Trim bloated encoding table

cloudflare · Nov 6, 2024 · e3d52ff · e3d52ff
1 parent 16a4a49
commit e3d52ff
Show file tree

Hide file tree

Showing 2 changed files with 12 additions and 204 deletions.
diff --git a/src/base/encoding.rs b/src/base/encoding.rs
@@ -5,193 +5,31 @@ use std::sync::Arc;
 
 /// This serves as a map from integer to [`Encoding`], which allows more efficient
 /// sets/gets of the [`SharedEncoding`].
-static ALL_ENCODINGS: [&Encoding; 228] = [
-    &encoding_rs::WINDOWS_1252_INIT,
-    &encoding_rs::ISO_8859_2_INIT,
-    &encoding_rs::ISO_8859_3_INIT,
-    &encoding_rs::ISO_8859_4_INIT,
-    &encoding_rs::WINDOWS_1254_INIT,
-    &encoding_rs::ISO_8859_10_INIT,
-    &encoding_rs::ISO_8859_15_INIT,
-    &encoding_rs::IBM866_INIT,
-    &encoding_rs::MACINTOSH_INIT,
-    &encoding_rs::KOI8_R_INIT,
-    &encoding_rs::GBK_INIT,
-    &encoding_rs::BIG5_INIT,
+static ALL_ENCODINGS: [&Encoding; 40] = [
     &encoding_rs::UTF_8_INIT,
-    &encoding_rs::KOI8_R_INIT,
     &encoding_rs::SHIFT_JIS_INIT,
-    &encoding_rs::UTF_16LE_INIT,
-    &encoding_rs::SHIFT_JIS_INIT,
-    &encoding_rs::IBM866_INIT,
-    &encoding_rs::UTF_8_INIT,
-    &encoding_rs::WINDOWS_1252_INIT,
-    &encoding_rs::WINDOWS_1252_INIT,
-    &encoding_rs::GBK_INIT,
-    &encoding_rs::ISO_8859_7_INIT,
-    &encoding_rs::WINDOWS_1250_INIT,
-    &encoding_rs::WINDOWS_1251_INIT,
-    &encoding_rs::WINDOWS_1252_INIT,
-    &encoding_rs::GBK_INIT,
-    &encoding_rs::WINDOWS_1252_INIT,
-    &encoding_rs::ISO_8859_2_INIT,
-    &encoding_rs::WINDOWS_1253_INIT,
-    &encoding_rs::ISO_8859_3_INIT,
-    &encoding_rs::WINDOWS_1254_INIT,
-    &encoding_rs::ISO_8859_4_INIT,
-    &encoding_rs::WINDOWS_1255_INIT,
     &encoding_rs::BIG5_INIT,
-    &encoding_rs::WINDOWS_1254_INIT,
-    &encoding_rs::UTF_16LE_INIT,
-    &encoding_rs::WINDOWS_1256_INIT,
-    &encoding_rs::IBM866_INIT,
-    &encoding_rs::ISO_8859_10_INIT,
-    &encoding_rs::WINDOWS_1257_INIT,
-    &encoding_rs::WINDOWS_1258_INIT,
-    &encoding_rs::ISO_8859_7_INIT,
-    &encoding_rs::WINDOWS_1252_INIT,
-    &encoding_rs::ISO_8859_6_INIT,
-    &encoding_rs::ISO_8859_8_INIT,
-    &encoding_rs::EUC_KR_INIT,
     &encoding_rs::EUC_JP_INIT,
-    &encoding_rs::KOI8_R_INIT,
-    &encoding_rs::KOI8_R_INIT,
     &encoding_rs::EUC_KR_INIT,
-    &encoding_rs::SHIFT_JIS_INIT,
-    &encoding_rs::KOI8_U_INIT,
-    &encoding_rs::ISO_8859_8_INIT,
-    &encoding_rs::WINDOWS_874_INIT,
     &encoding_rs::GB18030_INIT,
-    &encoding_rs::EUC_KR_INIT,
     &encoding_rs::GBK_INIT,
-    &encoding_rs::WINDOWS_874_INIT,
-    &encoding_rs::BIG5_INIT,
-    &encoding_rs::UTF_16LE_INIT,
-    &encoding_rs::GBK_INIT,
-    &encoding_rs::ISO_8859_8_I_INIT,
-    &encoding_rs::KOI8_R_INIT,
-    &encoding_rs::EUC_KR_INIT,
-    &encoding_rs::KOI8_U_INIT,
-    &encoding_rs::WINDOWS_1250_INIT,
-    &encoding_rs::EUC_KR_INIT,
-    &encoding_rs::WINDOWS_1251_INIT,
-    &encoding_rs::WINDOWS_1252_INIT,
-    &encoding_rs::GBK_INIT,
-    &encoding_rs::WINDOWS_1252_INIT,
-    &encoding_rs::ISO_8859_2_INIT,
-    &encoding_rs::WINDOWS_1253_INIT,
-    &encoding_rs::ISO_8859_3_INIT,
-    &encoding_rs::ISO_8859_6_INIT,
-    &encoding_rs::WINDOWS_1254_INIT,
-    &encoding_rs::ISO_8859_4_INIT,
-    &encoding_rs::WINDOWS_1255_INIT,
-    &encoding_rs::ISO_8859_5_INIT,
-    &encoding_rs::BIG5_INIT,
-    &encoding_rs::WINDOWS_1256_INIT,
     &encoding_rs::IBM866_INIT,
-    &encoding_rs::ISO_8859_6_INIT,
-    &encoding_rs::WINDOWS_1257_INIT,
-    &encoding_rs::ISO_8859_7_INIT,
-    &encoding_rs::ISO_8859_6_INIT,
-    &encoding_rs::ISO_8859_7_INIT,
-    &encoding_rs::ISO_8859_7_INIT,
-    &encoding_rs::WINDOWS_1258_INIT,
-    &encoding_rs::ISO_8859_8_INIT,
-    &encoding_rs::WINDOWS_1254_INIT,
-    &encoding_rs::ISO_8859_5_INIT,
-    &encoding_rs::UTF_16BE_INIT,
-    &encoding_rs::UTF_16LE_INIT,
-    &encoding_rs::WINDOWS_1252_INIT,
-    &encoding_rs::SHIFT_JIS_INIT,
-    &encoding_rs::EUC_JP_INIT,
-    &encoding_rs::ISO_8859_10_INIT,
-    &encoding_rs::WINDOWS_1252_INIT,
-    &encoding_rs::WINDOWS_874_INIT,
-    &encoding_rs::ISO_8859_2_INIT,
-    &encoding_rs::ISO_8859_3_INIT,
-    &encoding_rs::ISO_8859_13_INIT,
-    &encoding_rs::ISO_8859_4_INIT,
-    &encoding_rs::ISO_8859_14_INIT,
-    &encoding_rs::ISO_8859_5_INIT,
-    &encoding_rs::ISO_8859_15_INIT,
-    &encoding_rs::ISO_8859_6_INIT,
-    &encoding_rs::ISO_8859_7_INIT,
-    &encoding_rs::ISO_8859_8_INIT,
-    &encoding_rs::GBK_INIT,
-    &encoding_rs::WINDOWS_1254_INIT,
-    &encoding_rs::UTF_16LE_INIT,
-    &encoding_rs::MACINTOSH_INIT,
-    &encoding_rs::SHIFT_JIS_INIT,
-    &encoding_rs::SHIFT_JIS_INIT,
-    &encoding_rs::WINDOWS_1252_INIT,
-    &encoding_rs::ISO_8859_10_INIT,
-    &encoding_rs::ISO_8859_4_INIT,
-    &encoding_rs::GBK_INIT,
-    &encoding_rs::WINDOWS_1252_INIT,
-    &encoding_rs::WINDOWS_1252_INIT,
-    &encoding_rs::ISO_8859_2_INIT,
-    &encoding_rs::WINDOWS_874_INIT,
-    &encoding_rs::ISO_8859_2_INIT,
     &encoding_rs::ISO_8859_2_INIT,
-    &encoding_rs::REPLACEMENT_INIT,
-    &encoding_rs::ISO_8859_3_INIT,
     &encoding_rs::ISO_8859_3_INIT,
-    &encoding_rs::ISO_8859_13_INIT,
     &encoding_rs::ISO_8859_4_INIT,
-    &encoding_rs::ISO_8859_4_INIT,
-    &encoding_rs::ISO_8859_14_INIT,
-    &encoding_rs::ISO_8859_5_INIT,
     &encoding_rs::ISO_8859_5_INIT,
-    &encoding_rs::ISO_8859_5_INIT,
-    &encoding_rs::ISO_8859_15_INIT,
-    &encoding_rs::ISO_8859_6_INIT,
     &encoding_rs::ISO_8859_6_INIT,
     &encoding_rs::ISO_8859_7_INIT,
-    &encoding_rs::ISO_8859_7_INIT,
-    &encoding_rs::ISO_8859_7_INIT,
-    &encoding_rs::ISO_8859_6_INIT,
-    &encoding_rs::ISO_8859_10_INIT,
-    &encoding_rs::ISO_8859_8_INIT,
-    &encoding_rs::ISO_8859_8_INIT,
+    &encoding_rs::ISO_8859_8_I_INIT,
     &encoding_rs::ISO_8859_8_INIT,
-    &encoding_rs::WINDOWS_1254_INIT,
-    &encoding_rs::WINDOWS_1254_INIT,
-    &encoding_rs::WINDOWS_1254_INIT,
-    &encoding_rs::ISO_8859_3_INIT,
-    &encoding_rs::EUC_KR_INIT,
-    &encoding_rs::BIG5_INIT,
-    &encoding_rs::SHIFT_JIS_INIT,
     &encoding_rs::ISO_8859_10_INIT,
-    &encoding_rs::WINDOWS_874_INIT,
-    &encoding_rs::WINDOWS_1252_INIT,
-    &encoding_rs::ISO_8859_2_INIT,
     &encoding_rs::ISO_8859_13_INIT,
-    &encoding_rs::ISO_8859_3_INIT,
     &encoding_rs::ISO_8859_14_INIT,
-    &encoding_rs::WINDOWS_874_INIT,
-    &encoding_rs::ISO_8859_4_INIT,
-    &encoding_rs::ISO_8859_15_INIT,
     &encoding_rs::ISO_8859_15_INIT,
-    &encoding_rs::WINDOWS_1254_INIT,
     &encoding_rs::ISO_8859_16_INIT,
-    &encoding_rs::ISO_8859_10_INIT,
-    &encoding_rs::EUC_KR_INIT,
-    &encoding_rs::ISO_8859_15_INIT,
-    &encoding_rs::ISO_8859_6_INIT,
-    &encoding_rs::ISO_8859_8_INIT,
-    &encoding_rs::UTF_16BE_INIT,
-    &encoding_rs::UTF_16LE_INIT,
-    &encoding_rs::MACINTOSH_INIT,
-    &encoding_rs::ISO_8859_6_INIT,
-    &encoding_rs::ISO_8859_8_I_INIT,
-    &encoding_rs::SHIFT_JIS_INIT,
+    &encoding_rs::KOI8_R_INIT,
+    &encoding_rs::KOI8_U_INIT,
     &encoding_rs::MACINTOSH_INIT,
-    &encoding_rs::REPLACEMENT_INIT,
-    &encoding_rs::ISO_2022_JP_INIT,
-    &encoding_rs::ISO_2022_JP_INIT,
-    &encoding_rs::REPLACEMENT_INIT,
-    &encoding_rs::REPLACEMENT_INIT,
-    &encoding_rs::REPLACEMENT_INIT,
     &encoding_rs::WINDOWS_1250_INIT,
     &encoding_rs::WINDOWS_1251_INIT,
     &encoding_rs::WINDOWS_1252_INIT,
@@ -201,39 +39,14 @@ static ALL_ENCODINGS: [&Encoding; 228] = [
     &encoding_rs::WINDOWS_1256_INIT,
     &encoding_rs::WINDOWS_1257_INIT,
     &encoding_rs::WINDOWS_1258_INIT,
-    &encoding_rs::ISO_8859_6_INIT,
-    &encoding_rs::ISO_8859_8_INIT,
-    &encoding_rs::ISO_8859_6_INIT,
-    &encoding_rs::ISO_8859_8_I_INIT,
-    &encoding_rs::ISO_8859_7_INIT,
-    &encoding_rs::EUC_KR_INIT,
-    &encoding_rs::UTF_8_INIT,
-    &encoding_rs::UTF_8_INIT,
-    &encoding_rs::EUC_KR_INIT,
-    &encoding_rs::WINDOWS_1252_INIT,
-    &encoding_rs::EUC_KR_INIT,
+    &encoding_rs::WINDOWS_874_INIT,
     &encoding_rs::X_MAC_CYRILLIC_INIT,
     &encoding_rs::X_USER_DEFINED_INIT,
-    &encoding_rs::GBK_INIT,
-    &encoding_rs::UTF_16LE_INIT,
-    &encoding_rs::WINDOWS_1252_INIT,
-    &encoding_rs::ISO_8859_2_INIT,
-    &encoding_rs::ISO_8859_6_INIT,
-    &encoding_rs::ISO_8859_7_INIT,
-    &encoding_rs::ISO_8859_3_INIT,
-    &encoding_rs::ISO_8859_4_INIT,
-    &encoding_rs::ISO_8859_5_INIT,
-    &encoding_rs::ISO_8859_8_INIT,
-    &encoding_rs::UTF_8_INIT,
-    &encoding_rs::WINDOWS_1254_INIT,
-    &encoding_rs::ISO_8859_7_INIT,
-    &encoding_rs::X_MAC_CYRILLIC_INIT,
+    // non-ASCII-compatible
     &encoding_rs::REPLACEMENT_INIT,
-    &encoding_rs::ISO_8859_6_INIT,
-    &encoding_rs::ISO_8859_8_INIT,
-    &encoding_rs::UTF_8_INIT,
-    &encoding_rs::ISO_8859_5_INIT,
-    &encoding_rs::EUC_JP_INIT,
+    &encoding_rs::UTF_16BE_INIT,
+    &encoding_rs::UTF_16LE_INIT,
+    &encoding_rs::ISO_2022_JP_INIT,
 ];
 
 fn encoding_to_index(encoding: AsciiCompatibleEncoding) -> usize {

diff --git a/src/rewriter/mod.rs b/src/rewriter/mod.rs
@@ -31,17 +31,12 @@ impl AsciiCompatibleEncoding {
     /// Returns `Some` if `Encoding` is ascii-compatible, or `None` otherwise.
     #[must_use]
     pub fn new(encoding: &'static Encoding) -> Option<Self> {
-        if encoding.is_ascii_compatible() {
-            Some(Self(encoding))
-        } else {
-            None
-        }
+        encoding.is_ascii_compatible().then_some(Self(encoding))
     }
 
     fn from_mimetype(mime: &Mime) -> Option<Self> {
-        mime.get_param("charset")
-            .and_then(|cs| Encoding::for_label_no_replacement(cs.as_str().as_bytes()))
-            .and_then(Self::new)
+        let cs = mime.get_param("charset")?;
+        Self::new(Encoding::for_label_no_replacement(cs.as_str().as_bytes())?)
     }
 
     /// Returns the most commonly used UTF-8 encoding.