mirrord/src/proxy.rs

use actix_web::{HttpRequest, HttpResponse};
use rand::prelude::*;
use std::{
    path::{Path, PathBuf},
    sync::Arc,
};

use crate::config::Config;

// todo : download from many mirrors and compare results if one acts badly
// todo : better logging
// todo : better cache management

pub struct Mirror {
    mirrors: Vec<Arc<String>>,
    data_dir: String,
    no_cache: regex::Regex,
    only_allow: Option<regex::Regex>,
    config: Config,
}

impl Mirror {
    pub fn new(config: &Config) -> Self {
        let mirrors = config.mirrors.clone();
        Self {
            mirrors: mirrors.into_iter().map(Arc::new).collect(),
            data_dir: config.cache_dir.clone(),
            no_cache: regex::Regex::new(&config.no_cache).unwrap(),
            only_allow: config
                .only_allow
                .clone()
                .map(|x| regex::Regex::new(&x).unwrap()),
            config: config.clone(),
        }
    }

    /// Creates a cache directory at the specified path.
    ///
    /// If the specified path points to a file, it renames the file by appending ".tmp" to its name,
    /// creates the cache directory, and moves the file into the cache directory with the name "index".
    ///
    /// If the specified path does not exist or points to a directory, it creates the directory.
    ///
    /// # Arguments
    ///
    /// * `dir` - A reference to the path where the cache directory should be created.
    ///
    /// # Panics
    ///
    /// This function panics if any of the file system operations fail, such as renaming files or creating directories.
    pub fn create_cache_dir(dir: &Path) {
        if dir.is_file() {
            let tmp_file_path = dir.with_extension("tmp");
            std::fs::rename(dir, &tmp_file_path).expect("Failed to rename file");

            std::fs::create_dir_all(dir).expect("Failed to create directory");

            let index_file_path = dir.join("index");
            std::fs::rename(&tmp_file_path, index_file_path)
                .expect("Failed to move file into directory");
        } else {
            std::fs::create_dir_all(dir).unwrap();
        }
    }

    /// Checks if the cached data at the specified path is invalid based on the TTL (time-to-live) setting.
    ///
    /// # Arguments
    ///
    /// * `p` - A reference to the path of the cached data file.
    ///
    /// # Returns
    ///
    /// * `true` if the cached data is invalid (older than TTL), `false` otherwise.
    pub fn is_cache_invalid(&self, p: &Path) -> bool {
        if self.config.ttl.is_none() {
            return false;
        }

        let try_is_cache_invalid = || {
            let modified = p.metadata().ok()?.modified().ok()?;
            let current_time = std::time::SystemTime::now();
            let elapsed_time = current_time.duration_since(modified).ok()?;
            let threshold_duration =
                std::time::Duration::from_secs((self.config.ttl.unwrap() * 60).try_into().unwrap());

            if elapsed_time > threshold_duration {
                log::info!(
                    "Cached file is {} minutes old. Older than TTL {}.",
                    (elapsed_time.as_secs() / 60),
                    self.config.ttl.unwrap()
                );
                Some(true)
            } else {
                Some(false)
            }
        };

        try_is_cache_invalid().unwrap_or(false)
    }

    /// Asynchronously retrieves content from the specified path, either from cache or mirrors.
    ///
    /// This function attempts to retrieve content from the specified `path`. If caching is enabled
    /// and the content exists in the cache, it is returned directly from the cache. If not found in
    /// the cache or caching is disabled, the function attempts to fetch the content from a list of mirrors.
    ///
    /// # Arguments
    ///
    /// * `path` - The path from which to retrieve content.
    /// * `req` - The HTTP request associated with the retrieval operation.
    ///
    /// # Returns
    ///
    /// An `Option` containing an `HttpResponse` if content is successfully retrieved, or `None` if the
    /// content could not be found or fetched from any source.
    pub async fn get(&self, path: &str, req: &HttpRequest) -> Option<HttpResponse> {
        let p = std::path::Path::new(&path[1..]);
        let p = std::path::Path::new(&self.data_dir).join(p);

        // check if path is in scope
        if let Some(only_allow) = &self.only_allow {
            if !only_allow.is_match(path) {
                return Some(HttpResponse::NotFound().finish());
            }
        }

        // check if cache should be used
        Self::create_cache_dir(p.parent().unwrap());
        if !self.no_cache.is_match(path) && !self.is_cache_invalid(&p) {
            // use cache if present
            if let Some(resp) = self.fetch_cache(&p, req).await {
                log::info!("Returning {path} from cache");
                return Some(resp);
            }
        }

        // fetch from network, if no response (offline) then use cache
        if let Some(resp) = self.fetch_network(path, &p).await {
            Some(resp)
        } else {
            log::info!("Returning {path} from cache");
            self.fetch_cache(&p, req).await
        }
    }

    /// Asynchronously fetches cached data from the specified path and returns an HTTP response if the data exists.
    ///
    /// # Arguments
    ///
    /// * `p` - A reference to the path of the cached data file.
    /// * `req` - A reference to the HTTP request associated with the fetch operation.
    ///
    /// # Returns
    ///
    /// An optional HTTP response containing the fetched data if it exists, or `None` if the data does not exist.
    pub async fn fetch_cache(&self, p: &PathBuf, req: &HttpRequest) -> Option<HttpResponse> {
        if p.exists() {
            if p.is_dir() {
                return Some(
                    actix_files::NamedFile::open_async(p.join("index"))
                        .await
                        .ok()?
                        .into_response(req),
                );
            }

            return Some(
                actix_files::NamedFile::open_async(&p)
                    .await
                    .ok()?
                    .into_response(req),
            );
        }

        None
    }

    /// Asynchronously fetches data from network mirrors and returns an HTTP response if successful.
    ///
    /// # Arguments
    ///
    /// * `path` - The path of the resource to fetch from network mirrors.
    /// * `local` - A reference to the local path where the fetched data will be stored.
    ///
    /// # Returns
    ///
    /// An optional HTTP response containing the fetched data if successful, or `None` if fetching fails.
    pub async fn fetch_network(&self, path: &str, local: &PathBuf) -> Option<HttpResponse> {
        let mut mirrors = self.mirrors.clone();
        mirrors.shuffle(&mut rand::thread_rng());

        log::info!("Fetching {path} from mirrors");
        for mirror in mirrors {
            let url = format!("{mirror}{path}");
            if let Some(response) = self.get_url(&url, local).await {
                if response.status().is_success() {
                    return Some(response);
                }
            }
        }

        None
    }

    pub fn index_page(&self) -> Option<HttpResponse> {
        if let Some(index) = &self.config.index {
            return match index.as_str() {
                "no" => None,
                file => Some(
                    HttpResponse::Ok()
                        .content_type("text/html")
                        .body(std::fs::read_to_string(file).unwrap()),
                ),
            };
        }

        return None;
    }

    /// Asynchronously fetches content from the specified URL and saves it to the provided file path.
    ///
    /// This function sends an HTTP GET request to the URL specified by `path`, retrieves the response,
    /// and saves the response body to the file specified by `save`. If the HTTP request is successful
    /// (status code 2xx), the response body is saved to the file.
    ///
    /// # Arguments
    ///
    /// * `path` - The URL from which to fetch content.
    /// * `save` - The file path where the fetched content should be saved.
    ///
    /// # Returns
    ///
    /// An `Option` containing an `HttpResponse` if the request was successful and the response was received,
    /// or `None` if there was an error during the request or response retrieval.
    pub async fn get_url(&self, path: &str, save: &PathBuf) -> Option<HttpResponse> {
        log::info!("Fetching {path}");
        let response = reqwest::get(path).await.ok()?;
        let status_code = response.status();
        let body_bytes = response.bytes().await.ok()?;
        if status_code.is_success() {
            log::debug!("Writing request to {save:?}");
            std::fs::write(save, &body_bytes).unwrap();
        }
        let mut http_response = HttpResponse::build(
            actix_web::http::StatusCode::from_u16(status_code.as_u16()).unwrap(),
        );
        let http_response = http_response.body(body_bytes);
        Some(http_response)
    }
}