Merge pull request #3747 from mvanhorn/fix/3697-bat-auto-detect-language-from-stdin

chore: improve auto-detection of language from stdin via first-line patterns and add tests

src/assets.rs

@@ -363,11 +363,11 @@ impl HighlightingAssets {
         reader: &mut InputReader,
     ) -> Result<Option<SyntaxReferenceInSet<'_>>> {
         let syntax_set = self.get_syntax_set()?;
-        Ok(String::from_utf8(reader.first_line.clone())
+        Ok(std::str::from_utf8(&reader.first_line)
             .ok()
             .and_then(|l| {
                 // Strip UTF-8 BOM if present
-                let line = l.strip_prefix('\u{feff}').unwrap_or(&l);
+                let line = l.strip_prefix('\u{feff}').unwrap_or(l);
                 syntax_set.find_syntax_by_first_line(line)
             })
             .map(|syntax| SyntaxReferenceInSet { syntax, syntax_set }))

tests/integration_tests.rs

@@ -2559,6 +2559,145 @@ fn no_first_line_fallback_when_mapping_to_invalid_syntax() {
         .stderr(predicate::str::contains("unknown syntax: 'InvalidSyntax'"));
 }
 
+#[test]
+fn stdin_detects_bash_from_first_line() {
+    let content = "#!/bin/bash\necho hi\n";
+
+    let detected_output = bat()
+        .arg("--color=always")
+        .arg("--style=plain")
+        .write_stdin(content)
+        .assert()
+        .success()
+        .get_output()
+        .stdout
+        .clone();
+
+    let explicit_output = bat()
+        .arg("--color=always")
+        .arg("--style=plain")
+        .arg("--language=bash")
+        .write_stdin(content)
+        .assert()
+        .success()
+        .get_output()
+        .stdout
+        .clone();
+
+    assert_eq!(
+        from_utf8(&detected_output).expect("output is valid utf-8"),
+        from_utf8(&explicit_output).expect("output is valid utf-8")
+    );
+}
+
+#[test]
+fn stdin_detects_diff_from_first_line() {
+    let content = "diff --git a/x b/y\n--- a/x\n+++ b/y\n@@ -1 +1 @@\n-old\n+new\n";
+
+    let detected_output = bat()
+        .arg("--color=always")
+        .arg("--style=plain")
+        .write_stdin(content)
+        .assert()
+        .success()
+        .get_output()
+        .stdout
+        .clone();
+
+    let explicit_output = bat()
+        .arg("--color=always")
+        .arg("--style=plain")
+        .arg("--language=diff")
+        .write_stdin(content)
+        .assert()
+        .success()
+        .get_output()
+        .stdout
+        .clone();
+
+    assert_eq!(
+        from_utf8(&detected_output).expect("output is valid utf-8"),
+        from_utf8(&explicit_output).expect("output is valid utf-8")
+    );
+}
+
+#[test]
+fn empty_stdin_does_not_detect_syntax() {
+    bat()
+        .arg("--color=always")
+        .arg("--style=plain")
+        .write_stdin("")
+        .assert()
+        .success()
+        .stdout("")
+        .stderr("");
+}
+
+#[test]
+fn binary_stdin_does_not_detect_syntax_from_invalid_utf8_first_line() {
+    let content = b"#!/bin/bash\xff\necho hi\n";
+
+    let detected_output = bat()
+        .arg("--binary=as-text")
+        .arg("--color=always")
+        .arg("--style=plain")
+        .write_stdin(content.as_slice())
+        .assert()
+        .success()
+        .get_output()
+        .stdout
+        .clone();
+
+    let plain_text_output = bat()
+        .arg("--binary=as-text")
+        .arg("--color=always")
+        .arg("--style=plain")
+        .arg("--language=txt")
+        .write_stdin(content.as_slice())
+        .assert()
+        .success()
+        .get_output()
+        .stdout
+        .clone();
+
+    assert_eq!(
+        from_utf8(&detected_output).expect("output is valid utf-8"),
+        from_utf8(&plain_text_output).expect("output is valid utf-8")
+    );
+}
+
+#[test]
+fn explicit_language_overrides_stdin_first_line_detection() {
+    let content = "diff --git a/x b/y\n--- a/x\n+++ b/y\n@@ -1 +1 @@\n-old\n+new\n";
+
+    let detected_output = bat()
+        .arg("--color=always")
+        .arg("--style=plain")
+        .write_stdin(content)
+        .assert()
+        .success()
+        .get_output()
+        .stdout
+        .clone();
+
+    let explicit_output = bat()
+        .arg("--color=always")
+        .arg("--style=plain")
+        .arg("--language=json")
+        .arg("-")
+        .write_stdin(content)
+        .assert()
+        .success()
+        .get_output()
+        .stdout
+        .clone();
+
+    assert_ne!(
+        from_utf8(&detected_output).expect("output is valid utf-8"),
+        from_utf8(&explicit_output).expect("output is valid utf-8")
+    );
+}
+
 #[test]
 fn fallback_syntax_is_used_when_no_syntax_is_detected() {
     let content = "# comment\nfoo=bar\n";