diff options
author | Quentin <quentin@dufour.io> | 2024-01-06 10:38:37 +0000 |
---|---|---|
committer | Quentin <quentin@dufour.io> | 2024-01-06 10:38:37 +0000 |
commit | 44ca458c5cf666246e472dea9be70b745a130e8c (patch) | |
tree | b78cca747e5c2bc004cb93b93536623f7abb6ef5 /src/imap | |
parent | bcf6de83419b405fea95b740869f98d43586ea7c (diff) | |
parent | 53dbf82cbce3cb17cbcffd09558677faf8702f54 (diff) | |
download | aerogramme-44ca458c5cf666246e472dea9be70b745a130e8c.tar.gz aerogramme-44ca458c5cf666246e472dea9be70b745a130e8c.zip |
Merge pull request 'Aerogramme refactoring' (#57) from feat/more-imap-qol into main
Reviewed-on: https://git.deuxfleurs.fr/Deuxfleurs/aerogramme/pulls/57
Diffstat (limited to 'src/imap')
-rw-r--r-- | src/imap/attributes.rs | 49 | ||||
-rw-r--r-- | src/imap/command/examined.rs | 2 | ||||
-rw-r--r-- | src/imap/command/selected.rs | 14 | ||||
-rw-r--r-- | src/imap/flags.rs | 30 | ||||
-rw-r--r-- | src/imap/imf_view.rs | 97 | ||||
-rw-r--r-- | src/imap/index.rs | 95 | ||||
-rw-r--r-- | src/imap/mail_view.rs | 247 | ||||
-rw-r--r-- | src/imap/mailbox_view.rs | 1165 | ||||
-rw-r--r-- | src/imap/mime_view.rs | 538 | ||||
-rw-r--r-- | src/imap/mod.rs | 7 | ||||
-rw-r--r-- | src/imap/search.rs | 130 |
11 files changed, 1349 insertions, 1025 deletions
diff --git a/src/imap/attributes.rs b/src/imap/attributes.rs new file mode 100644 index 0000000..7a55632 --- /dev/null +++ b/src/imap/attributes.rs @@ -0,0 +1,49 @@ +use imap_codec::imap_types::fetch::{MacroOrMessageDataItemNames, MessageDataItemName}; + +/// Internal decisions based on fetched attributes +/// passed by the client + +pub struct AttributesProxy { + pub attrs: Vec<MessageDataItemName<'static>>, +} +impl AttributesProxy { + pub fn new(attrs: &MacroOrMessageDataItemNames<'static>, is_uid_fetch: bool) -> Self { + // Expand macros + let mut fetch_attrs = match attrs { + MacroOrMessageDataItemNames::Macro(m) => { + use imap_codec::imap_types::fetch::Macro; + use MessageDataItemName::*; + match m { + Macro::All => vec![Flags, InternalDate, Rfc822Size, Envelope], + Macro::Fast => vec![Flags, InternalDate, Rfc822Size], + Macro::Full => vec![Flags, InternalDate, Rfc822Size, Envelope, Body], + _ => { + tracing::error!("unimplemented macro"); + vec![] + } + } + } + MacroOrMessageDataItemNames::MessageDataItemNames(a) => a.clone(), + }; + + // Handle uids + if is_uid_fetch && !fetch_attrs.contains(&MessageDataItemName::Uid) { + fetch_attrs.push(MessageDataItemName::Uid); + } + + Self { attrs: fetch_attrs } + } + + pub fn need_body(&self) -> bool { + self.attrs.iter().any(|x| { + matches!( + x, + MessageDataItemName::Body + | MessageDataItemName::BodyExt { .. } + | MessageDataItemName::Rfc822 + | MessageDataItemName::Rfc822Text + | MessageDataItemName::BodyStructure + ) + }) + } +} diff --git a/src/imap/command/examined.rs b/src/imap/command/examined.rs index 0d688c0..ec16973 100644 --- a/src/imap/command/examined.rs +++ b/src/imap/command/examined.rs @@ -125,7 +125,7 @@ impl<'a> ExaminedContext<'a> { } pub async fn noop(self) -> Result<(Response<'static>, flow::Transition)> { - self.mailbox.mailbox.force_sync().await?; + self.mailbox.0.mailbox.force_sync().await?; let updates = self.mailbox.update().await?; Ok(( diff --git a/src/imap/command/selected.rs b/src/imap/command/selected.rs index c8cc680..35c3eb4 100644 --- a/src/imap/command/selected.rs +++ b/src/imap/command/selected.rs @@ -136,21 +136,23 @@ impl<'a> SelectedContext<'a> { pub async fn search( self, - _charset: &Option<Charset<'a>>, - _criteria: &SearchKey<'a>, - _uid: &bool, + charset: &Option<Charset<'a>>, + criteria: &SearchKey<'a>, + uid: &bool, ) -> Result<(Response<'static>, flow::Transition)> { + let found = self.mailbox.search(charset, criteria, *uid).await?; Ok(( Response::build() .to_req(self.req) - .message("Not implemented") - .bad()?, + .set_body(found) + .message("SEARCH completed") + .ok()?, flow::Transition::None, )) } pub async fn noop(self) -> Result<(Response<'static>, flow::Transition)> { - self.mailbox.mailbox.force_sync().await?; + self.mailbox.0.mailbox.force_sync().await?; let updates = self.mailbox.update().await?; Ok(( diff --git a/src/imap/flags.rs b/src/imap/flags.rs new file mode 100644 index 0000000..0f6ec64 --- /dev/null +++ b/src/imap/flags.rs @@ -0,0 +1,30 @@ +use imap_codec::imap_types::core::Atom; +use imap_codec::imap_types::flag::{Flag, FlagFetch}; + +pub fn from_str(f: &str) -> Option<FlagFetch<'static>> { + match f.chars().next() { + Some('\\') => match f { + "\\Seen" => Some(FlagFetch::Flag(Flag::Seen)), + "\\Answered" => Some(FlagFetch::Flag(Flag::Answered)), + "\\Flagged" => Some(FlagFetch::Flag(Flag::Flagged)), + "\\Deleted" => Some(FlagFetch::Flag(Flag::Deleted)), + "\\Draft" => Some(FlagFetch::Flag(Flag::Draft)), + "\\Recent" => Some(FlagFetch::Recent), + _ => match Atom::try_from(f.strip_prefix('\\').unwrap().to_string()) { + Err(_) => { + tracing::error!(flag=%f, "Unable to encode flag as IMAP atom"); + None + } + Ok(a) => Some(FlagFetch::Flag(Flag::system(a))), + }, + }, + Some(_) => match Atom::try_from(f.to_string()) { + Err(_) => { + tracing::error!(flag=%f, "Unable to encode flag as IMAP atom"); + None + } + Ok(a) => Some(FlagFetch::Flag(Flag::keyword(a))), + }, + None => None, + } +} diff --git a/src/imap/imf_view.rs b/src/imap/imf_view.rs new file mode 100644 index 0000000..4297769 --- /dev/null +++ b/src/imap/imf_view.rs @@ -0,0 +1,97 @@ +use imap_codec::imap_types::core::{IString, NString}; +use imap_codec::imap_types::envelope::{Address, Envelope}; + +use eml_codec::imf; + +/// Envelope rules are defined in RFC 3501, section 7.4.2 +/// https://datatracker.ietf.org/doc/html/rfc3501#section-7.4.2 +/// +/// Some important notes: +/// +/// If the Sender or Reply-To lines are absent in the [RFC-2822] +/// header, or are present but empty, the server sets the +/// corresponding member of the envelope to be the same value as +/// the from member (the client is not expected to know to do +/// this). Note: [RFC-2822] requires that all messages have a valid +/// From header. Therefore, the from, sender, and reply-to +/// members in the envelope can not be NIL. +/// +/// If the Date, Subject, In-Reply-To, and Message-ID header lines +/// are absent in the [RFC-2822] header, the corresponding member +/// of the envelope is NIL; if these header lines are present but +/// empty the corresponding member of the envelope is the empty +/// string. + +//@FIXME return an error if the envelope is invalid instead of panicking +//@FIXME some fields must be defaulted if there are not set. +pub fn message_envelope(msg: &imf::Imf) -> Envelope<'static> { + let from = msg.from.iter().map(convert_mbx).collect::<Vec<_>>(); + + Envelope { + date: NString( + msg.date + .as_ref() + .map(|d| IString::try_from(d.to_rfc3339()).unwrap()), + ), + subject: NString( + msg.subject + .as_ref() + .map(|d| IString::try_from(d.to_string()).unwrap()), + ), + sender: msg + .sender + .as_ref() + .map(|v| vec![convert_mbx(v)]) + .unwrap_or(from.clone()), + reply_to: if msg.reply_to.is_empty() { + from.clone() + } else { + convert_addresses(&msg.reply_to) + }, + from, + to: convert_addresses(&msg.to), + cc: convert_addresses(&msg.cc), + bcc: convert_addresses(&msg.bcc), + in_reply_to: NString( + msg.in_reply_to + .iter() + .next() + .map(|d| IString::try_from(d.to_string()).unwrap()), + ), + message_id: NString( + msg.msg_id + .as_ref() + .map(|d| IString::try_from(d.to_string()).unwrap()), + ), + } +} + +pub fn convert_addresses(addrlist: &Vec<imf::address::AddressRef>) -> Vec<Address<'static>> { + let mut acc = vec![]; + for item in addrlist { + match item { + imf::address::AddressRef::Single(a) => acc.push(convert_mbx(a)), + imf::address::AddressRef::Many(l) => acc.extend(l.participants.iter().map(convert_mbx)), + } + } + return acc; +} + +pub fn convert_mbx(addr: &imf::mailbox::MailboxRef) -> Address<'static> { + Address { + name: NString( + addr.name + .as_ref() + .map(|x| IString::try_from(x.to_string()).unwrap()), + ), + // SMTP at-domain-list (source route) seems obsolete since at least 1991 + // https://www.mhonarc.org/archive/html/ietf-822/1991-06/msg00060.html + adl: NString(None), + mailbox: NString(Some( + IString::try_from(addr.addrspec.local_part.to_string()).unwrap(), + )), + host: NString(Some( + IString::try_from(addr.addrspec.domain.to_string()).unwrap(), + )), + } +} diff --git a/src/imap/index.rs b/src/imap/index.rs new file mode 100644 index 0000000..01dd2ef --- /dev/null +++ b/src/imap/index.rs @@ -0,0 +1,95 @@ +use std::num::NonZeroU32; + +use anyhow::{anyhow, bail, Result}; +use imap_codec::imap_types::sequence::{self, SequenceSet}; + +use crate::mail::uidindex::{ImapUid, UidIndex}; +use crate::mail::unique_ident::UniqueIdent; + +pub struct Index<'a>(pub &'a UidIndex); +impl<'a> Index<'a> { + pub fn fetch( + self: &Index<'a>, + sequence_set: &SequenceSet, + by_uid: bool, + ) -> Result<Vec<MailIndex<'a>>> { + let mail_vec = self + .0 + .idx_by_uid + .iter() + .map(|(uid, uuid)| (*uid, *uuid)) + .collect::<Vec<_>>(); + + let mut mails = vec![]; + + if by_uid { + if mail_vec.is_empty() { + return Ok(vec![]); + } + let iter_strat = sequence::Strategy::Naive { + largest: mail_vec.last().unwrap().0, + }; + + let mut i = 0; + for uid in sequence_set.iter(iter_strat) { + while mail_vec.get(i).map(|mail| mail.0 < uid).unwrap_or(false) { + i += 1; + } + if let Some(mail) = mail_vec.get(i) { + if mail.0 == uid { + mails.push(MailIndex { + i: NonZeroU32::try_from(i as u32 + 1).unwrap(), + uid: mail.0, + uuid: mail.1, + flags: self + .0 + .table + .get(&mail.1) + .ok_or(anyhow!("mail is missing from index"))? + .1 + .as_ref(), + }); + } + } else { + break; + } + } + } else { + if mail_vec.is_empty() { + bail!("No such message (mailbox is empty)"); + } + + let iter_strat = sequence::Strategy::Naive { + largest: NonZeroU32::try_from((mail_vec.len()) as u32).unwrap(), + }; + + for i in sequence_set.iter(iter_strat) { + if let Some(mail) = mail_vec.get(i.get() as usize - 1) { + mails.push(MailIndex { + i, + uid: mail.0, + uuid: mail.1, + flags: self + .0 + .table + .get(&mail.1) + .ok_or(anyhow!("mail is missing from index"))? + .1 + .as_ref(), + }); + } else { + bail!("No such mail: {}", i); + } + } + } + + Ok(mails) + } +} + +pub struct MailIndex<'a> { + pub i: NonZeroU32, + pub uid: ImapUid, + pub uuid: UniqueIdent, + pub flags: &'a Vec<String>, +} diff --git a/src/imap/mail_view.rs b/src/imap/mail_view.rs new file mode 100644 index 0000000..de9bfe3 --- /dev/null +++ b/src/imap/mail_view.rs @@ -0,0 +1,247 @@ +use std::num::NonZeroU32; + +use anyhow::{anyhow, bail, Result}; +use chrono::{Offset, TimeZone, Utc}; + +use imap_codec::imap_types::core::NString; +use imap_codec::imap_types::datetime::DateTime; +use imap_codec::imap_types::fetch::{ + MessageDataItem, MessageDataItemName, Section as FetchSection, +}; +use imap_codec::imap_types::flag::Flag; +use imap_codec::imap_types::response::Data; + +use eml_codec::{ + imf, + part::{composite::Message, AnyPart}, +}; + +use crate::mail::query::QueryResult; + +use crate::imap::attributes::AttributesProxy; +use crate::imap::flags; +use crate::imap::imf_view::message_envelope; +use crate::imap::index::MailIndex; +use crate::imap::mime_view; +use crate::imap::response::Body; + +pub struct MailView<'a> { + pub in_idx: MailIndex<'a>, + pub query_result: &'a QueryResult<'a>, + pub content: FetchedMail<'a>, +} + +impl<'a> MailView<'a> { + pub fn new(query_result: &'a QueryResult<'a>, in_idx: MailIndex<'a>) -> Result<MailView<'a>> { + Ok(Self { + in_idx, + query_result, + content: match query_result { + QueryResult::FullResult { content, .. } => { + let (_, parsed) = + eml_codec::parse_message(&content).or(Err(anyhow!("Invalid mail body")))?; + FetchedMail::new_from_message(parsed) + } + QueryResult::PartialResult { metadata, .. } => { + let (_, parsed) = eml_codec::parse_imf(&metadata.headers) + .or(Err(anyhow!("unable to parse email headers")))?; + FetchedMail::Partial(parsed) + } + QueryResult::IndexResult { .. } => FetchedMail::IndexOnly, + }, + }) + } + + fn uid(&self) -> MessageDataItem<'static> { + MessageDataItem::Uid(self.in_idx.uid.clone()) + } + + fn flags(&self) -> MessageDataItem<'static> { + MessageDataItem::Flags( + self.in_idx + .flags + .iter() + .filter_map(|f| flags::from_str(f)) + .collect(), + ) + } + + fn rfc_822_size(&self) -> Result<MessageDataItem<'static>> { + let sz = self + .query_result + .metadata() + .ok_or(anyhow!("mail metadata are required"))? + .rfc822_size; + Ok(MessageDataItem::Rfc822Size(sz as u32)) + } + + fn rfc_822_header(&self) -> Result<MessageDataItem<'static>> { + let hdrs: NString = self + .query_result + .metadata() + .ok_or(anyhow!("mail metadata are required"))? + .headers + .to_vec() + .try_into()?; + Ok(MessageDataItem::Rfc822Header(hdrs)) + } + + fn rfc_822_text(&self) -> Result<MessageDataItem<'static>> { + let txt: NString = self.content.as_full()?.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()?; + Ok(MessageDataItem::Rfc822(full)) + } + + fn envelope(&self) -> MessageDataItem<'static> { + MessageDataItem::Envelope(message_envelope(self.content.imf().clone())) + } + + fn body(&self) -> Result<MessageDataItem<'static>> { + Ok(MessageDataItem::Body(mime_view::bodystructure( + self.content.as_full()?.child.as_ref(), + )?)) + } + + fn body_structure(&self) -> Result<MessageDataItem<'static>> { + Ok(MessageDataItem::Body(mime_view::bodystructure( + self.content.as_full()?.child.as_ref(), + )?)) + } + + /// maps to BODY[<section>]<<partial>> and BODY.PEEK[<section>]<<partial>> + /// peek does not implicitly set the \Seen flag + /// eg. BODY[HEADER.FIELDS (DATE FROM)] + /// eg. BODY[]<0.2048> + fn body_ext( + &self, + section: &Option<FetchSection<'static>>, + partial: &Option<(u32, NonZeroU32)>, + peek: &bool, + ) -> Result<(MessageDataItem<'static>, SeenFlag)> { + // Manage Seen flag + let mut seen = SeenFlag::DoNothing; + let seen_flag = Flag::Seen.to_string(); + if !peek && !self.in_idx.flags.iter().any(|x| *x == seen_flag) { + // Add \Seen flag + //self.mailbox.add_flags(uuid, &[seen_flag]).await?; + seen = SeenFlag::MustAdd; + } + + // Process message + let (text, origin) = + match mime_view::body_ext(self.content.as_anypart()?, section, partial)? { + mime_view::BodySection::Full(body) => (body, None), + mime_view::BodySection::Slice { body, origin_octet } => (body, Some(origin_octet)), + }; + + let data: NString = text.to_vec().try_into()?; + + return Ok(( + MessageDataItem::BodyExt { + section: section.as_ref().map(|fs| fs.clone()), + origin, + data, + }, + seen, + )); + } + + fn internal_date(&self) -> Result<MessageDataItem<'static>> { + let dt = Utc + .fix() + .timestamp_opt( + i64::try_from( + self.query_result + .metadata() + .ok_or(anyhow!("mail metadata were not fetched"))? + .internaldate + / 1000, + )?, + 0, + ) + .earliest() + .ok_or(anyhow!("Unable to parse internal date"))?; + Ok(MessageDataItem::InternalDate(DateTime::unvalidated(dt))) + } + + pub fn filter(&self, ap: &AttributesProxy) -> Result<(Body<'static>, SeenFlag)> { + let mut seen = SeenFlag::DoNothing; + let res_attrs = ap + .attrs + .iter() + .map(|attr| match attr { + MessageDataItemName::Uid => Ok(self.uid()), + MessageDataItemName::Flags => Ok(self.flags()), + MessageDataItemName::Rfc822Size => self.rfc_822_size(), + MessageDataItemName::Rfc822Header => self.rfc_822_header(), + MessageDataItemName::Rfc822Text => self.rfc_822_text(), + MessageDataItemName::Rfc822 => self.rfc822(), + MessageDataItemName::Envelope => Ok(self.envelope()), + MessageDataItemName::Body => self.body(), + MessageDataItemName::BodyStructure => self.body_structure(), + MessageDataItemName::BodyExt { + section, + partial, + peek, + } => { + let (body, has_seen) = self.body_ext(section, partial, peek)?; + seen = has_seen; + Ok(body) + } + MessageDataItemName::InternalDate => self.internal_date(), + }) + .collect::<Result<Vec<_>, _>>()?; + + Ok(( + Body::Data(Data::Fetch { + seq: self.in_idx.i, + items: res_attrs.try_into()?, + }), + seen, + )) + } +} + +pub enum SeenFlag { + DoNothing, + MustAdd, +} + +// ------------------- + +pub enum FetchedMail<'a> { + IndexOnly, + Partial(imf::Imf<'a>), + Full(AnyPart<'a>), +} +impl<'a> FetchedMail<'a> { + pub fn new_from_message(msg: Message<'a>) -> Self { + Self::Full(AnyPart::Msg(msg)) + } + + fn as_anypart(&self) -> Result<&AnyPart<'a>> { + match self { + FetchedMail::Full(x) => Ok(&x), + _ => bail!("The full message must be fetched, not only its headers"), + } + } + + fn as_full(&self) -> Result<&Message<'a>> { + match self { + FetchedMail::Full(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) -> &imf::Imf<'a> { + match self { + FetchedMail::Full(AnyPart::Msg(x)) => &x.imf, + FetchedMail::Partial(x) => &x, + _ => panic!("Can't contain AnyPart that is not a message"), + } + } +} diff --git a/src/imap/mailbox_view.rs b/src/imap/mailbox_view.rs index 861d27c..e4ffdcd 100644 --- a/src/imap/mailbox_view.rs +++ b/src/imap/mailbox_view.rs @@ -1,35 +1,28 @@ -use std::borrow::Cow; -use std::iter::zip; use std::num::NonZeroU32; use std::sync::Arc; -use anyhow::{anyhow, bail, Error, Result}; -use chrono::{Offset, TimeZone, Utc}; +use anyhow::{anyhow, Error, Result}; use futures::stream::{FuturesOrdered, StreamExt}; -use imap_codec::imap_types::body::{BasicFields, Body as FetchBody, BodyStructure, SpecificFields}; -use imap_codec::imap_types::core::{AString, Atom, IString, NString, NonEmptyVec}; -use imap_codec::imap_types::datetime::DateTime; -use imap_codec::imap_types::envelope::{Address, Envelope}; -use imap_codec::imap_types::fetch::{ - MacroOrMessageDataItemNames, MessageDataItem, MessageDataItemName, Section as FetchSection, -}; +use imap_codec::imap_types::core::Charset; +use imap_codec::imap_types::fetch::{MacroOrMessageDataItemNames, MessageDataItem}; use imap_codec::imap_types::flag::{Flag, FlagFetch, FlagPerm, StoreResponse, StoreType}; use imap_codec::imap_types::response::{Code, Data, Status}; -use imap_codec::imap_types::sequence::{self, SequenceSet}; - -use eml_codec::{ - header, imf, mime, - mime::r#type::Deductible, - part::{composite::Message, AnyPart}, -}; - -use crate::cryptoblob::Key; +use imap_codec::imap_types::search::SearchKey; +use imap_codec::imap_types::sequence::SequenceSet; + +use crate::mail::mailbox::Mailbox; +use crate::mail::query::QueryScope; +use crate::mail::snapshot::FrozenMailbox; +use crate::mail::uidindex::{ImapUid, ImapUidvalidity}; + +use crate::imap::attributes::AttributesProxy; +use crate::imap::flags; +use crate::imap::index::Index; +use crate::imap::mail_view::{MailView, SeenFlag}; use crate::imap::response::Body; -use crate::mail::mailbox::{MailMeta, Mailbox}; -use crate::mail::uidindex::{ImapUid, ImapUidvalidity, UidIndex}; -use crate::mail::unique_ident::UniqueIdent; +use crate::imap::search; const DEFAULT_FLAGS: [Flag; 5] = [ Flag::Seen, @@ -39,377 +32,6 @@ const DEFAULT_FLAGS: [Flag; 5] = [ Flag::Draft, ]; -enum FetchedMail<'a> { - Partial(imf::Imf<'a>), - Full(AnyPart<'a>), -} -impl<'a> FetchedMail<'a> { - fn new_from_message(msg: Message<'a>) -> Self { - FetchedMail::Full(AnyPart::Msg(msg)) - } - - /*fn new_from_header(hdr: imf::Imf<'a>) -> Self { - FetchedMail::Partial(hdr) - }*/ - - fn as_anypart(&self) -> Result<&AnyPart<'a>> { - match self { - FetchedMail::Full(x) => Ok(&x), - _ => bail!("The full message must be fetched, not only its headers"), - } - } - - fn as_full(&self) -> Result<&Message<'a>> { - match self { - FetchedMail::Full(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) -> &imf::Imf<'a> { - match self { - FetchedMail::Full(AnyPart::Msg(x)) => &x.imf, - FetchedMail::Partial(x) => &x, - _ => panic!("Can't contain AnyPart that is not a message"), - } - } -} - -pub struct AttributesProxy { - attrs: Vec<MessageDataItemName<'static>>, -} -impl AttributesProxy { - fn new(attrs: &MacroOrMessageDataItemNames<'static>, is_uid_fetch: bool) -> Self { - // Expand macros - let mut fetch_attrs = match attrs { - MacroOrMessageDataItemNames::Macro(m) => { - use imap_codec::imap_types::fetch::Macro; - use MessageDataItemName::*; - match m { - Macro::All => vec![Flags, InternalDate, Rfc822Size, Envelope], - Macro::Fast => vec![Flags, InternalDate, Rfc822Size], - Macro::Full => vec![Flags, InternalDate, Rfc822Size, Envelope, Body], - _ => { - tracing::error!("unimplemented macro"); - vec![] - } - } - } - MacroOrMessageDataItemNames::MessageDataItemNames(a) => a.clone(), - }; - - // Handle uids - if is_uid_fetch && !fetch_attrs.contains(&MessageDataItemName::Uid) { - fetch_attrs.push(MessageDataItemName::Uid); - } - - Self { attrs: fetch_attrs } - } - - fn need_body(&self) -> bool { - self.attrs.iter().any(|x| { - matches!( - x, - MessageDataItemName::Body - | MessageDataItemName::BodyExt { .. } - | MessageDataItemName::Rfc822 - | MessageDataItemName::Rfc822Text - | MessageDataItemName::BodyStructure - ) - }) - } -} - -pub struct MailIdentifiers { - i: NonZeroU32, - uid: ImapUid, - uuid: UniqueIdent, -} -struct MailIdentifiersList(Vec<MailIdentifiers>); - -impl MailIdentifiersList { - fn uuids(&self) -> Vec<UniqueIdent> { - self.0.iter().map(|mi| mi.uuid).collect() - } -} - -pub struct MailView<'a> { - ids: &'a MailIdentifiers, - meta: &'a MailMeta, - flags: &'a Vec<String>, - content: FetchedMail<'a>, -} - -enum SeenFlag { - DoNothing, - MustAdd, -} - -impl<'a> MailView<'a> { - fn uid(&self) -> MessageDataItem<'static> { - MessageDataItem::Uid(self.ids.uid.clone()) - } - - fn flags(&self) -> MessageDataItem<'static> { - MessageDataItem::Flags( - self.flags - .iter() - .filter_map(|f| string_to_flag(f)) - .collect(), - ) - } - - fn rfc_822_size(&self) -> MessageDataItem<'static> { - MessageDataItem::Rfc822Size(self.meta.rfc822_size as u32) - } - - fn rfc_822_header(&self) -> MessageDataItem<'static> { - MessageDataItem::Rfc822Header(NString( - self.meta - .headers - .to_vec() - .try_into() - .ok() - .map(IString::Literal), - )) - } - - fn rfc_822_text(&self) -> Result<MessageDataItem<'static>> { - Ok(MessageDataItem::Rfc822Text(NString( - self.content - .as_full()? - .raw_body - .to_vec() - .try_into() - .ok() - .map(IString::Literal), - ))) - } - - fn rfc822(&self) -> Result<MessageDataItem<'static>> { - Ok(MessageDataItem::Rfc822(NString( - self.content - .as_full()? - .raw_part - .to_vec() - .try_into() - .ok() - .map(IString::Literal), - ))) - } - - fn envelope(&self) -> MessageDataItem<'static> { - MessageDataItem::Envelope(message_envelope(self.content.imf().clone())) - } - - fn body(&self) -> Result<MessageDataItem<'static>> { - Ok(MessageDataItem::Body(build_imap_email_struct( - self.content.as_full()?.child.as_ref(), - )?)) - } - - fn body_structure(&self) -> Result<MessageDataItem<'static>> { - Ok(MessageDataItem::Body(build_imap_email_struct( - self.content.as_full()?.child.as_ref(), - )?)) - } - - /// maps to BODY[<section>]<<partial>> and BODY.PEEK[<section>]<<partial>> - /// peek does not implicitly set the \Seen flag - /// eg. BODY[HEADER.FIELDS (DATE FROM)] - /// eg. BODY[]<0.2048> - fn body_ext<'b>( - &self, - section: &Option<FetchSection<'b>>, - partial: &Option<(u32, NonZeroU32)>, - peek: &bool, - ) -> Result<(MessageDataItem<'b>, SeenFlag)> { - let mut seen = SeenFlag::DoNothing; - - // Extract message section - let text = get_message_section(self.content.as_anypart()?, section)?; - - let seen_flag = Flag::Seen.to_string(); - if !peek && !self.flags.iter().any(|x| *x == seen_flag) { - // Add \Seen flag - //self.mailbox.add_flags(uuid, &[seen_flag]).await?; - seen = SeenFlag::MustAdd; - } - - // Handle <<partial>> which cut the message bytes - let (text, origin) = apply_partial(partial, &text); - - let data = NString(text.to_vec().try_into().ok().map(IString::Literal)); - - return Ok(( - MessageDataItem::BodyExt { - section: section.as_ref().map(|fs| fs.clone()), - origin, - data, - }, - seen, - )); - } - - fn internal_date(&self) -> Result<MessageDataItem<'static>> { - let dt = Utc - .fix() - .timestamp_opt(i64::try_from(self.meta.internaldate / 1000)?, 0) - .earliest() - .ok_or(anyhow!("Unable to parse internal date"))?; - Ok(MessageDataItem::InternalDate(DateTime::unvalidated(dt))) - } - - fn filter<'b>(&self, ap: &AttributesProxy) -> Result<(Body<'static>, SeenFlag)> { - let mut seen = SeenFlag::DoNothing; - let res_attrs = ap - .attrs - .iter() - .map(|attr| match attr { - MessageDataItemName::Uid => Ok(self.uid()), - MessageDataItemName::Flags => Ok(self.flags()), - MessageDataItemName::Rfc822Size => Ok(self.rfc_822_size()), - MessageDataItemName::Rfc822Header => Ok(self.rfc_822_header()), - MessageDataItemName::Rfc822Text => self.rfc_822_text(), - MessageDataItemName::Rfc822 => self.rfc822(), - MessageDataItemName::Envelope => Ok(self.envelope()), - MessageDataItemName::Body => self.body(), - MessageDataItemName::BodyStructure => self.body_structure(), - MessageDataItemName::BodyExt { - section, - partial, - peek, - } => { - let (body, has_seen) = self.body_ext(section, partial, peek)?; - seen = has_seen; - Ok(body) - } - MessageDataItemName::InternalDate => self.internal_date(), - }) - .collect::<Result<Vec<_>, _>>()?; - - Ok(( - Body::Data(Data::Fetch { - seq: self.ids.i, - items: res_attrs.try_into()?, - }), - seen, - )) - } -} - -fn apply_partial<'a>( - partial: &'_ Option<(u32, NonZeroU32)>, - text: &'a [u8], -) -> (&'a [u8], Option<u32>) { - match partial { - Some((begin, len)) => { - if *begin as usize > text.len() { - (&[][..], Some(*begin)) - } else if (begin + len.get()) as usize >= text.len() { - (&text[*begin as usize..], Some(*begin)) - } else { - ( - &text[*begin as usize..(begin + len.get()) as usize], - Some(*begin), - ) - } - } - None => (&text[..], None), - } -} - -pub struct BodyIdentifier<'a> { - msg_uuid: &'a UniqueIdent, - msg_key: &'a Key, -} - -#[derive(Default)] -pub struct MailSelectionBuilder<'a> { - //attrs: AttributeProxy, - mail_count: usize, - need_body: bool, - mi: &'a [MailIdentifiers], - meta: &'a [MailMeta], - flags: &'a [&'a Vec<String>], - bodies: &'a [Vec<u8>], -} - -impl<'a> MailSelectionBuilder<'a> { - fn new(need_body: bool, mail_count: usize) -> Self { - Self { - mail_count, - need_body, - ..MailSelectionBuilder::default() - } - } - - fn with_mail_identifiers(&mut self, mi: &'a [MailIdentifiers]) -> &mut Self { - self.mi = mi; - self - } - - fn with_metadata(&mut self, meta: &'a [MailMeta]) -> &mut Self { - self.meta = meta; - self - } - - fn with_flags(&mut self, flags: &'a [&'a Vec<String>]) -> &mut Self { - self.flags = flags; - self - } - - fn bodies_to_collect(&self) -> Vec<BodyIdentifier> { - if !self.need_body { - return vec![]; - } - zip(self.mi, self.meta) - .map(|(mi, meta)| BodyIdentifier { - msg_uuid: &mi.uuid, - msg_key: &meta.message_key, - }) - .collect::<Vec<_>>() - } - - fn with_bodies(&mut self, rbodies: &'a [Vec<u8>]) -> &mut Self { - self.bodies = rbodies; - self - } - - fn build(&self) -> Result<Vec<MailView<'a>>> { - let mut bodies = vec![]; - - if !self.need_body { - for m in self.meta.iter() { - let (_, hdrs) = - eml_codec::parse_imf(&m.headers).or(Err(anyhow!("Invalid mail headers")))?; - bodies.push(FetchedMail::Partial(hdrs)); - } - } else { - for rb in self.bodies.iter() { - let (_, p) = eml_codec::parse_message(&rb).or(Err(anyhow!("Invalid mail body")))?; - bodies.push(FetchedMail::new_from_message(p)); - } - } - - if self.mi.len() != self.mail_count && self.meta.len() != self.mail_count - || self.flags.len() != self.mail_count - || bodies.len() != self.mail_count - { - return Err(anyhow!("Can't build a mail view selection as parts were not correctly registered into the builder.")); - } - - Ok(zip(self.mi, zip(self.meta, zip(self.flags, bodies))) - .map(|(ids, (meta, (flags, content)))| MailView { - ids, - meta, - flags, - content, - }) - .collect()) - } -} - /// A MailboxView is responsible for giving the client the information /// it needs about a mailbox, such as an initial summary of the mailbox's /// content and continuous updates indicating when the content @@ -417,20 +39,12 @@ impl<'a> MailSelectionBuilder<'a> { /// To do this, it keeps a variable `known_state` that corresponds to /// what the client knows, and produces IMAP messages to be sent to the /// client that go along updates to `known_state`. -pub struct MailboxView { - pub(crate) mailbox: Arc<Mailbox>, - known_state: UidIndex, -} +pub struct MailboxView(pub FrozenMailbox); impl MailboxView { /// Creates a new IMAP view into a mailbox. pub async fn new(mailbox: Arc<Mailbox>) -> Self { - let state = mailbox.current_uid_index().await; - - Self { - mailbox, - known_state: state, - } + Self(mailbox.frozen().await) } /// Create an updated view, useful to make a diff @@ -440,11 +54,8 @@ impl MailboxView { /// This does NOT trigger a sync, it bases itself on what is currently /// loaded in RAM by Bayou. pub async fn update(&mut self) -> Result<Vec<Body<'static>>> { - let old_view: &mut Self = self; - let new_view = Self { - mailbox: old_view.mailbox.clone(), - known_state: old_view.mailbox.current_uid_index().await, - }; + let old_snapshot = self.0.update().await; + let new_snapshot = &self.0.snapshot; let mut data = Vec::<Body>::new(); @@ -465,8 +76,8 @@ impl MailboxView { // - notify client of expunged mails let mut n_expunge = 0; - for (i, (_uid, uuid)) in old_view.known_state.idx_by_uid.iter().enumerate() { - if !new_view.known_state.table.contains_key(uuid) { + for (i, (_uid, uuid)) in old_snapshot.idx_by_uid.iter().enumerate() { + if !new_snapshot.table.contains_key(uuid) { data.push(Body::Data(Data::Expunge( NonZeroU32::try_from((i + 1 - n_expunge) as u32).unwrap(), ))); @@ -475,21 +86,21 @@ impl MailboxView { } // - if new mails arrived, notify client of number of existing mails - if new_view.known_state.table.len() != old_view.known_state.table.len() - n_expunge - || new_view.known_state.uidvalidity != old_view.known_state.uidvalidity + if new_snapshot.table.len() != old_snapshot.table.len() - n_expunge + || new_snapshot.uidvalidity != old_snapshot.uidvalidity { - data.push(new_view.exists_status()?); + data.push(self.exists_status()?); } - if new_view.known_state.uidvalidity != old_view.known_state.uidvalidity { + if new_snapshot.uidvalidity != old_snapshot.uidvalidity { // TODO: do we want to push less/more info than this? - data.push(new_view.uidvalidity_status()?); - data.push(new_view.uidnext_status()?); + data.push(self.uidvalidity_status()?); + data.push(self.uidnext_status()?); } else { // - if flags changed for existing mails, tell client - for (i, (_uid, uuid)) in new_view.known_state.idx_by_uid.iter().enumerate() { - let old_mail = old_view.known_state.table.get(uuid); - let new_mail = new_view.known_state.table.get(uuid); + for (i, (_uid, uuid)) in new_snapshot.idx_by_uid.iter().enumerate() { + let old_mail = old_snapshot.table.get(uuid); + let new_mail = new_snapshot.table.get(uuid); if old_mail.is_some() && old_mail != new_mail { if let Some((uid, flags)) = new_mail { data.push(Body::Data(Data::Fetch { @@ -497,7 +108,7 @@ impl MailboxView { items: vec![ MessageDataItem::Uid(*uid), MessageDataItem::Flags( - flags.iter().filter_map(|f| string_to_flag(f)).collect(), + flags.iter().filter_map(|f| flags::from_str(f)).collect(), ), ] .try_into()?, @@ -506,7 +117,6 @@ impl MailboxView { } } } - *old_view = new_view; Ok(data) } @@ -532,21 +142,21 @@ impl MailboxView { flags: &[Flag<'a>], is_uid_store: &bool, ) -> Result<Vec<Body<'static>>> { - self.mailbox.opportunistic_sync().await?; + self.0.sync().await?; let flags = flags.iter().map(|x| x.to_string()).collect::<Vec<_>>(); - let mails = self.get_mail_ids(sequence_set, *is_uid_store)?; + let mails = self.index().fetch(sequence_set, *is_uid_store)?; for mi in mails.iter() { match kind { StoreType::Add => { - self.mailbox.add_flags(mi.uuid, &flags[..]).await?; + self.0.mailbox.add_flags(mi.uuid, &flags[..]).await?; } StoreType::Remove => { - self.mailbox.del_flags(mi.uuid, &flags[..]).await?; + self.0.mailbox.del_flags(mi.uuid, &flags[..]).await?; } StoreType::Replace => { - self.mailbox.set_flags(mi.uuid, &flags[..]).await?; + self.0.mailbox.set_flags(mi.uuid, &flags[..]).await?; } } } @@ -556,10 +166,10 @@ impl MailboxView { } pub async fn expunge(&mut self) -> Result<Vec<Body<'static>>> { - self.mailbox.opportunistic_sync().await?; + self.0.sync().await?; + let state = self.0.peek().await; let deleted_flag = Flag::Deleted.to_string(); - let state = self.mailbox.current_uid_index().await; let msgs = state .table .iter() @@ -567,7 +177,7 @@ impl MailboxView { .map(|(uuid, _)| *uuid); for msg in msgs { - self.mailbox.delete(msg).await?; + self.0.mailbox.delete(msg).await?; } self.update().await @@ -579,11 +189,11 @@ impl MailboxView { to: Arc<Mailbox>, is_uid_copy: &bool, ) -> Result<(ImapUidvalidity, Vec<(ImapUid, ImapUid)>)> { - let mails = self.get_mail_ids(sequence_set, *is_uid_copy)?; + let mails = self.index().fetch(sequence_set, *is_uid_copy)?; let mut new_uuids = vec![]; for mi in mails.iter() { - new_uuids.push(to.copy_from(&self.mailbox, mi.uuid).await?); + new_uuids.push(to.copy_from(&self.0.mailbox, mi.uuid).await?); } let mut ret = vec![]; @@ -606,21 +216,18 @@ impl MailboxView { to: Arc<Mailbox>, is_uid_copy: &bool, ) -> Result<(ImapUidvalidity, Vec<(ImapUid, ImapUid)>, Vec<Body<'static>>)> { - let mails = self.get_mail_ids(sequence_set, *is_uid_copy)?; + let mails = self.index().fetch(sequence_set, *is_uid_copy)?; - let mut new_uuids = vec![]; for mi in mails.iter() { - let copy_action = to.copy_from(&self.mailbox, mi.uuid).await?; - new_uuids.push(copy_action); - self.mailbox.delete(mi.uuid).await? + to.move_from(&self.0.mailbox, mi.uuid).await?; } let mut ret = vec![]; let to_state = to.current_uid_index().await; - for (mi, new_uuid) in mails.iter().zip(new_uuids.iter()) { + for mi in mails.iter() { let dest_uid = to_state .table - .get(new_uuid) + .get(&mi.uuid) .ok_or(anyhow!("moved mail not in destination mailbox"))? .0; ret.push((mi.uid, dest_uid)); @@ -639,152 +246,100 @@ impl MailboxView { attributes: &'b MacroOrMessageDataItemNames<'static>, is_uid_fetch: &bool, ) -> Result<Vec<Body<'static>>> { + // [1/6] Pre-compute data + // a. what are the uuids of the emails we want? + // b. do we need to fetch the full body? let ap = AttributesProxy::new(attributes, *is_uid_fetch); + let query_scope = match ap.need_body() { + true => QueryScope::Full, + _ => QueryScope::Partial, + }; + let mail_idx_list = self.index().fetch(sequence_set, *is_uid_fetch)?; - // Prepare data - let mids = MailIdentifiersList(self.get_mail_ids(sequence_set, *is_uid_fetch)?); - let mail_count = mids.0.len(); - let uuids = mids.uuids(); - let meta = self.mailbox.fetch_meta(&uuids).await?; - let flags = uuids + // [2/6] Fetch the emails + let uuids = mail_idx_list .iter() - .map(|uuid| { - self.known_state - .table - .get(uuid) - .map(|(_uuid, f)| f) - .ok_or(anyhow!("missing email from the flag table")) - }) - .collect::<Result<Vec<_>, _>>()?; - - // Start filling data to build the view - let mut selection = MailSelectionBuilder::new(ap.need_body(), mail_count); - selection - .with_mail_identifiers(&mids.0) - .with_metadata(&meta) - .with_flags(&flags); + .map(|midx| midx.uuid) + .collect::<Vec<_>>(); + let query_result = self.0.query(&uuids, query_scope).fetch().await?; - // Asynchronously fetch full bodies (if needed) - let btc = selection.bodies_to_collect(); - let future_bodies = btc + // [3/6] Derive an IMAP-specific view from the results, apply the filters + let views = query_result .iter() - .map(|bi| async move { - let body = self.mailbox.fetch_full(*bi.msg_uuid, bi.msg_key).await?; - Ok::<_, anyhow::Error>(body) - }) - .collect::<FuturesOrdered<_>>(); - let bodies = future_bodies - .collect::<Vec<_>>() - .await - .into_iter() + .zip(mail_idx_list.into_iter()) + .map(|(qr, midx)| MailView::new(qr, midx)) .collect::<Result<Vec<_>, _>>()?; - // Add bodies - selection.with_bodies(bodies.as_slice()); - - // Build mail selection views - let views = selection.build()?; - - // Filter views to build the result - // Also identify what must be put as seen - let filtered_view = views + // [4/6] Apply the IMAP transformation, bubble up any error + // We get 2 results: + // - The one we send to the client + // - The \Seen flags we must set internally + let (flag_mgmt, imap_ret): (Vec<_>, Vec<_>) = views .iter() - .filter_map(|mv| mv.filter(&ap).ok().map(|(body, seen)| (mv, body, seen))) - .collect::<Vec<_>>(); - // Register seen flags - let future_flags = filtered_view + .map(|mv| mv.filter(&ap).map(|(body, seen)| ((mv, seen), body))) + .collect::<Result<Vec<_>, _>>()? + .into_iter() + .unzip(); + + // [5/6] Register the \Seen flags + flag_mgmt .iter() - .filter(|(_mv, _body, seen)| matches!(seen, SeenFlag::MustAdd)) - .map(|(mv, _body, _seen)| async move { + .filter(|(_mv, seen)| matches!(seen, SeenFlag::MustAdd)) + .map(|(mv, _seen)| async move { let seen_flag = Flag::Seen.to_string(); - self.mailbox.add_flags(mv.ids.uuid, &[seen_flag]).await?; + self.0 + .mailbox + .add_flags(*mv.query_result.uuid(), &[seen_flag]) + .await?; Ok::<_, anyhow::Error>(()) }) - .collect::<FuturesOrdered<_>>(); - - future_flags + .collect::<FuturesOrdered<_>>() .collect::<Vec<_>>() .await .into_iter() .collect::<Result<_, _>>()?; - let command_body = filtered_view - .into_iter() - .map(|(_mv, body, _seen)| body) - .collect::<Vec<_>>(); - - Ok(command_body) + // [6/6] Build the final result that will be sent to the client. + Ok(imap_ret) } - // ---- - - // Gets the IMAP ID, the IMAP UIDs and, the Aerogramme UUIDs of mails identified by a SequenceSet of - // sequence numbers (~ IMAP selector) - fn get_mail_ids( + /// A naive search implementation... + pub async fn search<'a>( &self, - sequence_set: &SequenceSet, - by_uid: bool, - ) -> Result<Vec<MailIdentifiers>> { - let mail_vec = self - .known_state - .idx_by_uid - .iter() - .map(|(uid, uuid)| (*uid, *uuid)) - .collect::<Vec<_>>(); + _charset: &Option<Charset<'a>>, + search_key: &SearchKey<'a>, + uid: bool, + ) -> Result<Vec<Body<'static>>> { + // 1. Compute the subset of sequence identifiers we need to fetch + // based on the search query + let crit = search::Criteria(search_key); + let (seq_set, seq_type) = crit.to_sequence_set(); - let mut mails = vec![]; + // 2. Get the selection + let selection = self.index().fetch(&seq_set, seq_type.is_uid())?; - if by_uid { - if mail_vec.is_empty() { - return Ok(vec![]); - } - let iter_strat = sequence::Strategy::Naive { - largest: mail_vec.last().unwrap().0, - }; - - let mut i = 0; - for uid in sequence_set.iter(iter_strat) { - while mail_vec.get(i).map(|mail| mail.0 < uid).unwrap_or(false) { - i += 1; - } - if let Some(mail) = mail_vec.get(i) { - if mail.0 == uid { - mails.push(MailIdentifiers { - i: NonZeroU32::try_from(i as u32 + 1).unwrap(), - uid: mail.0, - uuid: mail.1, - }); - } - } else { - break; - } - } - } else { - if mail_vec.is_empty() { - bail!("No such message (mailbox is empty)"); - } + // 3. Filter the selection based on the ID / UID / Flags - let iter_strat = sequence::Strategy::Naive { - largest: NonZeroU32::try_from((mail_vec.len()) as u32).unwrap(), - }; - - for i in sequence_set.iter(iter_strat) { - if let Some(mail) = mail_vec.get(i.get() as usize - 1) { - mails.push(MailIdentifiers { - i, - uid: mail.0, - uuid: mail.1, - }); - } else { - bail!("No such mail: {}", i); - } - } - } + // 4. If needed, filter the selection based on the metadata + let _need_meta = crit.need_meta(); - Ok(mails) + // 5. If needed, filter the selection based on the body + let _need_body = crit.need_body(); + + // 6. Format the result according to the client's taste: + // either return UID or ID. + let selection_fmt = match uid { + true => selection.into_iter().map(|in_idx| in_idx.uid).collect(), + _ => selection.into_iter().map(|in_idx| in_idx.i).collect(), + }; + + Ok(vec![Body::Data(Data::Search(selection_fmt))]) } // ---- + fn index<'a>(&'a self) -> Index<'a> { + Index(&self.0.snapshot) + } /// Produce an OK [UIDVALIDITY _] message corresponding to `known_state` fn uidvalidity_status(&self) -> Result<Body<'static>> { @@ -798,7 +353,7 @@ impl MailboxView { } pub(crate) fn uidvalidity(&self) -> ImapUidvalidity { - self.known_state.uidvalidity + self.0.snapshot.uidvalidity } /// Produce an OK [UIDNEXT _] message corresponding to `known_state` @@ -813,7 +368,7 @@ impl MailboxView { } pub(crate) fn uidnext(&self) -> ImapUid { - self.known_state.uidnext + self.0.snapshot.uidnext } /// Produce an EXISTS message corresponding to the number of mails @@ -823,7 +378,7 @@ impl MailboxView { } pub(crate) fn exists(&self) -> Result<u32> { - Ok(u32::try_from(self.known_state.idx_by_uid.len())?) + Ok(u32::try_from(self.0.snapshot.idx_by_uid.len())?) } /// Produce a RECENT message corresponding to the number of @@ -834,7 +389,8 @@ impl MailboxView { pub(crate) fn recent(&self) -> Result<u32> { let recent = self - .known_state + .0 + .snapshot .idx_by_flag .get(&"\\Recent".to_string()) .map(|os| os.len()) @@ -850,10 +406,11 @@ impl MailboxView { // 1. Collecting all the possible flags in the mailbox // 1.a Fetch them from our index let mut known_flags: Vec<Flag> = self - .known_state + .0 + .snapshot .idx_by_flag .flags() - .filter_map(|f| match string_to_flag(f) { + .filter_map(|f| match flags::from_str(f) { Some(FlagFetch::Flag(fl)) => Some(fl), _ => None, }) @@ -889,9 +446,10 @@ impl MailboxView { } pub(crate) fn unseen_count(&self) -> usize { - let total = self.known_state.table.len(); + let total = self.0.snapshot.table.len(); let seen = self - .known_state + .0 + .snapshot .idx_by_flag .get(&Flag::Seen.to_string()) .map(|x| x.len()) @@ -900,456 +458,25 @@ impl MailboxView { } } -fn string_to_flag(f: &str) -> Option<FlagFetch<'static>> { - match f.chars().next() { - Some('\\') => match f { - "\\Seen" => Some(FlagFetch::Flag(Flag::Seen)), - "\\Answered" => Some(FlagFetch::Flag(Flag::Answered)), - "\\Flagged" => Some(FlagFetch::Flag(Flag::Flagged)), - "\\Deleted" => Some(FlagFetch::Flag(Flag::Deleted)), - "\\Draft" => Some(FlagFetch::Flag(Flag::Draft)), - "\\Recent" => Some(FlagFetch::Recent), - _ => match Atom::try_from(f.strip_prefix('\\').unwrap().to_string()) { - Err(_) => { - tracing::error!(flag=%f, "Unable to encode flag as IMAP atom"); - None - } - Ok(a) => Some(FlagFetch::Flag(Flag::system(a))), - }, - }, - Some(_) => match Atom::try_from(f.to_string()) { - Err(_) => { - tracing::error!(flag=%f, "Unable to encode flag as IMAP atom"); - None - } - Ok(a) => Some(FlagFetch::Flag(Flag::keyword(a))), - }, - None => None, - } -} - -/// Envelope rules are defined in RFC 3501, section 7.4.2 -/// https://datatracker.ietf.org/doc/html/rfc3501#section-7.4.2 -/// -/// Some important notes: -/// -/// If the Sender or Reply-To lines are absent in the [RFC-2822] -/// header, or are present but empty, the server sets the -/// corresponding member of the envelope to be the same value as -/// the from member (the client is not expected to know to do -/// this). Note: [RFC-2822] requires that all messages have a valid -/// From header. Therefore, the from, sender, and reply-to -/// members in the envelope can not be NIL. -/// -/// If the Date, Subject, In-Reply-To, and Message-ID header lines -/// are absent in the [RFC-2822] header, the corresponding member -/// of the envelope is NIL; if these header lines are present but -/// empty the corresponding member of the envelope is the empty -/// string. - -//@FIXME return an error if the envelope is invalid instead of panicking -//@FIXME some fields must be defaulted if there are not set. -fn message_envelope(msg: &imf::Imf) -> Envelope<'static> { - let from = msg.from.iter().map(convert_mbx).collect::<Vec<_>>(); - - Envelope { - date: NString( - msg.date - .as_ref() - .map(|d| IString::try_from(d.to_rfc3339()).unwrap()), - ), - subject: NString( - msg.subject - .as_ref() - .map(|d| IString::try_from(d.to_string()).unwrap()), - ), - sender: msg - .sender - .as_ref() - .map(|v| vec![convert_mbx(v)]) - .unwrap_or(from.clone()), - reply_to: if msg.reply_to.is_empty() { - from.clone() - } else { - convert_addresses(&msg.reply_to) - }, - from, - to: convert_addresses(&msg.to), - cc: convert_addresses(&msg.cc), - bcc: convert_addresses(&msg.bcc), - in_reply_to: NString( - msg.in_reply_to - .iter() - .next() - .map(|d| IString::try_from(d.to_string()).unwrap()), - ), - message_id: NString( - msg.msg_id - .as_ref() - .map(|d| IString::try_from(d.to_string()).unwrap()), - ), - } -} - -fn convert_addresses(addrlist: &Vec<imf::address::AddressRef>) -> Vec<Address<'static>> { - let mut acc = vec![]; - for item in addrlist { - match item { - imf::address::AddressRef::Single(a) => acc.push(convert_mbx(a)), - imf::address::AddressRef::Many(l) => acc.extend(l.participants.iter().map(convert_mbx)), - } - } - return acc; -} - -fn convert_mbx(addr: &imf::mailbox::MailboxRef) -> Address<'static> { - Address { - name: NString( - addr.name - .as_ref() - .map(|x| IString::try_from(x.to_string()).unwrap()), - ), - // SMTP at-domain-list (source route) seems obsolete since at least 1991 - // https://www.mhonarc.org/archive/html/ietf-822/1991-06/msg00060.html - adl: NString(None), - mailbox: NString(Some( - IString::try_from(addr.addrspec.local_part.to_string()).unwrap(), - )), - host: NString(Some( - IString::try_from(addr.addrspec.domain.to_string()).unwrap(), - )), - } -} - -/* ---CAPTURE-- -b fetch 29878:29879 (BODY) -* 29878 FETCH (BODY (("text" "plain" ("charset" "utf-8") NIL NIL "quoted-printable" 3264 82)("text" "html" ("charset" "utf-8") NIL NIL "quoted-printable" 31834 643) "alternative")) -* 29879 FETCH (BODY ("text" "html" ("charset" "us-ascii") NIL NIL "7bit" 4107 131)) - ^^^^^^^^^^^^^^^^^^^^^^ ^^^ ^^^ ^^^^^^ ^^^^ ^^^ - | | | | | | number of lines - | | | | | size - | | | | content transfer encoding - | | | description - | | id - | parameter list -b OK Fetch completed (0.001 + 0.000 secs). -*/ - -fn build_imap_email_struct<'a>(part: &AnyPart<'a>) -> Result<BodyStructure<'static>> { - match part { - AnyPart::Mult(x) => { - let itype = &x.mime.interpreted_type; - let subtype = IString::try_from(itype.subtype.to_string()) - .unwrap_or(unchecked_istring("alternative")); - - let inner_bodies = x - .children - .iter() - .filter_map(|inner| build_imap_email_struct(&inner).ok()) - .collect::<Vec<_>>(); - NonEmptyVec::validate(&inner_bodies)?; - let bodies = NonEmptyVec::unvalidated(inner_bodies); - - Ok(BodyStructure::Multi { - bodies, - subtype, - extension_data: None, - /*Some(MultipartExtensionData { - parameter_list: vec![], - disposition: None, - language: None, - location: None, - extension: vec![], - })*/ - }) - } - AnyPart::Txt(x) => { - let mut basic = basic_fields(&x.mime.fields, x.body.len())?; - - // Get the interpreted content type, set it - let itype = match &x.mime.interpreted_type { - Deductible::Inferred(v) | Deductible::Explicit(v) => v, - }; - let subtype = - IString::try_from(itype.subtype.to_string()).unwrap_or(unchecked_istring("plain")); - - // Add charset to the list of parameters if we know it has been inferred as it will be - // missing from the parsed content. - if let Deductible::Inferred(charset) = &itype.charset { - basic.parameter_list.push(( - unchecked_istring("charset"), - IString::try_from(charset.to_string()).unwrap_or(unchecked_istring("us-ascii")), - )); - } - - Ok(BodyStructure::Single { - body: FetchBody { - basic, - specific: SpecificFields::Text { - subtype, - number_of_lines: nol(x.body), - }, - }, - extension_data: None, - }) - } - AnyPart::Bin(x) => { - let basic = basic_fields(&x.mime.fields, x.body.len())?; - - let default = mime::r#type::NaiveType { - main: &b"application"[..], - sub: &b"octet-stream"[..], - params: vec![], - }; - let ct = x.mime.fields.ctype.as_ref().unwrap_or(&default); - - let r#type = - IString::try_from(String::from_utf8_lossy(ct.main).to_string()).or(Err( - anyhow!("Unable to build IString from given Content-Type type given"), - ))?; - - let subtype = - IString::try_from(String::from_utf8_lossy(ct.sub).to_string()).or(Err(anyhow!( - "Unable to build IString from given Content-Type subtype given" - )))?; - - Ok(BodyStructure::Single { - body: FetchBody { - basic, - specific: SpecificFields::Basic { r#type, subtype }, - }, - extension_data: None, - }) - } - AnyPart::Msg(x) => { - let basic = basic_fields(&x.mime.fields, x.raw_part.len())?; - - Ok(BodyStructure::Single { - body: FetchBody { - basic, - specific: SpecificFields::Message { - envelope: Box::new(message_envelope(&x.imf)), - body_structure: Box::new(build_imap_email_struct(x.child.as_ref())?), - number_of_lines: nol(x.raw_part), - }, - }, - extension_data: None, - }) - } - } -} - -fn nol(input: &[u8]) -> u32 { - input - .iter() - .filter(|x| **x == b'\n') - .count() - .try_into() - .unwrap_or(0) -} - -/// s is set to static to ensure that only compile time values -/// checked by developpers are passed. -fn unchecked_istring(s: &'static str) -> IString { - IString::try_from(s).expect("this value is expected to be a valid imap-codec::IString") -} - -fn basic_fields(m: &mime::NaiveMIME, sz: usize) -> Result<BasicFields<'static>> { - let parameter_list = m - .ctype - .as_ref() - .map(|x| { - x.params - .iter() - .map(|p| { - ( - IString::try_from(String::from_utf8_lossy(p.name).to_string()), - IString::try_from(p.value.to_string()), - ) - }) - .filter(|(k, v)| k.is_ok() && v.is_ok()) - .map(|(k, v)| (k.unwrap(), v.unwrap())) - .collect() - }) - .unwrap_or(vec![]); - - Ok(BasicFields { - parameter_list, - id: NString( - m.id.as_ref() - .and_then(|ci| IString::try_from(ci.to_string()).ok()), - ), - description: NString( - m.description - .as_ref() - .and_then(|cd| IString::try_from(cd.to_string()).ok()), - ), - content_transfer_encoding: match m.transfer_encoding { - mime::mechanism::Mechanism::_8Bit => unchecked_istring("8bit"), - mime::mechanism::Mechanism::Binary => unchecked_istring("binary"), - mime::mechanism::Mechanism::QuotedPrintable => unchecked_istring("quoted-printable"), - mime::mechanism::Mechanism::Base64 => unchecked_istring("base64"), - _ => unchecked_istring("7bit"), - }, - // @FIXME we can't compute the size of the message currently... - size: u32::try_from(sz)?, - }) -} - -/// Extract message section for section identifier passed by the FETCH BODY[<section>]<<partial>> -/// request -/// -/// Example of message sections: -/// -/// ``` -/// HEADER ([RFC-2822] header of the message) -/// TEXT ([RFC-2822] text body of the message) MULTIPART/MIXED -/// 1 TEXT/PLAIN -/// 2 APPLICATION/OCTET-STREAM -/// 3 MESSAGE/RFC822 -/// 3.HEADER ([RFC-2822] header of the message) -/// 3.TEXT ([RFC-2822] text body of the message) MULTIPART/MIXED -/// 3.1 TEXT/PLAIN -/// 3.2 APPLICATION/OCTET-STREAM -/// 4 MULTIPART/MIXED -/// 4.1 IMAGE/GIF -/// 4.1.MIME ([MIME-IMB] header for the IMAGE/GIF) -/// 4.2 MESSAGE/RFC822 -/// 4.2.HEADER ([RFC-2822] header of the message) -/// 4.2.TEXT ([RFC-2822] text body of the message) MULTIPART/MIXED -/// 4.2.1 TEXT/PLAIN -/// 4.2.2 MULTIPART/ALTERNATIVE -/// 4.2.2.1 TEXT/PLAIN -/// 4.2.2.2 TEXT/RICHTEXT -/// ``` -fn get_message_section<'a>( - parsed: &'a AnyPart<'a>, - section: &Option<FetchSection>, -) -> Result<Cow<'a, [u8]>> { - let msg = parsed - .as_message() - .ok_or(anyhow!("Part must be a message"))?; - match section { - Some(FetchSection::Text(None)) => Ok(msg.raw_body.into()), - Some(FetchSection::Text(Some(part))) => map_subpart(parsed, part.0.as_ref(), |part_msg| { - Ok(part_msg - .as_message() - .ok_or(Error::msg( - "Not a message/rfc822 part while expected by request (TEXT)", - ))? - .raw_body - .into()) - }), - Some(FetchSection::Header(part)) => map_subpart( - parsed, - part.as_ref().map(|p| p.0.as_ref()).unwrap_or(&[]), - |part_msg| { - Ok(part_msg - .as_message() - .ok_or(Error::msg( - "Not a message/rfc822 part while expected by request (HEADER)", - ))? - .raw_headers - .into()) - }, - ), - Some( - FetchSection::HeaderFields(part, fields) | FetchSection::HeaderFieldsNot(part, fields), - ) => { - let invert = matches!(section, Some(FetchSection::HeaderFieldsNot(_, _))); - let fields = fields - .as_ref() - .iter() - .map(|x| match x { - AString::Atom(a) => a.inner().as_bytes(), - AString::String(IString::Literal(l)) => l.as_ref(), - AString::String(IString::Quoted(q)) => q.inner().as_bytes(), - }) - .collect::<Vec<_>>(); - - map_subpart( - parsed, - part.as_ref().map(|p| p.0.as_ref()).unwrap_or(&[]), - |part_msg| { - let mut ret = vec![]; - let mime = match &part_msg { - AnyPart::Msg(msg) => msg.child.mime(), - other => other.mime(), - }; - for f in mime.kv.iter() { - let (k, v) = match f { - header::Field::Good(header::Kv2(k, v)) => (k, v), - _ => continue, - }; - if fields.as_slice().iter().any(|x| (x == k) ^ invert) { - ret.extend(*k); - ret.extend(b": "); - ret.extend(*v); - ret.extend(b"\r\n"); - } - } - ret.extend(b"\r\n"); - Ok(ret.into()) - }, - ) - } - Some(FetchSection::Part(part)) => map_subpart(parsed, part.0.as_ref(), |part| { - let bytes = match &part { - AnyPart::Txt(p) => p.body, - AnyPart::Bin(p) => p.body, - AnyPart::Msg(p) => p.raw_part, - AnyPart::Mult(_) => bail!("Multipart part has no body"), - }; - Ok(bytes.to_vec().into()) - }), - Some(FetchSection::Mime(part)) => map_subpart(parsed, part.0.as_ref(), |part| { - let bytes = match &part { - AnyPart::Txt(p) => p.mime.fields.raw, - AnyPart::Bin(p) => p.mime.fields.raw, - AnyPart::Msg(p) => p.child.mime().raw, - AnyPart::Mult(p) => p.mime.fields.raw, - }; - Ok(bytes.to_vec().into()) - }), - None => Ok(msg.raw_part.into()), - } -} - -/// Fetch a MIME SubPart -/// -/// eg. FETCH BODY[4.2.2.1] -> [4, 2, 2, 1] -fn map_subpart<'a, F, R>(part: &AnyPart<'a>, path: &[NonZeroU32], f: F) -> Result<R> -where - F: FnOnce(&AnyPart<'a>) -> Result<R>, -{ - if path.is_empty() { - f(part) - } else { - match part { - AnyPart::Mult(x) => map_subpart( - x.children - .get(path[0].get() as usize - 1) - .ok_or(anyhow!("Unable to resolve subpath {:?}, current multipart has only {} elements", path, x.children.len()))?, - &path[1..], - f), - AnyPart::Msg(x) => map_subpart(x.child.as_ref(), path, f), - _ => bail!("You tried to access a subpart on an atomic part (text or binary). Unresolved subpath {:?}", path), - } - } -} - #[cfg(test)] mod tests { use super::*; - use crate::cryptoblob; - use crate::mail::unique_ident; use imap_codec::encode::Encoder; + use imap_codec::imap_types::core::NonEmptyVec; use imap_codec::imap_types::fetch::Section; + use imap_codec::imap_types::fetch::{MacroOrMessageDataItemNames, MessageDataItemName}; use imap_codec::imap_types::response::Response; use imap_codec::ResponseCodec; use std::fs; + use crate::cryptoblob; + use crate::imap::index::MailIndex; + use crate::imap::mail_view::MailView; + use crate::imap::mime_view; + use crate::mail::mailbox::MailMeta; + use crate::mail::query::QueryResult; + use crate::mail::unique_ident; + #[test] fn mailview_body_ext() -> Result<()> { let ap = AttributesProxy::new( @@ -1363,7 +490,6 @@ mod tests { false, ); - let flags = vec![]; let key = cryptoblob::gen_key(); let meta = MailMeta { internaldate: 0u64, @@ -1371,20 +497,23 @@ mod tests { message_key: key, rfc822_size: 8usize, }; - let ids = MailIdentifiers { + + let index_entry = (NonZeroU32::MIN, vec![]); + let mail_in_idx = MailIndex { i: NonZeroU32::MIN, - uid: NonZeroU32::MIN, + uid: index_entry.0, uuid: unique_ident::gen_ident(), + flags: &index_entry.1, }; let rfc822 = b"Subject: hello\r\nFrom: a@a.a\r\nTo: b@b.b\r\nDate: Thu, 12 Oct 2023 08:45:28 +0000\r\n\r\nhello world"; - let content = FetchedMail::new_from_message(eml_codec::parse_message(rfc822)?.1); - - let mv = MailView { - ids: &ids, - content, - meta: &meta, - flags: &flags, + let qr = QueryResult::FullResult { + uuid: mail_in_idx.uuid.clone(), + index: &index_entry, + metadata: meta, + content: rfc822.to_vec(), }; + + let mv = MailView::new(&qr, mail_in_idx)?; let (res_body, _seen) = mv.filter(&ap)?; let fattr = match res_body { @@ -1452,7 +581,7 @@ mod tests { let test_repr = Response::Data(Data::Fetch { seq: NonZeroU32::new(1).unwrap(), - items: NonEmptyVec::from(MessageDataItem::Body(build_imap_email_struct( + items: NonEmptyVec::from(MessageDataItem::Body(mime_view::bodystructure( &message.child, )?)), }); diff --git a/src/imap/mime_view.rs b/src/imap/mime_view.rs new file mode 100644 index 0000000..1f36c47 --- /dev/null +++ b/src/imap/mime_view.rs @@ -0,0 +1,538 @@ +use std::borrow::Cow; +use std::collections::HashSet; +use std::num::NonZeroU32; + +use anyhow::{anyhow, bail, Result}; + +use imap_codec::imap_types::body::{BasicFields, Body as FetchBody, BodyStructure, SpecificFields}; +use imap_codec::imap_types::core::{AString, IString, NString, NonEmptyVec}; +use imap_codec::imap_types::fetch::{Part as FetchPart, Section as FetchSection}; + +use eml_codec::{ + header, mime, mime::r#type::Deductible, part::composite, part::discrete, part::AnyPart, +}; + +use crate::imap::imf_view::message_envelope; + +pub enum BodySection<'a> { + Full(Cow<'a, [u8]>), + Slice { + body: Cow<'a, [u8]>, + origin_octet: u32, + }, +} + +/// Logic for BODY[<section>]<<partial>> +/// Works in 3 times: +/// 1. Find the section (RootMime::subset) +/// 2. Apply the extraction logic (SelectedMime::extract), like TEXT, HEADERS, etc. +/// 3. Keep only the given subset provided by partial +/// +/// Example of message sections: +/// +/// ``` +/// HEADER ([RFC-2822] header of the message) +/// TEXT ([RFC-2822] text body of the message) MULTIPART/MIXED +/// 1 TEXT/PLAIN +/// 2 APPLICATION/OCTET-STREAM +/// 3 MESSAGE/RFC822 +/// 3.HEADER ([RFC-2822] header of the message) +/// 3.TEXT ([RFC-2822] text body of the message) MULTIPART/MIXED +/// 3.1 TEXT/PLAIN +/// 3.2 APPLICATION/OCTET-STREAM +/// 4 MULTIPART/MIXED +/// 4.1 IMAGE/GIF +/// 4.1.MIME ([MIME-IMB] header for the IMAGE/GIF) +/// 4.2 MESSAGE/RFC822 +/// 4.2.HEADER ([RFC-2822] header of the message) +/// 4.2.TEXT ([RFC-2822] text body of the message) MULTIPART/MIXED +/// 4.2.1 TEXT/PLAIN +/// 4.2.2 MULTIPART/ALTERNATIVE +/// 4.2.2.1 TEXT/PLAIN +/// 4.2.2.2 TEXT/RICHTEXT +/// ``` +pub fn body_ext<'a>( + part: &'a AnyPart<'a>, + section: &'a Option<FetchSection<'a>>, + partial: &'a Option<(u32, NonZeroU32)>, +) -> Result<BodySection<'a>> { + let root_mime = NodeMime(part); + let (extractor, path) = SubsettedSection::from(section); + let selected_mime = root_mime.subset(path)?; + let extracted_full = selected_mime.extract(&extractor)?; + Ok(extracted_full.to_body_section(partial)) +} + +/// Logic for BODY and BODYSTRUCTURE +/// +/// ```raw +/// b fetch 29878:29879 (BODY) +/// * 29878 FETCH (BODY (("text" "plain" ("charset" "utf-8") NIL NIL "quoted-printable" 3264 82)("text" "html" ("charset" "utf-8") NIL NIL "quoted-printable" 31834 643) "alternative")) +/// * 29879 FETCH (BODY ("text" "html" ("charset" "us-ascii") NIL NIL "7bit" 4107 131)) +/// ^^^^^^^^^^^^^^^^^^^^^^ ^^^ ^^^ ^^^^^^ ^^^^ ^^^ +/// | | | | | | number of lines +/// | | | | | size +/// | | | | content transfer encoding +/// | | | description +/// | | id +/// | parameter list +/// b OK Fetch completed (0.001 + 0.000 secs). +/// ``` +pub fn bodystructure(part: &AnyPart) -> Result<BodyStructure<'static>> { + NodeMime(part).structure() +} + +/// NodeMime +/// +/// Used for recursive logic on MIME. +/// See SelectedMime for inspection. +struct NodeMime<'a>(&'a AnyPart<'a>); +impl<'a> NodeMime<'a> { + /// A MIME object is a tree of elements. + /// The path indicates which element must be picked. + /// This function returns the picked element as the new view + fn subset(self, path: Option<&'a FetchPart>) -> Result<SelectedMime<'a>> { + match path { + None => Ok(SelectedMime(self.0)), + Some(v) => self.rec_subset(v.0.as_ref()), + } + } + + fn rec_subset(self, path: &'a [NonZeroU32]) -> Result<SelectedMime> { + if path.is_empty() { + Ok(SelectedMime(self.0)) + } else { + match self.0 { + AnyPart::Mult(x) => { + let next = Self(x.children + .get(path[0].get() as usize - 1) + .ok_or(anyhow!("Unable to resolve subpath {:?}, current multipart has only {} elements", path, x.children.len()))?); + next.rec_subset(&path[1..]) + }, + AnyPart::Msg(x) => { + let next = Self(x.child.as_ref()); + next.rec_subset(path) + }, + _ => bail!("You tried to access a subpart on an atomic part (text or binary). Unresolved subpath {:?}", path), + } + } + } + + fn structure(&self) -> Result<BodyStructure<'static>> { + match self.0 { + AnyPart::Txt(x) => NodeTxt(self, x).structure(), + AnyPart::Bin(x) => NodeBin(self, x).structure(), + AnyPart::Mult(x) => NodeMult(self, x).structure(), + AnyPart::Msg(x) => NodeMsg(self, x).structure(), + } + } +} + +//---------------------------------------------------------- + +/// A FetchSection must be handled in 2 times: +/// - First we must extract the MIME part +/// - Then we must process it as desired +/// The given struct mixes both work, so +/// we separate this work here. +enum SubsettedSection<'a> { + Part, + Header, + HeaderFields(&'a NonEmptyVec<AString<'a>>), + HeaderFieldsNot(&'a NonEmptyVec<AString<'a>>), + Text, + Mime, +} +impl<'a> SubsettedSection<'a> { + fn from(section: &'a Option<FetchSection>) -> (Self, Option<&'a FetchPart>) { + match section { + Some(FetchSection::Text(maybe_part)) => (Self::Text, maybe_part.as_ref()), + Some(FetchSection::Header(maybe_part)) => (Self::Header, maybe_part.as_ref()), + Some(FetchSection::HeaderFields(maybe_part, fields)) => { + (Self::HeaderFields(fields), maybe_part.as_ref()) + } + Some(FetchSection::HeaderFieldsNot(maybe_part, fields)) => { + (Self::HeaderFieldsNot(fields), maybe_part.as_ref()) + } + Some(FetchSection::Mime(part)) => (Self::Mime, Some(part)), + Some(FetchSection::Part(part)) => (Self::Part, Some(part)), + None => (Self::Part, None), + } + } +} + +/// Used for current MIME inspection +/// +/// See NodeMime for recursive logic +struct SelectedMime<'a>(&'a AnyPart<'a>); +impl<'a> SelectedMime<'a> { + /// 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. + fn extract(&self, extractor: &SubsettedSection<'a>) -> Result<ExtractedFull<'a>> { + match extractor { + SubsettedSection::Text => self.text(), + SubsettedSection::Header => self.header(), + SubsettedSection::HeaderFields(fields) => self.header_fields(fields, false), + SubsettedSection::HeaderFieldsNot(fields) => self.header_fields(fields, true), + SubsettedSection::Part => self.part(), + SubsettedSection::Mime => self.mime(), + } + } + + fn mime(&self) -> Result<ExtractedFull<'a>> { + let bytes = match &self.0 { + AnyPart::Txt(p) => p.mime.fields.raw, + AnyPart::Bin(p) => p.mime.fields.raw, + AnyPart::Msg(p) => p.child.mime().raw, + AnyPart::Mult(p) => p.mime.fields.raw, + }; + Ok(ExtractedFull(bytes.into())) + } + + fn part(&self) -> Result<ExtractedFull<'a>> { + let bytes = match &self.0 { + AnyPart::Txt(p) => p.body, + AnyPart::Bin(p) => p.body, + AnyPart::Msg(p) => p.raw_part, + AnyPart::Mult(_) => bail!("Multipart part has no body"), + }; + Ok(ExtractedFull(bytes.to_vec().into())) + } + + /// 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. + /// HEADER.FIELDS and HEADER.FIELDS.NOT are followed by a list of + /// field-name (as defined in [RFC-2822]) names, and return a + /// subset of the header. The subset returned by HEADER.FIELDS + /// contains only those header fields with a field-name that + /// matches one of the names in the list; similarly, the subset + /// returned by HEADER.FIELDS.NOT contains only the header fields + /// with a non-matching field-name. The field-matching is + /// case-insensitive but otherwise exact. + fn header_fields( + &self, + fields: &'a NonEmptyVec<AString<'a>>, + invert: bool, + ) -> Result<ExtractedFull<'a>> { + // Build a lowercase ascii hashset with the fields to fetch + let index = fields + .as_ref() + .iter() + .map(|x| { + match x { + AString::Atom(a) => a.inner().as_bytes(), + AString::String(IString::Literal(l)) => l.as_ref(), + AString::String(IString::Quoted(q)) => q.inner().as_bytes(), + } + .to_ascii_lowercase() + }) + .collect::<HashSet<_>>(); + + // Extract MIME headers + let mime = match &self.0 { + AnyPart::Msg(msg) => msg.child.mime(), + other => other.mime(), + }; + + // Filter our MIME headers based on the field index + // 1. Keep only the correctly formatted headers + // 2. Keep only based on the index presence or absence + // 3. Reduce as a byte vector + let buffer = mime + .kv + .iter() + .filter_map(|field| match field { + header::Field::Good(header::Kv2(k, v)) => Some((k, v)), + _ => None, + }) + .filter(|(k, _)| index.contains(&k.to_ascii_lowercase()) ^ invert) + .fold(vec![], |mut acc, (k, v)| { + acc.extend(*k); + acc.extend(b": "); + acc.extend(*v); + acc.extend(b"\r\n"); + acc + }); + + Ok(ExtractedFull(buffer.into())) + } + + /// The HEADER [...] part specifiers refer to the [RFC-2822] header of the message or of + /// an encapsulated [MIME-IMT] MESSAGE/RFC822 message. + /// ```raw + /// HEADER ([RFC-2822] header of the message) + /// ``` + fn header(&self) -> Result<ExtractedFull<'a>> { + let msg = self + .0 + .as_message() + .ok_or(anyhow!("Selected part must be a message/rfc822"))?; + Ok(ExtractedFull(msg.raw_headers.into())) + } + + /// The TEXT part specifier refers to the text body of the message, omitting the [RFC-2822] header. + fn text(&self) -> Result<ExtractedFull<'a>> { + let msg = self + .0 + .as_message() + .ok_or(anyhow!("Selected part must be a message/rfc822"))?; + Ok(ExtractedFull(msg.raw_body.into())) + } + + // ------------ + + /// Basic field of a MIME part that is + /// common to all parts + fn basic_fields(&self) -> Result<BasicFields<'static>> { + let sz = match self.0 { + AnyPart::Txt(x) => x.body.len(), + AnyPart::Bin(x) => x.body.len(), + AnyPart::Msg(x) => x.raw_part.len(), + AnyPart::Mult(_) => 0, + }; + let m = self.0.mime(); + let parameter_list = m + .ctype + .as_ref() + .map(|x| { + x.params + .iter() + .map(|p| { + ( + IString::try_from(String::from_utf8_lossy(p.name).to_string()), + IString::try_from(p.value.to_string()), + ) + }) + .filter(|(k, v)| k.is_ok() && v.is_ok()) + .map(|(k, v)| (k.unwrap(), v.unwrap())) + .collect() + }) + .unwrap_or(vec![]); + + Ok(BasicFields { + parameter_list, + id: NString( + m.id.as_ref() + .and_then(|ci| IString::try_from(ci.to_string()).ok()), + ), + description: NString( + m.description + .as_ref() + .and_then(|cd| IString::try_from(cd.to_string()).ok()), + ), + content_transfer_encoding: match m.transfer_encoding { + mime::mechanism::Mechanism::_8Bit => unchecked_istring("8bit"), + mime::mechanism::Mechanism::Binary => unchecked_istring("binary"), + mime::mechanism::Mechanism::QuotedPrintable => { + unchecked_istring("quoted-printable") + } + mime::mechanism::Mechanism::Base64 => unchecked_istring("base64"), + _ => unchecked_istring("7bit"), + }, + // @FIXME we can't compute the size of the message currently... + size: u32::try_from(sz)?, + }) + } +} + +// --------------------------- +struct NodeMsg<'a>(&'a NodeMime<'a>, &'a composite::Message<'a>); +impl<'a> NodeMsg<'a> { + fn structure(&self) -> Result<BodyStructure<'static>> { + let basic = SelectedMime(self.0 .0).basic_fields()?; + + Ok(BodyStructure::Single { + body: FetchBody { + basic, + specific: SpecificFields::Message { + envelope: Box::new(message_envelope(&self.1.imf)), + body_structure: Box::new(NodeMime(&self.1.child).structure()?), + number_of_lines: nol(self.1.raw_part), + }, + }, + extension_data: None, + }) + } +} +struct NodeMult<'a>(&'a NodeMime<'a>, &'a composite::Multipart<'a>); +impl<'a> NodeMult<'a> { + fn structure(&self) -> Result<BodyStructure<'static>> { + let itype = &self.1.mime.interpreted_type; + let subtype = IString::try_from(itype.subtype.to_string()) + .unwrap_or(unchecked_istring("alternative")); + + let inner_bodies = self + .1 + .children + .iter() + .filter_map(|inner| NodeMime(&inner).structure().ok()) + .collect::<Vec<_>>(); + + NonEmptyVec::validate(&inner_bodies)?; + let bodies = NonEmptyVec::unvalidated(inner_bodies); + + Ok(BodyStructure::Multi { + bodies, + subtype, + extension_data: None, + /*Some(MultipartExtensionData { + parameter_list: vec![], + disposition: None, + language: None, + location: None, + extension: vec![], + })*/ + }) + } +} +struct NodeTxt<'a>(&'a NodeMime<'a>, &'a discrete::Text<'a>); +impl<'a> NodeTxt<'a> { + fn structure(&self) -> Result<BodyStructure<'static>> { + let mut basic = SelectedMime(self.0 .0).basic_fields()?; + + // Get the interpreted content type, set it + let itype = match &self.1.mime.interpreted_type { + Deductible::Inferred(v) | Deductible::Explicit(v) => v, + }; + let subtype = + IString::try_from(itype.subtype.to_string()).unwrap_or(unchecked_istring("plain")); + + // Add charset to the list of parameters if we know it has been inferred as it will be + // missing from the parsed content. + if let Deductible::Inferred(charset) = &itype.charset { + basic.parameter_list.push(( + unchecked_istring("charset"), + IString::try_from(charset.to_string()).unwrap_or(unchecked_istring("us-ascii")), + )); + } + + Ok(BodyStructure::Single { + body: FetchBody { + basic, + specific: SpecificFields::Text { + subtype, + number_of_lines: nol(self.1.body), + }, + }, + extension_data: None, + }) + } +} + +struct NodeBin<'a>(&'a NodeMime<'a>, &'a discrete::Binary<'a>); +impl<'a> NodeBin<'a> { + fn structure(&self) -> Result<BodyStructure<'static>> { + let basic = SelectedMime(self.0 .0).basic_fields()?; + + let default = mime::r#type::NaiveType { + main: &b"application"[..], + sub: &b"octet-stream"[..], + params: vec![], + }; + let ct = self.1.mime.fields.ctype.as_ref().unwrap_or(&default); + + let r#type = IString::try_from(String::from_utf8_lossy(ct.main).to_string()).or(Err( + anyhow!("Unable to build IString from given Content-Type type given"), + ))?; + + let subtype = IString::try_from(String::from_utf8_lossy(ct.sub).to_string()).or(Err( + anyhow!("Unable to build IString from given Content-Type subtype given"), + ))?; + + Ok(BodyStructure::Single { + body: FetchBody { + basic, + specific: SpecificFields::Basic { r#type, subtype }, + }, + extension_data: None, + }) + } +} + +// --------------------------- + +struct ExtractedFull<'a>(Cow<'a, [u8]>); +impl<'a> ExtractedFull<'a> { + /// It is possible to fetch a substring of the designated text. + /// This is done by appending an open angle bracket ("<"), the + /// octet position of the first desired octet, a period, the + /// maximum number of octets desired, and a close angle bracket + /// (">") to the part specifier. If the starting octet is beyond + /// the end of the text, an empty string is returned. + /// + /// Any partial fetch that attempts to read beyond the end of the + /// text is truncated as appropriate. A partial fetch that starts + /// at octet 0 is returned as a partial fetch, even if this + /// truncation happened. + /// + /// Note: This means that BODY[]<0.2048> of a 1500-octet message + /// will return BODY[]<0> with a literal of size 1500, not + /// BODY[]. + /// + /// Note: A substring fetch of a HEADER.FIELDS or + /// HEADER.FIELDS.NOT part specifier is calculated after + /// subsetting the header. + fn to_body_section(self, partial: &'_ Option<(u32, NonZeroU32)>) -> BodySection<'a> { + match partial { + Some((begin, len)) => self.partialize(*begin, *len), + None => BodySection::Full(self.0), + } + } + + fn partialize(self, begin: u32, len: NonZeroU32) -> BodySection<'a> { + // Asked range is starting after the end of the content, + // returning an empty buffer + if begin as usize > self.0.len() { + return BodySection::Slice { + body: Cow::Borrowed(&[][..]), + origin_octet: begin, + }; + } + + // Asked range is ending after the end of the content, + // slice only the beginning of the buffer + if (begin + len.get()) as usize >= self.0.len() { + return BodySection::Slice { + body: match self.0 { + Cow::Borrowed(body) => Cow::Borrowed(&body[begin as usize..]), + Cow::Owned(body) => Cow::Owned(body[begin as usize..].to_vec()), + }, + origin_octet: begin, + }; + } + + // Range is included inside the considered content, + // this is the "happy case" + BodySection::Slice { + body: match self.0 { + Cow::Borrowed(body) => { + Cow::Borrowed(&body[begin as usize..(begin + len.get()) as usize]) + } + Cow::Owned(body) => { + Cow::Owned(body[begin as usize..(begin + len.get()) as usize].to_vec()) + } + }, + origin_octet: begin, + } + } +} + +/// ---- LEGACY + +/// s is set to static to ensure that only compile time values +/// checked by developpers are passed. +fn unchecked_istring(s: &'static str) -> IString { + IString::try_from(s).expect("this value is expected to be a valid imap-codec::IString") +} + +// Number Of Lines +fn nol(input: &[u8]) -> u32 { + input + .iter() + .filter(|x| **x == b'\n') + .count() + .try_into() + .unwrap_or(0) +} diff --git a/src/imap/mod.rs b/src/imap/mod.rs index aac1fd3..4142ef9 100644 --- a/src/imap/mod.rs +++ b/src/imap/mod.rs @@ -1,8 +1,15 @@ +mod attributes; mod capability; mod command; +mod flags; mod flow; +mod imf_view; +mod index; +mod mail_view; mod mailbox_view; +mod mime_view; mod response; +mod search; mod session; use std::net::SocketAddr; diff --git a/src/imap/search.rs b/src/imap/search.rs new file mode 100644 index 0000000..b3c6b05 --- /dev/null +++ b/src/imap/search.rs @@ -0,0 +1,130 @@ +use imap_codec::imap_types::core::NonEmptyVec; +use imap_codec::imap_types::search::SearchKey; +use imap_codec::imap_types::sequence::{SeqOrUid, Sequence, SequenceSet}; +use std::num::NonZeroU32; + +pub enum SeqType { + Undefined, + NonUid, + Uid, +} +impl SeqType { + pub fn is_uid(&self) -> bool { + matches!(self, Self::Uid) + } +} + +pub struct Criteria<'a>(pub &'a SearchKey<'a>); +impl<'a> Criteria<'a> { + /// Returns a set of email identifiers that is greater or equal + /// to the set of emails to return + pub fn to_sequence_set(&self) -> (SequenceSet, SeqType) { + match self.0 { + SearchKey::All => (sequence_set_all(), SeqType::Undefined), + SearchKey::SequenceSet(seq_set) => (seq_set.clone(), SeqType::NonUid), + SearchKey::Uid(seq_set) => (seq_set.clone(), SeqType::Uid), + SearchKey::Not(_inner) => { + tracing::debug!( + "using NOT in a search request is slow: it selects all identifiers" + ); + (sequence_set_all(), SeqType::Undefined) + } + SearchKey::Or(left, right) => { + tracing::debug!("using OR in a search request is slow: no deduplication is done"); + let (base, base_seqtype) = Self(&left).to_sequence_set(); + let (ext, ext_seqtype) = Self(&right).to_sequence_set(); + + // Check if we have a UID/ID conflict in fetching: now we don't know how to handle them + match (base_seqtype, ext_seqtype) { + (SeqType::Uid, SeqType::NonUid) | (SeqType::NonUid, SeqType::Uid) => { + (sequence_set_all(), SeqType::Undefined) + } + (SeqType::Undefined, x) | (x, _) => { + let mut new_vec = base.0.into_inner(); + new_vec.extend_from_slice(ext.0.as_ref()); + let seq = SequenceSet( + NonEmptyVec::try_from(new_vec) + .expect("merging non empty vec lead to non empty vec"), + ); + (seq, x) + } + } + } + SearchKey::And(search_list) => { + tracing::debug!( + "using AND in a search request is slow: no intersection is performed" + ); + search_list + .as_ref() + .iter() + .map(|crit| Self(&crit).to_sequence_set()) + .min_by(|(x, _), (y, _)| { + let x_size = approx_sequence_set_size(x); + let y_size = approx_sequence_set_size(y); + x_size.cmp(&y_size) + }) + .unwrap_or((sequence_set_all(), SeqType::Undefined)) + } + _ => (sequence_set_all(), SeqType::Undefined), + } + } + + /// Not really clever as we can have cases where we filter out + /// the email before needing to inspect its meta. + /// But for now we are seeking the most basic/stupid algorithm. + pub fn need_meta(&self) -> bool { + use SearchKey::*; + match self.0 { + // IMF Headers + Bcc(_) | Cc(_) | From(_) | Header(..) | SentBefore(_) | SentOn(_) | SentSince(_) + | Subject(_) | To(_) => true, + // Internal Date is also stored in MailMeta + Before(_) | On(_) | Since(_) => true, + // Message size is also stored in MailMeta + Larger(_) | Smaller(_) => true, + And(and_list) => and_list.as_ref().iter().any(|sk| Criteria(sk).need_meta()), + Not(inner) => Criteria(inner).need_meta(), + Or(left, right) => Criteria(left).need_meta() || Criteria(right).need_meta(), + _ => false, + } + } + + pub fn need_body(&self) -> bool { + use SearchKey::*; + match self.0 { + Text(_) | Body(_) => true, + And(and_list) => and_list.as_ref().iter().any(|sk| Criteria(sk).need_body()), + Not(inner) => Criteria(inner).need_body(), + Or(left, right) => Criteria(left).need_body() || Criteria(right).need_body(), + _ => false, + } + } +} + +fn sequence_set_all() -> SequenceSet { + SequenceSet::from(Sequence::Range( + SeqOrUid::Value(NonZeroU32::MIN), + SeqOrUid::Asterisk, + )) +} + +// This is wrong as sequences can overlap +fn approx_sequence_set_size(seq_set: &SequenceSet) -> u64 { + seq_set.0.as_ref().iter().fold(0u64, |acc, seq| { + acc.saturating_add(approx_sequence_size(seq)) + }) +} + +// This is wrong as sequence UID can have holes, +// as we don't know the number of messages in the mailbox also +fn approx_sequence_size(seq: &Sequence) -> u64 { + match seq { + Sequence::Single(_) => 1, + Sequence::Range(SeqOrUid::Asterisk, _) | Sequence::Range(_, SeqOrUid::Asterisk) => u64::MAX, + Sequence::Range(SeqOrUid::Value(x1), SeqOrUid::Value(x2)) => { + let x2 = x2.get() as i64; + let x1 = x1.get() as i64; + (x2 - x1).abs().try_into().unwrap_or(1) + } + } +} |