Merge pull request #132 from http-rs/infer-type

Add mime sniffing to Body::from_file

Author: yoshuawuyts

Cargo.toml

@@ -42,3 +42,4 @@ serde_urlencoded = "0.6.1"
 
 [dev-dependencies]
 http = "0.2.0"
+async-std = { version = "1.4.0", features = ["unstable", "attributes"] }

src/body.rs

@@ -358,12 +358,26 @@ impl Body {
     /// # Ok(()) }) }
     /// ```
     #[cfg(feature = "async_std")]
-    pub async fn from_file<P>(file: P) -> io::Result<Self>
+    pub async fn from_file<P>(path: P) -> io::Result<Self>
     where
         P: AsRef<Path>,
     {
-        let file = fs::read(file.as_ref()).await?;
-        Ok(file.into())
+        let path = path.as_ref();
+        let mut file = fs::File::open(path).await?;
+        let len = file.metadata().await?.len();
+
+        // Look at magic bytes first, look at extension second, fall back to
+        // octet stream.
+        let mime = peek_mime(&mut file)
+            .await?
+            .or_else(|| guess_ext(path))
+            .unwrap_or(mime::BYTE_STREAM);
+
+        Ok(Self {
+            mime,
+            length: Some(len as usize),
+            reader: Box::new(io::BufReader::new(file)),
+        })
     }
 
     /// Get the length of the body in bytes.
@@ -448,3 +462,32 @@ impl BufRead for Body {
         Pin::new(&mut self.reader).consume(amt)
     }
 }
+
+/// Look at first few bytes of a file to determine the mime type.
+/// This is used for various binary formats such as images and videos.
+#[cfg(feature = "async_std")]
+async fn peek_mime(file: &mut async_std::fs::File) -> io::Result<Option<Mime>> {
+    // We need to read the first 300 bytes to correctly infer formats such as tar.
+    let mut buf = [0_u8; 300];
+    file.read(&mut buf).await?;
+    let mime = Mime::sniff(&buf).ok();
+
+    // Reset the file cursor back to the start.
+    file.seek(io::SeekFrom::Start(0)).await?;
+    Ok(mime)
+}
+
+/// Look at the extension of a file to determine the mime type.
+/// This is useful for plain-text formats such as HTML and CSS.
+#[cfg(feature = "async_std")]
+fn guess_ext(path: &Path) -> Option<Mime> {
+    let ext = path.extension().map(|p| p.to_str()).flatten();
+    match ext {
+        Some("html") => Some(mime::HTML),
+        Some("js") | Some("mjs") | Some("jsonp") => Some(mime::JAVASCRIPT),
+        Some("json") => Some(mime::JSON),
+        Some("css") => Some(mime::CSS),
+        Some("svg") => Some(mime::SVG),
+        None | Some(_) => None,
+    }
+}

src/mime/constants.rs

@@ -86,6 +86,77 @@ pub const HTML: Mime = Mime {
     static_subtype: Some("html"),
 };
 
+/// Content-Type for SVG.
+///
+/// # Mime Type
+///
+/// ```txt
+/// image/svg+xml
+/// ```
+pub const SVG: Mime = Mime {
+    static_essence: Some("image/svg+xml"),
+    essence: String::new(),
+    basetype: String::new(),
+    subtype: String::new(),
+    params: None,
+    static_basetype: Some("image"),
+    static_subtype: Some("svg+xml"),
+};
+
+/// Content-Type for ICO icons.
+///
+/// # Mime Type
+///
+/// ```txt
+/// image/x-icon
+/// ```
+// There are multiple `.ico` mime types known, but `image/x-icon`
+// is what most browser use. See:
+// https://en.wikipedia.org/wiki/ICO_%28file_format%29#MIME_type
+pub const ICO: Mime = Mime {
+    static_essence: Some("image/x-icon"),
+    essence: String::new(),
+    basetype: String::new(),
+    subtype: String::new(),
+    params: None,
+    static_basetype: Some("image"),
+    static_subtype: Some("x-icon"),
+};
+
+/// Content-Type for PNG images.
+///
+/// # Mime Type
+///
+/// ```txt
+/// image/png
+/// ```
+pub const PNG: Mime = Mime {
+    static_essence: Some("image/png"),
+    essence: String::new(),
+    basetype: String::new(),
+    subtype: String::new(),
+    params: None,
+    static_basetype: Some("image"),
+    static_subtype: Some("png"),
+};
+
+/// Content-Type for JPEG images.
+///
+/// # Mime Type
+///
+/// ```txt
+/// image/jpeg
+/// ```
+pub const JPEG: Mime = Mime {
+    static_essence: Some("image/jpeg"),
+    essence: String::new(),
+    basetype: String::new(),
+    subtype: String::new(),
+    params: None,
+    static_basetype: Some("image"),
+    static_subtype: Some("jpeg"),
+};
+
 /// Content-Type for Server Sent Events
 ///
 /// # Mime Type
@@ -170,3 +241,20 @@ pub const MULTIPART_FORM: Mime = Mime {
     static_subtype: Some("form-data"),
     params: None,
 };
+
+/// Content-Type for webassembly.
+///
+/// # Mime Type
+///
+/// ```txt
+/// application/wasm
+/// ```
+pub const WASM: Mime = Mime {
+    static_essence: Some("application/wasm"),
+    essence: String::new(),
+    basetype: String::new(),
+    subtype: String::new(),
+    static_basetype: Some("application"),
+    static_subtype: Some("wasm"),
+    params: None,
+};

src/mime/mod.rs

@@ -98,6 +98,20 @@ impl Mime {
     }
 }
 
+impl PartialEq<Mime> for Mime {
+    fn eq(&self, other: &Mime) -> bool {
+        let left = match self.static_essence {
+            Some(essence) => essence,
+            None => &self.essence,
+        };
+        let right = match other.static_essence {
+            Some(essence) => essence,
+            None => &other.essence,
+        };
+        left == right
+    }
+}
+
 impl Display for Mime {
     fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
         parse::format(self, f)
@@ -188,7 +202,7 @@ impl PartialEq<str> for ParamValue {
 
 /// This is a hack that allows us to mark a trait as utf8 during compilation. We
 /// can remove this once we can construct HashMap during compilation.
-#[derive(Debug, Clone)]
+#[derive(Debug, Clone, PartialEq, Eq)]
 pub(crate) enum ParamKind {
     Utf8,
     Vec(Vec<(ParamName, ParamValue)>),

src/response.rs

@@ -367,6 +367,11 @@ impl Response {
         }
     }
 
+    /// Get the current content type
+    pub fn content_type(&self) -> Option<Mime> {
+        self.header(CONTENT_TYPE)?.last().as_str().parse().ok()
+    }
+
     /// Get the length of the body stream, if it has been set.
     ///
     /// This value is set when passing a fixed-size object into as the body. E.g. a string, or a

tests/fixtures/empty.custom

@@ -0,0 +1,7 @@
+<html>
+
+<head></head>
+
+<body>This is a fixture!</body>
+
+</html>

tests/fixtures/index.html

@@ -0,0 +1 @@
+this is an unknown text format

tests/fixtures/nori.png

@@ -0,0 +1,44 @@
+use async_std::fs;
+use async_std::io;
+use http_types::{mime, Body, Response};
+
+#[async_std::test]
+async fn guess_plain_text_mime() -> io::Result<()> {
+    let body = Body::from_file("tests/fixtures/index.html").await?;
+    let mut res = Response::new(200);
+    res.set_body(body);
+    assert_eq!(res.content_type(), Some(mime::HTML));
+    Ok(())
+}
+
+#[async_std::test]
+async fn guess_binary_mime() -> http_types::Result<()> {
+    let body = Body::from_file("tests/fixtures/nori.png").await?;
+    let mut res = Response::new(200);
+    res.set_body(body);
+    assert_eq!(res.content_type(), Some(mime::PNG));
+
+    // Assert the file is correctly reset after we've peeked the bytes
+    let left = fs::read("tests/fixtures/nori.png").await?;
+    let right = res.body_bytes().await?;
+    assert_eq!(left, right);
+    Ok(())
+}
+
+#[async_std::test]
+async fn guess_mime_fallback() -> io::Result<()> {
+    let body = Body::from_file("tests/fixtures/unknown.custom").await?;
+    let mut res = Response::new(200);
+    res.set_body(body);
+    assert_eq!(res.content_type(), Some(mime::BYTE_STREAM));
+    Ok(())
+}
+
+#[async_std::test]
+async fn parse_empty_files() -> http_types::Result<()> {
+    let body = Body::from_file("tests/fixtures/empty.custom").await?;
+    let mut res = Response::new(200);
+    res.set_body(body);
+    assert_eq!(res.content_type(), Some(mime::BYTE_STREAM));
+    Ok(())
+}