I added source code output for F# arrays. They use a semicolon instead of a comma as the delimiter, and each literal needs to have the uy suffix, or the F# compiler will complain. That's why I had to special case the new f array format.

~
❯ touch test

~
❯ echo "Hello, world." > test

~
❯ time hx -ar test
^CCommand terminated by signal 2
0.00user 0.00system 0:04.46elapsed 0%CPU (0avgtext+0avgdata 2528maxresident)k
0inputs+0outputs (0major+107minor)pagefaults 0swaps

~ took 4s
❯ time hx test -ar
let ARRAY: [u8; 14] = [
    0x48, 0x65, 0x6c, 0x6c, 0x6f, 0x2c, 0x20, 0x77, 0x6f, 0x72,
    0x6c, 0x64, 0x2e, 0x0a
];
0.00user 0.00system 0:00.00elapsed 93%CPU (0avgtext+0avgdata 2796maxresident)k
0inputs+0outputs (0major+112minor)pagefaults 0swaps

~```

Hi there,

I made several changes to your tool to make it fit my needs, and looking at the issues I figured other people might want some of these changes as well.

• Allow reading from stdin. This lets me use hx in sequence with other command line tools.
• Make it streaming. So it doesn't read the entire resource in and try to hold it in memory.

These are just some changes I made to make it work; if you dont' think they should be merged or needs other changes, feel free to let me know or close it.

When used in a pipeline hex will crash if a downstream program exits before hex finishes printing.

To Reproduce

$ dd if=/dev/random bs=512 count=10 | RUST_BACKTRACE=1 hx | head -n 10

10+0 records in
10+0 records out
5120 bytes (5.1 kB, 5.0 KiB) copied, 0.001144 s, 4.5 MB/s
0x000000: 0x24 0xfc 0xad 0xc8 0x27 0x0f 0x30 0x33 0x45 0x33 $...'.03E3
0x00000a: 0x4c 0x49 0xcc 0x33 0x47 0x9e 0x6d 0x38 0xc6 0x6f LI.3G.m8.o
0x000014: 0x86 0x7d 0xc2 0xd3 0x03 0x31 0xf9 0xbf 0xec 0xa4 .}...1....
0x00001e: 0x36 0x33 0xf1 0x6e 0x8b 0x43 0xe6 0xab 0x7c 0xef 63.n.C..|.
0x000028: 0x52 0xb4 0xdb 0xa1 0x63 0xe8 0x96 0x61 0xd8 0x2b R...c..a.+
0x000032: 0x47 0xc6 0xad 0x86 0x86 0xf6 0x52 0xe3 0x54 0x58 G.....R.TX
0x00003c: 0xfb 0x1b 0x80 0x50 0x79 0x80 0x11 0x31 0xa6 0x72 ...Py..1.r
0x000046: 0x50 0xf8 0x77 0x96 0x50 0x13 0x39 0x17 0xfe 0xc7 P.w.P.9...
0x000050: 0xef 0x40 0x78 0xe8 0xd5 0xfb 0xc6 0x4c 0x62 0x36 [email protected]
0x00005a: 0xd9 0x32 0x18 0xe5 0x1d 0xc5 0xf0 0x95 0x35 0xdf .2......5.
thread 'main' panicked at 'failed printing to stdout: Broken pipe (os error 32)', src/libstd/io/stdio.rs:822:9
stack backtrace:
   0: <std::sys_common::backtrace::_print::DisplayBacktrace as core::fmt::Display>::fmt
   1: core::fmt::write
   2: std::io::Write::write_fmt
   3: std::panicking::default_hook::{{closure}}
   4: std::panicking::default_hook
   5: std::panicking::rust_panic_with_hook
   6: rust_begin_unwind
   7: std::panicking::begin_panic_fmt
   8: std::io::stdio::_print
   9: hx::lib::run
  10: hx::main
  11: std::rt::lang_start::{{closure}}
  12: std::rt::lang_start_internal
  13: main
note: Some details are omitted, run with `RUST_BACKTRACE=full` for a verbose backtrace.

In this situation head reads 10 lines from hex, prints them, and exits. This causes a SIGPIPE to be sent to hex which crashes. The expected behavior would be to handle the signal, stop printing, and cleanly exit.

Environment

Version: v0.3.0 (installed via cargo install hx) OS: macOS 10.14.6

Hey! hex can be installed from Arch Linux [community] repository: https://archlinux.org/packages/community/x86_64/hex/ This PR adds installation instruction to README.md

Fixes #24. Fixes #22.

#24 is essentially a duplicates of #22, the root cause is the ordering of the checks inside the is_stdin function.

I also simplified some parts of ìs_stdin.

Fixes #26. Fixes #23.

With this PR hx is now able to properly report an error when it is given a directory, an invalid filename or a broken pipe:

# All those cases crashed previously.

❯ ./target/release/hx filename-missing
error = "No such file or directory (os error 2)"

❯ ./target/release/hx src
error = "Is a directory (os error 21)"

❯ ./target/release/hx symlink_to_dir
error = "Is a directory (os error 21)"

❯ dd if=/dev/random bs=512 count=10 | RUST_BACKTRACE=1 ./target/debug/hx | head -n 10
10+0 records in
10+0 records out
5120 bytes transferred in 0.000036 secs (142217460 bytes/sec)
0x000000: 0x6c 0xdc 0x4f 0x43 0xfd 0x5e 0xb3 0xa7 0x79 0xaa l.OC.^..y.
0x00000a: 0x6b 0x2c 0x81 0x47 0x00 0x10 0xa6 0x40 0xb5 0x6a k,[email protected]
0x000014: 0x4d 0x31 0x5b 0x11 0x15 0x71 0xb8 0xa1 0xff 0x91 M1[..q....
0x00001e: 0x2e 0xaa 0x69 0x77 0x52 0x2a 0x20 0xa6 0x52 0xe5 ..iwR* .R.
0x000028: 0x41 0x2b 0x83 0x55 0x21 0x41 0x33 0xae 0xf8 0x38 A+.U!A3..8
0x000032: 0xba 0x74 0xfe 0x23 0xc2 0xc9 0xf5 0x3d 0x28 0x8e .t.#...=(.
0x00003c: 0xbc 0x0e 0xe1 0xa3 0xbf 0xda 0x81 0x92 0xc6 0xe9 ..........
0x000046: 0x9f 0xb7 0x13 0x72 0x79 0xd3 0x1e 0x4c 0xd8 0x5d ...ry..L.]
0x000050: 0x85 0x36 0x91 0x9f 0x13 0x7e 0x55 0x5f 0x7b 0xa6 .6...~U_{.
0x00005a: 0x23 0xe7 0xed 0x54 0x47 0x9a 0x1c 0x54 0x91 0x85 #..TG..T..
error = "Broken pipe (os error 32)"

Describe the bug

Passing an invalid filename to hx causes it to crash on a Result:

To Reproduce Steps to reproduce the behavior:

$ hx toto
thread 'main' panicked at 'called `Result::unwrap()` on an `Err` value: Os { code: 2, kind: NotFound, message: "No such file or directory" }', $HOME/.config/rust/cargo/registry/src/github.com-1ecc6299db9ec823/hx-0.3.0/src/lib.rs:238:17
note: run with `RUST_BACKTRACE=1` environment variable to display a backtrace

Expected behavior

I expected the program to tell me I made a mistake.

Desktop (please complete the following information):

• OS: macOS 10.15.7

Additional context

$ hx --version
hx 0.3.0

I'll probably have the time to open a PR for a fix tomorrow if wanted.

Describe the bug Calling hx with the -l or --len argument causes it to hang indefinitely.

To Reproduce Steps to reproduce the behavior: hx -l myfile.txt

Desktop (please complete the following information):

• OS: Arch linux

Additional context Thanks for this great tool :)

support stdin

• dynamically detect stdin, file path and parameters
• remove unused verbose param
• refactor output_array() own function
• add clippy command to Makefile
• update edition

Is your feature request related to a problem? Please describe.

The hexdump command and many Unix/Linux utilities allow operating on stdin either naturally cat FILE | util or by supplying special file name "-" like cat FILE | util -.

A clear and concise description of what the problem is. Ex. I'm always frustrated when [...]

Describe the solution you'd like

If file name does not exist and it has name "-", read and process input from stdin

Describe alternatives you've considered

If file name is not supplied, check if there is an input on stdin and process it instead.

Additional context

Since hx is basically filter, those two invocations are the first ones I have tried and none has worked for me. It is not a deal breaker, but it might be considered as lack of polish that will hinder adoption of hx tool.

Especially since in the second case I get panic instead of nicely formatted error message:

$ cat FILE | hx -
thread 'main' panicked at 'called `Result::unwrap()` on an `Err` value: Os { code: 2, kind: NotFound, message: "No such file or directory" }', /home/haku/.cargo/registry/src/github.com-1ecc6299db9ec823/hx-0.2.1/src/lib.rs:223:17
note: run with `RUST_BACKTRACE=1` environment variable to display a backtrace

Is your feature request related to a problem? Please describe. No

Describe the solution you'd like A --length and --address to specify the memory range to print

Describe alternatives you've considered Less with search is terrible

Additional context It helps developers

I was unable to install hex because https://github.com/helix-editor/helix has hx.

$ cargo install hx
error: binary `hx` already exists in destination as part of `helix-term v0.6.0 (/home/x/helix/helix-term)`
Add --force to overwrite

Is your feature request related to a problem? Please describe. I'm always frustrated when I try to use hx for a very large file. For example, when trying to load a 3.1 GB binary file, it eventually used up 16 GB of RAM before I stopped the process.

Describe the solution you'd like I think that a potential solution would involve replacing the fs::File::open with an unsafe memmap call. Just that would probably save a lot of memory usage.

Describe alternatives you've considered Note that I'm not very familiar with the code at all (just started reading it today). That said, another alternative would be to not append the given line to the page.body for all of the bytes ... and maybe the truncate_len used in buf_to_array should be set to the console's number of lines somehow? The difficulty with that second approach is that at each console refresh, another part of the file will need to be parsed.

In my system, I have also installed installed the helix editor.

• https://github.com/helix-editor/helix

It is currently not yet on Crates

• https://crates.io/crates/helix-editor
• https://crates.io/crates/helix-term

It provides a wrapper that uses the same binary name than hx, which therefore clashes.

Is there a way to assign another name to the binary during build, e.g. hex? Then choosing an abbreviated version remains as simple as creating a symlink.

Is your feature request related to a problem? Please describe. Currently, there's no options to disable prefixes like 0x as far as I know. The hexdump command doesn't show 0x so I can show more digits in a line.

Describe the solution you'd like It would be better to add an option to disable prefixes.

Describe alternatives you've considered The option -f, --format could have values to disable prefixes.

Additional context