tpcpr/src/parser/mod.rs

229 lines
9.5 KiB
Rust
Raw Normal View History

2020-05-13 00:03:45 +02:00
extern crate bitfield;
2020-05-21 00:54:56 +02:00
extern crate byteorder;
2020-05-13 00:03:45 +02:00
extern crate eui48;
mod packet_handler;
use pcap::{Capture, Linktype};
use regex::bytes::Regex;
use std::str;
2020-06-04 21:13:48 +02:00
//use std::thread::{spawn, JoinHandle};
//use std::sync::mpsc::{channel, Receiver};
2020-05-13 00:03:45 +02:00
2020-05-21 00:54:56 +02:00
/* protocol ids, LittleEndian */
const ETH_P_IPV6: usize = 0xDD86;
const ETH_P_IP: usize = 0x08;
const TCP: usize = 0x06;
const UDP: usize = 0x11;
2020-05-31 00:44:47 +02:00
const ETH_P_ARP: usize = 0x0608;
const ETH_P_RARP: usize = 0x3580;
2020-05-19 19:34:19 +02:00
/*
QryData could be written in the sense of QryData{ ... frame: .., packet: .., segment:.. }
On the one hand, only the actual type of frame/packet/segment would be contained in the resulting struct.
So, increased benefit in serialization/cpu time, could result in less data to be serialized, depending on layout.
On the other hand, each datagram::type needs to implement traits which would need to be dynamically dispatched by returning any of these types per iso level from a single function each. The result would be a performance decrease.
See: https://doc.rust-lang.org/book/ch17-02-trait-objects.html#trait-objects-perform-dynamic-dispatch
Then again, parser logic would be fewer lines + more unified using the latter method. Maybe better optimizable as well? Maybe this is a nice tradeoff?
TODO: Implement and benchmark dynamically dispatched packet data in conjunction with restructured QryData.
*/
2020-05-13 00:03:45 +02:00
#[derive(Debug, Clone)]
2020-05-21 00:54:56 +02:00
pub struct QryData {
pub id: i32,
pub time: f64,
pub data: Option<Vec<u8>>,
pub ether_header: Option<packet_handler::EtherHeader>,
2020-05-21 00:54:56 +02:00
pub ipv4_header: Option<packet_handler::IpV4Header>,
pub ipv6_header: Option<packet_handler::IpV6Header>,
pub tcp_header: Option<packet_handler::TcpHeader>,
pub udp_header: Option<packet_handler::UdpHeader>,
2020-05-31 00:44:47 +02:00
pub arp_header: Option<packet_handler::ArpHeader>,
pub reg_res: Option<String>,
2020-05-13 00:03:45 +02:00
}
2020-06-07 23:56:53 +02:00
fn init_qrydata( ) -> Result<QryData, core::fmt::Error> {
Ok(QryData {
id: 0,
time: 0.0,
data: None,
ether_header: None::<packet_handler::EtherHeader>,
2020-06-07 23:56:53 +02:00
ipv4_header: None::<packet_handler::IpV4Header>,
ipv6_header: None::<packet_handler::IpV6Header>,
tcp_header: None::<packet_handler::TcpHeader>,
udp_header: None::<packet_handler::UdpHeader>,
arp_header: None::<packet_handler::ArpHeader>,
reg_res: None::<String>,
})
}
enum EncapsulationType { // pcap::Linktype::get_name() is unsafe.
EN10MB = 1, // See: https://docs.rs/pcap/0.7.0/src/pcap/lib.rs.html#247-261
RAW = 101, // Would this be an issue?
}
2020-06-07 23:56:53 +02:00
impl QryData {
// This is not cool!
// This will get modularized into subfunctions
fn encap_ether( packet_data: &[u8] ) -> Option<QryData> {
let mut pkg: QryData = init_qrydata().unwrap();
pkg.ether_header = Some(packet_handler::ethernet_handler(packet_data)).unwrap();
match pkg.ether_header.unwrap().ether_type as usize {
ETH_P_IP => {
pkg.ipv4_header = Some(packet_handler::ip_handler(packet_data)).unwrap();
match pkg.ipv4_header.unwrap().ip_protocol as usize {
TCP => {
pkg.tcp_header = Some(packet_handler::tcp_handler(
pkg.ipv4_header.unwrap().ip_ihl,
packet_data,
))
.unwrap();
pkg.data = Some(packet_handler::payload_handler(
pkg.ipv4_header.unwrap().ip_ihl,
pkg.tcp_header.unwrap().data_offset,
packet_data,
)).unwrap();
}
UDP => {
pkg.udp_header = Some(packet_handler::udp_handler(
pkg.ipv4_header.unwrap().ip_ihl,
packet_data,
))
.unwrap();
pkg.data = Some(packet_handler::payload_handler(
pkg.ipv4_header.unwrap().ip_ihl,
7,
packet_data,
)).unwrap();
}
_ => println!("Transport layer protocol not implemented"),
}
}
ETH_P_IPV6 => {
pkg.ipv6_header = Some(packet_handler::ipv6_handler(packet_data)).unwrap();
match pkg.ipv6_header.unwrap().next_header as usize {
TCP => {
pkg.tcp_header = Some(packet_handler::tcp_handler(10, packet_data)).unwrap();
pkg.data = Some(packet_handler::payload_handler(
10,
pkg.tcp_header.unwrap().data_offset,
packet_data,
)).unwrap();
}
UDP => {
pkg.udp_header = Some(packet_handler::udp_handler(10, packet_data)).unwrap();
pkg.data = Some(packet_handler::payload_handler(10, 7, packet_data)).unwrap();
}
_ => println!("Transport layer protocol not implemented"),
}
}
2020-05-31 00:44:47 +02:00
ETH_P_ARP | ETH_P_RARP => {
pkg.arp_header = Some(packet_handler::arp_handler(packet_data)).unwrap();
2020-05-31 00:44:47 +02:00
}
_ => println!("Network protocol not implemented"),
}
Some(pkg)
}
fn encap_rawip ( packet_data: &[u8] ) -> Option<QryData> {
let mut pkg: QryData = init_qrydata().unwrap();
pkg.ipv4_header = Some(packet_handler::ip_handler(packet_data)).unwrap();
match pkg.ipv4_header.unwrap().ip_protocol as usize {
TCP => {
pkg.tcp_header = Some(packet_handler::tcp_handler(
pkg.ipv4_header.unwrap().ip_ihl,
packet_data,
))
.unwrap();
pkg.data = Some(packet_handler::payload_handler(
pkg.ipv4_header.unwrap().ip_ihl,
pkg.tcp_header.unwrap().data_offset,
packet_data,
)).unwrap();
}
UDP => {
pkg.udp_header = Some(packet_handler::udp_handler(
pkg.ipv4_header.unwrap().ip_ihl,
packet_data,
))
.unwrap();
pkg.data = Some(packet_handler::payload_handler(
pkg.ipv4_header.unwrap().ip_ihl,
7,
packet_data,
)).unwrap();
}
_ => println!("Transport layer protocol not implemented"),
}
Some(pkg)
}
}
/* Regex parse _complete_ package */
fn flag_carnage(re: &Regex, payload: &[u8]) -> Option<String> {
let mut flags: String = String::new();
for mat in re.find_iter(payload) {
// TODO: Test benchmark format! vs. push_str()
// flags.push_str(&format!("{} ",std::str::from_utf8(mat.as_bytes()).unwrap()));
// See: https://github.com/hoodie/concatenation_benchmarks-rs
flags.push_str(std::str::from_utf8(mat.as_bytes()).unwrap());
flags.push_str(";");
}
match 0 < flags.len() {
false => None,
true => Some(flags),
}
}
pub fn parse(parse_file: &std::path::Path, filter_str: &str, regex_filter: &str, encap: u16) -> Vec<QryData> {
let mut me: QryData = init_qrydata().unwrap();
let mut v: Vec<QryData> = Vec::new();
let mut cap = Capture::from_file(parse_file).unwrap();
Capture::filter(&mut cap, &filter_str).unwrap();
let re = Regex::new(regex_filter).unwrap();
while let Ok(packet) = cap.next() {
match encap {
// Syntax is clunky, but no num_derive + num_traits dependencies.
encap if encap == EncapsulationType::EN10MB as u16 => me = QryData::encap_ether(packet.data).unwrap(),
encap if encap == EncapsulationType::RAW as u16 => me = QryData::encap_rawip(packet.data).unwrap(),
_ => (),
};
me.time = (packet.header.ts.tv_usec as f64 / 1000000.0) + packet.header.ts.tv_sec as f64;
//me.data = Some(packet.data.to_vec());
me.reg_res = flag_carnage(&re, packet.data); // Regex overhead is between 4-9% --single threaded-- on complete packet [u8] data
v.push(me.clone());
2020-05-13 00:03:45 +02:00
}
v
}
2020-06-04 21:13:48 +02:00
/* This could need some love */
2020-06-06 03:29:38 +02:00
pub fn parse_device(parse_device: &str, filter_str: &str, insert_max: &usize, regex_filter: &str) -> Vec<QryData> {
2020-06-07 23:56:53 +02:00
let mut me: QryData = init_qrydata ( ).unwrap();
2020-05-13 00:03:45 +02:00
let mut v: Vec<QryData> = Vec::new();
let mut cap = Capture::from_device(parse_device).unwrap().open().unwrap();
Capture::filter(&mut cap, &filter_str).unwrap();
let linktype = cap.get_datalink();
2020-06-06 03:29:38 +02:00
let re = Regex::new(regex_filter).unwrap();
2020-05-21 00:54:56 +02:00
'parse: while let Ok(packet) = cap.next() {
match linktype {
Linktype(1) => me = QryData::encap_ether(packet.data).unwrap(),
Linktype(101) => me = QryData::encap_rawip(packet.data).unwrap(),
_ => (),
}
me.time = (packet.header.ts.tv_usec as f64 / 1000000.0) + packet.header.ts.tv_sec as f64;
me.reg_res = flag_carnage(&re, packet.data);
v.push(me.clone());
2020-05-13 00:03:45 +02:00
if &v.len() >= insert_max {
break 'parse;
}
}
v
}