fif/src/mimedb.rs

#[cfg(any(all(not(target_os = "linux"), not(feature = "xdg-mime-backend")), all(target_os = "linux", feature = "infer-backend")))]
use std::str::FromStr;

use mime_guess::Mime;

pub trait MimeDb {
	fn init() -> Self;
	fn get_type(&self, data: &[u8]) -> Option<Mime>;
}

#[cfg(any(all(not(target_os = "linux"), not(feature = "xdg-mime-backend")), all(target_os = "linux", feature = "infer-backend")))]
pub struct InferDb {
	db: infer::Infer,
}

#[cfg(any(all(not(target_os = "linux"), not(feature = "xdg-mime-backend")), all(target_os = "linux", feature = "infer-backend")))]
fn open_document_check(buf: &[u8], kind: &str) -> bool {
	let mime = format!("application/vnd.oasis.opendocument.{}", kind);
	let mime = mime.as_bytes();

	buf.len() > 38 + mime.len() && buf.starts_with(b"PK\x03\x04") && buf[38..mime.len() + 38] == mime[..]
}

#[cfg(any(all(not(target_os = "linux"), not(feature = "xdg-mime-backend")), all(target_os = "linux", feature = "infer-backend")))]
impl MimeDb for InferDb {
	fn init() -> Self {
		let mut info = infer::Infer::new();

		// jpeg2000 support because why the stinch not
		info.add("image/jpeg2000", ".jp2", |buf| {
			buf.len() > 23 && buf[..23] == b"\x00\x00\x00\x0C\x6A\x50\x20\x20\x0D\x0A\x87\x0A\x6A\x70\x32\x20"[..]
		});

		info.add("application/vnd.oasis.opendocument.text", "odt", |buf| {
			open_document_check(buf, "text")
		});

		info.add("application/vnd.oasis.opendocument.spreadsheet", "ods", |buf| {
			open_document_check(buf, "spreadsheet")
		});

		info.add("application/vnd.oasis.opendocument.presentation", "odp", |buf| {
			open_document_check(buf, "presentation")
		});

		info.add("image/svg+xml", "svg", |buf| {
			// before doing the moderately expensive SVG check, we should make sure that the input is actually SGML-ish
			// by "SGML-ish", i mean starts with anywhere from zero to ∞-1 whitespace characters, and then a less than sign,
			// and then there's some other stuff we don't care about right now

			// so, here comes our fancy pants """""SGML-ish validator"""""
			for c in buf {
				match c {
					// whitespace (according to https://www.w3.org/TR/xml/#NT-S)
					b'\t' | b'\r' | b'\n' | b'\x20' => continue,
					b'<' => break,
					_ => return false,
				}
			}

			// finally, to check whether or not the file is an SVG:
			// - split the buffer up into chunks separated by the less than sign
			// - check to see if this chunk starts with any of these identifiers:
			let identifiers: Vec<&[u8]> = vec![b"svg", b"SVG", b"!DOCTYPE svg", b"!DOCTYPE SVG"];
			// - if it does, the nested `any` will short circuit and immediately return true, causing the parent `any` to do
			//   the same
			// - and finally, if none of the chunks match, we'll return false

			// TODO: this is kind of messy, i'd like to clean it up somehow :(
			buf
				.split(|c| *c == b'<')
				.any(|buf| identifiers.iter().any(|id| buf.starts_with(id)))
		});

		// unmut
		let info = info;

		Self { db: info }
	}

	fn get_type(&self, data: &[u8]) -> Option<Mime> {
		self.db.get(data).map(|f| Mime::from_str(f.mime_type()).unwrap())
	}
}

#[cfg(any(all(target_os = "linux", not(feature = "infer-backend")), all(not(target_os = "linux"), not(feature = "xdg-mime-backend"))))]
pub struct XdgDb {
	db: xdg_mime::SharedMimeInfo,
}

#[cfg(any(all(target_os = "linux", not(feature = "infer-backend")), all(not(target_os = "linux"), not(feature = "xdg-mime-backend"))))]
impl MimeDb for XdgDb {
	fn init() -> Self {
		Self {
			db: xdg_mime::SharedMimeInfo::new(),
		}
	}

	fn get_type(&self, data: &[u8]) -> Option<Mime> {
		self.db.get_mime_type_for_data(&data).map(|m| m.0)
	}
}