Merge pull request #2626 from rhysd/encoding_rs

Replace unmaintained `encoding` dependency with `encoding_rs`
sharkdp · Sep 1, 2023 · 1e52785 · 1e52785
2 parents 9ba3b6e + d33b2e1
commit 1e52785
Show file tree

Hide file tree

Showing 3 changed files with 21 additions and 79 deletions.
diff --git a/Cargo.lock b/Cargo.lock
diff --git a/Cargo.toml b/Cargo.toml
@@ -49,7 +49,6 @@ once_cell = "1.17"
 thiserror = "1.0"
 wild = { version = "2.1", optional = true }
 content_inspector = "0.2.4"
-encoding = "0.2"
 shell-words = { version = "1.1.0", optional = true }
 unicode-width = "0.1.10"
 globset = "0.4"
@@ -64,6 +63,7 @@ grep-cli = { version = "0.1.9", optional = true }
 regex = { version = "1.8.3", optional = true }
 walkdir = { version = "2.3", optional = true }
 bytesize = { version = "1.2.0" }
+encoding_rs = "0.8.32"
 
 [dependencies.git2]
 version = "0.18"

diff --git a/src/printer.rs b/src/printer.rs
@@ -15,8 +15,7 @@ use syntect::parsing::SyntaxSet;
 
 use content_inspector::ContentType;
 
-use encoding::all::{UTF_16BE, UTF_16LE};
-use encoding::{DecoderTrap, Encoding};
+use encoding_rs::{UTF_16BE, UTF_16LE};
 
 use unicode_width::UnicodeWidthChar;
 
@@ -431,27 +430,25 @@ impl<'a> Printer for InteractivePrinter<'a> {
                 self.config.tab_width,
                 self.config.nonprintable_notation,
             )
+            .into()
         } else {
-            let line = match self.content_type {
+            match self.content_type {
                 Some(ContentType::BINARY) | None => {
                     return Ok(());
                 }
-                Some(ContentType::UTF_16LE) => UTF_16LE
-                    .decode(line_buffer, DecoderTrap::Replace)
-                    .map_err(|_| "Invalid UTF-16LE")?,
-                Some(ContentType::UTF_16BE) => UTF_16BE
-                    .decode(line_buffer, DecoderTrap::Replace)
-                    .map_err(|_| "Invalid UTF-16BE")?,
-                _ => String::from_utf8_lossy(line_buffer).to_string(),
-            };
-            // Remove byte order mark from the first line if it exists
-            if line_number == 1 {
-                match line.strip_prefix('\u{feff}') {
-                    Some(stripped) => stripped.to_string(),
-                    None => line,
+                Some(ContentType::UTF_16LE) => UTF_16LE.decode_with_bom_removal(line_buffer).0,
+                Some(ContentType::UTF_16BE) => UTF_16BE.decode_with_bom_removal(line_buffer).0,
+                _ => {
+                    let line = String::from_utf8_lossy(line_buffer);
+                    if line_number == 1 {
+                        match line.strip_prefix('\u{feff}') {
+                            Some(stripped) => stripped.to_string().into(),
+                            None => line,
+                        }
+                    } else {
+                        line
+                    }
                 }
-            } else {
-                line
             }
         };