Merge pull request #3517 from akirk/strip-overstriking

Improve native man pages and command help syntax highlighting by stripping overstriking

CHANGELOG.md

@@ -2,8 +2,9 @@
 
 ## Features
 
-## Bugfixes
+- Improve native man pages and command help syntax highlighting by stripping overstriking, see #3517 (@akirk)
 
+## Bugfixes
 - `--help` now correctly honors `--pager=builtin`. See #3516 (@keith-hall)
 - `--help` now correctly honors custom themes. See #3524 (@keith-hall)
 

README.md

@@ -195,18 +195,13 @@ bat main.cpp | xclip
 `MANPAGER` environment variable:
 
 ```bash
-export MANPAGER="sh -c 'awk '\''{ gsub(/\x1B\[[0-9;]*m/, \"\", \$0); gsub(/.\x08/, \"\", \$0); print }'\'' | bat -p -lman'"
+export MANPAGER="bat -plman"
 man 2 select
 ```
 (replace `bat` with `batcat` if you are on Debian or Ubuntu)
 
 If you prefer to have this bundled in a new command, you can also use [`batman`](https://github.com/eth-p/bat-extras/blob/master/doc/batman.md).
 
-> [!WARNING]  
-> This will [not work](https://github.com/sharkdp/bat/issues/1145) out of the box with Mandoc's `man` implementation.
->
-> Please either use `batman`, or convert the shell script to a [shebang executable](https://en.wikipedia.org/wiki/Shebang_(Unix)) and point `MANPAGER` to that.
-
 Note that the [Manpage syntax](assets/syntaxes/02_Extra/Manpage.sublime-syntax) is developed in this repository and still needs some work.
 
 #### `prettier` / `shfmt` / `rustfmt`

src/preprocessor.rs

@@ -149,6 +149,35 @@ pub fn strip_ansi(line: &str) -> String {
     buffer
 }
 
+/// Strips overstrike sequences (backspace formatting) from input.
+///
+/// Overstrike formatting is used by man pages and some help output:
+/// - Bold: `X\x08X` (character, backspace, same character)
+/// - Underline: `_\x08X` (underscore, backspace, character)
+///
+/// This function removes these sequences, keeping only the visible character.
+/// `first_backspace` is the position of the first backspace in the line.
+pub fn strip_overstrike(line: &str, first_backspace: usize) -> String {
+    let mut output = String::with_capacity(line.len());
+    output.push_str(&line[..first_backspace]);
+    output.pop();
+
+    let mut remaining = &line[first_backspace + 1..];
+
+    loop {
+        if let Some(pos) = remaining.find('\x08') {
+            output.push_str(&remaining[..pos]);
+            output.pop();
+            remaining = &remaining[pos + 1..];
+        } else {
+            output.push_str(remaining);
+            break;
+        }
+    }
+
+    output
+}
+
 #[derive(Debug, PartialEq, Clone, Copy, Default)]
 pub enum StripAnsiMode {
     #[default]
@@ -211,3 +240,24 @@ fn test_strip_ansi() {
         "multiple sequences"
     );
 }
+
+#[test]
+fn test_strip_overstrike() {
+    // Bold: X\x08X (same char repeated)
+    assert_eq!(strip_overstrike("H\x08Hello", 1), "Hello");
+
+    // Underline: _\x08X (underscore before char)
+    assert_eq!(strip_overstrike("_\x08Hello", 1), "Hello");
+
+    // Multiple overstrike sequences
+    assert_eq!(strip_overstrike("B\x08Bo\x08ol\x08ld\x08d", 1), "Bold");
+
+    // Backspace at start of line (nothing to pop)
+    assert_eq!(strip_overstrike("\x08Hello", 0), "Hello");
+
+    // Multiple consecutive backspaces
+    assert_eq!(strip_overstrike("ABC\x08\x08\x08XYZ", 3), "XYZ");
+
+    // Unicode with overstrike
+    assert_eq!(strip_overstrike("ä\x08äöü", 2), "äöü");
+}

src/printer.rs

@@ -29,8 +29,7 @@ use crate::error::*;
 use crate::input::OpenedInput;
 use crate::line_range::{MaxBufferedLineNumber, RangeCheckResult};
 use crate::output::OutputHandle;
-use crate::preprocessor::strip_ansi;
-use crate::preprocessor::{expand_tabs, replace_nonprintable};
+use crate::preprocessor::{expand_tabs, replace_nonprintable, strip_ansi, strip_overstrike};
 use crate::style::StyleComponent;
 use crate::terminal::{as_terminal_escaped, to_ansi_color};
 use crate::vscreen::{AnsiStyle, EscapeSequence, EscapeSequenceIterator};
@@ -199,6 +198,7 @@ pub(crate) struct InteractivePrinter<'a> {
     background_color_highlight: Option<Color>,
     consecutive_empty_lines: usize,
     strip_ansi: bool,
+    strip_overstrike: bool,
 }
 
 impl<'a> InteractivePrinter<'a> {
@@ -263,17 +263,22 @@ impl<'a> InteractivePrinter<'a> {
             || matches!(config.binary, BinaryBehavior::AsText))
             && (config.colored_output || config.strip_ansi == StripAnsiMode::Auto);
 
-        let (is_plain_text, highlighter_from_set) = if needs_to_match_syntax {
+        let (is_plain_text, strip_overstrike, highlighter_from_set) = if needs_to_match_syntax {
             // Determine the type of syntax for highlighting
             const PLAIN_TEXT_SYNTAX: &str = "Plain Text";
+            const MANPAGE_SYNTAX: &str = "Manpage";
+            const COMMAND_HELP_SYNTAX: &str = "Command Help";
             match assets.get_syntax(config.language, input, &config.syntax_mapping) {
                 Ok(syntax_in_set) => (
                     syntax_in_set.syntax.name == PLAIN_TEXT_SYNTAX,
+                    syntax_in_set.syntax.name == MANPAGE_SYNTAX
+                        || syntax_in_set.syntax.name == COMMAND_HELP_SYNTAX,
                     Some(HighlighterFromSet::new(syntax_in_set, theme)),
                 ),
 
                 Err(Error::UndetectedSyntax(_)) => (
                     true,
+                    false,
                     Some(
                         assets
                             .find_syntax_by_name(PLAIN_TEXT_SYNTAX)?
@@ -285,7 +290,7 @@ impl<'a> InteractivePrinter<'a> {
                 Err(e) => return Err(e),
             }
         } else {
-            (false, None)
+            (false, false, None)
         };
 
         // Determine when to strip ANSI sequences
@@ -310,6 +315,7 @@ impl<'a> InteractivePrinter<'a> {
             background_color_highlight,
             consecutive_empty_lines: 0,
             strip_ansi,
+            strip_overstrike,
         })
     }
 
@@ -622,6 +628,12 @@ impl Printer for InteractivePrinter<'_> {
                 }
             };
 
+            if self.strip_overstrike {
+                if let Some(pos) = line.find('\x08') {
+                    line = strip_overstrike(&line, pos).into();
+                }
+            }
+
             // If ANSI escape sequences are supposed to be stripped, do it before syntax highlighting.
             if self.strip_ansi {
                 line = strip_ansi(&line).into()

tests/examples/git-commit.man

@@ -0,0 +1,30 @@
+GIT-COMMIT(1)                     Git Manual                     GIT-COMMIT(1)
+
+NNAAMMEE
+       git-commit - Record changes to the repository
+
+SSYYNNOOPPSSIISS
+       ggiitt ccoommmmiitt [--aa | ----iinntteerraaccttiivvee | ----ppaattcchh] [--ss] [--vv] [--uu[_<_m_o_d_e_>]] [----aammeenndd]
+               [----ddrryy--rruunn] [(--cc | --CC | ----ssqquuaasshh) _<_c_o_m_m_i_t_> | ----ffiixxuupp [(aammeenndd|rreewwoorrdd)::]_<_c_o_m_m_i_t_>]
+               [--FF _<_f_i_l_e_> | --mm _<_m_s_g_>] [----rreesseett--aauutthhoorr] [----aallllooww--eemmppttyy]
+               [----aallllooww--eemmppttyy--mmeessssaaggee] [----nnoo--vveerriiffyy] [--ee] [----aauutthhoorr==_<_a_u_t_h_o_r_>]
+               [----ddaattee==_<_d_a_t_e_>] [----cclleeaannuupp==_<_m_o_d_e_>] [----[nnoo--]ssttaattuuss]
+               [--ii | --oo] [----ppaatthhssppeecc--ffrroomm--ffiillee==_<_f_i_l_e_> [----ppaatthhssppeecc--ffiillee--nnuull]]
+               [(----ttrraaiilleerr _<_t_o_k_e_n_>[(==|::)_<_v_a_l_u_e_>])...] [--SS[_<_k_e_y_i_d_>]]
+               [----] [_<_p_a_t_h_s_p_e_c_>...]
+
+DDEESSCCRRIIPPTTIIOONN
+       Create a new commit containing the current contents of the index and
+       the given log message describing the changes. The new commit is a
+       direct child of HEAD, usually the tip of the current branch, and the
+       branch is updated to point to it (unless no branch is associated with
+       the working tree, in which case HHEEAADD is "detached" as described in ggiitt--
+       cchheecckkoouutt(1)).
+
+       The content to be committed can be specified in several ways:
+
+        1. by using ggiitt--aadddd(1) to incrementally "add" changes to the index
+           before using the ccoommmmiitt command (Note: even modified files must be
+           "added");
+
+        2. by using ggiitt--rrmm(1) to remove files from the working tree and the

tests/examples/overstrike.txt

@@ -0,0 +1 @@
+BBold tteexxtt and _u_n_d_e_r_l_i_n_e

tests/integration_tests.rs

@@ -2541,6 +2541,73 @@ fn binary_as_text() {
         .stderr("");
 }
 
+#[test]
+fn no_strip_overstrike_for_plain_text() {
+    // Overstrike is preserved for plain text files (no syntax highlighting)
+    bat()
+        .arg("--color=never")
+        .arg("--decorations=never")
+        .arg("overstrike.txt")
+        .assert()
+        .success()
+        .stdout("B\x08Bold t\x08te\x08ex\x08xt\x08t and _\x08u_\x08n_\x08d_\x08e_\x08r_\x08l_\x08i_\x08n_\x08e\n")
+        .stderr("");
+}
+
+#[test]
+fn strip_overstrike_with_syntax_highlighting() {
+    // Overstrike is stripped for certain syntax highlighting like command help.
+    bat()
+        .arg("--force-colorization")
+        .arg("--language=help")
+        .arg("overstrike.txt")
+        .assert()
+        .success()
+        .stdout(predicate::str::contains("Bold text and underline"))
+        .stderr("");
+}
+
+#[test]
+fn strip_overstrike_for_manpage_syntax() {
+    // Overstrike is stripped for .man files (Manpage syntax)
+    bat()
+        .arg("--force-colorization")
+        .arg("git-commit.man")
+        .assert()
+        .success()
+        .stdout(predicate::str::contains("NAME"))
+        .stdout(predicate::str::contains("git-commit - Record changes"))
+        .stdout(predicate::str::is_match(r"\x1b\[38;[0-9;]+m--interactive\x1b\[").unwrap())
+        .stderr("");
+}
+
+#[test]
+fn no_strip_overstrike_for_other_syntax() {
+    // Overstrike is NOT stripped for other syntaxes (e.g., Rust)
+    bat()
+        .arg("--force-colorization")
+        .arg("--language=rust")
+        .arg("overstrike.txt")
+        .assert()
+        .success()
+        .stdout(predicate::str::contains("\x08"))
+        .stderr("");
+}
+
+#[test]
+fn show_all_shows_backspace_with_caret_notation() {
+    // --show-all should display backspace characters (not strip them)
+    bat()
+        .arg("--show-all")
+        .arg("--nonprintable-notation=caret")
+        .arg("--decorations=never")
+        .arg("overstrike.txt")
+        .assert()
+        .success()
+        .stdout(predicate::str::contains("^H"))
+        .stderr("");
+}
+
 #[test]
 fn no_paging_arg() {
     bat()