//! Core of the crate, where the `compress_files` and `decompress_file` functions are implemented //! //! Also, where correctly call functions based on the detected `Command`. use std::{ io::{self, BufReader, BufWriter, Read, Write}, ops::ControlFlow, path::{Path, PathBuf}, }; use fs_err as fs; use utils::colors; use crate::{ archive, error::FinalError, extension::{ self, CompressionFormat::{self, *}, Extension, }, info, list::{self, ListOptions}, progress::Progress, utils::{ self, concatenate_os_str_list, dir_is_empty, nice_directory_display, to_utf, try_infer_extension, user_wants_to_continue_decompressing, }, warning, Opts, QuestionPolicy, Subcommand, }; // Used in BufReader and BufWriter to perform less syscalls const BUFFER_CAPACITY: usize = 1024 * 64; fn represents_several_files(files: &[PathBuf]) -> bool { let is_non_empty_dir = |path: &PathBuf| { let is_non_empty = || !dir_is_empty(path); path.is_dir().then(is_non_empty).unwrap_or_default() }; files.iter().any(is_non_empty_dir) || files.len() > 1 } /// Entrypoint of ouch, receives cli options and matches Subcommand to decide what to do pub fn run(args: Opts, question_policy: QuestionPolicy) -> crate::Result<()> { match args.cmd { Subcommand::Compress { mut files, output: output_path } => { // If the output_path file exists and is the same as some of the input files, warn the user and skip those inputs (in order to avoid compression recursion) if output_path.exists() { clean_input_files_if_needed(&mut files, &fs::canonicalize(&output_path)?); } // After cleaning, if there are no input files left, exit if files.is_empty() { return Err(FinalError::with_title("No files to compress").into()); } // Formats from path extension, like "file.tar.gz.xz" -> vec![Tar, Gzip, Lzma] let mut formats = extension::extensions_from_path(&output_path); if formats.is_empty() { let error = FinalError::with_title(format!("Cannot compress to '{}'.", to_utf(&output_path))) .detail("You shall supply the compression format") .hint("Try adding supported extensions (see --help):") .hint(format!(" ouch compress ... {}.tar.gz", to_utf(&output_path))) .hint(format!(" ouch compress ... {}.zip", to_utf(&output_path))) .hint("") .hint("Alternatively, you can overwrite this option by using the '--format' flag:") .hint(format!(" ouch compress ... {} --format tar.gz", to_utf(&output_path))); return Err(error.into()); } if !formats.get(0).map(Extension::is_archive).unwrap_or(false) && represents_several_files(&files) { // This piece of code creates a suggestion for compressing multiple files // It says: // Change from file.bz.xz // To file.tar.bz.xz let extensions_text: String = formats.iter().map(|format| format.to_string()).collect(); let output_path = to_utf(output_path); // Breaks if Lzma is .lz or .lzma and not .xz // Or if Bzip is .bz2 and not .bz let extensions_start_position = output_path.rfind(&extensions_text).unwrap(); let pos = extensions_start_position; let empty_range = pos..pos; let mut suggested_output_path = output_path.clone(); suggested_output_path.replace_range(empty_range, ".tar"); let error = FinalError::with_title(format!("Cannot compress to '{}'.", to_utf(&output_path))) .detail("You are trying to compress multiple files.") .detail(format!("The compression format '{}' cannot receive multiple files.", &formats[0])) .detail("The only supported formats that archive files into an archive are .tar and .zip.") .hint(format!("Try inserting '.tar' or '.zip' before '{}'.", &formats[0])) .hint(format!("From: {}", output_path)) .hint(format!("To: {}", suggested_output_path)); return Err(error.into()); } if let Some(format) = formats.iter().skip(1).find(|format| format.is_archive()) { let error = FinalError::with_title(format!("Cannot compress to '{}'.", to_utf(&output_path))) .detail(format!("Found the format '{}' in an incorrect position.", format)) .detail(format!("'{}' can only be used at the start of the file extension.", format)) .hint(format!("If you wish to compress multiple files, start the extension with '{}'.", format)) .hint(format!("Otherwise, remove the last '{}' from '{}'.", format, to_utf(&output_path))); return Err(error.into()); } if output_path.exists() && !utils::user_wants_to_overwrite(&output_path, question_policy)? { // User does not want to overwrite this file, skip and return without any errors return Ok(()); } let output_file = fs::File::create(&output_path)?; if !represents_several_files(&files) { // It's possible the file is already partially compressed so we don't want to compress it again // `ouch compress file.tar.gz file.tar.gz.xz` should produce `file.tar.gz.xz` and not `file.tar.gz.tar.gz.xz` let input_extensions = extension::extensions_from_path(&files[0]); // We calculate the formats that are left if we filter out a sublist at the start of what we have that's the same as the input formats let mut new_formats = Vec::with_capacity(formats.len()); for (inp_ext, out_ext) in input_extensions.iter().zip(&formats) { if inp_ext.compression_formats == out_ext.compression_formats { new_formats.push(out_ext.clone()); } else if inp_ext .compression_formats .iter() .zip(out_ext.compression_formats.iter()) .all(|(inp, out)| inp == out) { let new_ext = Extension::new( &out_ext.compression_formats[..inp_ext.compression_formats.len()], &out_ext.display_text, ); new_formats.push(new_ext); break; } } // If the input is a sublist at the start of `formats` then remove the extensions // Note: If input_extensions is empty then it will make `formats` empty too, which we don't want if !input_extensions.is_empty() && new_formats != formats { // Safety: // We checked above that input_extensions isn't empty, so files[0] has an extension. // // Path::extension says: "if there is no file_name, then there is no extension". // Contrapositive statement: "if there is extension, then there is file_name". info!( accessible, // important information "Partial compression detected. Compressing {} into {}", to_utf(files[0].as_path().file_name().unwrap()), to_utf(&output_path) ); formats = new_formats; } } let compress_result = compress_files(files, formats, output_file); // If any error occurred, delete incomplete file if compress_result.is_err() { // Print an extra alert message pointing out that we left a possibly // CORRUPTED FILE at `output_path` if let Err(err) = fs::remove_file(&output_path) { eprintln!("{red}FATAL ERROR:\n", red = *colors::RED); eprintln!(" Please manually delete '{}'.", to_utf(&output_path)); eprintln!(" Compression failed and we could not delete '{}'.", to_utf(&output_path),); eprintln!(" Error:{reset} {}{red}.{reset}\n", err, reset = *colors::RESET, red = *colors::RED); } } else { // this is only printed once, so it doesn't result in much text. On the other hand, // having a final status message is important especially in an accessibility context // as screen readers may not read a commands exit code, making it hard to reason // about whether the command succeeded without such a message info!(accessible, "Successfully compressed '{}'.", to_utf(output_path)); } compress_result?; } Subcommand::Decompress { files, output_dir } => { let mut output_paths = vec![]; let mut formats = vec![]; for path in files.iter() { let (file_output_path, file_formats) = extension::separate_known_extensions_from_name(path); output_paths.push(file_output_path); formats.push(file_formats); } if let ControlFlow::Break(_) = check_mime_type(&files, &mut formats, question_policy)? { return Ok(()); } let files_missing_format: Vec = files .iter() .zip(&formats) .filter(|(_, formats)| formats.is_empty()) .map(|(input_path, _)| PathBuf::from(input_path)) .collect(); if !files_missing_format.is_empty() { let error = FinalError::with_title("Cannot decompress files without extensions") .detail(format!( "Files without supported extensions: {}", concatenate_os_str_list(&files_missing_format) )) .detail("Decompression formats are detected automatically by the file extension") .hint("Provide a file with a supported extension:") .hint(" ouch decompress example.tar.gz") .hint("") .hint("Or overwrite this option with the '--format' flag:") .hint(format!(" ouch decompress {} --format tar.gz", to_utf(&files_missing_format[0]))); return Err(error.into()); } // The directory that will contain the output files // We default to the current directory if the user didn't specify an output directory with --dir let output_dir = if let Some(dir) = output_dir { if !utils::clear_path(&dir, question_policy)? { // User doesn't want to overwrite return Ok(()); } utils::create_dir_if_non_existent(&dir)?; dir } else { PathBuf::from(".") }; for ((input_path, formats), file_name) in files.iter().zip(formats).zip(output_paths) { let output_file_path = output_dir.join(file_name); // Path used by single file format archives decompress_file(input_path, formats, &output_dir, output_file_path, question_policy)?; } } Subcommand::List { archives: files, tree } => { let mut formats = vec![]; for path in files.iter() { let (_, file_formats) = extension::separate_known_extensions_from_name(path); formats.push(file_formats); } if let ControlFlow::Break(_) = check_mime_type(&files, &mut formats, question_policy)? { return Ok(()); } let not_archives: Vec = files .iter() .zip(&formats) .filter(|(_, formats)| !formats.get(0).map(Extension::is_archive).unwrap_or(false)) .map(|(path, _)| path.clone()) .collect(); if !not_archives.is_empty() { let error = FinalError::with_title("Cannot list archive contents") .detail("Only archives can have their contents listed") .detail(format!("Files are not archives: {}", concatenate_os_str_list(¬_archives))); return Err(error.into()); } let list_options = ListOptions { tree }; for (i, (archive_path, formats)) in files.iter().zip(formats).enumerate() { if i > 0 { println!(); } let formats = formats.iter().flat_map(Extension::iter).map(Clone::clone).collect(); list_archive_contents(archive_path, formats, list_options)?; } } } Ok(()) } // Compress files into an `output_file` // // files are the list of paths to be compressed: ["dir/file1.txt", "dir/file2.txt"] // formats contains each format necessary for compression, example: [Tar, Gz] (in compression order) // output_file is the resulting compressed file name, example: "compressed.tar.gz" fn compress_files(files: Vec, formats: Vec, output_file: fs::File) -> crate::Result<()> { // The next lines are for displaying the progress bar // If the input files contain a directory, then the total size will be underestimated let (total_input_size, precise) = files .iter() .map(|f| (f.metadata().expect("file exists").len(), f.is_file())) .fold((0, true), |(total_size, and_precise), (size, precise)| (total_size + size, and_precise & precise)); //NOTE: canonicalize is here to avoid a weird bug: // > If output_file_path is a nested path and it exists and the user overwrite it // >> output_file_path.exists() will always return false (somehow) // - canonicalize seems to fix this let output_file_path = output_file.path().canonicalize()?; let file_writer = BufWriter::with_capacity(BUFFER_CAPACITY, output_file); let mut writer: Box = Box::new(file_writer); // Grab previous encoder and wrap it inside of a new one let chain_writer_encoder = |format: &CompressionFormat, encoder: Box| -> crate::Result> { let encoder: Box = match format { Gzip => Box::new(flate2::write::GzEncoder::new(encoder, Default::default())), Bzip => Box::new(bzip2::write::BzEncoder::new(encoder, Default::default())), Lz4 => Box::new(lzzzz::lz4f::WriteCompressor::new(encoder, Default::default())?), Lzma => Box::new(xz2::write::XzEncoder::new(encoder, 6)), Zstd => { let zstd_encoder = zstd::stream::write::Encoder::new(encoder, Default::default()); // Safety: // Encoder::new() can only fail if `level` is invalid, but Default::default() // is guaranteed to be valid Box::new(zstd_encoder.unwrap().auto_finish()) } Tar | Zip => unreachable!(), }; Ok(encoder) }; for format in formats.iter().flat_map(Extension::iter).skip(1).collect::>().iter().rev() { writer = chain_writer_encoder(format, writer)?; } match formats[0].compression_formats[0] { Gzip | Bzip | Lz4 | Lzma | Zstd => { let _progress = Progress::new_accessible_aware( total_input_size, precise, Some(Box::new(move || output_file_path.metadata().expect("file exists").len())), ); writer = chain_writer_encoder(&formats[0].compression_formats[0], writer)?; let mut reader = fs::File::open(&files[0]).unwrap(); io::copy(&mut reader, &mut writer)?; } Tar => { let mut progress = Progress::new_accessible_aware( total_input_size, precise, Some(Box::new(move || output_file_path.metadata().expect("file exists").len())), ); archive::tar::build_archive_from_paths( &files, &mut writer, progress.as_mut().map(Progress::display_handle).unwrap_or(&mut io::stdout()), )?; writer.flush()?; } Zip => { eprintln!("{yellow}Warning:{reset}", yellow = *colors::YELLOW, reset = *colors::RESET); eprintln!("\tCompressing .zip entirely in memory."); eprintln!("\tIf the file is too big, your PC might freeze!"); eprintln!( "\tThis is a limitation for formats like '{}'.", formats.iter().map(|format| format.to_string()).collect::() ); eprintln!("\tThe design of .zip makes it impossible to compress via stream."); let mut vec_buffer = io::Cursor::new(vec![]); let current_position_fn = { let vec_buffer_ptr = { struct FlyPtr(*const io::Cursor>); unsafe impl Send for FlyPtr {} FlyPtr(&vec_buffer as *const _) }; Box::new(move || { let vec_buffer_ptr = &vec_buffer_ptr; // Safety: ptr is valid and vec_buffer is still alive unsafe { &*vec_buffer_ptr.0 }.position() }) }; let mut progress = Progress::new_accessible_aware(total_input_size, precise, Some(current_position_fn)); archive::zip::build_archive_from_paths( &files, &mut vec_buffer, progress.as_mut().map(Progress::display_handle).unwrap_or(&mut io::stdout()), )?; let vec_buffer = vec_buffer.into_inner(); io::copy(&mut vec_buffer.as_slice(), &mut writer)?; } } Ok(()) } // Decompress a file // // File at input_file_path is opened for reading, example: "archive.tar.gz" // formats contains each format necessary for decompression, example: [Gz, Tar] (in decompression order) // output_dir it's where the file will be decompressed to, this function assumes that the directory exists // output_file_path is only used when extracting single file formats, not archive formats like .tar or .zip fn decompress_file( input_file_path: &Path, formats: Vec, output_dir: &Path, output_file_path: PathBuf, question_policy: QuestionPolicy, ) -> crate::Result<()> { assert!(output_dir.exists()); let total_input_size = input_file_path.metadata().expect("file exists").len(); let reader = fs::File::open(&input_file_path)?; // Zip archives are special, because they require io::Seek, so it requires it's logic separated // from decoder chaining. // // This is the only case where we can read and unpack it directly, without having to do // in-memory decompression/copying first. // // Any other Zip decompression done can take up the whole RAM and freeze ouch. if formats.len() == 1 && *formats[0].compression_formats == [Zip] { let zip_archive = zip::ZipArchive::new(reader)?; let files = if let ControlFlow::Continue(files) = smart_unpack( Box::new(move |output_dir| { let mut progress = Progress::new_accessible_aware(total_input_size, true, None); crate::archive::zip::unpack_archive( zip_archive, output_dir, progress.as_mut().map(Progress::display_handle).unwrap_or(&mut io::stdout()), ) }), output_dir, &output_file_path, question_policy, )? { files } else { return Ok(()); }; // this is only printed once, so it doesn't result in much text. On the other hand, // having a final status message is important especially in an accessibility context // as screen readers may not read a commands exit code, making it hard to reason // about whether the command succeeded without such a message info!( accessible, "Successfully decompressed archive in {} ({} files).", nice_directory_display(output_dir), files.len() ); return Ok(()); } // Will be used in decoder chaining let reader = BufReader::with_capacity(BUFFER_CAPACITY, reader); let mut reader: Box = Box::new(reader); // Grab previous decoder and wrap it inside of a new one let chain_reader_decoder = |format: &CompressionFormat, decoder: Box| -> crate::Result> { let decoder: Box = match format { Gzip => Box::new(flate2::read::GzDecoder::new(decoder)), Bzip => Box::new(bzip2::read::BzDecoder::new(decoder)), Lz4 => Box::new(lzzzz::lz4f::ReadDecompressor::new(decoder)?), Lzma => Box::new(xz2::read::XzDecoder::new(decoder)), Zstd => Box::new(zstd::stream::Decoder::new(decoder)?), Tar | Zip => unreachable!(), }; Ok(decoder) }; for format in formats.iter().flat_map(Extension::iter).skip(1).collect::>().iter().rev() { reader = chain_reader_decoder(format, reader)?; } let files_unpacked; match formats[0].compression_formats[0] { Gzip | Bzip | Lz4 | Lzma | Zstd => { reader = chain_reader_decoder(&formats[0].compression_formats[0], reader)?; let writer = utils::create_or_ask_overwrite(&output_file_path, question_policy)?; if writer.is_none() { // Means that the user doesn't want to overwrite return Ok(()); } let mut writer = writer.unwrap(); let current_position_fn = Box::new({ let output_file_path = output_file_path.clone(); move || output_file_path.clone().metadata().expect("file exists").len() }); let _progress = Progress::new_accessible_aware(total_input_size, true, Some(current_position_fn)); io::copy(&mut reader, &mut writer)?; files_unpacked = vec![output_file_path]; } Tar => { files_unpacked = if let ControlFlow::Continue(files) = smart_unpack( Box::new(move |output_dir| { let mut progress = Progress::new_accessible_aware(total_input_size, true, None); crate::archive::tar::unpack_archive( reader, output_dir, progress.as_mut().map(Progress::display_handle).unwrap_or(&mut io::stdout()), ) }), output_dir, &output_file_path, question_policy, )? { files } else { return Ok(()); }; } Zip => { eprintln!("Compressing first into .zip."); eprintln!("Warning: .zip archives with extra extensions have a downside."); eprintln!( "The only way is loading everything into the RAM while compressing, and then write everything down." ); eprintln!("this means that by compressing .zip with extra compression formats, you can run out of RAM if the file is too large!"); let mut vec = vec![]; io::copy(&mut reader, &mut vec)?; let zip_archive = zip::ZipArchive::new(io::Cursor::new(vec))?; files_unpacked = if let ControlFlow::Continue(files) = smart_unpack( Box::new(move |output_dir| { let mut progress = Progress::new_accessible_aware(total_input_size, true, None); crate::archive::zip::unpack_archive( zip_archive, output_dir, progress.as_mut().map(Progress::display_handle).unwrap_or(&mut io::stdout()), ) }), output_dir, &output_file_path, question_policy, )? { files } else { return Ok(()); }; } } // this is only printed once, so it doesn't result in much text. On the other hand, // having a final status message is important especially in an accessibility context // as screen readers may not read a commands exit code, making it hard to reason // about whether the command succeeded without such a message info!(accessible, "Successfully decompressed archive in {}.", nice_directory_display(output_dir)); info!(accessible, "Files unpacked: {}", files_unpacked.len()); Ok(()) } // File at input_file_path is opened for reading, example: "archive.tar.gz" // formats contains each format necessary for decompression, example: [Gz, Tar] (in decompression order) fn list_archive_contents( archive_path: &Path, formats: Vec, list_options: ListOptions, ) -> crate::Result<()> { let reader = fs::File::open(&archive_path)?; // Zip archives are special, because they require io::Seek, so it requires it's logic separated // from decoder chaining. // // This is the only case where we can read and unpack it directly, without having to do // in-memory decompression/copying first. // // Any other Zip decompression done can take up the whole RAM and freeze ouch. if let [Zip] = *formats.as_slice() { let zip_archive = zip::ZipArchive::new(reader)?; let files = crate::archive::zip::list_archive(zip_archive)?; list::list_files(archive_path, files, list_options); return Ok(()); } // Will be used in decoder chaining let reader = BufReader::with_capacity(BUFFER_CAPACITY, reader); let mut reader: Box = Box::new(reader); // Grab previous decoder and wrap it inside of a new one let chain_reader_decoder = |format: &CompressionFormat, decoder: Box| -> crate::Result> { let decoder: Box = match format { Gzip => Box::new(flate2::read::GzDecoder::new(decoder)), Bzip => Box::new(bzip2::read::BzDecoder::new(decoder)), Lz4 => Box::new(lzzzz::lz4f::ReadDecompressor::new(decoder)?), Lzma => Box::new(xz2::read::XzDecoder::new(decoder)), Zstd => Box::new(zstd::stream::Decoder::new(decoder)?), Tar | Zip => unreachable!(), }; Ok(decoder) }; for format in formats.iter().skip(1).rev() { reader = chain_reader_decoder(format, reader)?; } let files = match formats[0] { Tar => crate::archive::tar::list_archive(reader)?, Zip => { eprintln!("Listing files from zip archive."); eprintln!("Warning: .zip archives with extra extensions have a downside."); eprintln!("The only way is loading everything into the RAM while compressing, and then reading the archive contents."); eprintln!("this means that by compressing .zip with extra compression formats, you can run out of RAM if the file is too large!"); let mut vec = vec![]; io::copy(&mut reader, &mut vec)?; let zip_archive = zip::ZipArchive::new(io::Cursor::new(vec))?; crate::archive::zip::list_archive(zip_archive)? } Gzip | Bzip | Lz4 | Lzma | Zstd => { panic!("Not an archive! This should never happen, if it does, something is wrong with `CompressionFormat::is_archive()`. Please report this error!"); } }; list::list_files(archive_path, files, list_options); Ok(()) } /// Unpacks an archive with some heuristics /// - If the archive contains only one file, it will be extracted to the `output_dir` /// - If the archive contains multiple files, it will be extracted to a subdirectory of the output_dir named after the archive (given by `output_file_path`) /// Note: This functions assumes that `output_dir` exists fn smart_unpack( unpack_fn: Box crate::Result>>, output_dir: &Path, output_file_path: &Path, question_policy: QuestionPolicy, ) -> crate::Result>> { assert!(output_dir.exists()); let temp_dir = tempfile::tempdir_in(output_dir)?; let temp_dir_path = temp_dir.path(); info!( accessible, "Created temporary directory {} to hold decompressed elements.", nice_directory_display(temp_dir_path) ); // unpack the files let files = unpack_fn(temp_dir_path)?; let root_contains_only_one_element = fs::read_dir(&temp_dir_path)?.count() == 1; if root_contains_only_one_element { // Only one file in the root directory, so we can just move it to the output directory let file = fs::read_dir(&temp_dir_path)?.next().expect("item exists")?; let file_path = file.path(); let file_name = file_path.file_name().expect("Should be safe because paths in archives should not end with '..'"); let correct_path = output_dir.join(file_name); // One case to handle tough is we need to check if a file with the same name already exists if !utils::clear_path(&correct_path, question_policy)? { return Ok(ControlFlow::Break(())); } fs::rename(&file_path, &correct_path)?; info!( accessible, "Successfully moved {} to {}.", nice_directory_display(&file_path), nice_directory_display(&correct_path) ); } else { // Multiple files in the root directory, so: // Rename the temporary directory to the archive name, which is output_file_path // One case to handle tough is we need to check if a file with the same name already exists if !utils::clear_path(output_file_path, question_policy)? { return Ok(ControlFlow::Break(())); } fs::rename(&temp_dir_path, &output_file_path)?; info!( accessible, "Successfully moved {} to {}.", nice_directory_display(&temp_dir_path), nice_directory_display(&output_file_path) ); } Ok(ControlFlow::Continue(files)) } fn check_mime_type( files: &[PathBuf], formats: &mut Vec>, question_policy: QuestionPolicy, ) -> crate::Result> { for (path, format) in files.iter().zip(formats.iter_mut()) { if format.is_empty() { // File with no extension // Try to detect it automatically and prompt the user about it if let Some(detected_format) = try_infer_extension(path) { // Infering the file extension can have unpredicted consequences (e.g. the user just // mistyped, ...) which we should always inform the user about. info!(accessible, "Detected file: `{}` extension as `{}`", path.display(), detected_format); if user_wants_to_continue_decompressing(path, question_policy)? { format.push(detected_format); } else { return Ok(ControlFlow::Break(())); } } } else if let Some(detected_format) = try_infer_extension(path) { // File ending with extension // Try to detect the extension and warn the user if it differs from the written one let outer_ext = format.iter().next_back().unwrap(); if outer_ext != &detected_format { warning!( "The file extension: `{}` differ from the detected extension: `{}`", outer_ext, detected_format ); if !user_wants_to_continue_decompressing(path, question_policy)? { return Ok(ControlFlow::Break(())); } } } else { // NOTE: If this actually produces no false positives, we can upgrade it in the future // to a warning and ask the user if he wants to continue decompressing. info!(accessible, "Could not detect the extension of `{}`", path.display()); } } Ok(ControlFlow::Continue(())) } fn clean_input_files_if_needed(files: &mut Vec, output_path: &Path) { let mut idx = 0; while idx < files.len() { if files[idx] == output_path { warning!("The output file and the input file are the same: `{}`, skipping...", output_path.display()); files.remove(idx); } else { idx += 1; } } }