Merge pull request #3874 from sharkdp/dependabot/cargo/minus-5.7.2

build(deps): bump minus from 5.7.1 to 5.7.2

.gitmodules

@@ -281,3 +281,6 @@
 [submodule "assets/syntaxes/02_Extra/Kotlin"]
 	path = assets/syntaxes/02_Extra/Kotlin
 	url = https://github.com/guille/sublime-kotlin
+[submodule "assets/syntaxes/02_Extra/Caddy"]
+	path = assets/syntaxes/02_Extra/Caddy
+	url = https://github.com/caddyserver/sublimetext.git

CHANGELOG.md

@@ -3,11 +3,14 @@
 
 ## Other
 
+- Update Cargo dependencies to resolve current RustSec advisories, see #3861 (@TyceHerrman)
 - Add instructions for removing fish help abbreviations to README, see #3655 (@claw-explorer). Closes #3536
 - Add .NET slnx extension, see #3682 (@ltrzesniewski)
 
 ## Features
 
+- Add a `--sanitize=<auto|always|never>` flag for safe display of untrusted input. It implies `--strip-ansi` at the same value and additionally substitutes terminal-active control bytes (cursor moves, charset switches, beep, etc.) and Unicode bidi / zero-width formatting characters with the Unicode replacement character (U+FFFD). Mitigates Trojan-Source-style spoofing (CVE-2021-42574). See #3729 (@curious-rabbit)
+- Map justfile, Justfile, .justfile, and *.justfile to Makefile syntax highlighting, see #3623 (@zachvalenta)
 - Preserve `--diff` change markers and snip separators when `--plain` is set. Closes #3630, see #3643 (@mvanhorn)
 - Added support for `hidden_file_extensions` from `.sublime-syntax` files, see #3613 (@Matei02355)
 - Add word wrapping mode via `--wrap=word`, see #3597 (@veeceey)
@@ -21,6 +24,13 @@
 - Syntax highlighting for Python files using uv as script runner in shebang #3689 (@janlarres)
 
 ## Bugfixes
+- Fix `--list-languages` respecting `--paging=never`, see #3828 (@cyphercodes)
+- Fix `--sanitize` passing through the bidi control characters U+200E, U+200F and U+061C, see #3862 (@lenamonj)
+- `--strip-ansi`: also strip 8-bit C1 introducers (U+0090, U+0098, U+009B, U+009D, U+009E, U+009F) and DCS/SOS/PM/APC sequence bodies, which previously passed through. See #3729 (@curious-rabbit)
+- Fix `--ignored-suffix` not falling back to first-line/shebang detection when the ignored suffix is also a registered extension (e.g. `--ignored-suffix .txt` on a shebang script), see #2745 and #3816 (@adnrivera)
+- Fix `capacity overflow` panic when printing a snip separator at `--terminal-width=1` with multiple line ranges. Closes #3803, see #3804 (@leeewee)
+- Pass `--no-paging` to `bat` invocations inside the bash / zsh / fish / PowerShell shell completion scripts so that shell-level pager wiring (e.g. `LESSOPEN='|-bat -f -pp %s'`) cannot inject ANSI escape sequences into the completion candidates. Closes #3760 (@mvanhorn)
+- Quote filenames before substituting them into `$LESSOPEN` / `$LESSCLOSE` templates, preventing shell injection when a filename contains shell metacharacters, see #3726 (@curious-rabbit)
 - Fix `--list-themes` unconditionally probing the terminal via OSC 10/11 even when `--theme` was set to an explicit value, see #3700 (regression introduced in bc42149a). (@optimistiCli)
 - Fix inverted `$LESSCLOSE` warning so bat warns on nonzero exit, not on success. See #3654 (@cuiweixie)
 - 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)
@@ -41,6 +51,8 @@
 - Fixed bug caused by using `--plain` and `--terminal-width=N` flags simultaneously, see #3529 (@H4k1l)
 - Fixed syntax tests path, see #3610 (@foxfromworld)
 - Fix zsh tab completion word-splitting language names containing spaces (e.g. `HTML (Jinja2)`, `Apache Conf`), see #3693 (@YoshKoz)
+- Fix zsh tab completion offering invalid `-l` arguments (file globs, paths, hidden filenames) sourced from the second column of `--list-languages`. Closes #3735, see #3737 (@truffle-dev)
+- Fix `usize` underflow in `--list-languages` when `--terminal-width` is smaller than the longest language name, see #3812 (@greymoth-jp)
 
 ## Other
 - Use git version of cross. See #3533 (@OctopusET)
@@ -48,6 +60,7 @@
 - Allow home and end keys to be used with builtin pager, see #3651 (@keith-hall)
 - Builtin syntax mapping: cleanup matcher glob parsing logic #3652 (@cyqsimon)
 - Statically link the CRT for MSVC builds via Cargo config to avoid runtime DLL dependencies. Closes #3634, see #3692 (@barry3406)
+- Replace `libgit2` with a pure Rust implementation of git called `gitoxide`, see PR #3703 (@blinxen)
 
 ## Syntaxes
 
@@ -59,6 +72,11 @@
 - Map several Google Cloud CLI config files to their appropriate syntax #3635 (@victor-gp)
 - Map all ignore dotfiles to Git Ignore syntax #3636 (@victor-gp)
 - Improved Kotlin syntax, see #3699 (@guille)
+- Include subdirectories in SSH Config syntax mapping, see #3758 (@injust)
+- Add Ghostty syntax mapping, see #3759 (@injust)
+- Add syntax highlighting for `Caddyfile` #3789 (@CosmicHorrorDev)
+- Include `.code-workspace` as a JSON extension #3809 (@dhruvkb)
+- Add syntax mapping for DNF repo configuration files, see #3814 (@injust)
 
 ## Themes
 

Cargo.toml

@@ -31,8 +31,7 @@ minimal-application = [
     "regex-onig",
     "wild",
 ]
-git = ["git2"] # Support indicating git modifications
+git = ["gix"] # Support indicating git modifications
-vendored-libgit2 = ["git2/vendored-libgit2"]
 paging = [ "shell-words", "grep-cli", "minus"] # Support applying a pager on the output
 lessopen = ["execute"] # Support $LESSOPEN preprocessor
 build-assets = ["syntect/yaml-load", "syntect/plist-load", "regex", "walkdir"]
@@ -52,7 +51,7 @@ thiserror = "2.0"
 wild = { version = "2.2", optional = true }
 content_inspector = "0.2.4"
 shell-words = { version = "1.1.1", optional = true }
-minus = { version = "5.6", optional = true, features = [
+minus = { version = "5.7", optional = true, features = [
   "dynamic_output",
   "search",
 ] }
@@ -64,22 +63,23 @@ serde_yaml = "0.9.28"
 semver = "1.0"
 path_abs = { version = "0.5", default-features = false }
 clircle = { version = "0.6.1", default-features = false }
-bugreport = { version = "0.5.0", optional = true }
+bugreport = { version = "0.6.0", optional = true }
 etcetera = { version = "0.11.0", optional = true }
 grep-cli = { version = "0.1.12", optional = true }
-regex = { version = "1.12.2", optional = true }
+regex = { version = "1.12.3", optional = true }
 walkdir = { version = "2.5", optional = true }
 bytesize = { version = "2.3.1" }
 encoding_rs = "0.8.35"
-execute = { version = "0.2.15", optional = true }
+execute = { version = "0.3.0", optional = true }
 terminal-colorsaurus = "1.0"
 unicode-segmentation = "1.13.2"
 itertools = "0.14.0"
 
-[dependencies.git2]
+[dependencies.gix]
-version = "0.20"
+version = "0.86"
 optional = true
 default-features = false
+features = ["sha1", "blob-diff"]
 
 [dependencies.syntect]
 version = "5.3.0"
@@ -87,18 +87,18 @@ default-features = false
 features = ["parsing"]
 
 [dependencies.clap]
-version = "4.5.60"
+version = "4.6.1"
 optional = true
 features = ["wrap_help", "cargo"]
 
 [target.'cfg(target_os = "macos")'.dependencies]
-plist = "1.7.0"
+plist = "1.9.0"
 
 [dev-dependencies]
 assert_cmd = "2.0.16"
 expect-test = "1.5.0"
 serial_test = { version = "2.0.0", default-features = false }
-predicates = "3.1.3"
+predicates = "3.1.4"
 wait-timeout = "0.2.1"
 tempfile = "3.27.0"
 serde = { version = "1.0", features = ["derive"] }
@@ -108,13 +108,13 @@ nix = { version = "0.31", default-features = false, features = ["term"] }
 
 [build-dependencies]
 anyhow = "1.0.97"
-indexmap = { version = "2.13.0", features = ["serde"] }
+indexmap = { version = "2.14.0", features = ["serde"] }
 itertools = "0.14.0"
 once_cell = "1.20"
 prettyplease = "0.2.37"
 proc-macro2 = "1.0.106"
 quote = "1.0.45"
-regex = "1.12.2"
+regex = "1.12.3"
 serde = "1.0"
 serde_derive = "1.0"
 serde_with = { version = "3.17.0", default-features = false, features = ["macros"] }
@@ -123,7 +123,7 @@ toml = { version = "1.1.1", features = ["preserve_order"] }
 walkdir = "2.5"
 
 [build-dependencies.clap]
-version = "4.5.60"
+version = "4.6.1"
 optional = true
 features = ["wrap_help", "cargo"]
 

README.md

@@ -307,7 +307,7 @@ pacman -S bat
 
 ### On Fedora
 
-You can install [the `bat` package](https://koji.fedoraproject.org/koji/packageinfo?packageID=27506) from the official [Fedora Modular](https://docs.fedoraproject.org/en-US/modularity/using-modules/) repository.
+You can install [the `bat` package](https://packages.fedoraproject.org/pkgs/rust-bat/bat/) from the official sources:
 
 ```bash
 dnf install bat
@@ -843,6 +843,10 @@ If your version of `bat` supports the `--strip-ansi=auto` option, it can be used
 before syntax highlighting. Alternatively, you may disable both syntax highlighting and wrapping by
 passing the `--color=never --wrap=never` options to `bat`.
 
+For untrusted input, the `--sanitize=auto|always|never` option additionally replaces terminal-active
+control bytes and Unicode bidi / zero-width formatting characters with the Unicode replacement
+character. It implies `--strip-ansi` at the same value.
+
 > [!NOTE]
 > The `auto` option of `--strip-ansi` avoids removing escape sequences when the syntax is plain text.
 

assets/completions/_bat.ps1.in

@@ -14,12 +14,12 @@ Register-ArgumentCompleter -Native -CommandName '{{PROJECT_EXECUTABLE}}' -Script
     $ArrayPrint      = @('unicode', 'caret')
 
     function Get-MyThemes(){
-        $themes = {{PROJECT_EXECUTABLE}} --list-themes | ForEach-Object {$_  -replace "^(.*)$", '''$1'''} | select-object
+        $themes = {{PROJECT_EXECUTABLE}} --no-paging --list-themes | ForEach-Object {$_  -replace "^(.*)$", '''$1'''} | select-object
         return $themes
     }
 
     function Get-MyLanguages(){
-        $themes = {{PROJECT_EXECUTABLE}} --list-languages | ForEach-Object{[pscustomobject]@{MyParameter=$_.Substring(0,$_.IndexOf(":")).Trim();MyDescription=$_.Substring($_.IndexOf(":")+1)}} | select-object
+        $themes = {{PROJECT_EXECUTABLE}} --no-paging --list-languages | ForEach-Object{[pscustomobject]@{MyParameter=$_.Substring(0,$_.IndexOf(":")).Trim();MyDescription=$_.Substring($_.IndexOf(":")+1)}} | select-object
         return $themes
     }
 
@@ -100,6 +100,11 @@ Register-ArgumentCompleter -Native -CommandName '{{PROJECT_EXECUTABLE}}' -Script
             ForEach-Object {[System.Management.Automation.CompletionResult]::new($_, $_, [CompletionResultType]::ParameterValue, $_)}
             break
         }
+        '*;--sanitize' {
+            $ArrayWhen |
+            ForEach-Object {[System.Management.Automation.CompletionResult]::new($_, $_, [CompletionResultType]::ParameterValue, $_)}
+            break
+        }
         '*;--strip-ansi' {
             $ArrayWhen |
             ForEach-Object {[System.Management.Automation.CompletionResult]::new($_, $_, [CompletionResultType]::ParameterValue, $_)}
@@ -158,6 +163,7 @@ Register-ArgumentCompleter -Native -CommandName '{{PROJECT_EXECUTABLE}}' -Script
             [CompletionResult]::new('--ignored-suffix'        , 'ignored-suffix'        , [CompletionResultType]::ParameterName, 'Ignore extension. For example: ''bat --ignored-suffix ".dev" my_file.json.dev'' will use JSON syntax, and ignore ''.dev''')
             [CompletionResult]::new('--squeeze-blank'         , 'squeeze-blank'         , [CompletionResultType]::ParameterName, 'Squeeze consecutive empty lines into a single empty line.')
             [CompletionResult]::new('--squeeze-limit'         , 'squeeze-limit'         , [CompletionResultType]::ParameterName, 'Set the maximum number of consecutive empty lines to be printed.')
+            [CompletionResult]::new('--sanitize'              , 'sanitize'              , [CompletionResultType]::ParameterName, 'Specify when to sanitize untrusted input for safe display. Implies --strip-ansi and also replaces terminal-active and bidi / zero-width bytes. (auto, always, *never*).')
             [CompletionResult]::new('--strip-ansi'            , 'strip-ansi'            , [CompletionResultType]::ParameterName, 'Specify when to strip ANSI escape sequences from the input. The automatic mode will remove escape sequences unless the syntax highlighting language is plain text. (auto, always, *never*).')
         #   [CompletionResult]::new('-p'                      , 'p'                     , [CompletionResultType]::ParameterName, 'Show plain style (alias for ''--style=plain'').')
             [CompletionResult]::new('--plain'                 , 'plain'                 , [CompletionResultType]::ParameterName, 'Show plain style (alias for ''--style=plain'').')

assets/completions/bat.bash.in

@@ -80,7 +80,7 @@ _bat() {
 	-l | --language)
 		local IFS=$'\n'
 		COMPREPLY=($(compgen -W "$(
-			"$1" --list-languages | while IFS=: read -r lang _; do
+			"$1" --no-paging --list-languages | while IFS=: read -r lang _; do
 				printf "%s\n" "$lang"
 			done
 		)" -- "$cur"))
@@ -132,6 +132,10 @@ _bat() {
 		COMPREPLY=($(compgen -W "auto never always" -- "$cur"))
 		return 0
 		;;
+	--sanitize)
+		COMPREPLY=($(compgen -W "auto never always" -- "$cur"))
+		return 0
+		;;
 	--completion)
 		COMPREPLY=($(compgen -W "bash fish zsh ps1" -- "$cur"))
 		return 0
@@ -150,14 +154,14 @@ _bat() {
 		;;
 	--theme)
 		local IFS=$'\n'
-		COMPREPLY=($(compgen -W "auto${IFS}auto:always${IFS}auto:system${IFS}dark${IFS}light${IFS}$("$1" --list-themes)" -- "$cur"))
+		COMPREPLY=($(compgen -W "auto${IFS}auto:always${IFS}auto:system${IFS}dark${IFS}light${IFS}$("$1" --no-paging --list-themes)" -- "$cur"))
 		__bat_escape_completions
 		return 0
 		;;
 	--theme-dark | \
 	--theme-light)
 		local IFS=$'\n'
-		COMPREPLY=($(compgen -W "$("$1" --list-themes)" -- "$cur"))
+		COMPREPLY=($(compgen -W "$("$1" --no-paging --list-themes)" -- "$cur"))
 		__bat_escape_completions
 		return 0
 		;;
@@ -221,6 +225,7 @@ _bat() {
 			--list-themes
 			--squeeze-blank
 			--squeeze-limit
+			--sanitize
 			--strip-ansi
 			--style
 			--line-range

assets/completions/bat.fish.in

@@ -15,11 +15,11 @@ function __bat_complete_files -a token
 end
 
 function __bat_complete_one_language -a comp
-    command $bat --list-languages | string split -f1 : | string match -e "$comp"
+    command $bat --no-paging --list-languages | string split -f1 : | string match -e "$comp"
 end
 
 function __bat_complete_list_languages
-    for spec in (command $bat --list-languages)
+    for spec in (command $bat --no-paging --list-languages)
         set -l name (string split -f1 : $spec)
         for ext in (string split -f2 : $spec | string split ,)
             test -n "$ext"; or continue
@@ -221,6 +221,7 @@ complete -c $bat -s s -l squeeze-blank -d "Squeeze consecutive empty lines into
 complete -c $bat -l squeeze-limit -x -d "Set the maximum number of consecutive empty lines to be printed" -n __bat_no_excl_args
 
 complete -c $bat -l strip-ansi -x -a "auto never always" -d "Specify when to strip ANSI escape sequences from the input" -n __bat_no_excl_args
+complete -c $bat -l sanitize -x -a "auto never always" -d "Specify when to sanitize untrusted input for safe display" -n __bat_no_excl_args
 
 complete -c $bat -s p -l plain -d "Disable decorations" -n __bat_no_excl_args
 
@@ -234,11 +235,11 @@ complete -c $bat -l tabs -x -a "$tabs_opts" -d "Set tab width" -n __bat_no_excl_
 
 complete -c $bat -l terminal-width -x -d "Set terminal <width>, +<offset>, or -<offset>" -n __bat_no_excl_args
 
-complete -c $bat -l theme -x -a "$special_themes(command $bat --list-themes | command cat)" -d "Set the syntax highlighting theme" -n __bat_no_excl_args
+complete -c $bat -l theme -x -a "$special_themes(command $bat --no-paging --list-themes | command cat)" -d "Set the syntax highlighting theme" -n __bat_no_excl_args
 
-complete -c $bat -l theme-dark -x -a "(command $bat --list-themes | command cat)" -d "Set the syntax highlighting theme for dark backgrounds" -n __bat_no_excl_args
+complete -c $bat -l theme-dark -x -a "(command $bat --no-paging --list-themes | command cat)" -d "Set the syntax highlighting theme for dark backgrounds" -n __bat_no_excl_args
 
-complete -c $bat -l theme-light -x -a "(command $bat --list-themes | command cat)" -d "Set the syntax highlighting theme for light backgrounds" -n __bat_no_excl_args
+complete -c $bat -l theme-light -x -a "(command $bat --no-paging --list-themes | command cat)" -d "Set the syntax highlighting theme for light backgrounds" -n __bat_no_excl_args
 
 complete -c $bat -s u -l unbuffered -d "Enable unbuffered input reading for streaming use cases" -n __bat_no_excl_args
 

assets/completions/bat.zsh.in

@@ -54,6 +54,7 @@ _{{PROJECT_EXECUTABLE}}_main() {
         --squeeze-blank'[squeeze consecutive empty lines into a single empty line]'
         --squeeze-limit='[set the maximum number of consecutive empty lines]:limit:'
         --strip-ansi='[specify when to strip ANSI escape sequences]:when:(auto never always)'
+        --sanitize='[specify when to sanitize untrusted input for safe display]:when:(auto never always)'
         --style='[comma-separated list of style elements to display]: : _values "style [default]"
             default auto full plain changes header header-filename header-filesize grid rule numbers snip'
         \*{-r+,--line-range=}'[only print the specified line range]:start\:end'
@@ -90,20 +91,25 @@ _{{PROJECT_EXECUTABLE}}_main() {
         languages)
             local IFS=$'\n'
             local -a languages
-            languages=( ${(f)"$({{PROJECT_EXECUTABLE}} --list-languages | awk -F':|,' '{ for (i = 1; i <= NF; ++i) printf("%s:%s\n", $i, $1) }')"} )
+            # `--list-languages` emits one `name:matchers` line per language,
+            # which `_describe` parses as `value:description`. Only the
+            # language name is offered as the completion value; the matchers
+            # show up as the menu description. See
+            # https://github.com/sharkdp/bat/issues/3735.
+            languages=( ${(f)"$({{PROJECT_EXECUTABLE}} --no-paging --color=never --decorations=never --list-languages)"} )
 
             _describe 'language' languages && ret=0
         ;;
 
         themes)
             local -a themes expl
-            themes=(${(f)"$(_call_program themes {{PROJECT_EXECUTABLE}} --list-themes)"} )
+            themes=(${(f)"$(_call_program themes {{PROJECT_EXECUTABLE}} --no-paging --color=never --decorations=never --list-themes)"} )
 
             _wanted themes expl 'theme' compadd -a themes && ret=0
         ;;
         theme_preferences)
             local -a themes expl
-            themes=(auto dark light auto:always auto:system ${(f)"$(_call_program themes {{PROJECT_EXECUTABLE}} --list-themes)"} )
+            themes=(auto dark light auto:always auto:system ${(f)"$(_call_program themes {{PROJECT_EXECUTABLE}} --no-paging --color=never --decorations=never --list-themes)"} )
 
             _wanted themes expl 'theme' compadd -a themes && ret=0
         ;;

doc/assets.md

@@ -97,7 +97,7 @@ The following files have been manually modified after converting from a `.tmLang
    as it is not kept in a standalone repository. The file is generated from
    https://github.com/open-policy-agent/opa/blob/master/misc/syntax/textmate/Rego.tmLanguage
 * `SML.sublime_syntax` has been added manually from
-   https://github.com/seanjames777/SML-Language-Definitiona as it is not
+   https://github.com/seanjames777/SML-Language-Definition as it is not
    kept in a standalone repository. The file generated is from
    https://github.com/seanjames777/SML-Language-Definition/blob/master/sml.tmLanguage
 * `Cabal.sublime_syntax` has been added manually from

doc/long-help.txt

@@ -164,6 +164,14 @@ Options:
           escape sequences unless the syntax highlighting language is plain text. Possible values:
           auto, always, *never*.
 
+      --sanitize <when>
+          Specify when to sanitize input bytes for safe terminal display. Implies --strip-ansi to
+          the same value, and additionally substitutes terminal-active control bytes (cursor moves,
+          charset switches, beep, etc.) and Unicode bidi / zero-width formatting characters with the
+          Unicode replacement character (U+FFFD). Tab, LF, FF, and CRLF pass through. Useful for
+          displaying untrusted file content (e.g. file-manager preview panes). Possible values:
+          auto, always, *never*.
+
       --style <components>
           Configure which elements (line numbers, file headers, grid borders, Git modifications, ..)
           to display in addition to the file contents. The argument is a comma-separated list of

src/assets.rs

@@ -347,15 +347,15 @@ impl HighlightingAssets {
         file_name: &OsStr,
         ignored_suffixes: &IgnoredSuffixes,
     ) -> Result<Option<SyntaxReferenceInSet<'_>>> {
-        let mut syntax = self.find_syntax_by_extension(Path::new(file_name).extension())?;
+        let stripped =
-        if syntax.is_none() {
-            syntax =
             ignored_suffixes.try_with_stripped_suffix(file_name, |stripped_file_name| {
-                    // Note: recursion
                 self.get_syntax_for_file_extension(stripped_file_name, ignored_suffixes)
+                    .map(Some)
             })?;
+        match stripped {
+            Some(syntax) => Ok(syntax),
+            None => self.find_syntax_by_extension(Path::new(file_name).extension()),
         }
-        Ok(syntax)
     }
 
     fn get_first_line_syntax(
@@ -686,6 +686,48 @@ mod tests {
         );
     }
 
+    #[test]
+    fn syntax_detection_ignored_suffix_falls_back_to_first_line() {
+        let mut test = SyntaxDetectionTest::new();
+
+        // By default a `.txt` file uses Plain Text, even with a shebang: the
+        // `.txt` extension wins and first-line detection is not reached.
+        assert_eq!(
+            test.syntax_for_file_with_content("test.txt", "#!/usr/bin/env bash"),
+            "Plain Text"
+        );
+
+        // Once `.txt` is an ignored suffix it is stripped before detection. The
+        // stripped name (`test`) has no extension, so detection falls back to the
+        // first line -- even though `.txt` is itself a registered extension that
+        // would otherwise match as Plain Text. See #2745.
+        test.syntax_mapping.insert_ignored_suffix(".txt");
+        assert_eq!(
+            test.syntax_for_file_with_content("test.txt", "#!/usr/bin/env bash"),
+            "Bourne Again Shell (bash)"
+        );
+        assert_eq!(
+            test.syntax_for_file_with_content("test.txt", "<?php"),
+            "PHP"
+        );
+
+        // A `.txt` file without a recognizable first line keeps no syntax (the
+        // caller renders it as plain text); we must not resurrect the shadowed
+        // `.txt` -> Plain Text match here.
+        assert_eq!(
+            test.syntax_for_file_with_content("notes.txt", "just some prose"),
+            "!no syntax!"
+        );
+
+        // Stripping that exposes a real extension still works: `.dev` is ignored,
+        // and the remaining `.json` extension is detected as usual.
+        test.syntax_mapping.insert_ignored_suffix(".dev");
+        assert_eq!(
+            test.syntax_for_file_with_content("config.json.dev", ""),
+            "JSON"
+        );
+    }
+
     #[test]
     fn syntax_detection_is_case_insensitive() {
         let mut test = SyntaxDetectionTest::new();

src/bin/bat/app.rs

@@ -38,6 +38,15 @@ pub fn env_no_color() -> bool {
     env::var_os("NO_COLOR").is_some_and(|x| !x.is_empty())
 }
 
+fn parse_strip_ansi_value(raw: Option<&str>, flag_name: &str) -> StripAnsiMode {
+    match raw {
+        Some("never") | None => StripAnsiMode::Never,
+        Some("always") => StripAnsiMode::Always,
+        Some("auto") => StripAnsiMode::Auto,
+        _ => unreachable!("other values for {flag_name} are not allowed"),
+    }
+}
+
 enum HelpType {
     Short,
     Long,
@@ -458,16 +467,32 @@ impl App {
                         4
                     },
                 ),
-            strip_ansi: match self
+            strip_ansi: {
-                .matches
+                let sanitize = parse_strip_ansi_value(
+                    self.matches
+                        .get_one::<String>("sanitize")
+                        .map(|s| s.as_str()),
+                    "--sanitize",
+                );
+                let strip_ansi = parse_strip_ansi_value(
+                    self.matches
                         .get_one::<String>("strip-ansi")
-                .map(|s| s.as_str())
+                        .map(|s| s.as_str()),
-            {
+                    "--strip-ansi",
-                Some("never") => StripAnsiMode::Never,
+                );
-                Some("always") => StripAnsiMode::Always,
+                // --sanitize implies --strip-ansi to the same value.
-                Some("auto") => StripAnsiMode::Auto,
+                if sanitize != StripAnsiMode::Never {
-                _ => unreachable!("other values for --strip-ansi are not allowed"),
+                    sanitize
+                } else {
+                    strip_ansi
+                }
             },
+            sanitize: parse_strip_ansi_value(
+                self.matches
+                    .get_one::<String>("sanitize")
+                    .map(|s| s.as_str()),
+                "--sanitize",
+            ),
             quiet_empty: self.matches.get_flag("quiet-empty"),
             unbuffered: self.matches.get_flag("unbuffered"),
             theme: theme(self.theme_options()).to_string(),

src/bin/bat/clap_app.rs

@@ -480,6 +480,24 @@ pub fn build_app(interactive_output: bool) -> Command {
                 language is plain text. Possible values: auto, always, *never*.")
                 .hide_short_help(true)
         )
+        .arg(
+            Arg::new("sanitize")
+                .long("sanitize")
+                .overrides_with("sanitize")
+                .value_name("when")
+                .value_parser(["auto", "always", "never"])
+                .default_value("never")
+                .hide_default_value(true)
+                .help("Sanitize untrusted input for safe display (auto, always, *never*)")
+                .long_help("Specify when to sanitize input bytes for safe terminal display. \
+                Implies --strip-ansi to the same value, and additionally substitutes \
+                terminal-active control bytes (cursor moves, charset switches, beep, etc.) \
+                and Unicode bidi / zero-width formatting characters with the Unicode \
+                replacement character (U+FFFD). Tab, LF, FF, and CRLF pass through. Useful \
+                for displaying untrusted file content (e.g. file-manager preview panes). \
+                Possible values: auto, always, *never*.")
+                .hide_short_help(true)
+        )
         .arg(
             Arg::new("style")
                 .long("style")

src/bin/bat/main.rs

@@ -152,7 +152,14 @@ pub fn get_languages(config: &Config, cache_dir: &Path) -> Result<String> {
         let comma_separator = ", ";
         let separator = " ";
         // Line-wrapping for the possible file extension overflow.
-        let desired_width = config.term_width - longest - separator.len();
+        // Clamp instead of subtracting: a tiny `--terminal-width` (smaller than the
+        // longest language name) would otherwise underflow `usize` and wrap to a huge
+        // value, silently disabling wrapping (and panicking in debug/overflow-checked
+        // builds). Mirrors the `saturating_sub` fix applied to `print_snip` in #3804.
+        let desired_width = config
+            .term_width
+            .saturating_sub(longest)
+            .saturating_sub(separator.len());
 
         let style = if config.colored_output {
             Green.normal()
@@ -420,11 +427,9 @@ fn run() -> Result<bool> {
             if app.matches.get_flag("list-languages") {
                 let languages: String = get_languages(&config, cache_dir)?;
                 let inputs: Vec<Input> = vec![Input::from_reader(Box::new(languages.as_bytes()))];
-                let plain_config = Config {
+                let mut plain_config = config.clone();
-                    style_components: StyleComponents::new(StyleComponent::Plain.components(false)),
+                plain_config.style_components =
-                    paging_mode: PagingMode::QuitIfOneScreen,
+                    StyleComponents::new(StyleComponent::Plain.components(false));
-                    ..Default::default()
-                };
                 run_controller(inputs, &plain_config, cache_dir)
             } else if app.matches.get_flag("list-themes") {
                 list_themes(&config, config_dir, cache_dir, app.theme_options())?;

src/config.rs

@@ -111,6 +111,9 @@ pub struct Config<'a> {
     // Whether or not to strip ANSI escape codes from the input
     pub strip_ansi: StripAnsiMode,
 
+    // Substitute terminal-active and spoofing-relevant bytes; implies strip_ansi.
+    pub sanitize: StripAnsiMode,
+
     /// Whether or not to produce no output when input is empty
     pub quiet_empty: bool,
 

src/diff.rs

@@ -1,12 +1,13 @@
 #![cfg(feature = "git")]
 
+use gix::diff::blob::pipeline::{Mode, WorktreeRoots};
+use gix::diff::blob::{Algorithm, HunkIter, ResourceKind};
+use gix::object::tree::EntryKind;
+use path_abs::PathInfo;
 use std::collections::HashMap;
-use std::fs;
 use std::path::Path;
 
-use git2::{DiffOptions, IntoCString, Repository};
+#[derive(Copy, Clone, Debug, PartialEq, Eq)]
-
-#[derive(Copy, Clone, Debug)]
 pub enum LineChange {
     Added,
     RemovedAbove,
@@ -16,68 +17,181 @@ pub enum LineChange {
 
 pub type LineChanges = HashMap<u32, LineChange>;
 
+fn collect_changes_from_hunks(hunks: HunkIter) -> Option<LineChanges> {
+    let mut changes: LineChanges = HashMap::new();
+    for hunk in hunks {
+        if hunk.before.is_empty() && !hunk.after.is_empty() {
+            for line in hunk.after {
+                changes.insert(line + 1, LineChange::Added);
+            }
+        } else if hunk.after.is_empty() && !hunk.before.is_empty() {
+            if hunk.after.start == 0 {
+                changes.insert(1, LineChange::RemovedAbove);
+            } else {
+                changes.insert(hunk.after.start, LineChange::RemovedBelow);
+            }
+        } else {
+            for line in hunk.after {
+                changes.insert(line + 1, LineChange::Modified);
+            }
+        }
+    }
+
+    Some(changes)
+}
+
 pub fn get_git_diff(filename: &Path) -> Option<LineChanges> {
-    let repo = Repository::discover(filename).ok()?;
+    let filepath_absolute = filename.canonicalize().ok()?;
-
+    let repository = gix::discover(filepath_absolute.parent().ok()?).ok()?;
-    let repo_path_absolute = fs::canonicalize(repo.workdir()?).ok()?;
+    let repo_path_absolute = repository.workdir()?.canonicalize().ok()?;
-
+    let filepath_relative_to_repo = gix::path::to_unix_separators_on_windows(gix::path::into_bstr(
-    let filepath_absolute = fs::canonicalize(filename).ok()?;
+        filepath_absolute.strip_prefix(&repo_path_absolute).ok()?,
-    let filepath_relative_to_repo = filepath_absolute.strip_prefix(&repo_path_absolute).ok()?;
+    ));
-
+    let index = repository.index_or_load_from_head_or_empty().ok()?;
-    let mut diff_options = DiffOptions::new();
+    let index_entry = index.entry_by_path(filepath_relative_to_repo.as_ref())?;
-    let pathspec = filepath_relative_to_repo.into_c_string().ok()?;
+    let mut cache = repository
-    diff_options.pathspec(pathspec);
+        .diff_resource_cache(
-    diff_options.context_lines(0);
+            Mode::ToGit,
-
+            WorktreeRoots {
-    let diff = repo
+                old_root: None,
-        .diff_index_to_workdir(None, Some(&mut diff_options))
+                new_root: repository.workdir().map(Path::to_path_buf),
+            },
+        )
         .ok()?;
-
+    cache
-    let mut line_changes: LineChanges = HashMap::new();
+        .set_resource(
-
+            index_entry.id,
-    let mark_section =
+            index_entry.mode.to_tree_entry_mode()?.kind(),
-        |line_changes: &mut LineChanges, start: u32, end: i32, change: LineChange| {
+            filepath_relative_to_repo.as_ref(),
-            for line in start..=end as u32 {
+            ResourceKind::OldOrSource,
-                line_changes.insert(line, change);
+            &repository,
-            }
+        )
-        };
+        .ok()?;
-
+    cache
-    let _ = diff.foreach(
+        .set_resource(
-        &mut |_, _| true,
+            repository.object_hash().null(),
-        None,
+            EntryKind::Blob,
-        Some(&mut |delta, hunk| {
+            filepath_relative_to_repo.as_ref(),
-            let path = delta.new_file().path().unwrap_or_else(|| Path::new(""));
+            ResourceKind::NewOrDestination,
-
+            &repository,
-            if filepath_relative_to_repo != path {
+        )
-                return false;
+        .ok()?;
-            }
+    let diff = gix::diff::blob::diff_with_slider_heuristics(
-
+        Algorithm::Histogram,
-            let old_lines = hunk.old_lines();
+        &cache.prepare_diff().ok()?.interned_input(),
-            let new_start = hunk.new_start();
-            let new_lines = hunk.new_lines();
-            let new_end = (new_start + new_lines) as i32 - 1;
-
-            if old_lines == 0 && new_lines > 0 {
-                mark_section(&mut line_changes, new_start, new_end, LineChange::Added);
-            } else if new_lines == 0 && old_lines > 0 {
-                if new_start == 0 {
-                    mark_section(&mut line_changes, 1, 1, LineChange::RemovedAbove);
-                } else {
-                    mark_section(
-                        &mut line_changes,
-                        new_start,
-                        new_start as i32,
-                        LineChange::RemovedBelow,
-                    );
-                }
-            } else {
-                mark_section(&mut line_changes, new_start, new_end, LineChange::Modified);
-            }
-
-            true
-        }),
-        None,
     );
 
-    Some(line_changes)
+    collect_changes_from_hunks(diff.hunks())
+}
+
+#[cfg(test)]
+mod tests {
+    use super::{get_git_diff, LineChange};
+    use std::path::{Path, PathBuf};
+    use std::process::Command;
+
+    fn git(repo: &Path, args: &[&str]) {
+        let output = Command::new("git")
+            .args(args)
+            .current_dir(repo)
+            .output()
+            .expect("git command can run");
+        assert!(
+            output.status.success(),
+            "git {args:?} failed\nstdout: {}\nstderr: {}",
+            String::from_utf8_lossy(&output.stdout),
+            String::from_utf8_lossy(&output.stderr)
+        );
+    }
+
+    fn setup_repo() -> tempfile::TempDir {
+        let dir = tempfile::tempdir().expect("can create temporary directory");
+        let repo = dir.path();
+        git(repo, &["init"]);
+        git(repo, &["config", "user.email", "test@test.test"]);
+        git(repo, &["config", "user.name", "Test"]);
+        dir
+    }
+
+    fn create_and_track_file(repo: &tempfile::TempDir, filename: &str) -> PathBuf {
+        let filepath = repo.path().join(filename);
+        std::fs::write(&filepath, "file\n").expect("can write file");
+        git(repo.path(), &["add", filename]);
+        git(repo.path(), &["commit", "-m", "initial"]);
+
+        filepath.to_owned()
+    }
+
+    #[test]
+    fn not_a_git_repository() {
+        let dir = tempfile::tempdir().expect("can create temporary directory");
+        let file = dir.path().join("file.txt");
+        std::fs::write(&file, "line 1\n").expect("can write file");
+
+        assert_eq!(get_git_diff(&file), None);
+    }
+
+    #[test]
+    fn diff_is_calculated_against_index() {
+        let repo = setup_repo();
+        let file = create_and_track_file(&repo, "file.txt");
+
+        std::fs::write(&file, "line 1\nline 2 modified\n").expect("can write file");
+        // Add modified file to index -> should find 0 changes but not none
+        git(repo.path(), &["add", "file.txt"]);
+
+        assert_eq!(get_git_diff(&file).expect("empty map").len(), 0);
+    }
+
+    #[test]
+    fn diff_is_calculated_against_index_2() {
+        let repo = setup_repo();
+        let file = create_and_track_file(&repo, "file.txt");
+
+        std::fs::write(&file, "line 1\nline 2 modified\n").expect("can write file");
+        git(repo.path(), &["add", "file.txt"]);
+        // modify the second line again which should show a single change
+        std::fs::write(&file, "line 1\nline 2 modified again\n").expect("can write file");
+
+        assert_eq!(get_git_diff(&file).expect("one change").len(), 1);
+    }
+
+    #[test]
+    fn diff_is_calculated_correctly() {
+        let repo = setup_repo();
+        create_and_track_file(&repo, "committed.txt");
+        let filename = "committed2.txt";
+        let file = create_and_track_file(&repo, filename);
+
+        std::fs::write(&file, "file\nline 2 added\n").expect("can write file");
+        let mut changes = get_git_diff(&file).expect("multiple changes");
+        assert_eq!(changes.get(&2), Some(&LineChange::Added));
+        git(repo.path(), &["add", filename]);
+
+        std::fs::write(&file, "file\nline 2 modified\n").expect("can write file");
+        changes = get_git_diff(&file).expect("multiple changes");
+        assert_eq!(changes.get(&2), Some(&LineChange::Modified));
+        git(repo.path(), &["add", filename]);
+
+        std::fs::write(&file, "line 2 modified\nline 3 added").expect("can write file");
+        changes = get_git_diff(&file).expect("multiple changes");
+        assert_eq!(changes.get(&1), Some(&LineChange::RemovedAbove));
+        assert_eq!(changes.get(&2), Some(&LineChange::Added));
+    }
+
+    #[test]
+    fn faulty_git_version_does_not_panic() {
+        let repo = setup_repo();
+        let file = create_and_track_file(&repo, "file.txt");
+        std::fs::write(&file, "line 1\nline 2 modified\n").expect("can write file");
+        // changes are detected
+        assert_eq!(get_git_diff(&file).expect("one change").len(), 2);
+        // write invalid repositoryformatversion
+        git(
+            repo.path(),
+            &["config", "core.repositoryformatversion", "one"],
+        );
+        // changes are no longer detected
+        assert_eq!(get_git_diff(&file), None);
+    }
 }

src/lessopen.rs

@@ -14,6 +14,53 @@ use crate::{
     input::{Input, InputKind, InputReader, OpenedInput, OpenedInputKind},
 };
 
+/// Wrap `s` in POSIX single quotes. An embedded `'` is encoded as `'\''`
+/// (close, escaped quote, reopen) since `'...'` has no escape character.
+fn shell_quote(s: &str) -> String {
+    if !s.contains('\'') {
+        let mut quoted = String::with_capacity(s.len() + 2);
+        quoted.push('\'');
+        quoted.push_str(s);
+        quoted.push('\'');
+        return quoted;
+    }
+    let mut quoted = String::with_capacity(s.len() + 2);
+    quoted.push('\'');
+    for c in s.chars() {
+        if c == '\'' {
+            quoted.push_str("'\\''");
+        } else {
+            quoted.push(c);
+        }
+    }
+    quoted.push('\'');
+    quoted
+}
+
+/// Substitute the first `%s` in a $LESSOPEN/$LESSCLOSE template with a quoted value.
+fn shell_substitute(template: &str, replacement: &str) -> String {
+    template.replacen("%s", &shell_quote(replacement), 1)
+}
+
+/// Substitute the first two `%s` occurrences positionally in one pass; chaining
+/// `shell_substitute` twice would mis-target if `first` itself contains `%s`.
+fn shell_substitute_two(template: &str, first: &str, second: &str) -> String {
+    let mut out = String::with_capacity(template.len() + first.len() + second.len() + 4);
+    let mut remaining = template;
+    let mut to_substitute = [Some(first), Some(second)];
+    for slot in &mut to_substitute {
+        if let Some(idx) = remaining.find("%s") {
+            out.push_str(&remaining[..idx]);
+            out.push_str(&shell_quote(slot.take().unwrap()));
+            remaining = &remaining[idx + 2..];
+        } else {
+            break;
+        }
+    }
+    out.push_str(remaining);
+    out
+}
+
 /// Preprocess files and/or stdin using $LESSOPEN and $LESSCLOSE
 pub(crate) struct LessOpenPreprocessor {
     lessopen: String,
@@ -87,7 +134,7 @@ impl LessOpenPreprocessor {
                     None => return input.open(stdin, stdout_identifier),
                 };
 
-                let mut lessopen_command = shell(self.lessopen.replacen("%s", path_str, 1));
+                let mut lessopen_command = shell(shell_substitute(&self.lessopen, path_str));
                 lessopen_command.stdout(Stdio::piped());
 
                 let lessopen_output = match lessopen_command.execute_output() {
@@ -122,7 +169,7 @@ impl LessOpenPreprocessor {
                     let mut stdin_buffer = Vec::new();
                     stdin.read_to_end(&mut stdin_buffer)?;
 
-                    let mut lessopen_command = shell(self.lessopen.replacen("%s", "-", 1));
+                    let mut lessopen_command = shell(shell_substitute(&self.lessopen, "-"));
                     lessopen_command.stdout(Stdio::piped());
 
                     let lessopen_output = match lessopen_command.execute_input_output(&stdin_buffer)
@@ -181,7 +228,7 @@ impl LessOpenPreprocessor {
                         lessclose: self
                             .lessclose
                             .as_ref()
-                            .map(|s| s.replacen("%s", &path_str, 1).replacen("%s", &stdout, 1)),
+                            .map(|s| shell_substitute_two(s, &path_str, &stdout)),
                     }
                 } else {
                     Preprocessed {
@@ -189,7 +236,7 @@ impl LessOpenPreprocessor {
                         lessclose: self
                             .lessclose
                             .as_ref()
-                            .map(|s| s.replacen("%s", &path_str, 1).replacen("%s", "-", 1)),
+                            .map(|s| shell_substitute_two(s, &path_str, "-")),
                     }
                 },
             ))?,
@@ -384,4 +431,65 @@ mod tests {
 
         Ok(())
     }
+
+    #[test]
+    fn shell_quote_plain_filename() {
+        assert_eq!(super::shell_quote("file.txt"), "'file.txt'");
+        assert_eq!(super::shell_quote("with space.txt"), "'with space.txt'");
+        assert_eq!(super::shell_quote(""), "''");
+        assert_eq!(super::shell_quote("-"), "'-'");
+    }
+
+    #[test]
+    fn shell_quote_metacharacters_neutralised() {
+        assert_eq!(super::shell_quote("; rm -rf ~ ;"), "'; rm -rf ~ ;'");
+        assert_eq!(super::shell_quote("$(payload)"), "'$(payload)'");
+        assert_eq!(super::shell_quote("`payload`"), "'`payload`'");
+        assert_eq!(super::shell_quote("a|b&c;d"), "'a|b&c;d'");
+        assert_eq!(super::shell_quote("..>/dev/null"), "'..>/dev/null'");
+        assert_eq!(super::shell_quote("\n\t"), "'\n\t'");
+    }
+
+    #[test]
+    fn shell_quote_embedded_single_quote() {
+        assert_eq!(super::shell_quote("it's"), "'it'\\''s'");
+        assert_eq!(super::shell_quote("'"), "''\\'''");
+        assert_eq!(super::shell_quote("a'b'c"), "'a'\\''b'\\''c'");
+    }
+
+    #[test]
+    fn shell_substitute_only_replaces_first_percent_s() {
+        assert_eq!(super::shell_substitute("echo %s", "x"), "echo 'x'");
+        assert_eq!(super::shell_substitute("echo %s %s", "x"), "echo 'x' %s");
+    }
+
+    #[test]
+    fn shell_substitute_protects_against_filename_injection() {
+        let template = "|preproc %s";
+        let attacker_filename = "; rm -rf ~ ;";
+        let result = super::shell_substitute(template, attacker_filename);
+        assert_eq!(result, "|preproc '; rm -rf ~ ;'");
+    }
+
+    #[test]
+    fn shell_substitute_two_replaces_first_two_placeholders() {
+        assert_eq!(
+            super::shell_substitute_two("echo %s and %s done", "a", "b"),
+            "echo 'a' and 'b' done"
+        );
+        // A third %s is left alone.
+        assert_eq!(
+            super::shell_substitute_two("echo %s %s %s", "a", "b"),
+            "echo 'a' 'b' %s"
+        );
+    }
+
+    #[test]
+    fn shell_substitute_two_handles_percent_s_in_first_argument() {
+        // The chained-substitute approach would replace the injected `%s` here.
+        assert_eq!(
+            super::shell_substitute_two("a %s b %s c", "%s", "second"),
+            "a '%s' b 'second' c"
+        );
+    }
 }

src/preprocessor.rs

@@ -139,16 +139,111 @@ pub fn replace_nonprintable(
 /// Strips ANSI escape sequences from the input.
 pub fn strip_ansi(line: &str) -> String {
     let mut buffer = String::with_capacity(line.len());
-
     for seq in EscapeSequenceOffsetsIterator::new(line) {
         if let EscapeSequenceOffsets::Text { .. } = seq {
             buffer.push_str(&line[seq.index_of_start()..seq.index_past_end()]);
         }
     }
-
     buffer
 }
 
+/// Strips ANSI escape sequences and substitutes terminal-active control bytes
+/// and visual-spoofing Unicode codepoints (bidi, zero-width) with U+FFFD.
+pub fn sanitize(line: &str) -> String {
+    let stripped = strip_ansi(line);
+    let mut buffer = String::with_capacity(stripped.len());
+    let bytes = stripped.as_bytes();
+    let mut start = 0;
+    let mut i = 0;
+    // Skip directly to the next trigger byte instead of testing each one.
+    while let Some(off) = bytes[i..].iter().position(|&b| is_sanitize_trigger(b)) {
+        i += off;
+        let len = sanitize_at(bytes, i, &stripped, &mut buffer, &mut start);
+        i += len;
+    }
+    buffer.push_str(&stripped[start..]);
+    buffer
+}
+
+#[inline]
+fn is_sanitize_trigger(b: u8) -> bool {
+    // C0 controls minus \t \n \f; DEL; UTF-8 leads with dangerous codepoints.
+    matches!(b, 0x00..=0x08 | 0x0B | 0x0D..=0x1F | 0x7F | 0xC2 | 0xD8 | 0xE2 | 0xEF)
+}
+
+/// Substitutes the byte/sequence at `bytes[i]` (or passes it through on
+/// false-alarm trigger), flushing the prefix from `start`. Returns bytes consumed.
+fn sanitize_at(
+    bytes: &[u8],
+    i: usize,
+    full: &str,
+    buffer: &mut String,
+    start: &mut usize,
+) -> usize {
+    buffer.push_str(&full[*start..i]);
+    let consumed = match bytes[i] {
+        b'\r' if bytes.get(i + 1) == Some(&b'\n') => {
+            buffer.push_str("\r\n");
+            2
+        }
+        // 0xC2 leads U+0080..U+00FF; filter the C1 range.
+        0xC2 if matches!(bytes.get(i + 1), Some(0x80..=0x9F)) => {
+            buffer.push('\u{FFFD}');
+            2
+        }
+        0xE2 if is_dangerous_e2(bytes, i) => {
+            buffer.push('\u{FFFD}');
+            3
+        }
+        // 0xEF 0xBB 0xBF = U+FEFF (BOM / zero-width no-break space).
+        0xEF if bytes.get(i + 1) == Some(&0xBB) && bytes.get(i + 2) == Some(&0xBF) => {
+            buffer.push('\u{FFFD}');
+            3
+        }
+        // 0xD8 0x9C = U+061C (Arabic letter mark, a bidi control). The rest of
+        // the 0xD8 block is ordinary Arabic text.
+        0xD8 if bytes.get(i + 1) == Some(&0x9C) => {
+            buffer.push('\u{FFFD}');
+            2
+        }
+        // False-alarm trigger: pass the full UTF-8 sequence through.
+        lead @ (0xC2 | 0xD8 | 0xE2 | 0xEF) => {
+            let n = utf8_len_from_lead(lead);
+            buffer.push_str(&full[i..i + n]);
+            n
+        }
+        _ => {
+            buffer.push('\u{FFFD}');
+            1
+        }
+    };
+    *start = i + consumed;
+    consumed
+}
+
+#[inline]
+fn is_dangerous_e2(bytes: &[u8], i: usize) -> bool {
+    // U+200B..D (zero-width), U+200E..F (LRM/RLM), U+202A..E (bidi embedding
+    // and override), U+2066..9 (bidi isolates).
+    matches!(
+        (bytes.get(i + 1), bytes.get(i + 2)),
+        (Some(0x80), Some(0x8B..=0x8F | 0xAA..=0xAE)) | (Some(0x81), Some(0xA6..=0xA9))
+    )
+}
+
+#[inline]
+fn utf8_len_from_lead(lead: u8) -> usize {
+    if lead < 0x80 {
+        1
+    } else if lead < 0xE0 {
+        2
+    } else if lead < 0xF0 {
+        3
+    } else {
+        4
+    }
+}
+
 /// 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 {
@@ -270,6 +365,113 @@ fn test_strip_ansi() {
     );
 }
 
+#[test]
+fn test_strip_ansi_8bit_c1_introducers() {
+    assert_eq!(strip_ansi("a\u{9B}31mRED\u{9B}0mb"), "aREDb");
+    assert_eq!(strip_ansi("a\x1bP1;0|payload\x1b\\b"), "ab");
+    assert_eq!(strip_ansi("a\u{90}body\u{9C}b"), "ab");
+}
+
+#[test]
+fn test_strip_ansi_single_char_esc() {
+    // strip_ansi must consume both bytes of single-byte ESC sequences (RIS, DECSC, keypad, VT52).
+    assert_eq!(strip_ansi("a\x1bcb"), "ab");
+    assert_eq!(strip_ansi("a\x1b7b\x1b8c"), "abc");
+    assert_eq!(strip_ansi("a\x1b=b\x1b>c"), "abc");
+    assert_eq!(strip_ansi("a\x1bZb"), "ab");
+}
+
+#[test]
+fn test_strip_ansi_preserves_control_bytes() {
+    // strip_ansi removes only ANSI escape sequences; control bytes pass through.
+    assert_eq!(strip_ansi("safe\rEVIL"), "safe\rEVIL");
+    assert_eq!(strip_ansi("a\x08b\x07c\x0E\x0Fd"), "a\x08b\x07c\x0E\x0Fd");
+}
+
+#[test]
+fn test_sanitize_substitutes_dangerous_bytes() {
+    let r = '\u{FFFD}';
+    assert_eq!(sanitize("safe\rEVIL"), format!("safe{r}EVIL"));
+    assert_eq!(sanitize("a\x08b"), format!("a{r}b"));
+    assert_eq!(sanitize("a\x07b"), format!("a{r}b"));
+    assert_eq!(sanitize("a\x0Bb"), format!("a{r}b"));
+    assert_eq!(sanitize("a\x0Eb\x0Fc"), format!("a{r}b{r}c"));
+    assert_eq!(sanitize("a\u{8D}b"), format!("a{r}b"));
+    assert_eq!(sanitize("a\u{85}b"), format!("a{r}b"));
+    assert_eq!(sanitize("trailing\r"), format!("trailing{r}"));
+    assert_eq!(sanitize("a\x7Fb"), format!("a{r}b"));
+}
+
+#[test]
+fn test_sanitize_substitutes_bidi_and_zero_width() {
+    let r = '\u{FFFD}';
+    // Trojan-Source bidi formatting: U+202A..U+202E
+    assert_eq!(sanitize("a\u{202A}b"), format!("a{r}b"));
+    assert_eq!(sanitize("a\u{202E}b"), format!("a{r}b"));
+    // Bidi isolates: U+2066..U+2069
+    assert_eq!(sanitize("a\u{2066}b"), format!("a{r}b"));
+    assert_eq!(sanitize("a\u{2069}b"), format!("a{r}b"));
+    // Zero-width: U+200B..U+200D
+    assert_eq!(sanitize("a\u{200B}b"), format!("a{r}b"));
+    assert_eq!(sanitize("a\u{200D}b"), format!("a{r}b"));
+    // BOM in middle of file: U+FEFF
+    assert_eq!(sanitize("a\u{FEFF}b"), format!("a{r}b"));
+}
+
+#[test]
+fn test_sanitize_substitutes_every_bidi_control() {
+    // Unicode Bidi_Control is exactly these 12 codepoints. Covering only some
+    // of them leaves a reordering attack available through the rest.
+    let r = '\u{FFFD}';
+    for c in [
+        '\u{061C}', '\u{200E}', '\u{200F}', '\u{202A}', '\u{202B}', '\u{202C}', '\u{202D}',
+        '\u{202E}', '\u{2066}', '\u{2067}', '\u{2068}', '\u{2069}',
+    ] {
+        assert_eq!(
+            sanitize(&format!("a{c}b")),
+            format!("a{r}b"),
+            "U+{:04X} was not substituted",
+            c as u32
+        );
+    }
+}
+
+#[test]
+fn test_sanitize_preserves_arabic_sharing_the_alm_lead_byte() {
+    // U+061C is reached via lead byte 0xD8, which also leads ordinary Arabic.
+    assert_eq!(
+        sanitize("\u{0600}\u{061F}\u{06FF}"),
+        "\u{0600}\u{061F}\u{06FF}"
+    );
+    assert_eq!(sanitize("مرحبا بالعالم"), "مرحبا بالعالم");
+}
+
+#[test]
+fn test_sanitize_preserves_legitimate_bytes() {
+    assert_eq!(sanitize("crlf\r\nline\r\n"), "crlf\r\nline\r\n");
+    assert_eq!(sanitize("a\tb\nc"), "a\tb\nc");
+    assert_eq!(sanitize("plain ascii"), "plain ascii");
+    assert_eq!(sanitize("üñíçödé"), "üñíçödé");
+    // FF (U+000C) passes through; section separator in C source / Emacs Lisp.
+    assert_eq!(sanitize("section1\x0Csection2"), "section1\x0Csection2");
+    // Common Unicode that shares a UTF-8 lead byte with dangerous codepoints
+    // must pass through unchanged.
+    assert_eq!(sanitize("snowman ☃ moon ☾"), "snowman ☃ moon ☾");
+    assert_eq!(sanitize("emoji 🎉 ☃"), "emoji 🎉 ☃");
+    assert_eq!(sanitize("0xC2 lead: ÿ ñ ç"), "0xC2 lead: ÿ ñ ç");
+    assert_eq!(sanitize("CJK 漢字 emoji 🦀"), "CJK 漢字 emoji 🦀");
+}
+
+#[test]
+fn test_sanitize_strips_ansi() {
+    let r = '\u{FFFD}';
+    // sanitize is a strict superset of strip_ansi.
+    assert_eq!(sanitize("a\x1B[31mb\x1B[0mc"), "abc");
+    assert_eq!(sanitize("a\u{9B}31mb\u{9B}0mc"), "abc");
+    // ANSI then dangerous byte: ANSI gone, byte substituted.
+    assert_eq!(sanitize("\x1B[31mhello\rEVIL"), format!("hello{r}EVIL"));
+}
+
 #[test]
 fn test_strip_overstrike() {
     // Bold: X\x08X (same char repeated)

src/pretty_printer.rs

@@ -192,6 +192,15 @@ impl<'a> PrettyPrinter<'a> {
         self
     }
 
+    /// Whether to sanitize untrusted input for safe display (default: never)
+    ///
+    /// Strips ANSI escape sequences and additionally substitutes terminal-active
+    /// control bytes and bidi / zero-width codepoints with U+FFFD.
+    pub fn sanitize(&mut self, mode: StripAnsiMode) -> &mut Self {
+        self.config.sanitize = mode;
+        self
+    }
+
     /// Text wrapping mode (default: do not wrap)
     pub fn wrapping_mode(&mut self, mode: WrappingMode) -> &mut Self {
         self.config.wrapping_mode = mode;

src/printer.rs

@@ -30,7 +30,8 @@ use crate::input::OpenedInput;
 use crate::line_range::{MaxBufferedLineNumber, RangeCheckResult};
 use crate::output::OutputHandle;
 use crate::preprocessor::{
-    expand_tabs, replace_nonprintable, sanitize_for_terminal, strip_ansi, strip_overstrike,
+    expand_tabs, replace_nonprintable, sanitize, sanitize_for_terminal, strip_ansi,
+    strip_overstrike,
 };
 use crate::style::StyleComponent;
 use crate::terminal::{as_terminal_escaped, to_ansi_color};
@@ -210,6 +211,7 @@ pub(crate) struct InteractivePrinter<'a> {
     background_color_highlight: Option<Color>,
     consecutive_empty_lines: usize,
     strip_ansi: bool,
+    sanitize: bool,
     strip_overstrike: bool,
 }
 
@@ -273,7 +275,9 @@ impl<'a> InteractivePrinter<'a> {
 
         let needs_to_match_syntax = (!is_printing_binary
             || matches!(config.binary, BinaryBehavior::AsText))
-            && (config.colored_output || config.strip_ansi == StripAnsiMode::Auto);
+            && (config.colored_output
+                || config.strip_ansi == StripAnsiMode::Auto
+                || config.sanitize == StripAnsiMode::Auto);
 
         let (is_plain_text, strip_overstrike, highlighter_from_set) = if needs_to_match_syntax {
             // Determine the type of syntax for highlighting
@@ -319,6 +323,14 @@ impl<'a> InteractivePrinter<'a> {
             _ => false,
         };
 
+        let sanitize = match config.sanitize {
+            _ if config.show_nonprintable => false,
+            StripAnsiMode::Always => true,
+            StripAnsiMode::Auto if is_plain_text => false,
+            StripAnsiMode::Auto => true,
+            _ => false,
+        };
+
         Ok(InteractivePrinter {
             panel_width,
             colors,
@@ -332,6 +344,7 @@ impl<'a> InteractivePrinter<'a> {
             background_color_highlight,
             consecutive_empty_lines: 0,
             strip_ansi,
+            sanitize,
             strip_overstrike,
         })
     }
@@ -603,11 +616,23 @@ impl Printer for InteractivePrinter<'_> {
         let title = "8<";
         let title_count = title.chars().count();
 
-        let snip_left = "─ ".repeat((self.config.term_width - panel_count - (title_count / 2)) / 4);
+        let snip_left = "─ ".repeat(
+            self.config
+                .term_width
+                .saturating_sub(panel_count)
+                .saturating_sub(title_count / 2)
+                / 4,
+        );
         let snip_left_count = snip_left.chars().count(); // Can't use .len() with Unicode.
 
-        let snip_right =
+        let snip_right = " ─".repeat(
-            " ─".repeat((self.config.term_width - panel_count - snip_left_count - title_count) / 2);
+            self.config
+                .term_width
+                .saturating_sub(panel_count)
+                .saturating_sub(snip_left_count)
+                .saturating_sub(title_count)
+                / 2,
+        );
 
         writeln!(
             handle,
@@ -663,8 +688,10 @@ impl Printer for InteractivePrinter<'_> {
                 }
             }
 
-            // If ANSI escape sequences are supposed to be stripped, do it before syntax highlighting.
+            // Sanitize is the strict superset; otherwise strip-ansi alone.
-            if self.strip_ansi {
+            if self.sanitize {
+                line = sanitize(&line).into()
+            } else if self.strip_ansi {
                 line = strip_ansi(&line).into()
             }
 

src/syntax_mapping/builtins/common/50-json.toml

@@ -1,3 +1,3 @@
 # JSON Lines is a simple variation of JSON #2535
 [mappings]
-"JSON" = ["*.jsonl", "*.jsonc", "*.jsonld", "*.geojson", "*.ndjson"]
+"JSON" = ["*.jsonl", "*.jsonc", "*.jsonld", "*.geojson", "*.ndjson", "*.code-workspace"]

src/syntax_mapping/builtins/common/50-just.toml

@@ -0,0 +1,2 @@
+[mappings]
+"Makefile" = ["justfile", "Justfile", ".justfile", "*.justfile"]

src/syntax_mapping/builtins/common/50-ssh.toml

@@ -1,2 +1,2 @@
 [mappings]
-"SSH Config" = ["**/.ssh/config"]
+"SSH Config" = ["**/.ssh/**/config"]

src/syntax_mapping/builtins/linux/50-dnf.toml

@@ -0,0 +1,2 @@
+[mappings]
+"INI" = ["/etc/dnf/dnf.conf", "/etc/yum.repos.d/*.repo"]

src/syntax_mapping/builtins/unix-family/50-ghostty.toml

@@ -0,0 +1,2 @@
+[mappings]
+"INI" = ["**/ghostty/**/*.ghostty", "**/ghostty/themes/*"]

src/vscreen.rs

@@ -386,12 +386,23 @@ impl<'a> EscapeSequenceOffsetsIterator<'a> {
     }
 
     fn next_text(&mut self) -> Option<EscapeSequenceOffsets> {
-        self.chars_take_while(|c| c != '\x1B')
+        self.chars_take_while(|c| !is_sequence_introducer(c))
             .map(|(start, end)| EscapeSequenceOffsets::Text { start, end })
     }
 
     fn next_sequence(&mut self) -> Option<EscapeSequenceOffsets> {
         let (start_sequence, c) = self.chars.next().expect("to not be finished");
+
+        // Handle 8-bit C1 introducers as their 7-bit `ESC <x>` equivalents.
+        match c {
+            '\u{9B}' => return self.next_csi_body(start_sequence),
+            '\u{9D}' => return self.next_osc_body(start_sequence),
+            '\u{90}' | '\u{98}' | '\u{9E}' | '\u{9F}' => {
+                return self.next_string_terminated_body(start_sequence)
+            }
+            _ => {}
+        }
+
         match self.chars.peek() {
             None => Some(EscapeSequenceOffsets::Unknown {
                 start: start_sequence,
@@ -400,12 +411,24 @@ impl<'a> EscapeSequenceOffsetsIterator<'a> {
 
             Some((_, ']')) => self.next_osc(start_sequence),
             Some((_, '[')) => self.next_csi(start_sequence),
-            Some((i, c)) => match c {
+            // 7-bit DCS/SOS/PM/APC: `ESC P/X/^/_` introduces a string body.
+            Some((_, 'P' | 'X' | '^' | '_')) => {
+                self.chars.next();
+                self.next_string_terminated_body(start_sequence)
+            }
+            Some(&(i, c)) => match c {
                 '\x20'..='\x2F' => self.next_nf(start_sequence),
-                c => Some(EscapeSequenceOffsets::Unknown {
+                c => {
+                    // Single-byte ESC sequence (RIS, DECSC/DECRC, keypad, VT52 etc.).
+                    let end = match self.chars.next() {
+                        Some((j, fc)) => j + fc.len_utf8(),
+                        None => i + c.len_utf8(),
+                    };
+                    Some(EscapeSequenceOffsets::Unknown {
                         start: start_sequence,
-                    end: i + c.len_utf8(),
+                        end,
-                }),
+                    })
+                }
             },
         }
     }
@@ -413,12 +436,27 @@ impl<'a> EscapeSequenceOffsetsIterator<'a> {
     fn next_osc(&mut self, start_sequence: usize) -> Option<EscapeSequenceOffsets> {
         let (osc_open_index, osc_open_char) = self.chars.next().expect("to not be finished");
         debug_assert_eq!(osc_open_char, ']');
+        let start_command = osc_open_index + osc_open_char.len_utf8();
+        Some(self.read_osc_body(start_sequence, start_command))
+    }
 
+    /// OSC body parser entered after the 8-bit introducer U+009D was consumed.
+    fn next_osc_body(&mut self, start_sequence: usize) -> Option<EscapeSequenceOffsets> {
+        let start_command = start_sequence + '\u{9D}'.len_utf8();
+        Some(self.read_osc_body(start_sequence, start_command))
+    }
+
+    fn read_osc_body(
+        &mut self,
+        start_sequence: usize,
+        start_command: usize,
+    ) -> EscapeSequenceOffsets {
         let mut start_terminator: usize;
         let mut end_sequence: usize;
 
         loop {
-            match self.chars_take_while(|c| !matches!(c, '\x07' | '\x1B')) {
+            // ST is BEL, ESC `\\`, or U+009C.
+            match self.chars_take_while(|c| !matches!(c, '\x07' | '\x1B' | '\u{9C}')) {
                 None => {
                     start_terminator = self.text.len();
                     end_sequence = start_terminator;
@@ -437,6 +475,11 @@ impl<'a> EscapeSequenceOffsetsIterator<'a> {
                     break;
                 }
 
+                Some((ti, '\u{9C}')) => {
+                    end_sequence = ti + '\u{9C}'.len_utf8();
+                    break;
+                }
+
                 Some((ti, '\x1B')) => {
                     match self.chars.next() {
                         Some((i, '\\')) => {
@@ -466,20 +509,82 @@ impl<'a> EscapeSequenceOffsetsIterator<'a> {
             }
         }
 
-        Some(EscapeSequenceOffsets::OSC {
+        EscapeSequenceOffsets::OSC {
             start_sequence,
-            start_command: osc_open_index + osc_open_char.len_utf8(),
+            start_command,
             start_terminator,
             end: end_sequence,
+        }
+    }
+
+    /// DCS/SOS/PM/APC body parser. Emitted as `Unknown` so the body is stripped.
+    fn next_string_terminated_body(
+        &mut self,
+        start_sequence: usize,
+    ) -> Option<EscapeSequenceOffsets> {
+        let mut end_sequence: usize;
+
+        loop {
+            match self.chars_take_while(|c| !matches!(c, '\x07' | '\x1B' | '\u{9C}')) {
+                None => {
+                    end_sequence = self.text.len();
+                    break;
+                }
+                Some((_, end)) => {
+                    end_sequence = end;
+                }
+            }
+
+            match self.chars.next() {
+                Some((ti, '\x07')) => {
+                    end_sequence = ti + '\x07'.len_utf8();
+                    break;
+                }
+                Some((ti, '\u{9C}')) => {
+                    end_sequence = ti + '\u{9C}'.len_utf8();
+                    break;
+                }
+                Some((ti, '\x1B')) => match self.chars.next() {
+                    Some((i, '\\')) => {
+                        end_sequence = i + '\\'.len_utf8();
+                        break;
+                    }
+                    None => {
+                        end_sequence = ti + '\x1B'.len_utf8();
+                        break;
+                    }
+                    _ => {}
+                },
+                None => break,
+                Some((_, tc)) => {
+                    panic!("this should not be reached: char {tc:?}")
+                }
+            }
+        }
+
+        Some(EscapeSequenceOffsets::Unknown {
+            start: start_sequence,
+            end: end_sequence,
         })
     }
 
     fn next_csi(&mut self, start_sequence: usize) -> Option<EscapeSequenceOffsets> {
         let (csi_open_index, csi_open_char) = self.chars.next().expect("to not be finished");
         debug_assert_eq!(csi_open_char, '[');
+        Some(self.read_csi_body(start_sequence, csi_open_index + csi_open_char.len_utf8()))
+    }
 
-        let start_parameters: usize = csi_open_index + csi_open_char.len_utf8();
+    /// CSI body parser entered after the 8-bit introducer U+009B was consumed.
+    fn next_csi_body(&mut self, start_sequence: usize) -> Option<EscapeSequenceOffsets> {
+        let start_parameters = start_sequence + '\u{9B}'.len_utf8();
+        Some(self.read_csi_body(start_sequence, start_parameters))
+    }
 
+    fn read_csi_body(
+        &mut self,
+        start_sequence: usize,
+        start_parameters: usize,
+    ) -> EscapeSequenceOffsets {
         // Keep iterating while within the range of `0x30-0x3F`.
         let mut start_intermediates: usize = start_parameters;
         if let Some((_, end)) = self.chars_take_while(|c| matches!(c, '\x30'..='\x3F')) {
@@ -498,13 +603,13 @@ impl<'a> EscapeSequenceOffsetsIterator<'a> {
             Some((i, c)) => i + c.len_utf8(),
         };
 
-        Some(EscapeSequenceOffsets::CSI {
+        EscapeSequenceOffsets::CSI {
             start_sequence,
             start_parameters,
             start_intermediates,
             start_final_byte,
             end: end_of_sequence,
-        })
+        }
     }
 
     fn next_nf(&mut self, start_sequence: usize) -> Option<EscapeSequenceOffsets> {
@@ -543,13 +648,25 @@ impl Iterator for EscapeSequenceOffsetsIterator<'_> {
     type Item = EscapeSequenceOffsets;
     fn next(&mut self) -> Option<Self::Item> {
         match self.chars.peek() {
-            Some((_, '\x1B')) => self.next_sequence(),
+            Some((_, c)) if is_sequence_introducer(*c) => self.next_sequence(),
             Some((_, _)) => self.next_text(),
             None => None,
         }
     }
 }
 
+/// True for ESC and the 8-bit C1 sequence introducers (DCS/SOS/CSI/OSC/PM/APC).
+#[inline]
+fn is_sequence_introducer(c: char) -> bool {
+    if (c as u32) < 0x80 {
+        return c == '\x1B';
+    }
+    matches!(
+        c,
+        '\u{90}' | '\u{98}' | '\u{9B}' | '\u{9D}' | '\u{9E}' | '\u{9F}'
+    )
+}
+
 /// An iterator over ANSI/VT escape sequences within a string.
 ///
 /// ## Example
@@ -717,6 +834,79 @@ mod tests {
         );
     }
 
+    #[test]
+    fn test_escape_sequence_offsets_iterator_parses_8bit_csi() {
+        let mut iter = EscapeSequenceOffsetsIterator::new("\u{9B}31m");
+        assert_eq!(
+            iter.next(),
+            Some(EscapeSequenceOffsets::CSI {
+                start_sequence: 0,
+                start_parameters: 2,
+                start_intermediates: 4,
+                start_final_byte: 4,
+                end: 5,
+            })
+        );
+        assert_eq!(iter.next(), None);
+    }
+
+    #[test]
+    fn test_escape_sequence_offsets_iterator_parses_8bit_osc_with_8bit_st() {
+        let mut iter = EscapeSequenceOffsetsIterator::new("\u{9D}0;title\u{9C}");
+        assert_eq!(
+            iter.next(),
+            Some(EscapeSequenceOffsets::OSC {
+                start_sequence: 0,
+                start_command: 2,
+                start_terminator: 9,
+                end: 11,
+            })
+        );
+        assert_eq!(iter.next(), None);
+    }
+
+    #[test]
+    fn test_escape_sequence_offsets_iterator_parses_7bit_dcs_consumes_body() {
+        let mut iter = EscapeSequenceOffsetsIterator::new("\x1BP1;0;|payload\x1B\\rest");
+        assert_eq!(
+            iter.next(),
+            Some(EscapeSequenceOffsets::Unknown { start: 0, end: 16 })
+        );
+        assert_eq!(
+            iter.next(),
+            Some(EscapeSequenceOffsets::Text { start: 16, end: 20 })
+        );
+        assert_eq!(iter.next(), None);
+    }
+
+    #[test]
+    fn test_escape_sequence_offsets_iterator_8bit_dcs_consumes_body() {
+        let mut iter = EscapeSequenceOffsetsIterator::new("\u{90}body\u{9C}rest");
+        assert_eq!(
+            iter.next(),
+            Some(EscapeSequenceOffsets::Unknown { start: 0, end: 8 })
+        );
+        assert_eq!(
+            iter.next(),
+            Some(EscapeSequenceOffsets::Text { start: 8, end: 12 })
+        );
+    }
+
+    #[test]
+    fn test_escape_sequence_offsets_iterator_truncated_dcs_consumes_to_eof() {
+        // Unterminated DCS must still be consumed, not emitted as `Text`.
+        let input = "\x1BPno-terminator";
+        let mut iter = EscapeSequenceOffsetsIterator::new(input);
+        assert_eq!(
+            iter.next(),
+            Some(EscapeSequenceOffsets::Unknown {
+                start: 0,
+                end: input.len(),
+            })
+        );
+        assert_eq!(iter.next(), None);
+    }
+
     #[test]
     fn test_escape_sequence_offsets_iterator_parses_csi() {
         let mut iter = EscapeSequenceOffsetsIterator::new("\x1B[m");

tests/examples/regression_tests/issue_2745.txt

@@ -0,0 +1,5 @@
+#!/usr/bin/env bash
+NAME="${0##*/}"
+for i in {1..3}; do
+    echo "hello $NAME $i"
+done

tests/integration_tests.rs

@@ -305,6 +305,19 @@ fn line_range_multiple() {
         .stdout("line 1\nline 2\nline 4\n");
 }
 
+#[test]
+fn snip_at_terminal_width_one_does_not_panic() {
+    bat()
+        .arg("multiline.txt")
+        .arg("--style=snip")
+        .arg("--color=always")
+        .arg("--terminal-width=1")
+        .arg("--line-range=1:2")
+        .arg("--line-range=4:4")
+        .assert()
+        .success();
+}
+
 #[test]
 fn line_range_multiple_with_context() {
     bat()
@@ -613,12 +626,18 @@ fn list_themes_to_piped_output() {
 }
 
 #[test]
+#[serial]
 fn list_languages() {
+    mocked_pagers::with_mocked_versions_of_more_and_most_in_path(|| {
         bat()
+            .env("PAGER", mocked_pagers::from("echo pager-output"))
             .arg("--list-languages")
+            .arg("--paging=never")
             .assert()
             .success()
-        .stdout(predicate::str::contains("Rust").normalize());
+            .stdout(predicate::str::contains("Rust").normalize())
+            .stdout(predicate::str::contains("pager-output").not());
+    });
 }
 
 #[test]
@@ -3831,6 +3850,158 @@ fn strip_ansi_auto_does_not_strip_ansi_when_plain_text_by_option() {
     assert!(output.contains("\x1B[33mYellow"))
 }
 
+#[test]
+fn sanitize_implies_strip_ansi() {
+    bat()
+        .arg("--style=plain")
+        .arg("--decorations=always")
+        .arg("--color=never")
+        .arg("--sanitize=always")
+        .write_stdin("\x1B[33mYellow\x1B[m")
+        .assert()
+        .success()
+        .stdout("Yellow");
+}
+
+#[test]
+fn sanitize_strips_osc_clipboard_hijack() {
+    // OSC 52 sets the system clipboard. A file containing this would silently
+    // overwrite the user's clipboard if displayed unfiltered.
+    bat()
+        .arg("--style=plain")
+        .arg("--decorations=always")
+        .arg("--color=never")
+        .arg("--sanitize=always")
+        .write_stdin("safe\x1B]52;c;cm0=\x07payload")
+        .assert()
+        .success()
+        .stdout("safepayload");
+}
+
+#[test]
+fn sanitize_strips_osc_8_hyperlink_spoof() {
+    // OSC 8 hyperlinks let displayed text point to an arbitrary URL.
+    bat()
+        .arg("--style=plain")
+        .arg("--decorations=always")
+        .arg("--color=never")
+        .arg("--sanitize=always")
+        .write_stdin("\x1B]8;;https://evil.example\x07click here\x1B]8;;\x07")
+        .assert()
+        .success()
+        .stdout("click here");
+}
+
+#[test]
+fn sanitize_strips_window_title_injection() {
+    // OSC 0/1/2 set the terminal window title.
+    bat()
+        .arg("--style=plain")
+        .arg("--decorations=always")
+        .arg("--color=never")
+        .arg("--sanitize=always")
+        .write_stdin("hello\x1B]0;evil-title\x07world")
+        .assert()
+        .success()
+        .stdout("helloworld");
+}
+
+#[test]
+fn sanitize_strips_8bit_csi() {
+    // 8-bit CSI introducer (U+009B) is the single-codepoint equivalent of ESC [.
+    bat()
+        .arg("--style=plain")
+        .arg("--decorations=always")
+        .arg("--color=never")
+        .arg("--sanitize=always")
+        .write_stdin("a\u{9B}31mRED\u{9B}0mb")
+        .assert()
+        .success()
+        .stdout("aREDb");
+}
+
+#[test]
+fn sanitize_substitutes_bare_cr() {
+    // Bare CR (not part of CRLF) is the line-overwrite forgery vector.
+    bat()
+        .arg("--style=plain")
+        .arg("--decorations=always")
+        .arg("--color=never")
+        .arg("--sanitize=always")
+        .write_stdin("safe\rEVIL")
+        .assert()
+        .success()
+        .stdout("safe\u{FFFD}EVIL");
+}
+
+#[test]
+fn sanitize_preserves_crlf() {
+    bat()
+        .arg("--style=plain")
+        .arg("--decorations=always")
+        .arg("--color=never")
+        .arg("--sanitize=always")
+        .write_stdin("line1\r\nline2\r\n")
+        .assert()
+        .success()
+        .stdout("line1\r\nline2\r\n");
+}
+
+#[test]
+fn sanitize_substitutes_bidi_controls() {
+    // Trojan-Source attack (CVE-2021-42574): U+202E (RLO) reorders display.
+    bat()
+        .arg("--style=plain")
+        .arg("--decorations=always")
+        .arg("--color=never")
+        .arg("--sanitize=always")
+        .write_stdin("admin\u{202E}check")
+        .assert()
+        .success()
+        .stdout("admin\u{FFFD}check");
+}
+
+#[test]
+fn sanitize_substitutes_zero_width() {
+    // Zero-width chars allow invisible content / identifier confusion.
+    bat()
+        .arg("--style=plain")
+        .arg("--decorations=always")
+        .arg("--color=never")
+        .arg("--sanitize=always")
+        .write_stdin("ad\u{200B}min")
+        .assert()
+        .success()
+        .stdout("ad\u{FFFD}min");
+}
+
+#[test]
+fn sanitize_preserves_form_feed_in_source() {
+    // FF (U+000C) is used as a section separator in C source and Emacs Lisp.
+    bat()
+        .arg("--style=plain")
+        .arg("--decorations=always")
+        .arg("--color=never")
+        .arg("--sanitize=always")
+        .write_stdin("section1\x0Csection2")
+        .assert()
+        .success()
+        .stdout("section1\x0Csection2");
+}
+
+#[test]
+fn sanitize_preserves_unicode_text() {
+    bat()
+        .arg("--style=plain")
+        .arg("--decorations=always")
+        .arg("--color=never")
+        .arg("--sanitize=always")
+        .write_stdin("snowman ☃ CJK 漢字 emoji 🦀")
+        .assert()
+        .success()
+        .stdout("snowman ☃ CJK 漢字 emoji 🦀");
+}
+
 // Tests that style components can be removed with `-component`.
 #[test]
 fn style_components_can_be_removed() {
@@ -4235,3 +4406,37 @@ fn tcl_shebang_detection_expect() {
         .assert()
         .success();
 }
+
+#[test]
+fn ignored_suffix_enables_first_line_detection() {
+    // A shebang shell script saved with a `.txt` extension is Plain Text by
+    // default (the extension wins). With `--ignored-suffix .txt` the suffix is
+    // stripped before detection, so it falls back to the first line and is
+    // highlighted exactly as if its language were forced to bash. See #2745.
+    let fixture = "regression_tests/issue_2745.txt";
+    let common = ["--color=always", "--decorations=never", "--style=plain"];
+
+    let stdout = |args: &[&str]| -> Vec<u8> {
+        let assert = bat()
+            .args(common)
+            .args(args)
+            .arg(fixture)
+            .assert()
+            .success();
+        assert.get_output().stdout.clone()
+    };
+
+    let forced_bash = stdout(&["--language", "bash"]);
+    let with_ignored_suffix = stdout(&["--ignored-suffix", ".txt"]);
+    let default = stdout(&[]);
+
+    // The fixture really is being highlighted (forcing bash is not a no-op).
+    assert!(
+        forced_bash.windows(2).any(|w| w == b"\x1b["),
+        "forced-bash output should contain ANSI color codes"
+    );
+    // With the ignored suffix, detection matches forced bash highlighting...
+    assert_eq!(with_ignored_suffix, forced_bash);
+    // ...while the default (extension wins) stays plain and differs.
+    assert_ne!(default, forced_bash);
+}

tests/syntax-tests/highlighted/Caddyfile/Caddyfile

@@ -0,0 +1,25 @@
+(logging) {
+    log {
+        output file /var/log/caddy.log
+    }
+}
+
+# a comment
+localhost:8080, example.com {
+    root * /var/www/site
+    file_server
+    reverse_proxy /.well-known/matrix/* localhost:8008 {
+        header_up Host {upstream_hostport}
+    }
+    import logging
+}
+
+www.example.com {
+    redir https://example.com{uri} permanent
+    import logging
+}
+
+status.example.com {
+    reverse_proxy localhost:3002
+    import logging
+}

tests/syntax-tests/source/Caddyfile/Caddyfile

@@ -0,0 +1,25 @@
+(logging) {
+    log {
+        output file /var/log/caddy.log
+    }
+}
+
+# a comment
+localhost:8080, example.com {
+    root * /var/www/site
+    file_server
+    reverse_proxy /.well-known/matrix/* localhost:8008 {
+        header_up Host {upstream_hostport}
+    }
+    import logging
+}
+
+www.example.com {
+    redir https://example.com{uri} permanent
+    import logging
+}
+
+status.example.com {
+    reverse_proxy localhost:3002
+    import logging
+}

tests/tester/mod.rs

@@ -1,15 +1,17 @@
+use gix::actor::SignatureRef;
+use gix::bstr::BString;
+use gix::bstr::ByteSlice;
+use gix::date::time::Format;
+use gix::date::Time;
+use gix::objs::tree;
 use std::env;
 use std::fs::{self, File};
 use std::io::Read;
-use std::path::{Path, PathBuf};
+use std::path::PathBuf;
 use std::process::Command;
 
 use tempfile::TempDir;
 
-use git2::build::CheckoutBuilder;
-use git2::Repository;
-use git2::Signature;
-
 pub struct BatTester {
     /// Temporary working directory
     temp_dir: TempDir,
@@ -59,35 +61,42 @@ impl Default for BatTester {
     }
 }
 
-fn create_sample_directory() -> Result<TempDir, git2::Error> {
+fn create_sample_directory() -> Result<TempDir, Box<dyn std::error::Error>> {
     // Create temp directory and initialize repository
     let temp_dir = TempDir::new().expect("Temp directory");
-    let repo = Repository::init(&temp_dir)?;
+    let repo = gix::init(&temp_dir)?;
+    let mut tree = gix::objs::Tree::empty();
 
-    // Copy over `sample.rs`
+    // Create sample.rs from snapshot file
-    let sample_path = temp_dir.path().join("sample.rs");
+    let blob_id = repo.write_blob_stream(File::open("tests/snapshots/sample.rs")?)?;
-    println!("{sample_path:?}");
+    let entry = tree::Entry {
-    fs::copy("tests/snapshots/sample.rs", &sample_path).expect("successful copy");
+        mode: tree::EntryMode::from(tree::EntryKind::Blob),
+        oid: blob_id.object()?.id,
+        filename: BString::from("sample.rs"),
+    };
+    tree.entries.push(entry);
+    let tree_id = repo.write_object(tree)?;
 
-    // Commit
+    let author = SignatureRef {
-    let mut index = repo.index()?;
+        name: "test".as_bytes().as_bstr(),
-    index.add_path(Path::new("sample.rs"))?;
+        email: "test@test.test".as_bytes().as_bstr(),
-
+        time: &Time::now_local_or_utc().format_or_unix(Format::Raw),
-    let oid = index.write_tree()?;
+    };
-    let signature = Signature::now("bat test runner", "bat@test.runner")?;
+    let commit_id = repo.commit_as(
-    let tree = repo.find_tree(oid)?;
+        author,
-    let _ = repo.commit(
+        author,
-        Some("HEAD"), //  point HEAD to our new commit
+        "HEAD",
-        &signature,   // author
-        &signature,   // committer
         "initial commit",
-        &tree,
+        tree_id,
-        &[],
+        gix::commit::NO_PARENT_IDS,
-    );
+    )?;
-    let mut opts = CheckoutBuilder::new();
+    assert_eq!(commit_id, repo.head_id()?);
-    repo.checkout_head(Some(opts.force()))?;
 
-    fs::copy("tests/snapshots/sample.modified.rs", &sample_path).expect("successful copy");
+    fs::copy(
+        "tests/snapshots/sample.modified.rs",
+        temp_dir.path().join("sample.rs"),
+    )
+    .expect("successful copy");
 
     Ok(temp_dir)
 }