Add Extension wrapper that lets us keep tgz instead of forcing tar.gz

src/commands.rs

@@ -17,6 +17,7 @@ use crate::{
     extension::{
         self,
         CompressionFormat::{self, *},
+        Extension,
     },
     info,
     utils::nice_directory_display,
@@ -56,9 +57,7 @@ pub fn run(args: Opts, skip_questions_positively: Option<bool>) -> crate::Result
                 return Err(Error::with_reason(reason));
             }
 
-            if !formats.get(0).map(CompressionFormat::is_archive_format).unwrap_or(false)
-                && represents_several_files(&files)
-            {
+            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
@@ -86,7 +85,7 @@ pub fn run(args: Opts, skip_questions_positively: Option<bool>) -> crate::Result
                 return Err(Error::with_reason(reason));
             }
 
-            if let Some(format) = formats.iter().skip(1).find(|format| format.is_archive_format()) {
+            if let Some(format) = formats.iter().skip(1).find(|format| format.is_archive()) {
                 let reason = 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))
@@ -108,12 +107,28 @@ pub fn run(args: Opts, skip_questions_positively: Option<bool>) -> crate::Result
                 // `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)
+                        .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 this counts as true
-                if !input_extensions.is_empty()
-                    && input_extensions.len() < formats.len()
-                    && input_extensions.iter().zip(&formats).all(|(inp, out)| inp == out)
-                {
+                // 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 a extension.
                     //
@@ -124,8 +139,7 @@ pub fn run(args: Opts, skip_questions_positively: Option<bool>) -> crate::Result
                         to_utf(files[0].as_path().file_name().unwrap()),
                         to_utf(&output_path)
                     );
-                    let drain_iter = formats.drain(..input_extensions.len());
-                    drop(drain_iter); // Remove the extensions from `formats`
+                    formats = new_formats;
                 }
             }
             let compress_result = compress_files(files, formats, output_file);
@@ -185,7 +199,7 @@ pub fn run(args: Opts, skip_questions_positively: Option<bool>) -> crate::Result
     Ok(())
 }
 
-fn compress_files(files: Vec<PathBuf>, formats: Vec<CompressionFormat>, output_file: fs::File) -> crate::Result<()> {
+fn compress_files(files: Vec<PathBuf>, formats: Vec<Extension>, output_file: fs::File) -> crate::Result<()> {
     let file_writer = BufWriter::with_capacity(BUFFER_CAPACITY, output_file);
 
     let mut writer: Box<dyn Write> = Box::new(file_writer);
@@ -208,13 +222,13 @@ fn compress_files(files: Vec<PathBuf>, formats: Vec<CompressionFormat>, output_f
         encoder
     };
 
-    for format in formats.iter().skip(1).rev() {
+    for format in formats.iter().flat_map(Extension::iter).skip(1).collect::<Vec<_>>().iter().rev() {
         writer = chain_writer_encoder(format, writer);
     }
 
-    match formats[0] {
+    match formats[0].compression_formats[0] {
         Gzip | Bzip | Lzma | Zstd => {
-            writer = chain_writer_encoder(&formats[0], writer);
+            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)?;
         }
@@ -248,7 +262,7 @@ fn compress_files(files: Vec<PathBuf>, formats: Vec<CompressionFormat>, output_f
 // file_name is only used when extracting single file formats, no archive formats like .tar or .zip
 fn decompress_file(
     input_file_path: &Path,
-    formats: Vec<extension::CompressionFormat>,
+    formats: Vec<Extension>,
     output_folder: Option<&Path>,
     file_name: &Path,
     skip_questions_positively: Option<bool>,
@@ -270,7 +284,7 @@ fn decompress_file(
     // 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() {
+    if formats.len() == 1 && *formats[0].compression_formats.as_slice() == [Zip] {
         utils::create_dir_if_non_existent(output_folder)?;
         let zip_archive = zip::ZipArchive::new(reader)?;
         let _files = crate::archive::zip::unpack_archive(zip_archive, output_folder, skip_questions_positively)?;
@@ -294,15 +308,15 @@ fn decompress_file(
         Ok(decoder)
     };
 
-    for format in formats.iter().skip(1).rev() {
+    for format in formats.iter().flat_map(Extension::iter).skip(1).collect::<Vec<_>>().iter().rev() {
         reader = chain_reader_decoder(format, reader)?;
     }
 
     utils::create_dir_if_non_existent(output_folder)?;
 
-    match formats[0] {
+    match formats[0].compression_formats[0] {
         Gzip | Bzip | Lzma | Zstd => {
-            reader = chain_reader_decoder(&formats[0], reader)?;
+            reader = chain_reader_decoder(&formats[0].compression_formats[0], reader)?;
 
             // TODO: improve error treatment
             let mut writer = fs::File::create(&output_path)?;

src/extension.rs

@@ -4,7 +4,40 @@ use std::{ffi::OsStr, fmt, path::Path};
 
 use self::CompressionFormat::*;
 
-#[derive(Clone, PartialEq, Eq, Debug)]
+/// A wrapper around `CompressionFormat` that allows combinations like `tgz`
+#[derive(Debug, Clone, PartialEq, Eq)]
+pub struct Extension {
+    pub compression_formats: Vec<CompressionFormat>,
+    pub display_text: String,
+}
+
+impl Extension {
+    /// # Panics:
+    ///   Will panic if `formats` is empty
+    pub fn new(formats: impl Into<Vec<CompressionFormat>>, text: impl Into<String>) -> Self {
+        let formats = formats.into();
+        assert!(!formats.is_empty());
+        Self { compression_formats: formats, display_text: text.into() }
+    }
+
+    /// Checks if the first format in `compression_formats` is an archive
+    pub fn is_archive(&self) -> bool {
+        // Safety: we check that `compression_formats` is not empty in `Self::new`
+        self.compression_formats[0].is_archive_format()
+    }
+
+    pub fn iter(&self) -> impl Iterator<Item = &CompressionFormat> {
+        self.compression_formats.iter()
+    }
+}
+
+impl fmt::Display for Extension {
+    fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
+        f.write_str(&self.display_text)
+    }
+}
+
+#[derive(Copy, Clone, PartialEq, Eq, Debug)]
 /// Accepted extensions for input and output
 pub enum CompressionFormat {
     Gzip, // .gz
@@ -45,7 +78,7 @@ impl fmt::Display for CompressionFormat {
     }
 }
 
-pub fn separate_known_extensions_from_name(mut path: &Path) -> (&Path, Vec<CompressionFormat>) {
+pub fn separate_known_extensions_from_name(mut path: &Path) -> (&Path, Vec<Extension>) {
     // // TODO: check for file names with the name of an extension
     // // TODO2: warn the user that currently .tar.gz is a .gz file named .tar
     //
@@ -58,17 +91,17 @@ pub fn separate_known_extensions_from_name(mut path: &Path) -> (&Path, Vec<Compr
 
     // While there is known extensions at the tail, grab them
     while let Some(extension) = path.extension().and_then(OsStr::to_str) {
-        extensions.append(&mut match extension {
-            "tar" => vec![Tar],
-            "tgz" => vec![Gzip, Tar],
-            "tbz" | "tbz2" => vec![Bzip, Tar],
-            "txz" | "tlz" | "tlzma" => vec![Lzma, Tar],
-            "tzst" => vec![Zstd, Tar],
-            "zip" => vec![Zip],
-            "bz" | "bz2" => vec![Bzip],
-            "gz" => vec![Gzip],
-            "xz" | "lzma" | "lz" => vec![Lzma],
-            "zst" => vec![Zstd],
+        extensions.push(match extension {
+            "tar" => Extension::new([Tar], extension),
+            "tgz" => Extension::new([Tar, Gzip], extension),
+            "tbz" | "tbz2" => Extension::new([Tar, Bzip], extension),
+            "txz" | "tlz" | "tlzma" => Extension::new([Tar, Lzma], extension),
+            "tzst" => Extension::new([Tar, Zstd], ".tzst"),
+            "zip" => Extension::new([Zip], extension),
+            "bz" | "bz2" => Extension::new([Bzip], extension),
+            "gz" => Extension::new([Gzip], extension),
+            "xz" | "lzma" | "lz" => Extension::new([Lzma], extension),
+            "zst" => Extension::new([Zstd], extension),
             _ => break,
         });
 
@@ -81,7 +114,7 @@ pub fn separate_known_extensions_from_name(mut path: &Path) -> (&Path, Vec<Compr
     (path, extensions)
 }
 
-pub fn extensions_from_path(path: &Path) -> Vec<CompressionFormat> {
+pub fn extensions_from_path(path: &Path) -> Vec<Extension> {
     let (_, extensions) = separate_known_extensions_from_name(path);
     extensions
 }