Add option to output result in C include file style (#242) (#246)

* feat: Implement "-i/-include" option

* make clippy happy

Suppress clippy warning for too many arguments in Printer struct

* add tests & use buffered reading for include mode to handle large files

* Update CHANGELOG

CHANGELOG.md

@@ -2,6 +2,7 @@
 
 ## Features
 
+- Add option to output result in C include file style, see #242 (@wpcwzy)
 - Add `--color-scheme` option, see #247 (@aticu)
 - Add `braille` character table, see #247 (@aticu)
 - Add command line argument to generate shell completion, see #155 (@friedz)

src/lib.rs

@@ -24,6 +24,13 @@ pub enum ByteCategory {
     NonAscii,
 }
 
+pub enum IncludeMode {
+    File(String), // filename
+    Stdin,
+    Slice,
+    Off,
+}
+
 #[derive(Copy, Clone, Debug, Default, ValueEnum)]
 #[non_exhaustive]
 pub enum CharacterTable {
@@ -267,6 +274,7 @@ pub struct PrinterBuilder<'a, Writer: Write> {
     base: Base,
     endianness: Endianness,
     character_table: CharacterTable,
+    include_mode: IncludeMode,
     color_scheme: ColorScheme,
 }
 
@@ -284,6 +292,7 @@ impl<'a, Writer: Write> PrinterBuilder<'a, Writer> {
             base: Base::Hexadecimal,
             endianness: Endianness::Big,
             character_table: CharacterTable::Default,
+            include_mode: IncludeMode::Off,
             color_scheme: ColorScheme::Default,
         }
     }
@@ -338,6 +347,11 @@ impl<'a, Writer: Write> PrinterBuilder<'a, Writer> {
         self
     }
 
+    pub fn include_mode(mut self, include: IncludeMode) -> Self {
+        self.include_mode = include;
+        self
+    }
+
     pub fn color_scheme(mut self, color_scheme: ColorScheme) -> Self {
         self.color_scheme = color_scheme;
         self
@@ -382,6 +396,7 @@ impl<'a, Writer: Write> PrinterBuilder<'a, Writer> {
                 Base::Hexadecimal => 2,
             },
             endianness: self.endianness,
+            include_mode: self.include_mode,
         }
     }
 }
@@ -412,6 +427,8 @@ pub struct Printer<'a, Writer: Write> {
     base_digits: u8,
     /// Whether to show groups in little or big endian format.
     endianness: Endianness,
+    /// Whether to output in C include file style.
+    include_mode: IncludeMode,
 }
 
 impl<'a, Writer: Write> Printer<'a, Writer> {
@@ -663,6 +680,36 @@ impl<'a, Writer: Write> Printer<'a, Writer> {
 
         let mut buf = BufReader::new(reader);
 
+        // special handler for include mode
+        match &self.include_mode {
+            // Input from a file
+            // Output like `unsigned char <filename>[] = { ... }; unsigned int <filename>_len = ...;`
+            IncludeMode::File(filename) => {
+                // convert non-alphanumeric characters to '_'
+                let var_name = filename
+                    .chars()
+                    .map(|c| if c.is_alphanumeric() { c } else { '_' })
+                    .collect::<String>();
+
+                writeln!(self.writer, "unsigned char {}[] = {{", var_name)?;
+
+                let total_bytes = self.print_bytes_in_include_style(&mut buf)?;
+
+                writeln!(self.writer, "}};")?;
+                writeln!(
+                    self.writer,
+                    "unsigned int {}_len = {};",
+                    var_name, total_bytes
+                )?;
+                return Ok(());
+            }
+            IncludeMode::Stdin | IncludeMode::Slice => {
+                self.print_bytes_in_include_style(&mut buf)?;
+                return Ok(());
+            }
+            IncludeMode::Off => {}
+        }
+
         let leftover = loop {
             // read a maximum of 8 * self.panels bytes from the reader
             if let Ok(n) = buf.read(&mut self.line_buf) {
@@ -803,6 +850,44 @@ impl<'a, Writer: Write> Printer<'a, Writer> {
 
         Ok(())
     }
+
+    /// Print the bytes in C include file style
+    /// Return the number of bytes read  
+    fn print_bytes_in_include_style<Reader: Read>(
+        &mut self,
+        buf: &mut BufReader<Reader>,
+    ) -> Result<usize, io::Error> {
+        let mut buffer = [0; 1024];
+        let mut total_bytes = 0;
+        let mut is_first_chunk = true;
+        let mut line_counter = 0;
+        loop {
+            match buf.read(&mut buffer) {
+                Ok(0) => break, // EOF
+                Ok(bytes_read) => {
+                    total_bytes += bytes_read;
+
+                    for &byte in &buffer[..bytes_read] {
+                        if line_counter % 12 == 0 {
+                            if !is_first_chunk || line_counter > 0 {
+                                writeln!(self.writer, ",")?;
+                            }
+                            // indentation of first line
+                            write!(self.writer, "  ")?;
+                            is_first_chunk = false;
+                        } else {
+                            write!(self.writer, ", ")?;
+                        }
+                        write!(self.writer, "0x{:02x}", byte)?;
+                        line_counter += 1;
+                    }
+                }
+                Err(e) => return Err(e),
+            }
+        }
+        writeln!(self.writer)?;
+        Ok(total_bytes)
+    }
 }
 
 #[cfg(test)]
@@ -825,6 +910,7 @@ mod tests {
             .with_base(Base::Hexadecimal)
             .endianness(Endianness::Big)
             .character_table(CharacterTable::Default)
+            .include_mode(IncludeMode::Off)
             .color_scheme(ColorScheme::Default)
             .build();
 
@@ -881,6 +967,7 @@ mod tests {
             .with_base(Base::Hexadecimal)
             .endianness(Endianness::Big)
             .character_table(CharacterTable::Default)
+            .include_mode(IncludeMode::Off)
             .color_scheme(ColorScheme::Default)
             .build();
         printer.display_offset(0xdeadbeef);
@@ -916,6 +1003,7 @@ mod tests {
             .with_base(Base::Hexadecimal)
             .endianness(Endianness::Big)
             .character_table(CharacterTable::Default)
+            .include_mode(IncludeMode::Off)
             .color_scheme(ColorScheme::Default)
             .build();
 
@@ -977,6 +1065,7 @@ mod tests {
             .with_base(Base::Hexadecimal)
             .endianness(Endianness::Big)
             .character_table(CharacterTable::Default)
+            .include_mode(IncludeMode::Off)
             .color_scheme(ColorScheme::Default)
             .build();
 
@@ -999,4 +1088,66 @@ mod tests {
         .to_owned();
         assert_print_all_output(input, expected_string);
     }
+
+    #[test]
+    fn include_mode_from_file() {
+        let input = io::Cursor::new(b"spamspamspamspamspam");
+        let expected_string = "unsigned char test_txt[] = {
+  0x73, 0x70, 0x61, 0x6d, 0x73, 0x70, 0x61, 0x6d, 0x73, 0x70, 0x61, 0x6d,
+  0x73, 0x70, 0x61, 0x6d, 0x73, 0x70, 0x61, 0x6d
+};
+unsigned int test_txt_len = 20;
+"
+        .to_owned();
+        let mut output = vec![];
+        let mut printer: Printer<Vec<u8>> = PrinterBuilder::new(&mut output)
+            .show_color(false)
+            .show_char_panel(true)
+            .show_position_panel(true)
+            .with_border_style(BorderStyle::Unicode)
+            .enable_squeezing(true)
+            .num_panels(2)
+            .group_size(1)
+            .with_base(Base::Hexadecimal)
+            .endianness(Endianness::Big)
+            .character_table(CharacterTable::Default)
+            .include_mode(IncludeMode::File("test.txt".to_owned()))
+            .color_scheme(ColorScheme::Default)
+            .build();
+
+        printer.print_all(input).unwrap();
+
+        let actual_string: &str = str::from_utf8(&output).unwrap();
+        assert_eq!(actual_string, expected_string)
+    }
+
+    #[test]
+    fn include_mode_from_stdin() {
+        let input = io::Cursor::new(b"spamspamspamspamspam");
+        let expected_string =
+            "  0x73, 0x70, 0x61, 0x6d, 0x73, 0x70, 0x61, 0x6d, 0x73, 0x70, 0x61, 0x6d,
+  0x73, 0x70, 0x61, 0x6d, 0x73, 0x70, 0x61, 0x6d
+"
+            .to_owned();
+        let mut output = vec![];
+        let mut printer: Printer<Vec<u8>> = PrinterBuilder::new(&mut output)
+            .show_color(false)
+            .show_char_panel(true)
+            .show_position_panel(true)
+            .with_border_style(BorderStyle::Unicode)
+            .enable_squeezing(true)
+            .num_panels(2)
+            .group_size(1)
+            .with_base(Base::Hexadecimal)
+            .endianness(Endianness::Big)
+            .character_table(CharacterTable::Default)
+            .include_mode(IncludeMode::Stdin)
+            .color_scheme(ColorScheme::Default)
+            .build();
+
+        printer.print_all(input).unwrap();
+
+        let actual_string: &str = str::from_utf8(&output).unwrap();
+        assert_eq!(actual_string, expected_string)
+    }
 }

src/main.rs

@@ -15,7 +15,9 @@ use thiserror::Error as ThisError;
 
 use terminal_size::terminal_size;
 
-use hexyl::{Base, BorderStyle, CharacterTable, ColorScheme, Endianness, Input, PrinterBuilder};
+use hexyl::{
+    Base, BorderStyle, CharacterTable, ColorScheme, Endianness, IncludeMode, Input, PrinterBuilder,
+};
 
 use hexyl::{
     COLOR_ASCII_OTHER, COLOR_ASCII_PRINTABLE, COLOR_ASCII_WHITESPACE, COLOR_NONASCII, COLOR_NULL,
@@ -194,6 +196,16 @@ struct Opt {
     #[arg(long)]
     print_color_table: bool,
 
+    /// Output in C include file style (similar to xxd -i).
+    #[arg(
+        short('i'),
+        long("include"),
+        help = "Output in C include file style",
+        conflicts_with("little_endian_format"),
+        conflicts_with("endianness")
+    )]
+    include_mode: bool,
+
     /// Show shell completion for a certain shell
     #[arg(long, value_name("SHELL"))]
     completion: Option<Shell>,
@@ -262,12 +274,12 @@ fn run() -> Result<()> {
 
     let stdin = io::stdin();
 
-    let mut reader = match opt.file {
+    let mut reader = match &opt.file {
         Some(filename) => {
             if filename.is_dir() {
                 bail!("'{}' is a directory.", filename.to_string_lossy());
             }
-            let file = File::open(&filename)?;
+            let file = File::open(filename)?;
 
             Input::File(file)
         }
@@ -449,6 +461,29 @@ fn run() -> Result<()> {
 
     let mut stdout = BufWriter::new(io::stdout().lock());
 
+    let include_mode = match opt.include_mode {
+        // include mode on
+        true => {
+            if opt.file.is_some() {
+                // input from a file
+                IncludeMode::File(
+                    opt.file
+                        .as_ref()
+                        .unwrap()
+                        .file_name()
+                        .and_then(|n| n.to_str())
+                        .unwrap_or("file")
+                        .to_string(),
+                )
+            } else {
+                // input from stdin
+                IncludeMode::Stdin
+            }
+        }
+        // include mode off
+        false => IncludeMode::Off,
+    };
+
     let mut printer = PrinterBuilder::new(&mut stdout)
         .show_color(show_color)
         .show_char_panel(show_char_panel)
@@ -460,6 +495,7 @@ fn run() -> Result<()> {
         .with_base(base)
         .endianness(endianness)
         .character_table(character_table)
+        .include_mode(include_mode)
         .color_scheme(color_scheme)
         .build();
     printer.display_offset(skip_offset + display_offset);