aria2-rust-pro/crates/aria2-rust-pro-protocol/src/torrent/metadata.rs

use sha1::{Digest, Sha1};

use super::{
    bencode::{BencodeValue, TorrentBencodeDict, dict_bytes, dict_int, parse_root_dict},
    model::{
        TorrentBootstrapModel, TorrentFileEntryModel, TorrentHashModel, TorrentInfoModel,
        TorrentMetadataModel, TorrentPieceModel, TorrentTrackerModel,
    },
    utils::{bytes_to_string, hex_encode, i64_to_u64, value_to_string},
};

/// Parses a `.torrent` payload into structured metadata.
///
/// # Errors
///
/// Returns an error when the bencoded payload is malformed or lacks the required info dictionary.
pub fn parse_torrent_metadata(input: &[u8]) -> Result<TorrentMetadataModel, String> {
    let (root, info_raw) = parse_root_dict(input)?;
    let info_map = match root.get("info") {
        Some(BencodeValue::Dict(map)) => map,
        Some(_) => return Err("torrent info must be a dictionary".to_owned()),
        None => return Err("missing torrent info dictionary".to_owned()),
    };
    let info_hash_hex = hex_encode(&Sha1::digest(
        info_raw.ok_or_else(|| "missing info bytes".to_owned())?,
    ));
    let info = parse_info_model(info_map, info_hash_hex);
    let announce = dict_bytes(&root, "announce").map(bytes_to_string);
    let creation_date = dict_bytes(&root, "creation date").map(bytes_to_string);
    let comment = dict_bytes(&root, "comment").map(bytes_to_string);

    Ok(TorrentMetadataModel {
        pieces: build_piece_models(&info),
        info,
        announce,
        trackers: parse_trackers(&root),
        peers: Vec::new(),
        dht_nodes: parse_dht_nodes(&root),
        creation_date,
        comment,
    })
}

/// Parses a `.torrent` payload and immediately shapes it into bootstrap-ready metadata.
///
/// # Errors
///
/// Returns an error when the `.torrent` payload is malformed or the derived bootstrap fields
/// cannot be shaped.
pub fn parse_torrent_bootstrap(input: &[u8]) -> Result<TorrentBootstrapModel, String> {
    parse_torrent_metadata(input)?.bootstrap()
}

/// Builds the higher-level torrent info model from the parsed `info` dictionary.
fn parse_info_model(info: &TorrentBencodeDict, info_hash_hex: String) -> TorrentInfoModel {
    let name = dict_bytes(info, "name")
        .map(bytes_to_string)
        .unwrap_or_default();
    let piece_length = i64_to_u64(dict_int(info, "piece length").unwrap_or_default());
    let pieces = dict_bytes(info, "pieces")
        .map(|bytes| {
            bytes
                .chunks_exact(20)
                .map(|chunk| {
                    let mut hash = [0_u8; 20];
                    hash.copy_from_slice(chunk);
                    hash
                })
                .collect()
        })
        .unwrap_or_default();
    let private = dict_int(info, "private").is_some_and(|value| value != 0);
    let hash = Some(TorrentHashModel {
        info_hash_hex,
        info_hash_base32: None,
    });
    let files = if let Some(BencodeValue::List(entries)) = info.get("files") {
        let mut offset = 0_u64;
        entries
            .iter()
            .filter_map(|entry| match entry {
                BencodeValue::Dict(file) => {
                    let length = i64_to_u64(dict_int(file, "length").unwrap_or_default());
                    let path = match file.get("path") {
                        Some(BencodeValue::List(parts)) => parts
                            .iter()
                            .map(value_to_string)
                            .collect::<Vec<_>>()
                            .join("/"),
                        _ => String::new(),
                    };
                    let item = TorrentFileEntryModel {
                        path,
                        length,
                        piece_offset: Some(offset),
                        selected: true,
                    };
                    offset = offset.saturating_add(length);
                    Some(item)
                }
                _ => None,
            })
            .collect()
    } else {
        vec![TorrentFileEntryModel {
            path: name.clone(),
            length: i64_to_u64(dict_int(info, "length").unwrap_or_default()),
            piece_offset: Some(0),
            selected: true,
        }]
    };

    TorrentInfoModel {
        name,
        piece_length,
        pieces,
        files,
        hash,
        private,
    }
}

/// Derives piece descriptors with offsets and effective lengths from torrent metadata.
fn build_piece_models(info: &TorrentInfoModel) -> Vec<TorrentPieceModel> {
    info.pieces
        .iter()
        .enumerate()
        .filter_map(|(index, hash)| {
            u32::try_from(index).ok().map(|index| TorrentPieceModel {
                index,
                hash: *hash,
                length: info.piece_length,
            })
        })
        .collect()
}

/// Parses the primary announce URL plus announce-list tiers into stable tracker entries.
fn parse_trackers(root: &TorrentBencodeDict) -> Vec<TorrentTrackerModel> {
    let mut trackers = Vec::new();
    if let Some(url) = dict_bytes(root, "announce").map(bytes_to_string) {
        trackers.push(TorrentTrackerModel {
            url,
            tier: Some(0),
            id: None,
            seeders: None,
            leechers: None,
        });
    }
    if let Some(BencodeValue::List(tiers)) = root.get("announce-list") {
        for (tier_index, tier) in tiers.iter().enumerate() {
            if let BencodeValue::List(urls) = tier {
                let tier = u32::try_from(tier_index)
                    .ok()
                    .and_then(|value| value.checked_add(1));
                for url in urls {
                    trackers.push(TorrentTrackerModel {
                        url: value_to_string(url),
                        tier,
                        id: None,
                        seeders: None,
                        leechers: None,
                    });
                }
            }
        }
    }
    trackers
}

/// Parses DHT bootstrap nodes from the optional `nodes` field.
fn parse_dht_nodes(root: &TorrentBencodeDict) -> Vec<String> {
    match root.get("nodes") {
        Some(BencodeValue::List(nodes)) => nodes
            .iter()
            .filter_map(|node| match node {
                BencodeValue::List(parts) => parts.first().zip(parts.get(1)).map(|(host, port)| {
                    format!("{}:{}", value_to_string(host), value_to_string(port))
                }),
                _ => None,
            })
            .collect(),
        _ => Vec::new(),
    }
}