Merge pull request #3640 from eyupcanakman/fix/binary-as-text-width-3631

Fix line wrapping for files with control characters

CHANGELOG.md

@@ -14,6 +14,7 @@
 - Add `--fallback-syntax`/`--fallback-language` to apply syntax highlighting only when auto-detection fails, see #1341 (@Xavrir)
 
 ## Bugfixes
+- Fix incorrect text width computation when using `--binary=as-text` with non-printable characters in caret notation, see #3640 and #3631 (@eyupcanakman)
 - Fix `BAT_CONFIG_DIR` pointing at system config directory causing duplicate flag errors. Closes #3589, see #3620 (@Xavrir)
 - Fix syntax highlighting for symlinked files when the symlink name has no extension but the target does. Closes #1001, see #3621 (@Xavrir)
 - Report error when pager is missing instead of silently falling back, see #3588 (@IMaloney)

src/printer.rs

@@ -37,6 +37,16 @@ use crate::wrapping::WrappingMode;
 use crate::BinaryBehavior;
 use crate::StripAnsiMode;
 
+// Return the displayed width of a character.
+//
+// Control characters (0x00..=0x1F and 0x7F) are rendered by the terminal
+// in caret notation (e.g. ^@, ^A, ..., ^?), which occupies two columns.
+// UnicodeWidthChar::width() returns None for these, so we map them to 2
+// here instead of the previous default of 0.
+fn char_width(c: char) -> usize {
+    c.width().unwrap_or(if c.is_control() { 2 } else { 0 })
+}
+
 const ANSI_UNDERLINE_ENABLE: EscapeSequence = EscapeSequence::CSI {
     raw_sequence: "\x1B[4m",
     parameters: "4",
@@ -793,7 +803,7 @@ impl Printer for InteractivePrinter<'_> {
 
                             for c in text.chars() {
                                 // calculate the displayed width for next character
-                                let cw = c.width().unwrap_or(0);
+                                let cw = char_width(c);
                                 current_width += cw;
 
                                 // Track whitespace positions for word wrapping.
@@ -868,10 +878,8 @@ impl Printer for InteractivePrinter<'_> {
                                     if let Some(rs) = rest_start {
                                         // Word wrap: carry remainder to next line.
                                         let remainder = line_buf[rs..].to_string();
-                                        let rem_width: usize = remainder
-                                            .chars()
-                                            .map(|ch| ch.width().unwrap_or(0))
-                                            .sum();
+                                        let rem_width: usize =
+                                            remainder.chars().map(char_width).sum();
                                         line_buf.clear();
                                         line_buf.push_str(&remainder);
                                         current_width = rem_width + cw;

tests/integration_tests.rs

@@ -2681,6 +2681,25 @@ fn binary_as_text() {
         .stderr("");
 }
 
+#[test]
+fn binary_as_text_control_char_width() {
+    // Control characters are displayed as caret notation (e.g. ^@) by the
+    // terminal, occupying 2 columns each. With 20 NUL bytes (40 columns) +
+    // "END" (3 columns) = 43 columns, wrapping at terminal width 40 must
+    // produce 2 lines, not 1. See #3631.
+    bat()
+        .arg("--binary=as-text")
+        .arg("--wrap=character")
+        .arg("--terminal-width=40")
+        .arg("--decorations=always")
+        .arg("--style=plain")
+        .arg("--color=never")
+        .arg("regression_tests/issue_3631.txt")
+        .assert()
+        .success()
+        .stdout(predicate::function(|s: &str| s.lines().count() == 2));
+}
+
 #[test]
 fn no_strip_overstrike_for_plain_text() {
     // Overstrike is preserved for plain text files (no syntax highlighting)