header::Field::Good(header::Kv2(k, v)) => Some((k, v)),
_ => None,
})
.find(|(k, _)| k.to_ascii_lowercase() == to_match)
.map(|(_, v)| v)
.copied()
}
/// The subsetted fetch section basically tells us the
/// extraction logic to apply on our selected MIME.
/// This function acts as a router for these logic.
@ -200,6 +215,13 @@ impl<'a> SelectedMime<'a> {
Ok(ExtractedFull(bytes.to_vec().into()))
}
fn eml_mime(&self) -> &eml_codec::mime::NaiveMIME<'_> {
match &self.0 {
AnyPart::Msg(msg) => msg.child.mime(),
other => other.mime(),
}
}
/// The [...] HEADER.FIELDS, and HEADER.FIELDS.NOT part
/// specifiers refer to the [RFC-2822] header of the message or of
/// an encapsulated [MIME-IMT] MESSAGE/RFC822 message.
@ -231,10 +253,7 @@ impl<'a> SelectedMime<'a> {
.collect::<HashSet<_>>();
// Extract MIME headers
let mime = match &self.0 {
AnyPart::Msg(msg) => msg.child.mime(),
other => other.mime(),
};
let mime = self.eml_mime();
// Filter our MIME headers based on the field index
// 1. Keep only the correctly formatted headers

View file

@ -220,19 +220,17 @@ impl<'a> Criteria<'a> {
Smaller(size_ref) => mail_view.query_result.metadata().expect("metadata were fetched").rfc822_size < *size_ref as usize,
// Filter on well-known headers
Bcc(_) => unimplemented!(),
Cc(_) => unimplemented!(),
From(_) => unimplemented!(),
Subject(_)=> unimplemented!(),
To(_) => unimplemented!(),
// Filter on arbitrary header
Header(..) => unimplemented!(),
Bcc(txt) => mail_view.is_header_contains_pattern(&b"bcc"[..], txt.as_ref()),
Cc(txt) => mail_view.is_header_contains_pattern(&b"cc"[..], txt.as_ref()),
From(txt) => mail_view.is_header_contains_pattern(&b"from"[..], txt.as_ref()),
Subject(txt)=> mail_view.is_header_contains_pattern(&b"subject"[..], txt.as_ref()),
To(txt) => mail_view.is_header_contains_pattern(&b"to"[..], txt.as_ref()),
Header(hdr, txt) => mail_view.is_header_contains_pattern(hdr.as_ref(), txt.as_ref()),
// Filter on Date header
SentBefore(_) => unimplemented!(),
SentOn(_) => unimplemented!(),
SentSince(_) => unimplemented!(),
SentBefore(search_naive) => mail_view.imf().map(|imf| imf.naive_date().ok()).flatten().map(|msg_naive| &msg_naive < search_naive.as_ref()).unwrap_or(false),
SentOn(search_naive) => mail_view.imf().map(|imf| imf.naive_date().ok()).flatten().map(|msg_naive| &msg_naive == search_naive.as_ref()).unwrap_or(false),
SentSince(search_naive) => mail_view.imf().map(|imf| imf.naive_date().ok()).flatten().map(|msg_naive| &msg_naive > search_naive.as_ref()).unwrap_or(false),
// Filter on the full content of the email