perf: deprecate mime_guess in favor of a lighter guessing function. (#1698)

7086cmd · web-flow · commit cbcb5116b327 · 2024-07-26T04:49:50.000Z
I wrapped the `mime`, and `infer` crate to [`mime_more`](https://github.com/7086cmd/mime_more) (benchmark result also in README). It also provides lighter mime guessing functions, with the benchmark from using `mime_guess` in about 289 ns to 37 ns and less rare extensions. Also, the behavior of the Data URL is aligned with esbuild (except the percent decoding).
diff --git a/Cargo.lock b/Cargo.lock
diff --git a/Cargo.toml b/Cargo.toml
@@ -117,11 +117,11 @@ lightningcss       = { version = "1.0.0-alpha.57" }
 memchr             = "2.7.2"
 mimalloc           = "0.1.42"
 mime               = "0.3.17"
-mime_guess         = "2.0.5"
 napi               = { version = "3.0.0-alpha", features = ["async"] }
 napi-build         = { version = "2.1.3" }
 napi-derive        = { version = "3.0.0-alpha", default-features = false, features = ["type-def"] }
 oxc_resolver       = { version = "1.9.0" }
+phf                = "0.11.2"
 rayon              = "1.10.0"
 regex              = "1.10.5"
 rustc-hash         = "2.0.0"
diff --git a/crates/rolldown_utils/Cargo.toml b/crates/rolldown_utils/Cargo.toml
@@ -19,8 +19,8 @@ futures     = { workspace = true }
 indexmap    = { workspace = true }
 infer       = { workspace = true }
 mime        = { workspace = true }
-mime_guess  = { workspace = true }
 oxc         = { workspace = true }
+phf         = { workspace = true }
 regex       = { workspace = true }
 rustc-hash  = { workspace = true }
 sugar_path  = { workspace = true }
diff --git a/crates/rolldown_utils/src/lib.rs b/crates/rolldown_utils/src/lib.rs
@@ -7,11 +7,13 @@ pub mod debug;
 pub mod ecma_script;
 pub mod futures;
 pub mod indexmap;
+pub mod light_guess;
 pub mod mime;
 pub mod path_buf_ext;
 pub mod path_ext;
 pub mod percent_encoding;
 pub mod rayon;
 pub mod rustc_hash;
 pub mod xxhash;
+
 pub use bitset::BitSet;
diff --git a/crates/rolldown_utils/src/light_guess.rs b/crates/rolldown_utils/src/light_guess.rs
@@ -0,0 +1,124 @@
+// Port from https://github.com/7086cmd/mime_more/blob/main/src/light_guess.rs
+use mime::Mime;
+use phf::{phf_map, Map};
+use std::str::FromStr;
+
+pub static MIME_TYPES: Map<&'static str, &'static str> = phf_map! {
+    // Text
+    "txt" => "text/plain",
+    "css" => "text/css",
+    "htm" => "text/html",
+    "html" => "text/html",
+    "js" => "text/javascript",
+    "mjs" => "text/javascript",
+    "jsx" => "text/javascript",
+    "json" => "application/json",
+    "yaml" => "text/x-yaml",
+    "yml" => "text/x-yaml",
+    "toml" => "text/x-toml",
+    "markdown" => "text/markdown",
+    "md" => "text/markdown",
+    "xml" => "text/xml",
+    "csv" => "text/csv",
+    "tsv" => "text/tab-separated-values",
+    // Images
+    "bmp" => "image/bmp",
+    "avif" => "image/avif",
+    "gif" => "image/gif",
+    "ico" => "image/x-icon",
+    "icns" => "image/x-icns",
+    "jpg" => "image/jpeg",
+    "jpeg" => "image/jpeg",
+    "png" => "image/png",
+    "svg" => "image/svg+xml",
+    "webp" => "image/webp",
+    // Fonts
+    "otf" => "font/otf",
+    "ttf" => "font/ttf",
+    "ttc" => "font/collection",
+    "woff" => "font/woff",
+    "woff2" => "font/woff2",
+    // Audios
+    "aac" => "audio/aac",
+    "midi" => "audio/midi",
+    "mid" => "audio/midi",
+    "mp3" => "audio/mpeg",
+    "ogg" => "audio/ogg",
+    "oga" => "audio/ogg",
+    "wav" => "audio/wav",
+    "weba" => "audio/webm",
+    "flac" => "audio/flac",
+    "m3u8" => "audio/x-mpegurl",
+    "m4a" => "audio/m4a",
+    // Videos
+    "avi" => "video/x-msvideo",
+    "mpeg" => "video/mpeg",
+    "ogv" => "video/ogg",
+    "ivf" => "video/x-ivf",
+    "webm" => "video/webm",
+    "mp4" => "video/mp4",
+    "flv" => "video/x-flv",
+    "ts" => "audio/vnd.dlna.mpeg-tts", // Though I write TypeScript, this is not TypeScript
+    "mov" => "video/quicktime",
+    "wmv" => "video/x-ms-wmv",
+    // Other
+    "pdf" => "application/pdf",
+    "wasm" => "application/wasm",
+    "webmanifest" => "application/manifest+json",
+};
+
+/// Adapted from:
+/// - https://github.com/rolldown/rolldown/pull/1406/files#diff-4b612e077c82ae0e05e50eb0d419e02c05a04b83c6ac5440c0d0c9d0c38af942
+/// - https://github.com/evanw/esbuild/blob/fc37c2fa9de2ad77476a6d4a8f1516196b90187e/internal/helpers/mime.go#L5
+///
+/// Thanks to @ikkz and @evanw for the inspiration.
+pub fn mime_type_by_extension(ext: &str) -> Option<&'static str> {
+  MIME_TYPES.get(ext).copied()
+}
+
+pub fn try_from_ext(ext: &str) -> anyhow::Result<Mime> {
+  mime_type_by_extension(ext)
+    .ok_or_else(|| anyhow::Error::msg(format!("No mime type found for extension: {ext}")))
+    .and_then(|mime| Ok(Mime::from_str(mime)?))
+}
+
+pub fn try_from_path(path: &std::path::Path) -> anyhow::Result<Mime> {
+  if let Some(ext) = path.extension().and_then(|ext| ext.to_str()) {
+    try_from_ext(ext)
+  } else {
+    anyhow::bail!("No extension found for path: {:?}", path);
+  }
+}
+
+#[cfg(test)]
+mod tests {
+  use super::*;
+
+  #[test]
+  fn normal_extensions() {
+    assert_eq!(mime_type_by_extension("txt").unwrap(), "text/plain");
+    assert_eq!(mime_type_by_extension("css").unwrap(), "text/css");
+    assert_eq!(mime_type_by_extension("html").unwrap(), "text/html");
+    assert_eq!(mime_type_by_extension("json").unwrap(), "application/json");
+    assert_eq!(mime_type_by_extension("png").unwrap(), "image/png");
+    assert_eq!(mime_type_by_extension("svg").unwrap(), "image/svg+xml");
+    assert_eq!(mime_type_by_extension("woff2").unwrap(), "font/woff2");
+    assert_eq!(mime_type_by_extension("aac").unwrap(), "audio/aac");
+    assert_eq!(mime_type_by_extension("avi").unwrap(), "video/x-msvideo");
+    assert_eq!(mime_type_by_extension("pdf").unwrap(), "application/pdf");
+    assert_eq!(mime_type_by_extension("wasm").unwrap(), "application/wasm");
+    assert_eq!(mime_type_by_extension("webmanifest").unwrap(), "application/manifest+json");
+  }
+
+  #[test]
+  fn unknown_extensions() {
+    assert!(mime_type_by_extension("unknown").is_none());
+  }
+
+  #[test]
+  fn try_from_exts() {
+    assert!(matches!(try_from_ext("png").unwrap().subtype(), mime::PNG));
+    assert!(matches!(try_from_ext("svg").unwrap().subtype(), mime::SVG));
+    assert!(matches!(try_from_ext("woff2").unwrap().type_(), mime::FONT));
+  }
+}
diff --git a/crates/rolldown_utils/src/mime.rs b/crates/rolldown_utils/src/mime.rs
@@ -1,3 +1,4 @@
+use crate::light_guess;
 use mime::Mime;
 use std::{path::Path, str::FromStr};
 
@@ -6,7 +7,7 @@ fn is_texture(data: &[u8]) -> bool {
 }
 
 pub fn guess_mime(path: &Path, data: &[u8]) -> anyhow::Result<Mime> {
-  if let Some(guessed) = mime_guess::from_path(path).first() {
+  if let Ok(guessed) = light_guess::try_from_path(path) {
     return Ok(guessed);
   }
 
diff --git a/cspell.json b/cspell.json
@@ -21,6 +21,7 @@
     "/CHANGELOG.md",
     "/packages/rolldown/src/rollup.d.ts",
     "/crates/rolldown/tests/esbuild/splitting/ignore-test.md",
+    "/crates/rolldown_utils/src/light_guess.rs",
     "/pnpm-lock.yaml",
     "*.wasm"
   ],

Original file line number	Diff line number	Diff line change
`@@ -1,3 +1,4 @@`
	`1`	`+use crate::light_guess;`
`1`	`2`	`use mime::Mime;`
`2`	`3`	`use std::{path::Path, str::FromStr};`
`3`	`4`
`@@ -6,7 +7,7 @@ fn is_texture(data: &[u8]) -> bool {`
`6`	`7`	`}`
`7`	`8`
`8`	`9`	`pub fn guess_mime(path: &Path, data: &[u8]) -> anyhow::Result<Mime> {`
`9`		`- if let Some(guessed) = mime_guess::from_path(path).first() {`
	`10`	`+ if let Ok(guessed) = light_guess::try_from_path(path) {`
`10`	`11`	`return Ok(guessed);`
`11`	`12`	`}`
`12`	`13`