From 9821a9888e70569b5d3b1dcd9220968d5ad76f00 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Rodolphe=20Br=C3=A9ard?= <rodolphe@what.tf>
Date: Mon, 27 Mar 2023 22:46:17 +0200
Subject: [PATCH] Replace the mailparse crate by a custom header pseudo-parser

---
 Cargo.toml            |  1 -
 src/main.rs           |  1 +
 src/message.rs        | 48 +++++++++++++++----------------
 src/parsed_message.rs | 66 +++++++++++++++++++++++++++++++++++++++++++
 4 files changed, 91 insertions(+), 25 deletions(-)
 create mode 100644 src/parsed_message.rs

diff --git a/Cargo.toml b/Cargo.toml
index 7ea5d3d..13e60ff 100644
--- a/Cargo.toml
+++ b/Cargo.toml
@@ -11,5 +11,4 @@ publish = false
 clap = { version = "4.1.13", default-features = false, features = ["std", "derive"] }
 env_logger = { version = "0.10.0", default-features = false }
 log = { version = "0.4.17", default-features = false }
-mailparse = { version = "0.14.0", default-features = false }
 nom = { version = "7.1.3", default-features = false }
diff --git a/src/main.rs b/src/main.rs
index 0b2380a..0d61563 100644
--- a/src/main.rs
+++ b/src/main.rs
@@ -5,6 +5,7 @@ mod entry;
 mod handshake;
 mod logs;
 mod message;
+mod parsed_message;
 mod stdin_reader;
 
 use algorithm::Algorithm;
diff --git a/src/message.rs b/src/message.rs
index 2276f94..b79a56f 100644
--- a/src/message.rs
+++ b/src/message.rs
@@ -1,6 +1,6 @@
 use crate::config::Config;
 use crate::entry::Entry;
-use mailparse::parse_mail;
+use crate::parsed_message::ParsedMessage;
 use std::io::{BufWriter, Write};
 
 pub const RETURN_SEP: &[u8] = b"|";
@@ -52,35 +52,35 @@ impl Message {
 	}
 
 	pub fn sign_and_return(&self, cnf: &Config) {
-		log::trace!("content:\n{}", crate::display_bytes!(&self.content));
-		match parse_mail(&self.content) {
+		log::trace!("content: {}", crate::display_bytes!(&self.content));
+		match ParsedMessage::from_bytes(&self.content) {
 			Ok(parsed_msg) => {
 				log::trace!("mail parsed");
-				for h in parsed_msg.get_headers() {
-					log::trace!("{:?}", h);
-				}
-				match self.get_body() {
-					Some(body) => {
-						log::trace!("MailBody:\n{}", crate::display_bytes!(body));
-						// TODO: sign the message using DKIM
-					}
-					None => {
-						log::error!("{}: unable to find the body", self.session_id);
-					}
+				for h in &parsed_msg.headers {
+					log::trace!(
+						"ParsedMessage: header: raw: {}",
+						crate::display_bytes!(h.raw)
+					);
+					log::trace!(
+						"ParsedMessage: header: name: {}",
+						crate::display_bytes!(h.name)
+					);
+					log::trace!(
+						"ParsedMessage: header: value: {}",
+						crate::display_bytes!(h.value)
+					);
 				}
+				log::trace!(
+					"ParsedMessage: body: {}",
+					crate::display_bytes!(parsed_msg.body)
+				);
+				// TODO: sign the message using DKIM
 			}
-			Err(e) => {
-				log::error!("{}: unable to parse message: {e}", self.session_id);
+			Err(_) => {
+				log::error!("{}: unable to parse message", self.session_id);
 			}
-		};
-		self.print_msg();
-	}
-
-	fn get_body(&self) -> Option<&[u8]> {
-		match self.content.windows(4).position(|w| w == b"\r\n\r\n") {
-			Some(body_index) => Some(&self.content[body_index + 4..]),
-			None => None,
 		}
+		self.print_msg();
 	}
 
 	fn print_msg(&self) {
diff --git a/src/parsed_message.rs b/src/parsed_message.rs
new file mode 100644
index 0000000..336234a
--- /dev/null
+++ b/src/parsed_message.rs
@@ -0,0 +1,66 @@
+pub struct ParsedMessage<'a> {
+	pub headers: Vec<ParsedHeader<'a>>,
+	pub body: &'a [u8],
+}
+
+impl<'a> ParsedMessage<'a> {
+	pub fn from_bytes(data: &'a [u8]) -> Result<Self, ()> {
+		let (mut raw_headers, body) = match data.windows(4).position(|w| w == b"\r\n\r\n") {
+			Some(body_index) => (&data[..body_index + 2], &data[body_index + 4..]),
+			None => return Err(()),
+		};
+		let mut headers = Vec::with_capacity(128);
+		while !raw_headers.is_empty() {
+			let end_index = header_end_pos(raw_headers)?;
+			let h = ParsedHeader::from_bytes(&raw_headers[..end_index])?;
+			headers.push(h);
+			raw_headers = &raw_headers[end_index..];
+		}
+		headers.shrink_to_fit();
+		Ok(Self { headers, body })
+	}
+}
+
+fn is_wsp(c: u8) -> bool {
+	c == b' ' || c == b'\t'
+}
+
+fn header_end_pos(data: &[u8]) -> Result<usize, ()> {
+	let mut ret = 0;
+	let max_len = data.len();
+	loop {
+		ret += data[ret..]
+			.windows(2)
+			.position(|w| w == b"\r\n")
+			.ok_or(())? + 2;
+		if ret == max_len {
+			return Ok(ret);
+		}
+		if !is_wsp(data[ret]) {
+			return Ok(ret);
+		}
+	}
+}
+
+pub struct ParsedHeader<'a> {
+	pub name: &'a [u8],
+	pub name_lower: String,
+	pub value: &'a [u8],
+	pub raw: &'a [u8],
+}
+
+impl<'a> ParsedHeader<'a> {
+	fn from_bytes(data: &'a [u8]) -> Result<Self, ()> {
+		let colon_pos = data.iter().position(|&w| w == b':').ok_or(())?;
+		let name = &data[..colon_pos];
+		let value = &data[colon_pos + 1..];
+		Ok(Self {
+			name,
+			name_lower: String::from_utf8(name.to_vec())
+				.map_err(|_| ())?
+				.to_lowercase(),
+			value,
+			raw: data,
+		})
+	}
+}