warren/backend/src/lib/outbound/file_system.rs

use std::os::unix::fs::MetadataExt;

use anyhow::{Context, anyhow, bail};
use rustix::fs::statx;
use tokio::{
    fs,
    io::{self, AsyncWriteExt as _},
};
use tokio_util::io::ReaderStream;

use crate::{
    config::Config,
    domain::warren::{
        models::file::{
            AbsoluteFilePath, CreateDirectoryError, CreateDirectoryRequest, CreateFileError,
            CreateFileRequest, DeleteDirectoryError, DeleteDirectoryRequest, DeleteFileError,
            DeleteFileRequest, FetchFileError, FetchFileRequest, File, FileMimeType, FileName,
            FilePath, FileStream, FileType, ListFilesError, ListFilesRequest, RenameEntryError,
            RenameEntryRequest,
        },
        ports::FileSystemRepository,
    },
};

const MAX_FILE_FETCH_BYTES: &str = "MAX_FILE_FETCH_BYTES";

#[derive(Debug, Clone)]
pub struct FileSystemConfig {
    base_directory: String,
    max_file_fetch_bytes: u64,
}

impl FileSystemConfig {
    pub fn new(base_directory: String, max_file_fetch_bytes: u64) -> Self {
        Self {
            base_directory,
            max_file_fetch_bytes,
        }
    }

    pub fn from_env(serve_dir: String) -> anyhow::Result<Self> {
        // 268435456 bytes = 0.25GB
        let max_file_fetch_bytes: u64 = match Config::load_env(MAX_FILE_FETCH_BYTES) {
            Ok(value) => value.parse()?,
            Err(_) => 268435456,
        };

        Ok(Self::new(serve_dir, max_file_fetch_bytes))
    }
}

#[derive(Debug, Clone)]
pub struct FileSystem {
    base_directory: FilePath,
    max_file_fetch_bytes: u64,
}

impl FileSystem {
    pub fn new(config: FileSystemConfig) -> anyhow::Result<Self> {
        let file_system = Self {
            base_directory: FilePath::new(&config.base_directory)?,
            max_file_fetch_bytes: config.max_file_fetch_bytes,
        };

        Ok(file_system)
    }

    /// Combines `self.base_directory` with the specified path
    ///
    /// * `path`: The absolute path (absolute in relation to the base directory)
    fn get_target_path(&self, path: &AbsoluteFilePath) -> FilePath {
        self.base_directory.join(&path.as_relative())
    }

    async fn get_all_files(&self, absolute_path: &AbsoluteFilePath) -> anyhow::Result<Vec<File>> {
        let directory_path = self.get_target_path(absolute_path);

        let mut dir = fs::read_dir(&directory_path).await?;

        let mut files = Vec::new();

        while let Ok(Some(entry)) = dir.next_entry().await {
            let name = entry
                .file_name()
                .into_string()
                .ok()
                .context("Failed to get file name")?;
            let file_type = {
                let file_type = entry.file_type().await?;

                if file_type.is_dir() {
                    FileType::Directory
                } else if file_type.is_file() {
                    FileType::File
                } else {
                    continue;
                }
            };

            // TODO: Use `DirEntry::metadata` once `target=x86_64-unknown-linux-musl` updates from musl 1.2.3 to 1.2.5
            // https://github.com/rust-lang/rust/pull/142682
            let created_at = unsafe {
                statx(
                    std::os::fd::BorrowedFd::borrow_raw(-100),
                    entry.path(),
                    rustix::fs::AtFlags::empty(),
                    rustix::fs::StatxFlags::BTIME,
                )
            }
            .ok()
            .map(|statx| statx.stx_btime.tv_sec as u64);

            let mime_type = match file_type {
                FileType::File => FileMimeType::from_name(&name),
                _ => None,
            };

            files.push(File::new(
                FileName::new(&name)?,
                file_type,
                mime_type,
                created_at,
            ));
        }

        Ok(files)
    }

    /// Actually created a directory in the underlying file system
    ///
    /// * `path`: The directory's absolute path (absolute not in relation to the root file system but `self.base_directory`)
    async fn create_dir(&self, path: &AbsoluteFilePath) -> io::Result<FilePath> {
        let file_path = self.get_target_path(path);

        if fs::try_exists(&file_path).await? {
            return Err(io::ErrorKind::AlreadyExists.into());
        }

        fs::create_dir(&file_path).await?;

        Ok(file_path)
    }

    /// Actually removes a directory from the underlying file system
    ///
    /// * `path`: The directory's absolute path (absolute not in relation to the root file system but `self.base_directory`)
    /// * `force`: Whether to delete directories that are not empty
    async fn remove_dir(&self, path: &AbsoluteFilePath, force: bool) -> io::Result<FilePath> {
        let file_path = self.get_target_path(path);

        if force {
            fs::remove_dir_all(&file_path).await?;
        } else {
            fs::remove_dir(&file_path).await?;
        }

        Ok(file_path)
    }

    async fn write_file(&self, path: &AbsoluteFilePath, data: &[u8]) -> anyhow::Result<FilePath> {
        let path = self.get_target_path(path);

        let mut file = fs::OpenOptions::new()
            .write(true)
            .create(true)
            .open(&path)
            .await?;

        file.write_all(data).await?;

        Ok(path)
    }

    async fn fetch_file(&self, path: &AbsoluteFilePath) -> anyhow::Result<FileStream> {
        let path = self.get_target_path(path);

        let file = fs::OpenOptions::new()
            .create(false)
            .write(false)
            .read(true)
            .open(&path)
            .await?;

        let file_size = file.metadata().await?.size();

        if file_size > self.max_file_fetch_bytes {
            bail!("File size exceeds configured limit");
        }

        let stream = FileStream::new(ReaderStream::new(file));

        Ok(stream)
    }

    /// Actually removes a file from the underlying file system
    ///
    /// * `path`: The file's absolute path (absolute not in relation to the root file system but `self.base_directory`)
    async fn remove_file(&self, path: &AbsoluteFilePath) -> anyhow::Result<FilePath> {
        let path = self.get_target_path(path);

        fs::remove_file(&path).await?;

        Ok(path)
    }

    async fn rename(
        &self,
        path: &AbsoluteFilePath,
        new_name: &FileName,
    ) -> anyhow::Result<FilePath> {
        let current_path = self.get_target_path(path);

        let new_path = {
            let mut c = current_path.to_string();
            let last_slash_index = c.rfind('/').unwrap();
            c.drain((last_slash_index + 1)..);
            c.push_str(new_name.as_str());

            FilePath::new(&c)?
        };

        if fs::try_exists(&new_path).await? {
            bail!("File exists");
        }

        fs::rename(current_path, &new_path).await?;

        Ok(new_path)
    }
}

impl FileSystemRepository for FileSystem {
    async fn list_files(&self, request: ListFilesRequest) -> Result<Vec<File>, ListFilesError> {
        let files = self.get_all_files(request.path()).await.map_err(|err| {
            anyhow!(err).context(format!(
                "Failed to get the files at path: {}",
                request.path()
            ))
        })?;

        Ok(files)
    }

    async fn create_directory(
        &self,
        request: CreateDirectoryRequest,
    ) -> Result<FilePath, CreateDirectoryError> {
        match self.create_dir(request.path()).await {
            Ok(path) => Ok(path),
            Err(e) => match e.kind() {
                std::io::ErrorKind::AlreadyExists => Err(CreateDirectoryError::Exists),
                _ => Err(anyhow!("Failed to create directory at path: {}", request.path()).into()),
            },
        }
    }

    async fn delete_directory(
        &self,
        request: DeleteDirectoryRequest,
    ) -> Result<FilePath, DeleteDirectoryError> {
        match self.remove_dir(request.path(), request.force()).await {
            Ok(deleted_path) => return Ok(deleted_path),
            Err(e) => match e.kind() {
                std::io::ErrorKind::NotFound => Err(DeleteDirectoryError::NotFound),
                std::io::ErrorKind::DirectoryNotEmpty => Err(DeleteDirectoryError::NotEmpty),
                _ => Err(anyhow!("Failed to delete directory at {}: {e:?}", request.path()).into()),
            },
        }
    }

    async fn create_file(&self, request: CreateFileRequest) -> Result<FilePath, CreateFileError> {
        let file_path = self
            .write_file(request.path(), request.data())
            .await
            .map_err(|e| {
                anyhow!(
                    "Failed to write {} byte(s) to path {}: {e:?}",
                    request.data().len(),
                    request.path()
                )
            })?;

        Ok(file_path)
    }

    async fn fetch_file(&self, request: FetchFileRequest) -> Result<FileStream, FetchFileError> {
        let contents = self
            .fetch_file(request.path())
            .await
            .map_err(|e| anyhow!("Failed to fetch file {}: {e:?}", request.path()))?;

        Ok(contents)
    }

    async fn delete_file(&self, request: DeleteFileRequest) -> Result<FilePath, DeleteFileError> {
        let deleted_path = self
            .remove_file(request.path())
            .await
            .context(format!("Failed to delete file at {}", request.path()))?;

        Ok(deleted_path)
    }

    async fn rename_entry(
        &self,
        request: RenameEntryRequest,
    ) -> Result<FilePath, RenameEntryError> {
        let new_path = self
            .rename(request.path(), request.new_name())
            .await
            .map_err(|e| {
                anyhow!(
                    "Failed to rename {} to {}: {e:?}",
                    request.path(),
                    request.new_name()
                )
            })?;

        Ok(new_path)
    }
}