Merge branch 'master' into makefile-syntax-for-justfiles

src/bin/bat/app.rs

@@ -590,7 +590,17 @@ impl App {
 
         // Plain if `--plain` is specified at least once.
         if self.matches.get_count("plain") > 0 {
-            return Some(StyleComponents(HashSet::from([StyleComponent::Plain])));
+            let mut components = HashSet::from([StyleComponent::Plain]);
+            // When --diff is active, preserve change markers and snip separators
+            // so that diff output remains visually useful.
+            if self.matches.try_contains_id("diff").unwrap_or_default()
+                && self.matches.get_flag("diff")
+            {
+                #[cfg(feature = "git")]
+                components.insert(StyleComponent::Changes);
+                components.insert(StyleComponent::Snip);
+            }
+            return Some(StyleComponents(components));
         }
 
         // Default behavior.

src/bin/bat/clap_app.rs

@@ -240,6 +240,7 @@ pub fn build_app(interactive_output: bool) -> Command {
         .arg(
             Arg::new("terminal-width")
                 .long("terminal-width")
+                .overrides_with("terminal-width")
                 .value_name("width")
                 .hide_short_help(true)
                 .allow_hyphen_values(true)
@@ -255,10 +256,13 @@ pub fn build_app(interactive_output: bool) -> Command {
                             })
                             .map_err(|e| e.to_string())
                     })
-                .help(
+                .help("Explicitly set the width of the terminal instead of determining it automatically.")
+                .long_help(
                     "Explicitly set the width of the terminal instead of determining it \
                      automatically. If prefixed with '+' or '-', the value will be treated \
-                     as an offset to the actual terminal width. See also: '--wrap'.",
+                     as an offset to the actual terminal width. This can also be configured \
+                     via the BAT_WIDTH environment variable (e.g. export BAT_WIDTH=\"100\"). \
+                     See also: '--wrap'.",
                 ),
         )
         .arg(

src/bin/bat/config.rs

@@ -153,6 +153,7 @@ fn get_args_from_str(content: &str) -> Result<Vec<OsString>, shell_words::ParseE
 pub fn get_args_from_env_vars() -> Vec<OsString> {
     [
         ("--tabs", "BAT_TABS"),
+        ("--terminal-width", "BAT_WIDTH"),
         ("--theme", bat::theme::env::BAT_THEME),
         ("--theme-dark", bat::theme::env::BAT_THEME_DARK),
         ("--theme-light", bat::theme::env::BAT_THEME_LIGHT),

src/bin/bat/main.rs

@@ -285,7 +285,28 @@ fn run_controller(inputs: Vec<Input>, config: &Config, cache_dir: &Path) -> Resu
 
 #[cfg(feature = "bugreport")]
 fn invoke_bugreport(app: &App, cache_dir: &Path) {
-    use bugreport::{bugreport, collector::*, format::Markdown};
+    use bugreport::{bugreport, collector::*, format::Markdown, report::ReportEntry};
+
+    struct ColorSchemeCollector;
+
+    impl Collector for ColorSchemeCollector {
+        fn description(&self) -> &str {
+            "Detected terminal color scheme"
+        }
+
+        fn collect(
+            &mut self,
+            _: &bugreport::CrateInfo,
+        ) -> std::result::Result<ReportEntry, CollectionError> {
+            let color_scheme = bat::theme::color_scheme(bat::theme::DetectColorScheme::Always);
+            let text = match color_scheme {
+                Some(bat::theme::ColorScheme::Dark) => "dark",
+                Some(bat::theme::ColorScheme::Light) => "light",
+                None => "not detected",
+            };
+            Ok(ReportEntry::Text(text.to_string()))
+        }
+    }
     let pager = bat::config::get_pager_executable(
         app.matches.get_one::<String>("pager").map(|s| s.as_str()),
     )
@@ -307,6 +328,9 @@ fn invoke_bugreport(app: &App, cache_dir: &Path) {
             "BAT_STYLE",
             "BAT_TABS",
             "BAT_THEME",
+            "BAT_WIDTH",
+            bat::theme::env::BAT_THEME_DARK,
+            bat::theme::env::BAT_THEME_LIGHT,
             "COLORTERM",
             "LANG",
             "LC_ALL",
@@ -326,6 +350,7 @@ fn invoke_bugreport(app: &App, cache_dir: &Path) {
             custom_assets_metadata,
         ))
         .info(DirectoryEntries::new("Custom assets", cache_dir))
+        .info(ColorSchemeCollector)
         .info(CompileTimeInformation::default());
 
     #[cfg(feature = "paging")]

src/input.rs

@@ -207,7 +207,7 @@ impl<'a> Input<'a> {
                     kind: OpenedInputKind::StdIn,
                     description,
                     metadata: self.metadata,
-                    reader: InputReader::new(stdin),
+                    reader: InputReader::try_new(stdin)?,
                 })
             }
 
@@ -236,14 +236,14 @@ impl<'a> Input<'a> {
                         file = input_identifier.into_inner().expect("The file was lost in the clircle::Identifier, this should not have happened...");
                     }
 
-                    InputReader::new(BufReader::new(file))
+                    InputReader::try_new(BufReader::new(file))?
                 },
             }),
             InputKind::CustomReader(reader) => Ok(OpenedInput {
                 description,
                 kind: OpenedInputKind::CustomReader,
                 metadata: self.metadata,
-                reader: InputReader::new(BufReader::new(reader)),
+                reader: InputReader::try_new(BufReader::new(reader))?,
             }),
         }
     }
@@ -257,28 +257,29 @@ pub(crate) struct InputReader<'a> {
 }
 
 impl<'a> InputReader<'a> {
-    pub(crate) fn new<R: BufRead + 'a>(mut reader: R) -> InputReader<'a> {
-        let mut first_line = vec![];
-        reader.read_until(b'\n', &mut first_line).ok();
+    #[cfg(test)]
+    pub(crate) fn new<R: BufRead + 'a>(reader: R) -> InputReader<'a> {
+        Self::try_new(reader).expect("reading the first line failed")
+    }
 
-        let content_type = if first_line.is_empty() {
-            None
-        } else {
-            Some(content_inspector::inspect(&first_line[..]))
-        };
+    pub(crate) fn try_new<R: BufRead + 'a>(mut reader: R) -> io::Result<InputReader<'a>> {
+        let mut first_line = vec![];
+        reader.read_until(b'\n', &mut first_line)?;
+
+        let content_type = inspect_content_type(&first_line);
 
         if content_type == Some(ContentType::UTF_16LE) {
-            read_utf16_line(&mut reader, &mut first_line, 0x00, 0x0A).ok();
+            read_utf16_line(&mut reader, &mut first_line, 0x00, 0x0A)?;
         } else if content_type == Some(ContentType::UTF_16BE) {
-            read_utf16_line(&mut reader, &mut first_line, 0x0A, 0x00).ok();
+            read_utf16_line(&mut reader, &mut first_line, 0x0A, 0x00)?;
         }
 
-        InputReader {
+        Ok(InputReader {
             inner: Box::new(reader),
             first_line,
             content_type,
             unbuffered: false,
-        }
+        })
     }
 
     pub(crate) fn read_line(&mut self, buf: &mut Vec<u8>) -> io::Result<bool> {
@@ -319,6 +320,25 @@ impl<'a> InputReader<'a> {
     }
 }
 
+fn inspect_content_type(first_line: &[u8]) -> Option<ContentType> {
+    if first_line.is_empty() {
+        return None;
+    }
+
+    let content_type = content_inspector::inspect(first_line);
+    if content_type == ContentType::UTF_8 && has_zip_signature(first_line) {
+        Some(ContentType::BINARY)
+    } else {
+        Some(content_type)
+    }
+}
+
+fn has_zip_signature(bytes: &[u8]) -> bool {
+    [b"PK\x03\x04", b"PK\x05\x06", b"PK\x07\x08"]
+        .into_iter()
+        .any(|signature| bytes.starts_with(signature))
+}
+
 fn read_utf16_line<R: BufRead>(
     reader: &mut R,
     buf: &mut Vec<u8>,
@@ -374,6 +394,43 @@ fn basic() {
     assert!(buffer.is_empty());
 }
 
+#[test]
+fn zip_magic_headers_are_treated_as_binary() {
+    for content in [b"PK\x03\x04hello", b"PK\x05\x06hello", b"PK\x07\x08hello"] {
+        let reader = InputReader::new(&content[..]);
+        assert_eq!(Some(ContentType::BINARY), reader.content_type);
+    }
+}
+
+#[test]
+fn non_zip_pk_prefix_is_not_treated_as_binary() {
+    assert_eq!(
+        Some(ContentType::UTF_8),
+        inspect_content_type(b"PK\x03\x03hello")
+    );
+}
+
+#[test]
+fn input_open_returns_initial_read_errors() {
+    struct FailingRead;
+
+    impl Read for FailingRead {
+        fn read(&mut self, _buf: &mut [u8]) -> io::Result<usize> {
+            Err(io::Error::other("initial read failed"))
+        }
+    }
+
+    let input = Input::from_reader(Box::new(FailingRead));
+    let result = input.open(io::empty(), None);
+
+    assert!(result.is_err());
+    assert!(result
+        .err()
+        .unwrap()
+        .to_string()
+        .contains("initial read failed"));
+}
+
 #[test]
 fn utf16le() {
     let content = b"\xFF\xFE\x73\x00\x0A\x00\x64\x00";

src/lessopen.rs

@@ -155,7 +155,7 @@ impl LessOpenPreprocessor {
 
         Ok(OpenedInput {
             kind,
-            reader: InputReader::new(BufReader::new(
+            reader: InputReader::try_new(BufReader::new(
                 if matches!(self.kind, LessOpenKind::TempFile) {
                     let lessopen_string = match String::from_utf8(lessopen_stdout) {
                         Ok(string) => string,
@@ -192,7 +192,7 @@ impl LessOpenPreprocessor {
                             .map(|s| s.replacen("%s", &path_str, 1).replacen("%s", "-", 1)),
                     }
                 },
-            )),
+            ))?,
             metadata: input.metadata,
             description: input.description,
         })

src/output.rs

@@ -23,6 +23,18 @@ pub struct BuiltinPager {
 impl BuiltinPager {
     fn new() -> Self {
         let pager = minus::Pager::new();
+
+        let mut input_register = minus::input::HashedEventRegister::default();
+        input_register.add_key_events(&["home"], |_, _| {
+            minus::input::InputEvent::UpdateUpperMark(0)
+        });
+        input_register.add_key_events(&["end"], |_, _| {
+            minus::input::InputEvent::UpdateUpperMark(usize::MAX)
+        });
+        pager
+            .set_input_classifier(Box::new(input_register))
+            .expect("failed to set input classifier on newly created pager");
+
         let handle = {
             let pager = pager.clone();
             Some(spawn(move || {

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;

src/syntax_mapping.rs

@@ -17,9 +17,16 @@ use ignored_suffixes::IgnoredSuffixes;
 mod builtin;
 pub mod ignored_suffixes;
 
-fn make_glob_matcher(from: &str) -> Result<GlobMatcher> {
+/// Whether a glob pattern should be matched case-sensitively or case-insensitively.
+#[derive(Debug, Clone, Copy, PartialEq, Eq)]
+pub(crate) enum Case {
+    Sensitive,
+    Insensitive,
+}
+
+fn make_glob_matcher(from: &str, case: Case) -> Result<GlobMatcher> {
     let matcher = GlobBuilder::new(from)
-        .case_insensitive(true)
+        .case_insensitive(matches!(case, Case::Insensitive))
         .literal_separator(true)
         .build()?
         .compile_matcher();
@@ -97,7 +104,14 @@ impl<'a> SyntaxMapping<'a> {
     }
 
     pub fn insert(&mut self, from: &str, to: MappingTarget<'a>) -> Result<()> {
-        let matcher = make_glob_matcher(from)?;
+        let matcher = make_glob_matcher(from, Case::Insensitive)?;
+        self.custom_mappings.push((matcher, to));
+        Ok(())
+    }
+
+    /// Like [`Self::insert`], but the glob pattern is matched case-sensitively.
+    pub fn insert_case_sensitive(&mut self, from: &str, to: MappingTarget<'a>) -> Result<()> {
+        let matcher = make_glob_matcher(from, Case::Sensitive)?;
         self.custom_mappings.push((matcher, to));
         Ok(())
     }
@@ -261,4 +275,41 @@ mod tests {
             Some(MappingTarget::MapTo("alpha"))
         );
     }
+
+    #[test]
+    fn case_sensitive_custom_mappings_work() {
+        let mut map = SyntaxMapping::new();
+        map.insert_case_sensitive("MY_SPECIAL_FILE", MappingTarget::MapTo("Python"))
+            .ok();
+
+        // Exact case matches
+        assert_eq!(
+            map.get_syntax_for("/path/to/MY_SPECIAL_FILE"),
+            Some(MappingTarget::MapTo("Python"))
+        );
+        // Different case should NOT match the case-sensitive rule
+        assert_eq!(map.get_syntax_for("/path/to/my_special_file"), None);
+        assert_eq!(map.get_syntax_for("/path/to/My_Special_File"), None);
+    }
+
+    #[test]
+    fn builtin_mappings_build_is_case_sensitive() {
+        let map = SyntaxMapping::new();
+
+        // "BUILD" (uppercase) should map to Python via case-sensitive builtin
+        assert_eq!(
+            map.get_syntax_for("/path/to/BUILD"),
+            Some(MappingTarget::MapTo("Python"))
+        );
+        // "build" (lowercase) should still map to MapToUnknown
+        assert_eq!(
+            map.get_syntax_for("/path/to/build"),
+            Some(MappingTarget::MapToUnknown)
+        );
+        // Mixed case should NOT match the Python rule
+        assert_eq!(
+            map.get_syntax_for("/path/to/Build"),
+            Some(MappingTarget::MapToUnknown)
+        );
+    }
 }

src/syntax_mapping/builtin.rs

@@ -3,7 +3,7 @@ use std::env;
 use globset::GlobMatcher;
 use once_cell::sync::Lazy;
 
-use crate::syntax_mapping::{make_glob_matcher, MappingTarget};
+use crate::syntax_mapping::{make_glob_matcher, Case, MappingTarget};
 
 // Static syntax mappings generated from /src/syntax_mapping/builtins/ by the
 // build script (/build/syntax_mapping.rs).
@@ -53,8 +53,8 @@ include!(concat!(
 /// A failure to compile is a fatal error.
 ///
 /// Used internally by `Lazy<Option<GlobMatcher>>`'s lazy evaluation closure.
-fn build_matcher_fixed(from: &str) -> GlobMatcher {
-    make_glob_matcher(from).expect("A builtin fixed glob matcher failed to compile")
+fn build_matcher_fixed(from: &str, case: Case) -> GlobMatcher {
+    make_glob_matcher(from, case).expect("A builtin fixed glob matcher failed to compile")
 }
 
 /// Join a list of matcher segments to create a glob string, replacing all
@@ -64,7 +64,7 @@ fn build_matcher_fixed(from: &str) -> GlobMatcher {
 /// to compile.
 ///
 /// Used internally by `Lazy<Option<GlobMatcher>>`'s lazy evaluation closure.
-fn build_matcher_dynamic(segs: &[MatcherSegment]) -> Option<GlobMatcher> {
+fn build_matcher_dynamic(segs: &[MatcherSegment], case: Case) -> Option<GlobMatcher> {
     // join segments
     let mut buf = String::new();
     for seg in segs {
@@ -77,7 +77,7 @@ fn build_matcher_dynamic(segs: &[MatcherSegment]) -> Option<GlobMatcher> {
         }
     }
     // compile glob matcher
-    let matcher = make_glob_matcher(&buf).ok()?;
+    let matcher = make_glob_matcher(&buf, case).ok()?;
     Some(matcher)
 }
 

src/syntax_mapping/builtins/README.md

@@ -20,12 +20,10 @@ syntax mappings defined by all TOML files, and embed them into the binary.
 
 ## File syntax
 
-Each TOML file should contain a single section named `mappings`, with each of
-its keys being a language identifier (first column of `bat -L`; also referred to
-as "target").
+Each TOML file should contain a single section named `mappings`, with each of its keys being a language
+identifier (first column of `bat -L`; also referred to as "target").
 
-The value of each key should be an array of strings, with each item being a glob
-matcher. We will call each of these items a "rule".
+The value of each key should be an array of "rules". The rules are expected to be objects with a `glob` string and a `case_sensitive` boolean. For simplification, a rule can be just a glob string, which is shorthand for the default case insensitive mode.
 
 For example, if `foo-application` uses both TOML and YAML configuration files,
 we could write something like this:
@@ -98,6 +96,15 @@ like this:
 ]
 ```
 
+### Case sensitivity
+
+By default, all glob patterns are matched case-insensitively. To match a pattern case-sensitively, use the object form of the rule with the `case_sensitive` option:
+
+```toml
+[mappings]
+"Python" = [{ glob = "BUILD", case_sensitive = true }]
+```
+
 ## Ordering
 
 At compile time, all TOML files applicable to the target are processed in

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

@@ -0,0 +1,2 @@
+[mappings]
+"Python" = [{ glob = "BUILD", case_sensitive = true }]

src/syntax_mapping/builtins/common/50-dotnet-xml.toml

@@ -1,2 +1,2 @@
 [mappings]
-"XML" = ["*.csproj", "*.vbproj", "*.props", "*.targets"]
+"XML" = ["*.csproj", "*.vbproj", "*.props", "*.targets", "*.slnx"]

src/syntax_mapping/builtins/common/50-gcloud-cli-config.toml

@@ -0,0 +1,3 @@
+[mappings]
+"INI" = [".boto", "**/gcloud/configurations/config_*"]
+"Git Ignore" = [".gcloudignore"]

src/syntax_mapping/builtins/common/90-ignore-files.toml

@@ -0,0 +1,2 @@
+[mappings]
+"Git Ignore" = [".?*ignore"]