build.rs

use std::collections::BTreeMap;
use std::fs::{self, File};

use std::env;
use std::io::{BufReader, Read};
use std::path::PathBuf;
use std::process::Command;

// Reads in a file with payloads based on port
pub fn main() {
    let mut file_path = env::current_dir().expect("cant find curr dir");
    file_path.push("./nmap-payloads");

    let mut data = String::new();
    let file = File::open(&file_path).expect("File not found.");
    let mut file_buf = BufReader::new(file);
    file_buf
        .read_to_string(&mut data)
        .expect("unable to read file");

    let mut fp_map: BTreeMap<i32, String> = BTreeMap::new();

    let mut count = 0;
    let mut capturing = false;
    let mut curr = String::new();

    for line in data.trim().split('\n') {
        // Strip a trailing carriage return so CRLF checkouts parse
        // identically to LF ones (otherwise `\r` pollutes port tokens and
        // blank lines stop looking blank).
        let line = line.strip_suffix('\r').unwrap_or(line);
        let line = strip_comment(line).trim_end();
        if line.trim_start().is_empty() {
            continue;
        }

        if line.starts_with("udp") {
            if !curr.is_empty() {
                fp_map.insert(count, curr);
                curr = String::new();
            }
            capturing = true;
            count += 1;
        }

        if capturing {
            if !curr.is_empty() {
                curr.push(' ');
            }
            curr.push_str(line);
        }
    }
    if !curr.is_empty() {
        fp_map.insert(count, curr);
    }

    let pb_linenr = ports_v(&fp_map);
    let payb_linenr = payloads_v(&fp_map);
    let map = port_payload_map(pb_linenr, payb_linenr);

    generate_code(map);
}

/// Removes a trailing `#` comment from a line of the payload file.
///
/// A `#` inside a quoted payload segment is payload data, not a comment.
///
/// # Arguments
///
/// * `line` - One line of the payload file
///
/// # Returns
///
/// The line up to, but not including, its first unquoted `#`
fn strip_comment(line: &str) -> &str {
    let mut quoted = false;
    let mut escaped = false;
    for (index, byte) in line.bytes().enumerate() {
        if escaped {
            escaped = false;
            continue;
        }
        match byte {
            b'\\' if quoted => escaped = true,
            b'"' => quoted = !quoted,
            b'#' if !quoted => return &line[..index],
            _ => {}
        }
    }
    line
}

/// Generates a file called Generated.rs and calls cargo fmt from the command line
///
/// # Arguments
///
/// * `port_payload_map` - A BTreeMap mapping port lists to their payload variants
fn generate_code(port_payload_map: BTreeMap<Vec<u16>, Vec<Vec<u8>>>) {
    let dest_path = PathBuf::from("src/generated.rs");

    let mut generated_code = String::new();
    generated_code.push_str("use std::collections::BTreeMap;\n");
    generated_code.push_str("use once_cell::sync::Lazy;\n\n");

    generated_code.push_str("fn generated_data() -> BTreeMap<Vec<u16>, Vec<Vec<u8>>> {\n");
    generated_code.push_str("    let mut map = BTreeMap::new();\n");

    for (ports, payloads) in port_payload_map {
        generated_code.push_str("    map.insert(vec![");
        generated_code.push_str(
            &ports
                .iter()
                .map(|&p| p.to_string())
                .collect::<Vec<_>>()
                .join(","),
        );
        generated_code.push_str("], vec![");
        for payload in payloads {
            generated_code.push_str("vec![");
            generated_code.push_str(
                &payload
                    .iter()
                    .map(|&p| p.to_string())
                    .collect::<Vec<_>>()
                    .join(","),
            );
            generated_code.push_str("],");
        }
        generated_code.push_str("]);\n");
    }

    generated_code.push_str("    map\n");
    generated_code.push_str("}\n\n");

    generated_code.push_str(
        "static PARSED_DATA: Lazy<BTreeMap<Vec<u16>, Vec<Vec<u8>>>> = Lazy::new(generated_data);\n",
    );
    generated_code
        .push_str("pub fn get_parsed_data() -> &'static BTreeMap<Vec<u16>, Vec<Vec<u8>>> {\n");
    generated_code.push_str("    &PARSED_DATA\n");
    generated_code.push_str("}\n");

    fs::write(dest_path, generated_code).unwrap();

    // format the generated code
    Command::new("cargo")
        .arg("fmt")
        .arg("--all")
        .output()
        .expect("Failed to execute cargo fmt");
}

/// Creates a BTreeMap of line numbers mapped to a Vec<u16> of ports
///
/// # Arguments
///
/// * `fp_map` - A BTreeMap containing the parsed file data
///
/// # Returns
///
/// A BTreeMap where keys are line numbers and values are vectors of ports
fn ports_v(fp_map: &BTreeMap<i32, String>) -> BTreeMap<i32, Vec<u16>> {
    let mut pb_linenr: BTreeMap<i32, Vec<u16>> = BTreeMap::new();
    let mut port_list: Vec<u16> = Vec::new();

    for (&line_nr, ports) in fp_map {
        if ports.contains("udp ") {
            let remain = &ports[4..];
            let mut start = remain.split(' ');

            let ports = start.next().unwrap();
            let port_segments: Vec<&str> = ports.split(',').collect();

            for segment in port_segments {
                if segment.contains('-') {
                    let range: Vec<&str> = segment.trim().split('-').collect();
                    let start = range[0].parse::<u16>().unwrap();
                    let end = range[1].parse::<u16>().unwrap();
                    assert!(start <= end, "reversed port range: {}", segment);

                    port_list.extend(start..=end);
                } else if !segment.is_empty() {
                    match segment.parse::<u16>() {
                        Ok(port) => port_list.push(port),
                        Err(_) => panic!("invalid port: {}", segment),
                    }
                }
            }
            port_list.sort_unstable();
            port_list.dedup();
        }

        pb_linenr.insert(line_nr, port_list.clone());
        port_list.clear();
    }

    pb_linenr
}

/// Parses out the Payloads into a BTreeMap of line numbers mapped to vectors of payload bytes
///
/// # Arguments
///
/// * `fp_map` - A BTreeMap containing the parsed file data
///
/// # Returns
///
/// A BTreeMap where keys are line numbers and values are vectors of payload bytes
fn payloads_v(fp_map: &BTreeMap<i32, String>) -> BTreeMap<i32, Vec<u8>> {
    let mut payb_linenr: BTreeMap<i32, Vec<u8>> = BTreeMap::new();

    for (&line_nr, data) in fp_map {
        if data.contains('"') {
            let start = data.find('"').expect("payload opening \" not found");
            // Pass from the opening quote: the decoder pairs quotes itself
            // to split multi-line segments.
            payb_linenr.insert(line_nr, parser(data[start..].trim()));
        }
    }

    payb_linenr
}

/// Converts a quoted Nmap payload block to bytes.
///
/// # Arguments
///
/// * `payload` - The raw block text starting at its first opening quote:
///   one or more `"..."` segments separated by whitespace
///
/// # Returns
///
/// The decoded probe bytes: segments are decoded with [`decode_segment`]
/// and concatenated with no separators, so multi-line probes reassemble
/// exactly.
fn parser(payload: &str) -> Vec<u8> {
    let mut bytes: Vec<u8> = Vec::new();
    let mut rest = payload.trim();
    while let Some(open) = rest.find('"') {
        let after_open = &rest[open + 1..];
        match split_segment(after_open) {
            Some((segment, remainder)) => {
                decode_segment(segment, &mut bytes);
                rest = remainder;
            }
            None => {
                // Unterminated trailing quote: decode what's left literally.
                decode_segment(after_open, &mut bytes);
                break;
            }
        }
    }

    bytes
}

/// Splits off the first `"..."` segment: `text` must start just after an
/// opening quote. Escaped quotes (`\"`) do not terminate the segment.
///
/// Returns the segment body and the remainder after the closing quote.
fn split_segment(text: &str) -> Option<(&str, &str)> {
    let mut escaped = false;
    for (i, c) in text.char_indices() {
        if escaped {
            escaped = false;
            continue;
        }
        if c == '\\' {
            escaped = true;
            continue;
        }
        if c == '"' {
            return Some((&text[..i], &text[i + 1..]));
        }
    }
    None
}

/// Decodes one quoted Nmap payload segment in C-style escape format:
/// `\xNN` becomes a byte, common single-char escapes (`\\`, `\"`,
/// `\0`, `\n`, `\r`, `\t`) decode to their value, and every other
/// character contributes its literal bytes.
///
/// Preserving literal text matters: several probes embed plain-text
/// protocol words (the SNMP `public` community string, NetBIOS names,
/// LDAP `objectClass`, SSDP headers). The previous hexdigits-only
/// decoding mangled them into wrong bytes, so agents never answered.
fn decode_segment(segment: &str, bytes: &mut Vec<u8>) {
    let mut chars = segment.chars();
    while let Some(c) = chars.next() {
        if c != '\\' {
            let mut buf = [0; 4];
            bytes.extend_from_slice(c.encode_utf8(&mut buf).as_bytes());
            continue;
        }
        match chars.next() {
            Some('x') => {
                let hi = chars.next().unwrap_or('0');
                let lo = chars.next().unwrap_or('0');
                let hex: String = [hi, lo].iter().collect();
                bytes.push(u8::from_str_radix(&hex, 16).unwrap_or(0));
            }
            Some('0') => bytes.push(0),
            Some('n') => bytes.push(b'\n'),
            Some('r') => bytes.push(b'\r'),
            Some('t') => bytes.push(b'\t'),
            Some('\\') => bytes.push(b'\\'),
            Some('"') => bytes.push(b'"'),
            // Unknown escape: keep the character literally, matching Nmap,
            // which passes unrecognized escapes through untouched.
            Some(other) => {
                let mut buf = [0; 4];
                bytes.extend_from_slice(other.encode_utf8(&mut buf).as_bytes());
            }
            None => bytes.push(b'\\'),
        }
    }
}

/// Combines the ports BTreeMap and the Payloads BTreeMap
///
/// # Arguments
///
/// * `pb_linenr` - A BTreeMap mapping line numbers to vectors of ports
/// * `payb_linenr` - A BTreeMap mapping line numbers to vectors of payload bytes
///
/// # Returns
///
/// A BTreeMap mapping vectors of ports to their payload variants, in file
/// order. Entries that share a port list keep every distinct payload.
fn port_payload_map(
    pb_linenr: BTreeMap<i32, Vec<u16>>,
    payb_linenr: BTreeMap<i32, Vec<u8>>,
) -> BTreeMap<Vec<u16>, Vec<Vec<u8>>> {
    let mut ppm_fin: BTreeMap<Vec<u16>, Vec<Vec<u8>>> = BTreeMap::new();

    for (port_linenr, ports) in pb_linenr {
        for (pay_linenr, payloads) in &payb_linenr {
            if pay_linenr == &port_linenr {
                let variants = ppm_fin.entry(ports.to_vec()).or_default();
                if !variants.contains(payloads) {
                    variants.push(payloads.to_vec());
                }
            }
        }
    }

    ppm_fin
}