rematch/src/main.rs

#![allow(dead_code)]

mod re;
mod text;
mod ext; use ext::*;

use color_eyre::{
    eyre::{
        self,
        eyre,
        WrapErr as _,
    },
    SectionExt as _, Help as _,
};

fn initialise() -> eyre::Result<()>
{
    color_eyre::install()?;
    Ok(())
}

#[inline] 
fn print_groups<'a, S: ?Sized, G, T: 'a, I>(to: &mut S, g: G, groups: I) -> std::io::Result<()>
where S: std::io::Write,
      G: IntoIterator<Item = &'a Option<T>> + Clone + Copy, // NOTE: Copy bound to ensure we're not accidentally doing deep clones of `g`.
//G: std::ops::Index<usize>, G::Output: std::borrow::Borrow<Option<T>>,
      T: std::borrow::Borrow<str>,
      I: IntoIterator<Item: std::borrow::Borrow<usize>/*, IntoIter: ExactSizeIterator*/>,
{
    use std::borrow::Borrow;
    let mut first = true;
    for group in groups.into_iter() {
	let group = group.borrow();
	// // Moved to into match group (skipping invalid groups.)
	// if !first {
	//     write!(to, "\t")?;
	// }
	let print_delim = || first.then_some("").unwrap_or("\t"); // If it's not the first iteration, print `\t`.
	match g.into_iter().nth(*group) {
	    Some(None) => write!(to, "{}", print_delim()),
	    Some(Some(g)) => write!(to, "{}{}", print_delim(), g.borrow()),

	    //TODO: What should be the behaviour of a non-existent group index here? (NOTE: This now corresponds to the previous `g.len() > group` check in caller.) // (NOTE: The original behaviour is to just ignore groups that are out of range entirely (i.e. no printing, no delimit char, no error,) maybe treat non-existent groups as non-matched groups and *just* print the delim char?)
	    // (NOTE: Moved out of branch, see above ^) // None if !first => write!(to, "\t"),
	    // XXX: Should this do what it does now...? Or should it `break` to prevent the checking for more groups...? Print a warning maybe...?
	    None => {
		eprintln!("Warning: Invalid group index {}!", group);
		continue; // Do not set `first = false` if it was an invalid index.
		//Ok(())
	    }, 
	}?;
	first = false;
    }
    // If `first == true`, no groups were printed, so we do not print the new-line.
    if !first {
	to.write_all(b"\n")
    } else {
	Ok(())
    }
}

fn main() -> eyre::Result<()>
{
    initialise().wrap_err("Fatal: Failed to install panic handle")?;
    
    //let cli = args::parse_cli();//.wrap_err("Error parsing command-line arguments")?;
    
    //eprintln!("{:#?}", cli);
    // return Ok(());

    let args: re::FrozenVec<re::FrozenString> = std::env::args().map(String::into_boxed_str).collect();

    if args.len() < 4 {
	use owo_colors::OwoColorize;
	use owo_colors::Stream;

	macro_rules! colour {
	    (in $name:ident: $fmt:expr => $col:ident) => {
		$fmt.if_supports_color(Stream::$name, |text| text.$col())
	    };
	    ($fmt:expr => $col:ident) => {
		colour!(in Stdout: $fmt => $col)
	    }
	}
	
	println!("rematch v{}: Regular-expression group matcher", env!("CARGO_PKG_VERSION"));
	println!("");
	println!("Usage: {} <str> <regex> <group>...", args[0]);
	println!("Pass `-' as `<str>' to read lines from stdin");
	println!("");
	println!("Enabled Features:");
	if cfg!(feature="perl") {
	    println!("{}\t\t\tEnable PCRE2 (extended) regular-expressions.\n\t\t\tNote that PCRE2 regex engine matches on *bytes*, not *characters*; meaning if a match cuts a vlid UTF8 codepoint into an invalid one, the output will replace the invalid characters with U+FFFD REPLACEMENT CHARACTER.", colour!(disjoint!["+", "perl"] => bright_red));
	} else {
	    println!("{}\t\t\tPCRE2 (extended) features are disabled; a faster but less featureful regular expression engine (that matches on UTF8 strings instead of raw bytes) is used instead.", colour!(disjoint!["-", "perl"] => blue));
	}
	if cfg!(feature="unstable") {
	    println!("{}\t\tUnstable optimisations evailable & enabled for build.", colour!(disjoint!["+", "unstable"] => red));
	} else {
	    println!("{}\t\tUnstable optimisations disabled / not available for build.", colour!(disjoint!["-", "unstable"] => bright_blue));
	}
	std::process::exit(1)
    } else {
	let re = re::Regex::compile(&args[2])?;
	let text = &args[1];

	let groups = &args[3..];

	if groups.len() < 1 {
	    eprintln!("Warning: No capture groups requested.");
	    // NOTE: Unexpected branch...
	    return Ok(());
	}
	
	let groups = groups.iter().enumerate()
	    .map(|(i, x)| x.parse()
		 .with_section(|| format!("{:?}", groups).header("Groups specified were"))
		 .with_section(|| x.clone().header("Specified capture group index was"))
		 .with_section(move || i.header("Argument index in provided groups")))
	    .collect::<Result<Box<[usize]>, _>>()
	    .wrap_err("Invalid group index specified")?;

	//TODO: XXX: How to handle multiple groups in `stdin_lines()` case?
	//let group = groups[0]; //args[3].parse().expect("Invalid group number.");

	use std::io::Write;
	let mut stdout = std::io::stdout();
	
	let stdout = if &text[..] == "-" {
	    let mut stdout = std::io::BufWriter::new(stdout.lock());
	    text::stdin_lines(|text| -> eyre::Result<bool> {
		match re.exec(&text)? {
		    Some(g) /*if g.len() > group*/ => // NOTE: This check branch has now been moved into `print_groups()`
			print_groups(&mut stdout, &g, &groups)?, //println!("{}", &g[group]),
		    _ => (),
		}
		Ok(true)
	    })?;

            Some(stdout)
	} else {

	    match re.exec(&text)? {
		Some(g) /*if g.len() > group*/ => print_groups(&mut stdout, &g[..], &groups)?,//println!("{}", &g.nth(group).unwrap().map(|x| x.as_ref()).unwrap_or("")),
		_ => (),
	    }

            None
	}.ok_or_else(move || stdout);
	unwrap_either!(mut stdout => stdout.flush()).unwrap();
    }
    Ok(())
}