ignore_unknown_exts flag, slightly more testing

CHANGELOG.md

@@ -3,6 +3,10 @@ Dates are given in YYYY-MM-DD format.
 
 ## v0.3
 ### v0.3.4 (2021-mm-dd)
+#### Features
+- Added `-I`/`--ignore-unknown-exts` flag for ignoring files with unknown extensions - for example, if fif doesn't know
+  what a ".fake" file is, setting this flag will prevent it from renaming "photo.fake" to "photo.jpg". This is useful
+  for avoiding the case where fif incorrectly mislabels an obscure format it isn't aware of as something else.
 #### Other
 - Refactored `formats.rs`
 - More accurate dependency versions in `Cargo.toml` to ensure that the MSRV stays supported

Cargo.lock

@@ -544,9 +544,9 @@ dependencies = [
 
 [[package]]
 name = "redox_syscall"
-version = "0.2.9"
+version = "0.2.10"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "5ab49abadf3f9e1c4bc499e8845e152ad87d2ad2d30371841171169e9d75feee"
+checksum = "8383f39639269cde97d255a32bdb68c047337295414940c68bdd30c2e13203ff"
 dependencies = [
  "bitflags",
 ]

src/formats.rs

@@ -33,10 +33,6 @@ macro_rules! writables {
 		$crate::formats::Writable::Newline
 	};
 
-	(@do Space) => {
-		$crate::formats::Writable::Space
-	};
-
 	(@do $arg:expr) => {
 		$arg.into()
 	}
@@ -57,7 +53,6 @@ type Entries<'a> = [Result<Findings, ScanError<'a>>];
 pub enum Writable<'a> {
 	String(&'a str),
 	Path(&'a Path),
-	Space,
 	Newline,
 }
 
@@ -80,7 +75,6 @@ fn smart_write<W: Write>(f: &mut W, writeables: &[Writable]) -> io::Result<()> {
 	// ehhhh
 	for writeable in writeables {
 		match writeable {
-			Writable::Space => write!(f, " ")?,
 			Writable::Newline => {
 				cfg_if! {
 					if #[cfg(windows)] {
@@ -189,7 +183,7 @@ impl Format for Shell {
 
 impl FormatSteps for Shell {
 	fn rename<W: Write>(&self, f: &mut W, from: &Path, to: &Path) -> io::Result<()> {
-		smart_write(f, writablesln!("mv -v -i -- ", from, Space, to))
+		smart_write(f, writablesln!("mv -v -i -- ", from, "\t", to))
 	}
 
 	fn no_known_extension<W: Write>(&self, f: &mut W, path: &Path) -> io::Result<()> {

src/main.rs

@@ -34,6 +34,7 @@ use crate::mime_db::MimeDb;
 use crate::parameters::{OutputFormat, ScanOpts};
 use crate::utils::{clap_long_version, os_name};
 use std::collections::BTreeSet;
+use mime_guess::from_ext;
 
 mod findings;
 mod formats;
@@ -217,6 +218,13 @@ fn wanted_file(
 		let ext = ext.to_string_lossy().to_lowercase();
 		let ext = ext.as_str();
 
+		if scan_opts.ignore_unknown_exts {
+			if from_ext(ext).is_empty() {
+				// unknown extension, skip.
+				return false;
+			}
+		}
+
 		if let Some(exts) = exts {
 			// only scan if the file has one of the specified extensions.
 			exts.contains(&ext)

src/parameters.rs

@@ -128,6 +128,11 @@ pub struct Parameters {
 	/// The directory to process.
 	#[clap(name = "DIR", default_value = ".", parse(from_os_str))]
 	pub dir: PathBuf,
+
+	/// Don't rename files with extensions unknown to fif.
+	/// For example, with this option, fif will not rename "image.unknown" to "image.jpg"
+	#[clap(short = 'I', long)]
+	pub ignore_unknown_exts: bool,
 }
 
 fn lowercase_exts(exts: &str) -> Result<(), String> {
@@ -147,6 +152,8 @@ pub struct ScanOpts {
 	pub extensionless: bool,
 	/// Should symlinks be followed?
 	pub follow_symlinks: bool,
+	/// Whether to rename files with unknown extensions.
+	pub ignore_unknown_exts: bool,
 }
 
 impl Parameters {
@@ -214,6 +221,7 @@ impl Parameters {
 			hidden: self.scan_hidden,
 			extensionless: self.scan_extensionless,
 			follow_symlinks: self.follow_symlinks,
+			ignore_unknown_exts: self.ignore_unknown_exts
 		}
 	}
 

src/tests/mod.rs

@@ -100,6 +100,7 @@ fn simple_directory() {
 	files.insert("test.pdf", PDF_BYTES);
 	files.insert("test.zip", ZIP_BYTES);
 	files.insert("wrong.jpg", PNG_BYTES);
+	files.insert("ignore.fake_ext", ZIP_BYTES);
 
 	let dir = tempdir().expect("Failed to create temporary directory.");
 	set_current_dir(dir.path()).expect("Failed to change directory.");
@@ -117,11 +118,13 @@ fn simple_directory() {
 		hidden: true,
 		extensionless: false,
 		follow_symlinks: false,
+		ignore_unknown_exts: true,
 	};
 
 	let entries = scan_directory(dir.path(), None, None, &scan_opts).expect("Directory scan failed.");
 
-	assert_eq!(entries.len(), files.len());
+	// there should be one file missing: "ignore.fake_ext"
+	assert_eq!(entries.len(), files.len() - 1);
 
 	// initialise global mime DB - this is needed because `scan_from_walkdir` expects it to be present.
 	crate::init_db();
@@ -209,6 +212,7 @@ fn argument_parsing() {
 			hidden: false,
 			extensionless: false,
 			follow_symlinks: true,
+			ignore_unknown_exts: false,
 		},
 		"ScanOpts are incorrect"
 	);
@@ -329,7 +333,7 @@ fn identify_random_bytes() {
 	for (mime, count) in &results {
 		println!("{}:\t{} counts", mime, count);
 	}
-	println!("No type found:\t{} counts", 500 - results.values().sum::<i32>());
+	println!("No type found:\t{} counts", results.values().len() as i32 - results.values().sum::<i32>());
 }
 
 #[test]
@@ -363,7 +367,7 @@ fn outputs_move_commands() {
 
 		// the output should contain a command like "mv -i misnamed_file.png misnamed_file.jpg"
 		assert!(
-			contents.contains("misnamed_file.jpg"),
+			contents.contains("misnamed_file.jpg") && contents.contains("misnamed_file.png"),
 			"{} output doesn't contain move command!\n===\n{}",
 			format,
 			contents
@@ -433,9 +437,8 @@ fn writables_is_correct() {
 			"henlo".into(),
 			Path::new("henlo").into(),
 			Writable::Newline,
-			Writable::Space
 		],
-		writables!["henlo", (Path::new("henlo")), Newline, Space]
+		writables!["henlo", (Path::new("henlo")), Newline]
 	);
 }