Merge pull request #3816 from adnrivera/fix/ignored-suffix-firstline

Fix --ignored-suffix to fall back to first-line detection

CHANGELOG.md

@@ -22,6 +22,7 @@
 - Syntax highlighting for Python files using uv as script runner in shebang #3689 (@janlarres)
 
 ## Bugfixes
+- 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)

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() {
+            ignored_suffixes.try_with_stripped_suffix(file_name, |stripped_file_name| {
-            syntax =
+                self.get_syntax_for_file_extension(stripped_file_name, ignored_suffixes)
-                ignored_suffixes.try_with_stripped_suffix(file_name, |stripped_file_name| {
+                    .map(Some)
-                    // Note: recursion
+            })?;
-                    self.get_syntax_for_file_extension(stripped_file_name, ignored_suffixes)
+        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();

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

@@ -4248,3 +4248,37 @@ 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);
+}