Merge pull request #3691 from curious-rabbit/master

sanitize filenames

CHANGELOG.md

@@ -21,6 +21,7 @@
 - Syntax highlighting for Python files using uv as script runner in shebang #3689 (@janlarres)
 
 ## Bugfixes
+- Sanitize control characters in filenames before displaying them in the file header, error messages, and the terminal title, preventing ANSI escape injection via crafted filenames. Closes #3054, see #3691 (@curious-rabbit)
 - Report initial input read errors instead of treating them as empty input. Closes #3002, see #3706 (@lawrence3699)
 - Treat ZIP archives as binary content based on their magic header, see #3686 (@officialasishkumar)
 - Fix i686 `.deb` package using incorrect architecture name (`i686` instead of `i386`), preventing installation on Debian. Closes #3611, see #3650 (@Sim-hu)

src/assets.rs

@@ -158,7 +158,9 @@ impl HighlightingAssets {
         let syntax_match = mapping.get_syntax_for(path);
 
         if let Some(MappingTarget::MapToUnknown) = syntax_match {
-            return Err(Error::UndetectedSyntax(path.to_string_lossy().into()));
+            return Err(Error::UndetectedSyntax(
+                crate::preprocessor::sanitize_for_terminal(&path.to_string_lossy()),
+            ));
         }
 
         if let Some(MappingTarget::MapTo(syntax_name)) = syntax_match {
@@ -175,13 +177,17 @@ impl HighlightingAssets {
         ) {
             (Some(syntax), _) => Ok(syntax),
 
-            (_, Some(MappingTarget::MapExtensionToUnknown)) => {
-                Err(Error::UndetectedSyntax(path.to_string_lossy().into()))
-            }
+            (_, Some(MappingTarget::MapExtensionToUnknown)) => Err(Error::UndetectedSyntax(
+                crate::preprocessor::sanitize_for_terminal(&path.to_string_lossy()),
+            )),
 
             _ => self
                 .get_syntax_for_file_extension(file_name, &mapping.ignored_suffixes)?
-                .ok_or_else(|| Error::UndetectedSyntax(path.to_string_lossy().into())),
+                .ok_or_else(|| {
+                    Error::UndetectedSyntax(crate::preprocessor::sanitize_for_terminal(
+                        &path.to_string_lossy(),
+                    ))
+                }),
         }
     }
 

src/bin/bat/main.rs

@@ -259,7 +259,9 @@ pub fn list_themes(
 fn set_terminal_title_to(new_terminal_title: String) {
     let osc_command_for_setting_terminal_title = "\x1b]0;";
     let osc_end_command = "\x07";
-    print!("{osc_command_for_setting_terminal_title}{new_terminal_title}{osc_end_command}");
+    // Prevent BEL/ESC/C1 bytes in the title from terminating or nesting the OSC.
+    let safe_title = bat::sanitize_for_terminal(&new_terminal_title);
+    print!("{osc_command_for_setting_terminal_title}{safe_title}{osc_end_command}");
     io::stdout().flush().unwrap();
 }
 

src/input.rs

@@ -216,20 +216,20 @@ impl<'a> Input<'a> {
                 description,
                 metadata: self.metadata,
                 reader: {
-                    let mut file = File::open(&path)
-                        .map_err(|e| format!("'{}': {e}", path.to_string_lossy()))?;
+                    let path_display =
+                        crate::preprocessor::sanitize_for_terminal(&path.to_string_lossy());
+                    let mut file =
+                        File::open(&path).map_err(|e| format!("'{path_display}': {e}"))?;
                     if file.metadata()?.is_dir() {
-                        return Err(format!("'{}' is a directory.", path.to_string_lossy()).into());
+                        return Err(format!("'{path_display}' is a directory.").into());
                     }
 
                     if let Some(stdout) = stdout_identifier {
-                        let input_identifier = Identifier::try_from(file).map_err(|e| {
-                            format!("{}: Error identifying file: {e}", path.to_string_lossy())
-                        })?;
+                        let input_identifier = Identifier::try_from(file)
+                            .map_err(|e| format!("{path_display}: Error identifying file: {e}"))?;
                         if stdout.surely_conflicts_with(&input_identifier) {
                             return Err(format!(
-                                "IO circle detected. The input from '{}' is also an output. Aborting to avoid infinite loop.",
-                                path.to_string_lossy()
+                                "IO circle detected. The input from '{path_display}' is also an output. Aborting to avoid infinite loop.",
                             )
                             .into());
                         }

src/lib.rs

@@ -54,6 +54,7 @@ mod vscreen;
 pub(crate) mod wrapping;
 
 pub use nonprintable_notation::{BinaryBehavior, NonprintableNotation};
+pub use preprocessor::sanitize_for_terminal;
 pub use preprocessor::StripAnsiMode;
 pub use pretty_printer::{Input, PrettyPrinter, Syntax};
 pub use syntax_mapping::{MappingTarget, SyntaxMapping};

src/preprocessor.rs

@@ -149,6 +149,35 @@ pub fn strip_ansi(line: &str) -> String {
     buffer
 }
 
+/// Escape C0, DEL, and C1 control characters so a string from an untrusted
+/// filename or path can be safely written to the terminal.
+pub fn sanitize_for_terminal(input: &str) -> String {
+    if !input
+        .chars()
+        .any(|c| matches!(c, '\x00'..='\x08' | '\x0A'..='\x1F' | '\x7F'..='\u{9F}'))
+    {
+        return input.to_owned();
+    }
+
+    let mut out = String::with_capacity(input.len() + 8);
+    for c in input.chars() {
+        match c {
+            '\t' => out.push('\t'),
+            '\x00'..='\x1F' => {
+                out.push('^');
+                out.push(char::from_u32(0x40 + c as u32).unwrap_or('?'));
+            }
+            '\x7F' => out.push_str("^?"),
+            '\u{80}'..='\u{9F}' => {
+                use std::fmt::Write as _;
+                let _ = write!(out, "\\u{{{:x}}}", c as u32);
+            }
+            other => out.push(other),
+        }
+    }
+    out
+}
+
 /// Strips overstrike sequences (backspace formatting) from input.
 ///
 /// Overstrike formatting is used by man pages and some help output:
@@ -261,3 +290,40 @@ fn test_strip_overstrike() {
     // Unicode with overstrike
     assert_eq!(strip_overstrike("ä\x08äöü", 2), "äöü");
 }
+
+#[test]
+fn test_sanitize_for_terminal_passthrough() {
+    assert_eq!(sanitize_for_terminal(""), "");
+    assert_eq!(sanitize_for_terminal("hello.txt"), "hello.txt");
+    assert_eq!(sanitize_for_terminal("résumé.pdf"), "résumé.pdf");
+    assert_eq!(sanitize_for_terminal("日本語.md"), "日本語.md");
+    assert_eq!(
+        sanitize_for_terminal("path/with spaces/file.log"),
+        "path/with spaces/file.log"
+    );
+    assert_eq!(sanitize_for_terminal("a\tb"), "a\tb");
+}
+
+#[test]
+fn test_sanitize_for_terminal_c0_controls() {
+    assert_eq!(
+        sanitize_for_terminal("\x1b[31mINJECTED\x1b[0m.txt"),
+        "^[[31mINJECTED^[[0m.txt"
+    );
+    assert_eq!(sanitize_for_terminal("bad\x07rest"), "bad^Grest");
+    assert_eq!(sanitize_for_terminal("\x00\x08\n\r\x7F"), "^@^H^J^M^?");
+    assert_eq!(sanitize_for_terminal("\u{9b}31m"), "\\u{9b}31m");
+    assert_eq!(
+        sanitize_for_terminal("\u{9d}0;pwned\x07"),
+        "\\u{9d}0;pwned^G"
+    );
+}
+
+#[test]
+fn test_sanitize_for_terminal_idempotent_on_sanitized() {
+    let dirty = "\x1b]0;pwned\x07file.txt";
+    let clean = sanitize_for_terminal(dirty);
+    assert_eq!(sanitize_for_terminal(&clean), clean);
+    assert!(!clean.contains('\x1b'));
+    assert!(!clean.contains('\x07'));
+}

src/printer.rs

@@ -29,7 +29,9 @@ use crate::error::*;
 use crate::input::OpenedInput;
 use crate::line_range::{MaxBufferedLineNumber, RangeCheckResult};
 use crate::output::OutputHandle;
-use crate::preprocessor::{expand_tabs, replace_nonprintable, strip_ansi, strip_overstrike};
+use crate::preprocessor::{
+    expand_tabs, replace_nonprintable, sanitize_for_terminal, strip_ansi, strip_overstrike,
+};
 use crate::style::StyleComponent;
 use crate::terminal::{as_terminal_escaped, to_ansi_color};
 use crate::vscreen::{AnsiStyle, EscapeSequence, EscapeSequenceIterator};
@@ -489,7 +491,7 @@ impl Printer for InteractivePrinter<'_> {
                      (but will be present if the output of 'bat' is piped). You can use 'bat -A' \
                      to show the binary file contents.",
                     Yellow.paint("[bat warning]"),
-                    input.description.summary(),
+                    sanitize_for_terminal(&input.description.summary()),
                 )?;
             } else if self.config.style_components.grid() {
                 self.print_horizontal_line(handle, '┬')?;
@@ -543,9 +545,11 @@ impl Printer for InteractivePrinter<'_> {
                         "{}{}{mode}",
                         description
                             .kind()
-                            .map(|kind| format!("{kind}: "))
+                            .map(|kind| format!("{}: ", sanitize_for_terminal(kind)))
                             .unwrap_or_else(|| "".into()),
-                        self.colors.header_value.paint(description.title()),
+                        self.colors
+                            .header_value
+                            .paint(sanitize_for_terminal(description.title())),
                     );
                     self.print_header_multiline_component(handle, &header_filename)
                 }