retrieve extensions from mime_guess, minor code changes, added a readme and such

Cargo.toml

@@ -6,6 +6,10 @@ authors = ["Lynnesbian <lynne@bune.city>"]
 edition = "2018"
 license = "GPL-3.0-or-later"
 rust-version = "1.43.0" # cached requires 1.42.0
+repository = "https://git.bune.city/lynnesbian/fif"
+readme = "README.md"
+keywords = ["mime", "mimetype", "utilities", "tools"]
+categories = ["command-line-utilities"]
 #resolver = "2"
 #license-file = "LICENSE"
 

README.md

@@ -0,0 +1,14 @@
+fif
+===
+A command-line tool for detecting and optionally correcting files with incorrect extensions.
+
+## Installation
+```bash
+cargo install --locked fif
+```
+
+## Usage
+<!-- TODO: documentation 0uo -->
+```bash
+fif --help
+```

src/extensionset.rs

@@ -13,21 +13,16 @@ pub enum ExtensionSet {
 impl ExtensionSet {
 	pub fn extensions(&self) -> Vec<&str> {
 		match self {
-			Self::Images => vec![
+			Self::Images => mime_guess::get_mime_extensions_str("image/*"),
-				"png", "jpg", "jpeg", "webp", "raw", "gif", "apng", "tga", "bmp", "tif", "tiff", "heif", "avif", "jp2", "mng",
+			Self::Videos => mime_guess::get_mime_extensions_str("video/*"),
-				"svg",
+			Self::Audio => mime_guess::get_mime_extensions_str("audio/*"),
-			],
+			Self::Documents => Some(&[
-			Self::Videos => vec![
-				"webm", "mp4", "mkv", "mov", "avi", "m4v", "wmv", "bik", "ogv", "qt", "3gp", "3g2", "divx",
-			],
-			Self::Audio => vec![
-				"ogg", "oga", "opus", "mp3", "m4a", "aac", "flac", "ape", "midi", "mid", "alac", "wav", "aiff", "aa3", "at3",
-			],
-			Self::Documents => vec![
 				"doc", "docx", "ppt", "pptx", "xls", "xlsx", "csv", "tsv", "pdf", "odt", "ods", "odp",
-			],
+			][..]),
-			Self::Archives => vec!["zip", "tar", "gz", "zst", "xz", "rar", "7z", "bz", "bz2"],
+			// many compressed file types follow the name scheme "application/x.+compressed.*" - maybe this can be used
+			// somehow to extract extensions for compressed files from mime_guess?
+			Self::Archives => Some(&["zip", "tar", "gz", "zst", "xz", "rar", "7z", "bz", "bz2"][..]),
 			_ => todo!(),
-		}
+		}.unwrap().to_vec()
 	}
 }

src/formats.rs

@@ -16,7 +16,7 @@ enum Writable<'a> {
 	String(&'a str),
 	Path(&'a PathBuf),
 	Space,
-	Newline
+	Newline,
 }
 
 // the lifetime of a lifetime
@@ -40,11 +40,9 @@ fn smart_write<W: Write>(f: &mut W, writeables: &[Writable]) -> io::Result<()> {
 			Writable::Newline => writeln!(f, )?,
 			Writable::String(s) => write!(f, "{}", s)?,
 			Writable::Path(path) => {
-				match path.to_str() {
+				if let Some(string) = path.to_str() {
-					Some(string) => {
 					write!(f, "{}", escape(string))?
-					}
+				} else {
-					None => {
 					write!(f, "'''")?;
 					#[cfg(unix)]
 						f.write_all(&*path.as_os_str().as_bytes())?;
@@ -56,7 +54,6 @@ fn smart_write<W: Write>(f: &mut W, writeables: &[Writable]) -> io::Result<()> {
 			}
 		}
 	}
-	}
 	Ok(())
 }
 

src/main.rs

@@ -117,10 +117,8 @@ fn scan_file(entry: &DirEntry) -> Result<Findings, (ScanError, PathBuf)> {
 	let valid = match known_exts {
 		// there is a known set of extensions for this mimetype, and the file has an extension
 		Some(e) if entry_ext.is_some() => e.contains(&entry_ext.unwrap().to_lowercase().into()),
-		// there is a known set of extensions for this mimetype, but the file has no extension
+		// either this file has no extension, or there is no known set of extensions for this mimetype :(
-		Some(_) => false,
+		Some(_) | None => false,
-		// there is no known set of extensions for this mimetype :(
-		None => false,
 	};
 
 	Ok(Findings {
@@ -130,7 +128,7 @@ fn scan_file(entry: &DirEntry) -> Result<Findings, (ScanError, PathBuf)> {
 	})
 }
 
-fn scan_from_walkdir(entries: Vec<DirEntry>) -> Vec<Result<Findings, (ScanError, PathBuf)>> {
+fn scan_from_walkdir(entries: &[DirEntry]) -> Vec<Result<Findings, (ScanError, PathBuf)>> {
 	#[cfg(feature = "multi-threaded")]
 	{
 		// rather than using a standard par_iter, split the entries into chunks of 32 first.
@@ -232,7 +230,7 @@ fn main() {
 
 	trace!("Found {} items to check", entries.len());
 
-	let results: Vec<_> = scan_from_walkdir(entries)
+	let results: Vec<_> = scan_from_walkdir(&entries)
 		.into_iter()
 		.filter(
 			|result| result.is_err() || !result.as_ref().unwrap().valid,