torrust_index/utils/

parse_torrent.rs

use std::error;

use derive_more::{Display, Error};
use serde::{Deserialize, Serialize};
use serde_bencode::value::Value;
use serde_bencode::{de, Error};
use sha1::{Digest, Sha1};

use crate::models::info_hash::InfoHash;
use crate::models::torrent_file::Torrent;

#[derive(Debug, Display, PartialEq, Eq, Error)]
pub enum DecodeTorrentFileError {
    #[display(fmt = "Torrent data could not be decoded from the bencoded format.")]
    InvalidBencodeData,

    #[display(fmt = "Torrent info dictionary key could not be decoded from the bencoded format.")]
    InvalidInfoDictionary,

    #[display(fmt = "Torrent has an invalid pieces key length. It should be a multiple of 20.")]
    InvalidTorrentPiecesLength,

    #[display(fmt = "Cannot bencode the parsed `info` dictionary again to generate the info-hash.")]
    CannotBencodeInfoDict,
}

/// It decodes and validate an array of bytes containing a torrent file.
///
/// It returns a tuple containing the decoded torrent and the original info hash.
/// The original info-hash migth not match the new one in the `Torrent` because
/// the info dictionary might have been modified. For example, ignoring some
/// non-standard fields.
///
/// # Errors
///
/// This function will return an error if
///
/// - The torrent file is not a valid bencoded file.
/// - The pieces key has a length that is not a multiple of 20.
pub fn decode_and_validate_torrent_file(bytes: &[u8]) -> Result<(Torrent, InfoHash), DecodeTorrentFileError> {
    let original_info_hash = calculate_info_hash(bytes)?;

    let torrent = decode_torrent(bytes).map_err(|_| DecodeTorrentFileError::InvalidBencodeData)?;

    // Make sure that the pieces key has a length that is a multiple of 20
    if let Some(pieces) = torrent.info.pieces.as_ref() {
        if pieces.as_ref().len() % 20 != 0 {
            return Err(DecodeTorrentFileError::InvalidTorrentPiecesLength);
        }
    }

    Ok((torrent, original_info_hash))
}

/// Decode a Torrent from Bencoded Bytes.
///
/// # Errors
///
/// This function will return an error if unable to parse bytes into torrent.
pub fn decode_torrent(bytes: &[u8]) -> Result<Torrent, Box<dyn error::Error>> {
    match de::from_bytes::<Torrent>(bytes) {
        Ok(torrent) => Ok(torrent),
        Err(e) => {
            println!("{e:?}");
            Err(e.into())
        }
    }
}

/// Encode a Torrent into Bencoded Bytes.
///
/// # Errors
///
/// This function will return an error if unable to bencode torrent.
pub fn encode_torrent(torrent: &Torrent) -> Result<Vec<u8>, Error> {
    match serde_bencode::to_bytes(torrent) {
        Ok(bencode_bytes) => Ok(bencode_bytes),
        Err(e) => {
            eprintln!("{e:?}");
            Err(e)
        }
    }
}

#[derive(Serialize, Deserialize, Debug, PartialEq)]
struct ParsedInfoDictFromMetainfoFile {
    pub info: Value,
}

/// Calculates the `InfoHash` from a the torrent file binary data.
///
/// # Errors
///
/// This function will return an error if:
///
/// - The torrent file is not a valid bencoded torrent file containing an `info`
///   dictionary key.
/// - The original torrent info-hash cannot be bencoded from the parsed `info`
///   dictionary is not a valid bencoded dictionary.
pub fn calculate_info_hash(bytes: &[u8]) -> Result<InfoHash, DecodeTorrentFileError> {
    // Extract the info dictionary
    let metainfo: ParsedInfoDictFromMetainfoFile =
        serde_bencode::from_bytes(bytes).map_err(|_| DecodeTorrentFileError::InvalidInfoDictionary)?;

    // Bencode the info dictionary
    let info_dict_bytes = serde_bencode::to_bytes(&metainfo.info).map_err(|_| DecodeTorrentFileError::CannotBencodeInfoDict)?;

    // Calculate the SHA-1 hash of the bencoded info dictionary
    let mut hasher = Sha1::new();
    hasher.update(&info_dict_bytes);
    let result = hasher.finalize();

    Ok(InfoHash::from_bytes(&result))
}

#[cfg(test)]
mod tests {
    use std::str::FromStr;

    use crate::models::info_hash::InfoHash;

    /// Returns a torrent file binary contents for a torrent with a custom key
    /// inside the `info` dictionary. A custom key means a key not included in
    /// official BEPs that we don't parse and store in the database.
    fn torrent_with_custom_info_dict_key() -> Vec<u8> {
        /* code-review:

        This is the contents of the torrent file in fixtures dir. After adding
        some tests using the following:

        `figment::Jail::expect_with(|jail| { ... });``

        some tests using relative paths for fixtures failed. It seems
        Figment::Jail changes the current dir. In the drop function it restores
        it but that could cause a problem when test are running in parallel.

        For the time being, I have replaced loading the torrent file with
        this function. This should make the test faster.

        I don't know why this happens only with these two tests.

        */

        // cspell:disable-next-line
        // "tests/fixtures/torrents/6c690018c5786dbbb00161f62b0712d69296df97_with_custom_info_dict_key.torrent"

        [
            100, 56, 58, 97, 110, 110, 111, 117, 110, 99, 101, 52, 49, 58, 104, 116, 116, 112, 115, 58, 47, 47, 97, 99, 97, 100,
            101, 109, 105, 99, 116, 111, 114, 114, 101, 110, 116, 115, 46, 99, 111, 109, 47, 97, 110, 110, 111, 117, 110, 99,
            101, 46, 112, 104, 112, 49, 51, 58, 97, 110, 110, 111, 117, 110, 99, 101, 45, 108, 105, 115, 116, 108, 108, 52, 49,
            58, 104, 116, 116, 112, 115, 58, 47, 47, 97, 99, 97, 100, 101, 109, 105, 99, 116, 111, 114, 114, 101, 110, 116, 115,
            46, 99, 111, 109, 47, 97, 110, 110, 111, 117, 110, 99, 101, 46, 112, 104, 112, 101, 108, 52, 54, 58, 104, 116, 116,
            112, 115, 58, 47, 47, 105, 112, 118, 54, 46, 97, 99, 97, 100, 101, 109, 105, 99, 116, 111, 114, 114, 101, 110, 116,
            115, 46, 99, 111, 109, 47, 97, 110, 110, 111, 117, 110, 99, 101, 46, 112, 104, 112, 101, 108, 52, 50, 58, 117, 100,
            112, 58, 47, 47, 116, 114, 97, 99, 107, 101, 114, 46, 111, 112, 101, 110, 116, 114, 97, 99, 107, 114, 46, 111, 114,
            103, 58, 49, 51, 51, 55, 47, 97, 110, 110, 111, 117, 110, 99, 101, 101, 108, 52, 52, 58, 117, 100, 112, 58, 47, 47,
            116, 114, 97, 99, 107, 101, 114, 46, 111, 112, 101, 110, 98, 105, 116, 116, 111, 114, 114, 101, 110, 116, 46, 99,
            111, 109, 58, 56, 48, 47, 97, 110, 110, 111, 117, 110, 99, 101, 101, 108, 51, 54, 58, 104, 116, 116, 112, 58, 47, 47,
            98, 116, 49, 46, 97, 114, 99, 104, 105, 118, 101, 46, 111, 114, 103, 58, 54, 57, 54, 57, 47, 97, 110, 110, 111, 117,
            110, 99, 101, 101, 108, 51, 54, 58, 104, 116, 116, 112, 58, 47, 47, 98, 116, 50, 46, 97, 114, 99, 104, 105, 118, 101,
            46, 111, 114, 103, 58, 54, 57, 54, 57, 47, 97, 110, 110, 111, 117, 110, 99, 101, 101, 101, 55, 58, 99, 111, 109, 109,
            101, 110, 116, 54, 51, 52, 58, 84, 104, 105, 115, 32, 99, 111, 110, 116, 101, 110, 116, 32, 104, 111, 115, 116, 101,
            100, 32, 97, 116, 32, 116, 104, 101, 32, 73, 110, 116, 101, 114, 110, 101, 116, 32, 65, 114, 99, 104, 105, 118, 101,
            32, 97, 116, 32, 104, 116, 116, 112, 115, 58, 47, 47, 97, 114, 99, 104, 105, 118, 101, 46, 111, 114, 103, 47, 100,
            101, 116, 97, 105, 108, 115, 47, 114, 97, 112, 112, 112, 105, 100, 45, 119, 101, 105, 103, 104, 116, 115, 46, 116,
            97, 114, 10, 70, 105, 108, 101, 115, 32, 109, 97, 121, 32, 104, 97, 118, 101, 32, 99, 104, 97, 110, 103, 101, 100,
            44, 32, 119, 104, 105, 99, 104, 32, 112, 114, 101, 118, 101, 110, 116, 115, 32, 116, 111, 114, 114, 101, 110, 116,
            115, 32, 102, 114, 111, 109, 32, 100, 111, 119, 110, 108, 111, 97, 100, 105, 110, 103, 32, 99, 111, 114, 114, 101,
            99, 116, 108, 121, 32, 111, 114, 32, 99, 111, 109, 112, 108, 101, 116, 101, 108, 121, 59, 32, 112, 108, 101, 97, 115,
            101, 32, 99, 104, 101, 99, 107, 32, 102, 111, 114, 32, 97, 110, 32, 117, 112, 100, 97, 116, 101, 100, 32, 116, 111,
            114, 114, 101, 110, 116, 32, 97, 116, 32, 104, 116, 116, 112, 115, 58, 47, 47, 97, 114, 99, 104, 105, 118, 101, 46,
            111, 114, 103, 47, 100, 111, 119, 110, 108, 111, 97, 100, 47, 114, 97, 112, 112, 112, 105, 100, 45, 119, 101, 105,
            103, 104, 116, 115, 46, 116, 97, 114, 47, 114, 97, 112, 112, 112, 105, 100, 45, 119, 101, 105, 103, 104, 116, 115,
            46, 116, 97, 114, 95, 97, 114, 99, 104, 105, 118, 101, 46, 116, 111, 114, 114, 101, 110, 116, 10, 78, 111, 116, 101,
            58, 32, 114, 101, 116, 114, 105, 101, 118, 97, 108, 32, 117, 115, 117, 97, 108, 108, 121, 32, 114, 101, 113, 117,
            105, 114, 101, 115, 32, 97, 32, 99, 108, 105, 101, 110, 116, 32, 116, 104, 97, 116, 32, 115, 117, 112, 112, 111, 114,
            116, 115, 32, 119, 101, 98, 115, 101, 101, 100, 105, 110, 103, 32, 40, 71, 101, 116, 82, 105, 103, 104, 116, 32, 115,
            116, 121, 108, 101, 41, 46, 10, 78, 111, 116, 101, 58, 32, 109, 97, 110, 121, 32, 73, 110, 116, 101, 114, 110, 101,
            116, 32, 65, 114, 99, 104, 105, 118, 101, 32, 116, 111, 114, 114, 101, 110, 116, 115, 32, 99, 111, 110, 116, 97, 105,
            110, 32, 97, 32, 39, 112, 97, 100, 32, 102, 105, 108, 101, 39, 32, 100, 105, 114, 101, 99, 116, 111, 114, 121, 46,
            32, 84, 104, 105, 115, 32, 100, 105, 114, 101, 99, 116, 111, 114, 121, 32, 97, 110, 100, 32, 116, 104, 101, 32, 102,
            105, 108, 101, 115, 32, 119, 105, 116, 104, 105, 110, 32, 105, 116, 32, 109, 97, 121, 32, 98, 101, 32, 101, 114, 97,
            115, 101, 100, 32, 111, 110, 99, 101, 32, 114, 101, 116, 114, 105, 101, 118, 97, 108, 32, 99, 111, 109, 112, 108,
            101, 116, 101, 115, 46, 10, 78, 111, 116, 101, 58, 32, 116, 104, 101, 32, 102, 105, 108, 101, 32, 114, 97, 112, 112,
            112, 105, 100, 45, 119, 101, 105, 103, 104, 116, 115, 46, 116, 97, 114, 95, 109, 101, 116, 97, 46, 120, 109, 108, 32,
            99, 111, 110, 116, 97, 105, 110, 115, 32, 109, 101, 116, 97, 100, 97, 116, 97, 32, 97, 98, 111, 117, 116, 32, 116,
            104, 105, 115, 32, 116, 111, 114, 114, 101, 110, 116, 39, 115, 32, 99, 111, 110, 116, 101, 110, 116, 115, 46, 49, 48,
            58, 99, 114, 101, 97, 116, 101, 100, 32, 98, 121, 49, 53, 58, 105, 97, 95, 109, 97, 107, 101, 95, 116, 111, 114, 114,
            101, 110, 116, 49, 51, 58, 99, 114, 101, 97, 116, 105, 111, 110, 32, 100, 97, 116, 101, 105, 49, 54, 56, 57, 50, 55,
            51, 55, 56, 55, 101, 52, 58, 105, 110, 102, 111, 100, 49, 49, 58, 99, 111, 108, 108, 101, 99, 116, 105, 111, 110,
            115, 108, 51, 49, 58, 111, 114, 103, 46, 97, 114, 99, 104, 105, 118, 101, 46, 114, 97, 112, 112, 112, 105, 100, 45,
            119, 101, 105, 103, 104, 116, 115, 46, 116, 97, 114, 101, 53, 58, 102, 105, 108, 101, 115, 108, 100, 53, 58, 99, 114,
            99, 51, 50, 56, 58, 53, 55, 100, 51, 51, 102, 99, 99, 54, 58, 108, 101, 110, 103, 116, 104, 105, 49, 49, 53, 50, 56,
            51, 50, 52, 101, 51, 58, 109, 100, 53, 51, 50, 58, 101, 57, 49, 98, 98, 52, 98, 97, 56, 50, 54, 57, 53, 49, 54, 49,
            98, 101, 54, 56, 102, 56, 98, 51, 51, 97, 101, 55, 54, 49, 52, 50, 53, 58, 109, 116, 105, 109, 101, 49, 48, 58, 49,
            54, 56, 57, 50, 55, 51, 55, 51, 48, 52, 58, 112, 97, 116, 104, 108, 50, 50, 58, 82, 65, 80, 80, 80, 73, 68, 32, 87,
            101, 105, 103, 104, 116, 115, 46, 116, 97, 114, 46, 103, 122, 101, 52, 58, 115, 104, 97, 49, 52, 48, 58, 52, 53, 57,
            55, 48, 101, 102, 51, 51, 99, 98, 51, 48, 52, 57, 97, 55, 97, 56, 54, 50, 57, 101, 52, 48, 99, 56, 102, 53, 101, 53,
            50, 54, 56, 100, 49, 100, 99, 53, 51, 101, 100, 53, 58, 99, 114, 99, 51, 50, 56, 58, 99, 54, 53, 56, 102, 100, 52,
            102, 54, 58, 108, 101, 110, 103, 116, 104, 105, 50, 48, 52, 56, 48, 101, 51, 58, 109, 100, 53, 51, 50, 58, 97, 55,
            56, 50, 98, 50, 97, 53, 51, 98, 97, 52, 57, 102, 48, 100, 52, 53, 102, 51, 100, 100, 54, 101, 51, 53, 101, 48, 100,
            53, 57, 51, 53, 58, 109, 116, 105, 109, 101, 49, 48, 58, 49, 54, 56, 57, 50, 55, 51, 55, 56, 51, 52, 58, 112, 97,
            116, 104, 108, 51, 49, 58, 114, 97, 112, 112, 112, 105, 100, 45, 119, 101, 105, 103, 104, 116, 115, 46, 116, 97, 114,
            95, 109, 101, 116, 97, 46, 115, 113, 108, 105, 116, 101, 101, 52, 58, 115, 104, 97, 49, 52, 48, 58, 98, 99, 98, 48,
            54, 98, 51, 49, 54, 52, 102, 49, 100, 50, 97, 98, 97, 50, 50, 101, 102, 54, 48, 52, 54, 101, 98, 56, 48, 102, 54, 53,
            50, 54, 52, 101, 57, 102, 98, 97, 101, 100, 53, 58, 99, 114, 99, 51, 50, 56, 58, 56, 49, 52, 48, 97, 53, 99, 55, 54,
            58, 108, 101, 110, 103, 116, 104, 105, 49, 48, 52, 52, 101, 51, 58, 109, 100, 53, 51, 50, 58, 49, 98, 97, 98, 50, 49,
            101, 53, 48, 101, 48, 54, 97, 98, 52, 50, 100, 51, 97, 55, 55, 100, 56, 55, 50, 98, 102, 50, 53, 50, 101, 53, 53, 58,
            109, 116, 105, 109, 101, 49, 48, 58, 49, 54, 56, 57, 50, 55, 51, 55, 54, 51, 52, 58, 112, 97, 116, 104, 108, 50, 56,
            58, 114, 97, 112, 112, 112, 105, 100, 45, 119, 101, 105, 103, 104, 116, 115, 46, 116, 97, 114, 95, 109, 101, 116, 97,
            46, 120, 109, 108, 101, 52, 58, 115, 104, 97, 49, 52, 48, 58, 98, 50, 102, 48, 102, 50, 98, 98, 101, 99, 51, 52, 97,
            97, 57, 49, 52, 48, 102, 98, 57, 97, 99, 51, 102, 99, 98, 49, 57, 48, 53, 56, 56, 97, 52, 57, 54, 97, 97, 51, 101,
            101, 52, 58, 110, 97, 109, 101, 49, 57, 58, 114, 97, 112, 112, 112, 105, 100, 45, 119, 101, 105, 103, 104, 116, 115,
            46, 116, 97, 114, 49, 50, 58, 112, 105, 101, 99, 101, 32, 108, 101, 110, 103, 116, 104, 105, 53, 50, 52, 50, 56, 56,
            101, 54, 58, 112, 105, 101, 99, 101, 115, 52, 54, 48, 58, 171, 236, 85, 110, 15, 123, 231, 211, 48, 12, 246, 104,
            140, 144, 109, 153, 12, 62, 50, 181, 44, 242, 182, 124, 12, 50, 82, 188, 114, 111, 7, 30, 115, 171, 118, 241, 188,
            50, 43, 252, 33, 212, 127, 26, 233, 114, 53, 64, 126, 195, 180, 137, 9, 43, 237, 75, 216, 176, 108, 101, 140, 39, 88,
            174, 251, 114, 117, 115, 68, 55, 136, 40, 32, 210, 148, 189, 164, 106, 248, 210, 166, 253, 2, 101, 28, 28, 223, 184,
            86, 109, 58, 210, 126, 167, 61, 202, 226, 73, 247, 54, 141, 23, 119, 110, 50, 173, 239, 165, 68, 194, 143, 182, 156,
            36, 86, 173, 232, 251, 123, 166, 113, 192, 129, 229, 67, 3, 145, 212, 79, 176, 166, 100, 202, 41, 27, 13, 29, 64,
            125, 57, 78, 118, 150, 235, 1, 24, 243, 245, 80, 142, 47, 250, 84, 252, 73, 102, 133, 216, 56, 135, 120, 155, 10,
            143, 122, 163, 44, 143, 114, 54, 173, 109, 116, 11, 252, 197, 87, 113, 134, 251, 243, 207, 202, 201, 218, 236, 97,
            98, 162, 42, 27, 167, 133, 137, 145, 143, 170, 192, 192, 203, 13, 87, 216, 183, 231, 100, 77, 242, 132, 115, 118,
            152, 251, 58, 23, 72, 215, 156, 1, 254, 202, 109, 31, 197, 151, 52, 5, 84, 57, 218, 194, 110, 23, 65, 17, 105, 243,
            70, 209, 125, 22, 211, 192, 135, 59, 195, 178, 12, 29, 224, 226, 73, 195, 5, 210, 76, 0, 90, 91, 120, 1, 18, 62, 191,
            82, 67, 73, 109, 26, 238, 35, 121, 210, 14, 40, 182, 132, 126, 197, 237, 121, 222, 100, 2, 237, 71, 113, 61, 147, 22,
            196, 162, 118, 24, 119, 84, 197, 49, 72, 150, 58, 81, 193, 74, 146, 144, 145, 243, 207, 72, 91, 36, 134, 85, 168,
            235, 12, 198, 45, 134, 226, 41, 86, 9, 44, 56, 11, 205, 193, 202, 69, 230, 100, 106, 71, 254, 187, 46, 71, 154, 119,
            69, 41, 233, 114, 25, 32, 111, 66, 121, 43, 55, 185, 83, 37, 237, 15, 41, 4, 213, 226, 150, 241, 222, 207, 153, 190,
            50, 170, 184, 10, 29, 11, 159, 185, 214, 171, 92, 80, 67, 120, 133, 65, 9, 1, 36, 207, 224, 137, 118, 91, 77, 169,
            202, 114, 192, 223, 146, 71, 15, 13, 206, 202, 150, 198, 126, 165, 65, 95, 43, 167, 187, 4, 204, 247, 68, 127, 148,
            30, 36, 210, 27, 23, 202, 24, 121, 144, 163, 214, 32, 117, 162, 150, 104, 6, 88, 90, 222, 245, 44, 26, 144, 34, 114,
            51, 142, 213, 178, 168, 250, 229, 233, 231, 105, 98, 2, 124, 9, 179, 76, 101, 54, 58, 108, 111, 99, 97, 108, 101, 50,
            58, 101, 110, 53, 58, 116, 105, 116, 108, 101, 49, 57, 58, 114, 97, 112, 112, 112, 105, 100, 45, 119, 101, 105, 103,
            104, 116, 115, 46, 116, 97, 114, 56, 58, 117, 114, 108, 45, 108, 105, 115, 116, 108, 50, 57, 58, 104, 116, 116, 112,
            115, 58, 47, 47, 97, 114, 99, 104, 105, 118, 101, 46, 111, 114, 103, 47, 100, 111, 119, 110, 108, 111, 97, 100, 47,
            52, 48, 58, 104, 116, 116, 112, 58, 47, 47, 105, 97, 57, 48, 50, 55, 48, 50, 46, 117, 115, 46, 97, 114, 99, 104, 105,
            118, 101, 46, 111, 114, 103, 47, 50, 50, 47, 105, 116, 101, 109, 115, 47, 52, 48, 58, 104, 116, 116, 112, 58, 47, 47,
            105, 97, 56, 48, 50, 55, 48, 50, 46, 117, 115, 46, 97, 114, 99, 104, 105, 118, 101, 46, 111, 114, 103, 47, 50, 50,
            47, 105, 116, 101, 109, 115, 47, 101, 101,
        ]
        .to_vec()
    }

    #[test]
    fn it_should_calculate_the_original_info_hash_using_all_fields_in_the_info_key_dictionary() {
        let original_info_hash = super::calculate_info_hash(&torrent_with_custom_info_dict_key()).unwrap();

        assert_eq!(
            original_info_hash,
            InfoHash::from_str("6c690018c5786dbbb00161f62b0712d69296df97").unwrap() // DevSkim: ignore DS173237
        );
    }

    #[test]
    fn it_should_calculate_the_new_info_hash_ignoring_non_standard_fields_in_the_info_key_dictionary() {
        let torrent = super::decode_torrent(&torrent_with_custom_info_dict_key()).unwrap();

        // The infohash is not the original infohash of the torrent file,
        // but the infohash of the info dictionary without the custom keys.
        assert_eq!(
            torrent.canonical_info_hash_hex(),
            "8aa01a4c816332045ffec83247ccbc654547fedf".to_string() // DevSkim: ignore DS173237
        );
    }
}