troll-patrol/src/extra_info.rs

/*! Fields we need from the extra-info documents for bridges...
Note, this is NOT a complete implementation of the document format.
(https://spec.torproject.org/dir-spec/extra-info-document-format.html) */

use chrono::DateTime;
use julianday::JulianDay;
use serde::{Deserialize, Serialize};
use std::collections::{BTreeMap, HashMap, HashSet};

/// Fields we need from extra-info document
#[derive(Eq, PartialEq, Hash, Serialize, Deserialize)]
pub struct ExtraInfo {
    /// Bridge nickname, probably unused
    pub nickname: String,

    /// Bridge fingerprint, a SHA-1 hash of the bridge ID
    pub fingerprint: [u8; 20],

    /// Date (in UTC) that this document covered (bridge-stats-end if
    /// available) or that the document was published (published), stored
    /// as a Julian date because we don't need to know more precisely than
    /// the day.
    pub date: u32,

    /// Map of country codes and how many users (rounded up to a multiple of
    /// 8) have connected to that bridge during the day.
    /// Uses BTreeMap instead of HashMap so ExtraInfo can implement Hash.
    pub bridge_ips: BTreeMap<String, u32>,
}

impl ExtraInfo {
    /// Converts a map of keys and values into an ExtraInfo if all necessary fields
    /// are represented.
    fn from_map(entry: &HashMap<String, String>) -> Result<Self, String> {
        if !entry.contains_key("nickname") || !entry.contains_key("fingerprint") {
            // How did we get here??
            return Err("Cannot parse extra-info: Missing nickname or fingerprint".to_string());
        }
        if !(entry.contains_key("bridge-stats-end") || entry.contains_key("published"))
            || !entry.contains_key("bridge-ips")
        {
            // Some extra-infos are missing data on connecting IPs...
            // But we can't do anything in that case.
            return Err(format!(
                "Failed to parse extra-info for {} {}",
                entry.get("nickname").unwrap(),
                entry.get("fingerprint").unwrap()
            ));
        }
        let nickname = entry.get("nickname").unwrap().to_string();
        let fingerprint_str = entry.get("fingerprint").unwrap();
        if fingerprint_str.len() != 40 {
            return Err("Fingerprint must be 20 bytes".to_string());
        }
        let fingerprint = array_bytes::hex2array(fingerprint_str).unwrap();
        let date: u32 = {
            let date_str = if entry.contains_key("bridge-stats-end") {
                let line = entry.get("bridge-stats-end").unwrap();
                // Parse out (86400 s) from end of line
                &line[..line.find("(").unwrap() - 1]
            } else {
                entry.get("published").unwrap().as_str()
            };
            JulianDay::from(
                DateTime::parse_from_str(&(date_str.to_owned() + " +0000"), "%F %T %z")
                    .unwrap()
                    .date_naive(),
            )
            .inner()
            .try_into()
            .unwrap()
        };
        let bridge_ips_str = entry.get("bridge-ips").unwrap();
        let mut bridge_ips: BTreeMap<String, u32> = BTreeMap::new();
        let countries: Vec<&str> = bridge_ips_str.split(',').collect();
        for country in countries {
            if country != "" {
                // bridge-ips may be empty
                let (cc, count) = country.split_once('=').unwrap();
                bridge_ips.insert(cc.to_string(), count.parse::<u32>().unwrap());
            }
        }

        Ok(Self {
            nickname,
            fingerprint,
            date,
            bridge_ips,
        })
    }

    /// Accepts a downloaded extra-infos file as a big string, returns a set of
    /// the ExtraInfos represented by the file.
    pub fn parse_file<'a>(extra_info_str: &str) -> HashSet<Self> {
        let mut set = HashSet::<Self>::new();
        let mut entry = HashMap::<String, String>::new();
        for line in extra_info_str.lines() {
            let line = line;
            if line.starts_with("@type bridge-extra-info ") {
                if !entry.is_empty() {
                    let extra_info = Self::from_map(&entry);
                    if extra_info.is_ok() {
                        set.insert(extra_info.unwrap());
                    } else {
                        // Just print the error and continue.
                        println!("{}", extra_info.err().unwrap());
                    }
                    entry = HashMap::<String, String>::new();
                }
            } else {
                if line.starts_with("extra-info ") {
                    // extra-info line has format:
                    // extra-info <nickname> <fingerprint>
                    let line_split: Vec<&str> = line.split(' ').collect();
                    if line_split.len() != 3 {
                        println!("Misformed extra-info line");
                    } else {
                        entry.insert("nickname".to_string(), line_split[1].to_string());
                        entry.insert("fingerprint".to_string(), line_split[2].to_string());
                    }
                } else {
                    let (key, value) = match line.split_once(' ') {
                        Some((k, v)) => (k, v),
                        None => (line, ""),
                    };
                    entry.insert(key.to_string(), value.to_string());
                }
            }
        }
        // Do for the last one
        let extra_info = Self::from_map(&entry);
        if extra_info.is_ok() {
            set.insert(extra_info.unwrap());
        } else {
            println!("{}", extra_info.err().unwrap());
        }
        set
    }
}