Cargo.toml

@@ -70,6 +70,12 @@ name = "ccc"
 path = "src/ccc/mod.rs"
 required-features = ["bin", "math"]
 
+[[bin]]
+name = "hedu"
+path = "src/hedu/mod.rs"
+required-features = ["bin", "bintols"]
+
+
 [dependencies]
 libpt-core = { workspace = true }
 libpt-bintols = { workspace = true }

members/libpt-bintols/src/hedu/mod.rs

@@ -0,0 +1,239 @@
+//! # Dump data
+//!
+//! This crate is part of [`pt`](../libpt/index.html), but can also be used as a standalone
+//! module.
+//!
+//! Hedu is made for hexdumping data. `libpt` offers a cli application using this module.
+
+use crate::display::humanbytes;
+use anyhow::{bail, Result};
+use libpt_log::{debug, error, trace, warn};
+use std::io::{prelude::*, Read, SeekFrom};
+
+pub const BYTES_PER_LINE: usize = 16;
+pub const LINE_SEP_HORIZ: char = '─';
+pub const LINE_SEP_VERT: char = '│';
+pub const CHAR_BORDER: &'static str = "|";
+
+#[derive(Debug)]
+pub struct Hedu {
+    pub chars: bool,
+    pub skip: usize,
+    pub show_identical: bool,
+    pub limit: usize,
+    stop: bool,
+    len: usize,
+    data_idx: usize,
+    rd_counter: usize,
+    buf: [[u8; BYTES_PER_LINE]; 2],
+    alt_buf: usize,
+    pub display_buf: String,
+    first_iter: bool,
+}
+
+impl Hedu {
+    pub fn new(chars: bool, skip: usize, show_identical: bool, limit: usize) -> Self {
+        Hedu {
+            chars,
+            skip,
+            show_identical,
+            limit,
+            stop: false,
+            len: 0,
+            data_idx: 0,
+            rd_counter: 0,
+            buf: [[0; BYTES_PER_LINE]; 2],
+            alt_buf: 0,
+            display_buf: String::new(),
+            first_iter: true,
+        }
+    }
+    #[inline]
+    pub fn display(&mut self) {
+        println!("{}", self.display_buf);
+        self.display_buf = String::new();
+    }
+    #[inline]
+    pub fn sep(&mut self) {
+        if self.chars {
+            self.display_buf += &format!("{LINE_SEP_HORIZ}").repeat(80);
+        } else {
+            self.display_buf += &format!("{LINE_SEP_HORIZ}").repeat(59);
+        }
+        self.display();
+    }
+    #[inline]
+    pub fn newline(&mut self) {
+        self.display_buf += "\n";
+        self.display();
+    }
+    fn dump_a_line(&mut self) {
+        self.display_buf += &format!("{:08X} {LINE_SEP_VERT} ", self.data_idx);
+        if self.len != 0 {
+            for i in 0..self.len {
+                if i as usize % BYTES_PER_LINE == BYTES_PER_LINE / 2 {
+                    self.display_buf += " ";
+                }
+                self.display_buf += &format!("{:02X} ", self.buf[self.alt_buf][i]);
+            }
+            if self.len == BYTES_PER_LINE / 2 {
+                self.display_buf += " "
+            }
+            for i in 0..(BYTES_PER_LINE - self.len) {
+                if i as usize % BYTES_PER_LINE == BYTES_PER_LINE / 2 {
+                    self.display_buf += " ";
+                }
+                self.display_buf += "   ";
+            }
+        } else {
+            self.display_buf += &format!("{:49}", "");
+        }
+        if self.chars {
+            self.display_buf += &format!("{LINE_SEP_VERT} ");
+            if self.len != 0 {
+                self.display_buf += CHAR_BORDER;
+                for i in 0..self.len {
+                    self.display_buf +=
+                        &format!("{}", mask_chars(self.buf[self.alt_buf][i] as char));
+                }
+                self.display_buf += CHAR_BORDER;
+            } else {
+                self.display_buf += &format!("{:^8}", "");
+            }
+        }
+        self.display();
+    }
+
+    fn skip_lines(&mut self, data: &mut dyn DataSource) -> Result<()> {
+        trace!(buf = format!("{:?}", self.buf), "found a duplicating line");
+        let start_line = self.data_idx;
+        while self.buf[0] == self.buf[1] && self.len == BYTES_PER_LINE {
+            self.rd_data(data)?;
+        }
+        self.display_buf += &format!(
+            "******** {LINE_SEP_VERT} {:<49}",
+            format!(
+                "(repeats {} lines)",
+                self.data_idx - start_line / (BYTES_PER_LINE) + 1
+            )
+        );
+        if self.chars {
+            self.display_buf += &format!("{LINE_SEP_VERT}");
+        }
+        trace!(
+            buf = format!("{:X?}", self.buf),
+            "dumping buf after line skip"
+        );
+        self.alt_buf ^= 1; // read into the other buf, so we can check for sameness
+        self.display();
+        Ok(())
+    }
+    pub fn dump(&mut self, data: &mut dyn DataSource) -> Result<()> {
+        // skip a given number of bytes
+        if self.skip > 0 {
+            data.skip(self.skip)?;
+            self.rd_counter += self.skip;
+            debug!(
+                data_idx = self.data_idx,
+                "Skipped {}",
+                humanbytes(self.skip)
+            );
+        }
+
+        // print the head
+        self.display_buf += &format!("DATA IDX {LINE_SEP_VERT} DATA AS HEX");
+        if self.chars {
+            self.display_buf += &format!("{:width$} {LINE_SEP_VERT} DATA AS CHAR", "", width = 37);
+        }
+        self.display();
+        self.sep();
+
+        // data dump loop
+        self.rd_data(data)?;
+        self.data_idx = 0;
+        while self.len > 0 || self.first_iter {
+            self.first_iter = false;
+
+            self.dump_a_line();
+
+            // loop breaker logic
+            if self.stop || self.len < BYTES_PER_LINE {
+                break;
+            }
+            self.rd_data(data)?;
+
+            // after line logic
+            if self.buf[0] == self.buf[1] && self.len == BYTES_PER_LINE && !self.show_identical {
+                self.skip_lines(data)?;
+            }
+        }
+        self.data_idx += BYTES_PER_LINE;
+
+        self.sep();
+        self.display_buf += &format!(
+            "{:08X} {LINE_SEP_VERT} read total:\t\t    {:<8} {:<15}",
+            self.rd_counter,
+            humanbytes(self.rd_counter),
+            format!("({} B)", self.rd_counter)
+        );
+        if self.chars {
+            self.display_buf += &format!("{LINE_SEP_VERT}");
+        }
+        self.display();
+        Ok(())
+    }
+    #[inline]
+    fn adjust_counters(&mut self) {
+        self.rd_counter += self.len;
+        self.data_idx += self.len;
+    }
+
+    fn rd_data(&mut self, data: &mut dyn DataSource) -> Result<()> {
+        match data.read(&mut self.buf[self.alt_buf]) {
+            Ok(mut len) => {
+                if self.limit != 0 && self.rd_counter + (BYTES_PER_LINE - 1) >= self.limit {
+                    len = self.limit % BYTES_PER_LINE;
+                    self.stop = true;
+                }
+                self.len = len;
+                self.adjust_counters();
+                return Ok(());
+            }
+            Err(err) => {
+                error!("error while reading data: {err}");
+                bail!(err)
+            }
+        }
+    }
+}
+
+pub trait DataSource: Read {
+    fn skip(&mut self, length: usize) -> std::io::Result<()>;
+}
+impl DataSource for std::io::Stdin {
+    fn skip(&mut self, _length: usize) -> std::io::Result<()> {
+        warn!("can't skip bytes for the stdin!");
+        Ok(())
+    }
+}
+impl DataSource for std::fs::File {
+    fn skip(&mut self, length: usize) -> std::io::Result<()> {
+        self.seek(SeekFrom::Current(length as i64))?;
+        // returns the new position from the start, we don't need it here.
+        Ok(())
+    }
+}
+
+fn mask_chars(c: char) -> char {
+    if c.is_ascii_graphic() {
+        return c;
+    } else if c == '\n' {
+        return '↩';
+    } else if c == ' ' {
+        return '␣';
+    } else if c == '\t' {
+        return '⭾';
+    } else {
+        return '<27>';
+    }
+}

members/libpt-bintols/src/lib.rs

@@ -25,3 +25,4 @@ pub const YOBI: u128 = 2u128.pow(80);
 // use libpt_core;
 pub mod datalayout;
 pub mod display;
+pub mod hedu;

members/libpt-log/src/error.rs

@@ -26,25 +26,45 @@ use tracing::subscriber::SetGlobalDefaultError;
 
 //// ENUMS /////////////////////////////////////////////////////////////////////////////////////////
 /// ## Errors for the [Logger](super::Logger)
-#[derive(Error, Debug)]
+#[derive(Error)]
 pub enum Error {
     /// Bad IO operation
     #[error("Bad IO operation")]
-    IO(#[from] std::io::Error),
+    IO(std::io::Error),
     /// Various errors raised when the messenger is used in a wrong way
     #[error("Bad usage")]
     Usage(String),
     /// Could not assign logger as the global default
-    #[error("Could not assign logger as global default")]
+    #[error("Could not assign as global default")] // TODO: make this more clear
-    SetGlobalDefaultFail(#[from] SetGlobalDefaultError),
+    SetGlobalDefaultFail(SetGlobalDefaultError),
-    /// any other error type, wrapped in [anyhow::Error](anyhow::Error)
-    #[error(transparent)]
-    Other(#[from] anyhow::Error),
 }
 
 //// STRUCTS ///////////////////////////////////////////////////////////////////////////////////////
 
 //// IMPLEMENTATION ////////////////////////////////////////////////////////////////////////////////
+impl From<std::io::Error> for Error {
+    fn from(value: std::io::Error) -> Self {
+        Error::IO(value)
+    }
+}
+
+////////////////////////////////////////////////////////////////////////////////////////////////////
+impl From<SetGlobalDefaultError> for Error {
+    fn from(value: SetGlobalDefaultError) -> Self {
+        Error::SetGlobalDefaultFail(value)
+    }
+}
+
+////////////////////////////////////////////////////////////////////////////////////////////////////
+impl std::fmt::Debug for Error {
+    fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
+        match self {
+            Error::IO(e) => write!(f, "<IO Error {e:?}>"),
+            Error::Usage(e) => write!(f, "<Usage Error {e:?}>"),
+            Error::SetGlobalDefaultFail(e) => write!(f, "<SetGlobalDefaultFail {e:?}>"),
+        }
+    }
+}
 
 //// PUBLIC FUNCTIONS //////////////////////////////////////////////////////////////////////////////
 

src/hedu/mod.rs

@@ -0,0 +1,174 @@
+//! # Executable for the hedu submodule
+//!
+//! Dump data to a fancy format.
+
+//// ATTRIBUTES ////////////////////////////////////////////////////////////////////////////////////
+// we want docs
+#![warn(missing_docs)]
+#![warn(rustdoc::missing_crate_level_docs)]
+// we want Debug everywhere.
+#![warn(missing_debug_implementations)]
+// enable clippy's extra lints, the pedantic version
+#![warn(clippy::pedantic)]
+
+//// IMPORTS ///////////////////////////////////////////////////////////////////////////////////////
+
+use libpt::{bintols::hedu::*, log::*};
+
+use clap::Parser;
+use clap_verbosity_flag::{InfoLevel, Verbosity};
+
+use std::{fs::File, io::IsTerminal, path::PathBuf};
+
+//// TYPES /////////////////////////////////////////////////////////////////////////////////////////
+
+//// CONSTANTS /////////////////////////////////////////////////////////////////////////////////////
+/// short about section displayed in help
+const ABOUT_ROOT: &'static str = r##"
+Dumps data in fancy formats.
+"##;
+/// longer about section displayed in help, is combined with [the short help](ABOUT_ROOT)
+static LONG_ABOUT_ROOT: &'static str = r##"
+
+    libpt is a personal general purpose library, offering this executable, a python module and a
+    dynamic library.
+"##;
+
+//// STATICS ///////////////////////////////////////////////////////////////////////////////////////
+
+//// MACROS ////////////////////////////////////////////////////////////////////////////////////////
+
+//// ENUMS /////////////////////////////////////////////////////////////////////////////////////////
+
+//// STRUCTS ///////////////////////////////////////////////////////////////////////////////////////
+/// defines CLI interface
+#[derive(Debug, Clone, Parser)]
+#[command(
+    author,
+    version,
+    about = ABOUT_ROOT,
+    long_about = format!("{}{}", ABOUT_ROOT ,LONG_ABOUT_ROOT),
+    help_template =
+r#"{about-section}
+{usage-heading} {usage}
+{all-args}{tab}
+
+libpt: {version}
+Author: {author-with-newline}
+"#
+    )]
+pub struct Cli {
+    // clap_verbosity_flag seems to make this a global option implicitly
+    /// set a verbosity, multiple allowed (f.e. -vvv)
+    #[command(flatten)]
+    pub verbose: Verbosity<InfoLevel>,
+
+    /// show additional logging meta data
+    #[arg(long)]
+    pub meta: bool,
+
+    /// show character representation
+    #[arg(short, long)]
+    pub chars: bool,
+
+    /// skip first N bytes
+    #[arg(short, long, default_value_t = 0)]
+    pub skip: usize,
+
+    /// only interpret N bytes (end after N)
+    #[arg(short, long, default_value_t = 0)]
+    pub limit: usize,
+
+    /// show identical lines
+    #[arg(short = 'i', long)]
+    pub show_identical: bool,
+
+    /// a data source, probably a file.
+    ///
+    /// If left empty or set as "-", the program will read from stdin.
+    // TODO: take many sources #60
+    pub data_source: Vec<String>,
+}
+
+//// IMPLEMENTATION ////////////////////////////////////////////////////////////////////////////////
+
+//// PUBLIC FUNCTIONS //////////////////////////////////////////////////////////////////////////////
+
+//// PRIVATE FUNCTIONS /////////////////////////////////////////////////////////////////////////////
+fn main() {
+    let mut cli = cli_parse();
+    let mut sources: Vec<Box<dyn DataSource>> = Vec::new();
+    if cli.data_source.len() > 0 && cli.data_source[0] != "-" {
+        for data_source in &cli.data_source {
+            let data_source: PathBuf = PathBuf::from(data_source);
+            if data_source.is_dir() {
+                warn!("Not a file {:?}, skipping", data_source);
+                // std::process::exit(1);
+                continue;
+            }
+            trace!("Trying to open '{:?}'", data_source);
+            match File::open(&data_source) {
+                Ok(file) => sources.push(Box::new(file)),
+                Err(err) => {
+                    error!("Could not open '{:?}': {err}", data_source);
+                    std::process::exit(1);
+                }
+            };
+        }
+    } else {
+        trace!("Trying to open stdin");
+        let stdin = std::io::stdin();
+        if stdin.is_terminal() {
+            warn!("Refusing to dump from interactive terminal");
+            std::process::exit(2)
+        }
+        // just for the little header
+        cli.data_source = Vec::new();
+        cli.data_source.push(format!("stdin"));
+        sources.push(Box::new(stdin));
+    }
+    for (i, source) in sources.iter_mut().enumerate() {
+        let mut config = Hedu::new(cli.chars, cli.skip, cli.show_identical, cli.limit);
+        // FIXME: find a better way to get the file name
+        // Currently, skipped sources make an extra newline here.
+        match config.chars {
+            false => {
+                println!("{:─^59}", format!(" {} ", cli.data_source[i]));
+            }
+            true => {
+                println!("{:─^80}", format!(" {} ", cli.data_source[i]));
+            }
+        }
+        match config.dump(&mut **source) {
+            Ok(_) => (),
+            Err(err) => {
+                error!("Could not dump data of file: {err}");
+                std::process::exit(3);
+            }
+        }
+        if i < cli.data_source.len() - 1 {
+            config.newline();
+        }
+    }
+}
+////////////////////////////////////////////////////////////////////////////////////////////////////
+fn cli_parse() -> Cli {
+    let cli = Cli::parse();
+    let ll: Level = match cli.verbose.log_level().unwrap().as_str() {
+        "TRACE" => Level::TRACE,
+        "DEBUG" => Level::DEBUG,
+        "INFO" => Level::INFO,
+        "WARN" => Level::WARN,
+        "ERROR" => Level::ERROR,
+        _ => {
+            unreachable!();
+        }
+    };
+    if cli.meta {
+        Logger::init(None, Some(ll), false).expect("could not initialize Logger");
+    } else {
+        // less verbose version
+        Logger::init_mini(Some(ll)).expect("could not initialize Logger");
+    }
+    return cli;
+}