fif 0.8.0

// SPDX-FileCopyrightText: 2021-2024 Lynnesbian
// SPDX-License-Identifier: GPL-3.0-or-later

use std::collections::{BTreeMap, HashMap};
use std::ffi::OsStr;
use std::path::{Path, PathBuf};

use clap::Parser;
use fif::files::{mime_extension_lookup, scan_directory, scan_from_walkdir, BUF_SIZE};
use fif::findings::Findings;
use fif::formats::{Format, PowerShell, Shell};
use fif::mime_db::MimeDb;
use fif::utils::APPLICATION_ZIP;
use fif::{String, MIMEDB};
use itertools::Itertools;
use maplit::{btreeset, hashmap};
use mime::{Mime, APPLICATION_OCTET_STREAM, APPLICATION_PDF, IMAGE_JPEG, IMAGE_PNG};

use crate::parameters::ExtensionSet;
use crate::parameters::Parameters;

const JPEG_BYTES: &[u8] = b"\xFF\xD8\xFF";
const PNG_BYTES: &[u8] = b"\x89\x50\x4E\x47\x0D\x0A\x1A\x0A";
const PDF_BYTES: &[u8] = b"%PDF-";
const ZIP_BYTES: &[u8] = b"PK\x03\x04";

#[test]
/// Ensure that `extension_from_path` successfully returns the extension from a set of paths.
fn get_ext() {
	let ext_checks: HashMap<_, Option<&OsStr>> = hashmap![
		Path::new("test.txt") => Some(OsStr::new("txt")),
		Path::new("test.zip") => Some(OsStr::new("zip")),
		Path::new("test.tar.gz") => Some(OsStr::new("gz")),
		Path::new("test.") => Some(OsStr::new("")),
		Path::new("test") => None,
		Path::new(".hidden") => None,
	];

	for (path, ext) in ext_checks {
		assert_eq!(path.extension(), ext);
	}
}

#[test]
/// Ensure that the MIME types for JPEG, PNG, PDF, and ZIP are detected from their magic numbers.
fn detect_type() {
	assert_eq!(MIMEDB.get_type(JPEG_BYTES), Some(IMAGE_JPEG));
	assert_eq!(MIMEDB.get_type(PNG_BYTES), Some(IMAGE_PNG));
	assert_eq!(MIMEDB.get_type(PDF_BYTES), Some(APPLICATION_PDF));
	assert_eq!(MIMEDB.get_type(ZIP_BYTES), Some(APPLICATION_ZIP.clone()));
}

#[test]
/// Ensure that `mime_extension_lookup` works as expected, and that the set of extensions for JPEG, PNG, PDF, and ZIP
/// contain "jpg", "png", "pdf", and "zip", respectively.
fn recommend_ext() {
	let tests = hashmap![
		&IMAGE_JPEG => "jpg",
		&IMAGE_PNG => "png",
		&APPLICATION_PDF => "pdf",
		&*APPLICATION_ZIP => "zip",
	];

	for (mime, ext) in tests {
		assert!(
			mime_extension_lookup(mime.essence_str().into())
				.unwrap()
				.contains(&String::from(ext)),
			"mime_extension_lookup for {} didn't contain {}!",
			mime.essence_str(),
			ext
		);
	}
}

#[test]
/// Create a simple directory with some files, run `scan_directory` on it, and ensure that the files have their
/// associated MIME types correctly deduced.
fn simple_directory() {
	use std::borrow::Borrow;
	use std::env::set_current_dir;
	use std::fs::{canonicalize, File};
	use std::io::Write;

	use tempfile::tempdir;

	use crate::parameters::ScanOpts;

	// set of files to scan. all but the last files have magic numbers corresponding to their extension, except for
	// "wrong.jpg", which is actually a png.
	let files = hashmap![
		"test.jpg" => JPEG_BYTES,
		"test.jpeg" => JPEG_BYTES,
		"test.png" => PNG_BYTES,
		"test.pdf" => PDF_BYTES,
		"test.zip" => ZIP_BYTES,
		"wrong.jpg" => PNG_BYTES,
		"ignore.fake_ext" => ZIP_BYTES,
	];

	let dir = tempdir().expect("Failed to create temporary directory.");
	set_current_dir(dir.path()).expect("Failed to change directory.");

	for (name, bytes) in &files {
		let mut file = File::create(dir.path().join(name)).unwrap_or_else(|_| panic!("Failed to create file: {name}"));

		file
			.write_all(bytes)
			.unwrap_or_else(|_| panic!("Failed to write to file: {name}"));
		drop(file);
	}

	let scan_opts = ScanOpts {
		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.");

	// there should be one file missing: "ignore.fake_ext"
	assert_eq!(entries.len(), files.len() - 1);

	let use_threads = cfg!(feature = "multi-threaded");

	let results = scan_from_walkdir(&entries, false, use_threads).0;
	let canonical_results = scan_from_walkdir(&entries, true, use_threads).0;
	assert_eq!(results.len(), canonical_results.len());

	for (result, canonical_result) in results.iter().zip(canonical_results.iter()) {
		// there should be no IO errors during this test. any IO errors encountered are outside the scope of this test.

		// paths should be canonical
		assert_eq!(canonicalize(&result.file).unwrap(), canonical_result.file);

		if !result.valid {
			// the only invalid file detected should be "wrong.jpg", which is a misnamed png file
			// 1. ensure detected extension is "jpg"
			assert_eq!(result.file.as_path().extension().unwrap(), OsStr::new("jpg"));
			// 2. ensure detected MIME type is IMAGE_PNG
			assert_eq!(result.mime, IMAGE_PNG);
			// 3. ensure the recommended extension for "wrong.jpg" is "png"
			assert_eq!(&result.recommended_extension().unwrap(), &String::from("png"));
			// 4. ensure the recommended filename for "wrong.jpg" is "wrong.png"
			assert_eq!(result.recommended_path().unwrap().file_name(), Some(OsStr::new("wrong.png")));
			continue;
		}

		// check if the recommended extension for this file is in the list of known extensions for its MIME type - for
		// example, if the file is determined to be an IMAGE_PNG, its recommended extension should be one of the extensions
		// returned by `mime_extension_lookup(IMAGE_PNG)`.
		assert!(mime_extension_lookup(result.mime.essence_str().into())
			.unwrap()
			.contains(&result.recommended_extension().unwrap()));

		// ensure that the recommended_name function outputs something beginning with "test"
		assert!(result
			.recommended_path()
			.unwrap()
			.file_name()
			.unwrap()
			.to_string_lossy()
			.starts_with("test"));

		// make sure the guessed MIME type is correct based on the extension of the scanned file
		// because we already know that the extensions match the MIME type (as we created these files ourselves earlier in
		// the test), all files with the "jpg" extension should be IMAGE_JPEGs, etc.
		let ext = result.file.as_path().extension().unwrap();
		assert_eq!(
			result.mime,
			match ext.to_string_lossy().borrow() {
				"jpg" | "jpeg" => IMAGE_JPEG,
				"png" => IMAGE_PNG,
				"pdf" => APPLICATION_PDF,
				"zip" => APPLICATION_ZIP.clone(),
				_ => APPLICATION_OCTET_STREAM, // general "fallback" type
			},
			"Incorrect MIME type detected - got {:?} for a {:?} file",
			result.mime,
			ext
		);
	}
}

#[test]
/// Ensure that command line argument parsing works correctly - flags are interpreted, booleans are set, and so on.
fn argument_parsing() {
	use crate::parameters::ScanOpts;

	// pass `-f`, which enables following symlinks, and `-E images`, which scans files with image extensions
	let args: Parameters = Parameters::parse_from(vec!["fif", "-f", "-E", "images"]);

	// check if "jpg" is in the list of extensions to be scanned
	assert!(
		args
			.extensions()
			.expect("args.extensions() should be Some(_)!")
			.contains(&"jpg"),
		"args.extensions() should contain the `images` set!"
	);

	// make sure "scan_hidden" is false
	assert!(!args.scan_hidden);

	// exts should be none
	assert!(args.exts.is_none());

	// there shouldn't be any excluded extensions
	assert!(args.excluded_extensions().is_none());

	// get the ScanOpts, and make sure they match expectations
	assert_eq!(
		args.get_scan_opts(),
		ScanOpts {
			hidden: false,
			extensionless: false,
			follow_symlinks: true,
			ignore_unknown_exts: false,
		},
		"ScanOpts are incorrect"
	);
}

#[test]
/// Ensure that `fif -e jpg dir` is interpreted as "scan for jpg files in dir" and not "scan for jpg and dir files"
fn positional_args() {
	for flag in ["-x", "-e", "-X", "-E"] {
		assert_eq!(
			Parameters::parse_from(vec!["fif", flag, "images", "directory"]).dir,
			PathBuf::from("directory")
		);
	}
}

#[test]
/// Ensure the `exclude` flag (`-x`) overrides `-e` and `-E`.
fn exclude_overrides() {
	// pass `-E images`, which includes many image extensions, and `-x jpg,png`, which should remove "jpg" and "png" from
	// the extensions list
	let args: Parameters = Parameters::parse_from(vec!["fif", "-x", "jpg,png", "-E", "images"]);
	let extensions = args.extensions();
	assert!(extensions.is_some(), "Extensions should contain the `images` set!");
	let extensions = extensions.unwrap();

	assert!(!extensions.contains(&"jpg"), "\"jpg\" should be excluded!");
	assert!(!extensions.contains(&"png"), "\"png\" should be excluded!");
	assert!(extensions.contains(&"jpeg"), "\"jpeg\" should be included!");

	// pass `-e abc,def,ghi,jkl` and `-x abc,def` -- extensions() should only contain "ghi" and "jkl"
	let args: Parameters = Parameters::parse_from(vec!["fif", "-e", "abc,def,ghi,jkl", "-x", "abc,def"]);
	let extensions = args.extensions();
	assert!(extensions.is_some(), "Extensions should be set!");
	assert_eq!(extensions, Some(btreeset!["ghi", "jkl"]));
}

#[test]
/// Ensure the `exclude_set` flag (`-X`) overrides `-e`.
fn exclude_set_overrides_includes() {
	// pass `-e jpg,flac` and `-X images` -- which should produce the equivalent of `-e flag`
	let args: Parameters = Parameters::parse_from(vec!["fif", "-e", "jpg,flac", "-X", "images"]);
	let extensions = args.extensions();
	assert!(extensions.is_some(), "Extensions should be set!");
	assert_eq!(extensions, Some(btreeset!["flac"]));
}

#[test]
/// Ensure the `exclude_set` flag (`-X`) overrides `-E`.
fn exclude_set_overrides_include_set() {
	// pass `-E media` and `-X images` -- which should produce the equivalent of `-E audio,video`
	let args: Parameters = Parameters::parse_from(vec!["fif", "-E", "media", "-X", "images"]);
	let extensions = args.extensions();
	assert!(extensions.is_some(), "Extensions should be set!");
	let extensions = extensions.unwrap();

	// ensure all of audio and video's extensions are here
	for &ext in ExtensionSet::Audio
		.extensions()
		.iter()
		.chain(ExtensionSet::Video.extensions().iter())
	{
		assert!(extensions.contains(&ext), "Extensions should contain {ext}!");
	}

	// ensure all of images' extensions are excluded
	for ext in ExtensionSet::Images.extensions() {
		assert!(!extensions.contains(&ext), "Extensions should not contain {ext}!");
	}
}

#[test]
/// Generate random series of bytes and try to identify them. This test makes no assertions and can only fail if the
/// mime database somehow panics or hangs.
fn identify_random_bytes() {
	use rand::prelude::*;
	let mut rng = rand::rng();
	let mut bytes: [u8; BUF_SIZE * 2] = [0; BUF_SIZE * 2];
	let mut results: BTreeMap<Mime, i32> = BTreeMap::new();

	for _ in 1..1000 {
		rng.fill_bytes(&mut bytes);
		if let Some(detected_type) = MIMEDB.get_type(&bytes) {
			*results.entry(detected_type).or_insert(0) += 1;
		}
	}

	for (mime, count) in &results {
		println!("{mime}:\t{count} counts");
	}
	println!("No type found:\t{} counts", 1000 - results.values().sum::<i32>());
}

#[test]
/// Ensure that, for a given file "wrong.bad", which should have extension "good", the shell output contains something
/// like "mv wrong.bad wrong.good".
fn outputs_move_commands() {
	use std::io::Read;

	// create an example finding stating that "misnamed_file.png" has been identified as a jpeg file
	let findings = vec![Findings {
		file: Path::new("misnamed_file.png").to_path_buf(),
		valid: false,
		mime: IMAGE_JPEG,
	}];

	for format in &["Shell", "PowerShell"] {
		let mut cursor = std::io::Cursor::new(Vec::new());
		let mut contents = std::string::String::new();

		match *format {
			"Shell" => Shell.write_all(&mut cursor, &findings, &[]),
			"PowerShell" => PowerShell.write_all(&mut cursor, &findings, &[]),
			_ => unreachable!(),
		}
		.expect("Failed to write to cursor");

		cursor.set_position(0);
		cursor
			.read_to_string(&mut contents)
			.expect("Failed to read from cursor to string");

		// 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.png"),
			"{format} output doesn't contain move command!\n===\n{contents}"
		);
	}
}

#[test]
#[cfg(feature = "json")]
/// Ensure JSON output is valid.
fn test_json() {
	use std::io::Read;

	use crate::formats::Json;
	// create an example finding stating that "misnamed_file.png" has been identified as a jpeg file
	let findings = vec![Findings {
		file: Path::new("misnamed_file.png").to_path_buf(),
		valid: false,
		mime: IMAGE_JPEG,
	}];

	let mut cursor = std::io::Cursor::new(Vec::new());
	let mut contents = std::string::String::new();

	Json
		.write_all(&mut cursor, &findings, &[])
		.expect("Failed to write to cursor");

	cursor.set_position(0);
	cursor
		.read_to_string(&mut contents)
		.expect("Failed to read from cursor to string");

	// the output should contain the file's MIME type
	assert!(
		contents.contains(IMAGE_JPEG.essence_str()),
		"JSON output doesn't contain move command!\n===\n{contents}"
	);
}

#[test]
/// Ensure that the Media extension set contains all (is a superset) of Audio, Video, and Images.
fn media_contains_audio_video_images() {
	use crate::parameters::ExtensionSet::{Audio, Images, Media, Video};
	let media_exts = Media.extensions();

	// assert every extension in the audio/video/image sets is contained in the media set
	[Audio.extensions(), Video.extensions(), Images.extensions()]
		.concat()
		.into_iter()
		.for_each(|ext| assert!(media_exts.contains(&ext)));

	assert_eq!(
		Parameters::parse_from(["fif", "-E", "media"]).extensions(),
		Parameters::parse_from(["fif", "-E", "audio,video,images"]).extensions()
	);
}

#[test]
/// Ensure that the `writables!` and `writablesln!` macros produce the output they should.
fn writables_is_correct() {
	use fif::formats::Writable;
	use fif::{writables, writablesln};

	assert_eq!(
		&["henlo".into(), Path::new("henlo").into(), Writable::Newline,],
		writables!["henlo", (Path::new("henlo")), Newline]
	);

	assert_eq!(
		&["henlo".into(), Path::new("henlo").into(), Writable::Newline, Writable::Newline],
		writablesln!["henlo", (Path::new("henlo")), Newline]
	);
}

#[test]
/// Test various combinations of verbosity flags.
fn verbosity() {
	use log::LevelFilter;
	assert!(
		Parameters::try_parse_from(["fif", "-q", "-v"]).is_err(),
		"Failed to reject usage of both -q and -v!"
	);

	let expected_results = hashmap![
		"-qqqqqqqq" => LevelFilter::Off,
		"-qqq" => LevelFilter::Off,
		"-qq" => LevelFilter::Error,
		"-q" => LevelFilter::Warn,
		"-s" => LevelFilter::Info,
		"-v" => LevelFilter::Debug,
		"-vv" => LevelFilter::Trace,
		"-vvv" => LevelFilter::Trace,
		"-vvvvvvvv" => LevelFilter::Trace,
	];

	for (flags, level) in expected_results {
		assert_eq!(Parameters::parse_from(["fif", flags]).get_verbosity(), level);
	}
}

#[test]
/// Ensures `os_name()`'s output is the same as [`std::env::consts::OS`], capitalisation notwithstanding
fn validate_os_name() {
	assert_eq!(fif::utils::os_name().to_lowercase(), std::env::consts::OS.to_lowercase());
}

#[test]
/// Ensures that [`Findings`] are sorted properly.
fn sort_findings() {
	let findings = [Findings {
			file: Path::new("ccc").to_path_buf(),
			valid: false,
			mime: IMAGE_JPEG,
		},
		Findings {
			file: Path::new("bbb.xyz").to_path_buf(),
			valid: true,
			mime: IMAGE_PNG,
		},
		Findings {
			file: Path::new("aaa").to_path_buf(),
			valid: true,
			mime: APPLICATION_PDF,
		}];
	let mut findings = findings.iter().sorted_unstable();

	assert_eq!(findings.next().unwrap().file, Path::new("aaa"));
	assert_eq!(findings.next().unwrap().file, Path::new("bbb.xyz"));
	assert_eq!(findings.next().unwrap().file, Path::new("ccc"));
	assert_eq!(findings.next(), None);
}