fif/src/mime_db.rs

114 lines
3.4 KiB
Rust

//! Backend-neutral Mime database implementation.
use cfg_if::cfg_if;
use mime::Mime;
pub trait MimeDb {
/// Initialise the database.
fn init() -> Self;
/// Given a slice of bytes, returns the inferred mimetype, if any.
fn get_type(&self, data: &[u8]) -> Option<Mime>;
}
cfg_if! {
if #[cfg(any(all(unix, feature = "infer-backend"), all(not(unix), not(feature = "xdg-mime-backend"))))] {
use std::str::FromStr;
pub struct InferDb {
db: infer::Infer,
}
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[..]
}
impl MimeDb for InferDb {
fn init() -> Self {
let mut info = infer::Infer::new();
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("application/x-rpa", "rpa", |buf| {
buf.len() >= 34 && buf.starts_with(b"RPA-") && buf[7] == b' ' && buf[24] ==b' '
});
info.add("audio/x-aiff", "aiff", |buf| {
// as added by https://github.com/bojand/infer/pull/48/files
// this should be removed when (if) that PR is accepted
buf.len() > 11
&& buf[0] == 0x46
&& buf[1] == 0x4F
&& buf[2] == 0x52
&& buf[3] == 0x4D
&& buf[8] == 0x41
&& buf[9] == 0x49
&& buf[10] == 0x46
&& buf[11] == 0x46
});
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 which 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)))
});
Self { db: info }
}
fn get_type(&self, data: &[u8]) -> Option<Mime> {
self.db.get(data).map(|f| Mime::from_str(f.mime_type()).unwrap())
}
}
} else {
pub struct XdgDb {
db: xdg_mime::SharedMimeInfo,
}
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)
}
}
}
}