Improve saving format

This commit is contained in:
Michael Mikovsky
2025-04-17 16:27:45 -06:00
parent f927719b06
commit ddfee03266
6 changed files with 103 additions and 81 deletions
-1
View File
@@ -4,7 +4,6 @@ version = "0.1.0"
edition = "2024" edition = "2024"
[dependencies] [dependencies]
bincode = { version = "2.0.1", features = ["serde"] }
byteorder = "1.5.0" byteorder = "1.5.0"
indicatif = "0.17.11" indicatif = "0.17.11"
memchr = "2.7.4" memchr = "2.7.4"
+47 -46
View File
@@ -1,12 +1,10 @@
use std::{net::IpAddr, sync::Arc, time::Instant}; use std::{net::IpAddr, sync::Arc, time::Instant};
use rocksdb::{Cache, ColumnFamily, DB, IteratorMode, Options, WriteBatch}; use rocksdb::{Cache, ColumnFamily, IteratorMode, Options, WriteBatch, DB};
use serde::{Deserialize, Serialize}; use serde::{Deserialize, Serialize};
use crate::port_scan::port_scan::ScanResult; use crate::port_scan::port_scan::ScanResult;
static COLUMN_COUNT: usize = 5;
// Global settings for optimal performance // Global settings for optimal performance
const BLOCK_CACHE_SIZE_MB: usize = 512; // 512MB block cache const BLOCK_CACHE_SIZE_MB: usize = 512; // 512MB block cache
const WRITE_BUFFER_SIZE_MB: usize = 64; // 64MB write buffer const WRITE_BUFFER_SIZE_MB: usize = 64; // 64MB write buffer
@@ -21,25 +19,49 @@ pub struct ResultDatabase {
#[derive(Debug, Serialize, Deserialize, Clone)] #[derive(Debug, Serialize, Deserialize, Clone)]
pub struct StringRow { pub struct StringRow {
pub id: String, // Row identifier pub id: String, // Row identifier
pub values: Vec<String>, // Array of string values pub ports: Vec<i32>, // Array of string values
} }
impl StringRow { impl StringRow {
pub fn to_string(&self) -> String { pub fn to_string(&self) -> String {
let mut str = "".to_string(); let mut str = "".to_string();
str += format!("Row ID: {}, Values: [", self.id).as_str(); str += format!("{} - ports: [{}]", self.id, join_nums(&self.ports, ",")).as_str();
for (i, value) in self.values.iter().enumerate() {
if i > 0 {
str += ", ";
}
str += format!("{}: \"{}\"", i, value).as_str();
}
str += "]";
str str
} }
pub fn ports_to_string(&self) -> String {
return join_nums(&self.ports, ",");
}
}
pub fn join_nums(nums: &Vec<i32>, sep: &str) -> String {
// 1. Convert numbers to strings
let str_nums: Vec<String> = nums
.iter()
.map(|n| n.to_string()) // map every integer to a string
.collect(); // collect the strings into the vector
// 2. Join the strings. There's already a function for this.
str_nums.join(sep)
}
pub fn split_nums(str: &str, sep: &str) -> Vec<i32> {
if str.is_empty() {
return vec![];
};
return str
.split(sep)
.map(|n| {
if let Ok(num) = n.parse::<i32>() {
return num;
} else {
return 0;
}
})
.collect();
} }
impl ResultDatabase { impl ResultDatabase {
@@ -74,13 +96,7 @@ impl ResultDatabase {
// Define column families for different indexes // Define column families for different indexes
let mut column_families = vec!["default".to_string()]; // Main data store let column_families = vec!["default".to_string(), "ports".to_string()];
// Add column families for each column index we might want to search by
// (for demo, we'll create indexes for 5 potential columns)
for i in 0..COLUMN_COUNT {
column_families.push(format!("col{}_idx", i).to_string());
}
Self { Self {
path: path.to_string(), path: path.to_string(),
@@ -98,7 +114,7 @@ impl ResultDatabase {
for result in results { for result in results {
string_rows.push(StringRow { string_rows.push(StringRow {
id: result.to_string(), id: result.to_string(),
values: vec![], ports: vec![],
}); });
} }
@@ -120,14 +136,8 @@ impl ResultDatabase {
pub fn save_rows(&self, string_rows: Vec<StringRow>) -> Result<(), Box<dyn std::error::Error>> { pub fn save_rows(&self, string_rows: Vec<StringRow>) -> Result<(), Box<dyn std::error::Error>> {
let db = Arc::new(DB::open_cf(&self.options, &self.path, &self.columns)?); let db = Arc::new(DB::open_cf(&self.options, &self.path, &self.columns)?);
let cf_default = db.cf_handle("default").unwrap(); let cf_default = db.cf_handle(&self.columns[0]).unwrap();
let cf_ports = db.cf_handle(&self.columns[1]).unwrap();
// Get handles to column index families
let mut cf_columns = Vec::new();
for i in 0..3 {
let cf = db.cf_handle(&format!("col{}_idx", i)).unwrap();
cf_columns.push(cf);
}
let start = Instant::now(); let start = Instant::now();
let length = string_rows.len(); let length = string_rows.len();
@@ -142,7 +152,6 @@ impl ResultDatabase {
let elapsed = { let elapsed = {
let db_ref = Arc::clone(&db); let db_ref = Arc::clone(&db);
let cf_default_ref = cf_default; let cf_default_ref = cf_default;
let cf_columns_ref = &cf_columns;
// Create batches in parallel but write them sequentially // Create batches in parallel but write them sequentially
let batches: Vec<WriteBatch> = chunks let batches: Vec<WriteBatch> = chunks
@@ -161,19 +170,9 @@ impl ResultDatabase {
// Store in main column family // Store in main column family
batch.put_cf(cf_default_ref, row.id.as_bytes(), &data); batch.put_cf(cf_default_ref, row.id.as_bytes(), &data);
// Create indexes only for searchable columns (0-2) let idx_key =
for (col_idx, value) in row.values.iter().enumerate() { format!("{}:{}", fast_escape(row.ports_to_string().as_str()), row.id);
if col_idx < cf_columns_ref.len() { batch.put_cf(cf_ports, idx_key.as_bytes(), row.id.as_bytes());
// Create search-friendly keys: value:rowid
// Use minimal escaping for better performance
let idx_key = format!("{}:{}", fast_escape(value), row.id);
batch.put_cf(
cf_columns_ref[col_idx],
idx_key.as_bytes(),
row.id.as_bytes(),
);
}
}
} }
batch batch
@@ -304,17 +303,19 @@ fn decode_row_binary(key: &str, data: &[u8]) -> Option<StringRow> {
Some(StringRow { Some(StringRow {
id: key.to_string(), id: key.to_string(),
values, ports: split_nums(values[0].as_str(), ","),
}) })
} }
// Binary encoding of row data for maximum performance // Binary encoding of row data for maximum performance
fn encode_row_binary(buf: &mut Vec<u8>, row: &StringRow) { fn encode_row_binary(buf: &mut Vec<u8>, row: &StringRow) {
let values = vec![row.ports_to_string()];
// Write number of values // Write number of values
buf.extend_from_slice(&(row.values.len() as u32).to_le_bytes()); buf.extend_from_slice(&(values.len() as u32).to_le_bytes());
// Write each value // Write each value
for value in &row.values { for value in vec![row.ports_to_string()] {
let value_bytes = value.as_bytes(); let value_bytes = value.as_bytes();
buf.extend_from_slice(&(value_bytes.len() as u32).to_le_bytes()); buf.extend_from_slice(&(value_bytes.len() as u32).to_le_bytes());
buf.extend_from_slice(value_bytes); buf.extend_from_slice(value_bytes);
+1
View File
@@ -164,6 +164,7 @@ fn scan(
]; ];
let tcp_results = tcp_scan::tcp_scan(up_hosts, tcp_ports, Duration::from_secs(3)); let tcp_results = tcp_scan::tcp_scan(up_hosts, tcp_ports, Duration::from_secs(3));
println!("Saving Data...");
let _ = database.add_tcp_results(&tcp_results); let _ = database.add_tcp_results(&tcp_results);
} }
} }
+1 -12
View File
@@ -24,18 +24,7 @@ impl PingResult {
pub fn to_string_row(&self) -> StringRow { pub fn to_string_row(&self) -> StringRow {
StringRow { StringRow {
id: self.host.to_string(), id: self.host.to_string(),
values: vec![ ports: vec![],
if self.is_up {
"up".to_string()
} else {
"down".to_string()
},
if self.response_time.is_some() {
self.response_time.unwrap().as_millis().to_string()
} else {
"None".to_string()
},
],
} }
} }
} }
+1 -12
View File
@@ -20,18 +20,7 @@ impl ScanResult {
pub fn to_string_row(&self) -> StringRow { pub fn to_string_row(&self) -> StringRow {
StringRow { StringRow {
id: self.ip.to_string(), id: self.ip.to_string(),
values: vec![join_nums(&self.open_ports, ",")], ports: (*self.open_ports).to_vec(),
} }
} }
} }
fn join_nums(nums: &Vec<i32>, sep: &str) -> String {
// 1. Convert numbers to strings
let str_nums: Vec<String> = nums
.iter()
.map(|n| n.to_string()) // map every integer to a string
.collect(); // collect the strings into the vector
// 2. Join the strings. There's already a function for this.
str_nums.join(sep)
}
+53 -10
View File
@@ -6,11 +6,12 @@ use std::thread;
use std::time::Duration; use std::time::Duration;
use indicatif::ProgressBar; use indicatif::ProgressBar;
use pnet::datalink::linux::interfaces;
use pnet::datalink::{self}; use pnet::datalink::{self};
use pnet::packet::ip::IpNextHeaderProtocols; use pnet::packet::ip::IpNextHeaderProtocols;
use pnet::packet::tcp::{MutableTcpPacket, TcpFlags, TcpPacket}; use pnet::packet::tcp::{MutableTcpPacket, TcpFlags, TcpPacket};
use pnet::packet::{Packet, tcp}; use pnet::packet::{tcp, Packet};
use pnet::transport::{self, TransportChannelType}; use pnet::transport::{self, TransportChannelType, TransportSender};
use rand::random_range; use rand::random_range;
use super::port_scan::ScanResult; use super::port_scan::ScanResult;
@@ -22,9 +23,20 @@ fn std_to_pnet_ipv4(previous: &IpAddr) -> Ipv4Addr {
// Main scanning function // Main scanning function
pub fn tcp_scan(targets: Vec<IpAddr>, ports: Vec<i32>, timeout: Duration) -> Vec<ScanResult> { pub fn tcp_scan(targets: Vec<IpAddr>, ports: Vec<i32>, timeout: Duration) -> Vec<ScanResult> {
// Find network interface // Find network interface
let interface = datalink::interfaces() let interfaces = datalink::interfaces();
println!("{}", interfaces.len());
let interface = interfaces
.into_iter() .into_iter()
.find(|iface| iface.is_up() && !iface.is_loopback() && !iface.ips.is_empty()) .find(|iface| {
iface.is_up()
&& !iface.is_loopback()
&& !iface.ips.is_empty()
&& !iface.is_dormant()
&& iface.is_running()
&& iface.is_point_to_point()
})
.expect("No valid network interface found"); .expect("No valid network interface found");
// Create transport channel for sending and receiving // Create transport channel for sending and receiving
@@ -52,13 +64,19 @@ pub fn tcp_scan(targets: Vec<IpAddr>, ports: Vec<i32>, timeout: Duration) -> Vec
let start_time = std::time::Instant::now(); let start_time = std::time::Instant::now();
while start_time.elapsed() < timeout { while start_time.elapsed() < timeout {
// println!("loop");
let mut iter = transport::tcp_packet_iter(&mut rx); let mut iter = transport::tcp_packet_iter(&mut rx);
match iter.next() { match iter.next_with_timeout(timeout) {
Ok((packet, addr)) => { Ok(Some((packet, addr))) => {
if let Some(tcp) = TcpPacket::new(packet.packet()) { if let Some(tcp) = TcpPacket::new(packet.packet()) {
// Check for SYN+ACK flags (indicating open port) // Check for SYN+ACK flags (indicating open port)
if tcp.get_flags() == TcpFlags::SYN | TcpFlags::ACK { if tcp.get_flags() == TcpFlags::SYN | TcpFlags::ACK {
println!(
"Discovered open port {} on {}",
tcp.get_source(),
addr.to_string()
);
let mut results_map = receiver_results.lock().unwrap(); let mut results_map = receiver_results.lock().unwrap();
if let Some(open_ports) = results_map.get_mut(&addr) { if let Some(open_ports) = results_map.get_mut(&addr) {
open_ports.push(tcp.get_source() as i32); open_ports.push(tcp.get_source() as i32);
@@ -66,6 +84,9 @@ pub fn tcp_scan(targets: Vec<IpAddr>, ports: Vec<i32>, timeout: Duration) -> Vec
} }
} }
} }
Ok(None) => {
break;
}
Err(_) => { Err(_) => {
// Just continue on errors // Just continue on errors
thread::sleep(Duration::from_millis(1)); thread::sleep(Duration::from_millis(1));
@@ -76,6 +97,8 @@ pub fn tcp_scan(targets: Vec<IpAddr>, ports: Vec<i32>, timeout: Duration) -> Vec
let pb = ProgressBar::new((targets.len() * ports.len()) as u64); let pb = ProgressBar::new((targets.len() * ports.len()) as u64);
println!("{:?}", interface.ips);
let source_ip = interface let source_ip = interface
.ips .ips
.iter() .iter()
@@ -83,6 +106,11 @@ pub fn tcp_scan(targets: Vec<IpAddr>, ports: Vec<i32>, timeout: Duration) -> Vec
.expect("No IPv4 address found") .expect("No IPv4 address found")
.ip(); .ip();
// let source_ip = IpAddr::V4(Ipv4Addr::new(127, 0, 0, 1));
// let source_ip = IpAddr::V4(Ipv4Addr::new(10, 0, 70, 4));
println!("Using IP: {}", source_ip.to_string());
for target in &targets { for target in &targets {
for port in &ports { for port in &ports {
// let source_ip = Ipv4Addr::from_bits(random_range(0..=(0xffffffff))); // let source_ip = Ipv4Addr::from_bits(random_range(0..=(0xffffffff)));
@@ -111,10 +139,7 @@ pub fn tcp_scan(targets: Vec<IpAddr>, ports: Vec<i32>, timeout: Duration) -> Vec
); );
tcp_header.set_checksum(checksum); tcp_header.set_checksum(checksum);
match tx.send_to(tcp_header, *target) { send_tcp_packet(&mut tx, tcp_header, target);
Ok(_) => {}
Err(e) => eprintln!("Failed to send packet: {}", e),
}
pb.inc(1); pb.inc(1);
thread::sleep(Duration::from_micros(100)); thread::sleep(Duration::from_micros(100));
@@ -122,6 +147,7 @@ pub fn tcp_scan(targets: Vec<IpAddr>, ports: Vec<i32>, timeout: Duration) -> Vec
} }
// Wait for receiver to finish // Wait for receiver to finish
// thread::sleep(timeout);
receiver_handle.join().unwrap(); receiver_handle.join().unwrap();
// Convert results to the return format // Convert results to the return format
@@ -131,6 +157,7 @@ pub fn tcp_scan(targets: Vec<IpAddr>, ports: Vec<i32>, timeout: Duration) -> Vec
.map(|ip| { .map(|ip| {
let mut open_ports = results_map.get(ip).cloned().unwrap_or_default(); let mut open_ports = results_map.get(ip).cloned().unwrap_or_default();
open_ports.sort(); open_ports.sort();
open_ports.dedup();
ScanResult { ScanResult {
ip: *ip, ip: *ip,
open_ports, open_ports,
@@ -138,3 +165,19 @@ pub fn tcp_scan(targets: Vec<IpAddr>, ports: Vec<i32>, timeout: Duration) -> Vec
}) })
.collect() .collect()
} }
fn send_tcp_packet(tx: &mut TransportSender, tcp_header: MutableTcpPacket<'_>, target: &IpAddr) {
match tx.send_to(&tcp_header, *target) {
Ok(_) => {}
Err(e) => {
if let Some(code) = e.raw_os_error() {
if code == 105 {
thread::sleep(Duration::from_millis(500));
send_tcp_packet(tx, tcp_header, target);
}
} else {
eprintln!("Failed to send packet: {}", e);
}
}
}
}