hb/src/args.rs

use clap::{Parser, ArgAction};
use glob::Pattern;

use crate::unit::Unit;

use std::fs::{symlink_metadata, Metadata};
use std::path::{Component, Path};
use std::slice::Iter;
use std::iter::once_with;

#[allow(clippy::struct_excessive_bools)]
#[derive(Parser, Debug, Clone)]
pub struct Args {
    #[arg(
        short, long, default_value_t = false,
        help = "keep going if an error occurs",
        long_help = "keep going if an error occurs (ex. unreadable subdirectories in a readable directory)"
    )]
    persistant: bool,

    #[arg(
        short, long,
        help = "minimize output",
        long_help = "print nothing but the total size for all directories, without a newline. Also supresses all error messages",
        conflicts_with = "total",
        default_value_t = false,
    )]
    minimal: bool,

    #[arg(
        short='T', long,
        help = "display in tree",
        default_value_t = false,
        conflicts_with = "minimal",
    )]
    tree: bool,

    #[arg(
        short, long,
        help = "display the total size",
        conflicts_with = "minimal",
        default_value_t = false,
    )]
    total: bool,

    #[arg(
        short='2', long,
        help = "alias for --unit 1024",
        default_value_t = false,
        conflicts_with_all = ["si","unit"],
    )]
    base_two: bool,

    #[arg(
        short='k', long,
        help = "alias for --unit 1000",
        default_value_t = false,
        conflicts_with_all = ["base_two","unit"],
    )]
    si: bool,

    #[arg(
        short, long,
        help = "unit to print in",
        long_help = "printing unit (case insensitive): b = bytes, kb = kilobytes, ki = kibibytes, gb = gigabytes, gi = gibibytes, tb = terabytes, ti = tibibytes",
        value_parser = Unit::parse,
        default_value_t = Unit::Byte,
        conflicts_with_all = ["base_two","si"],
    )]
    unit: Unit,

    #[arg(
        short='s', long,
        help = "follow symlinks",
        default_value_t = false,
    )]
    follow_links: bool,

    #[arg(
        short='x', long = "exclude",
        help = "include in search, but exclude from printing",
        long_help = "include in search, but exclude from printing. accepts glob syntax",
        default_values_t = once_with(|| Pattern::new(".*").unwrap()),
        value_parser = parse_glob,
        value_delimiter = ',',
        action = ArgAction::Append,
    )]
    exclude_print: Vec<Pattern>,

    #[arg(
        short='X', long,
        help = "exclude from search and printing",
        value_parser = parse_glob,
        value_delimiter = ',',
        action = ArgAction::Append,
    )]
    exclude_search: Vec<Pattern>,

    #[arg(
        short='H', long,
        help = "disable implicit hiding of results",
        long_help = "don't implicitly hide dotfiles and dot directories",
        conflicts_with = "exclude_print",
        default_value_t = false,
    )]
    show_hidden: bool,

    #[arg(
        value_parser = validate_path,
        help = "items to summate",
        action = ArgAction::Append,
        num_args = 1..
    )]
    path: Vec<String>,
}
impl Args {
    /// utility method to chuck default values on the end
    /// it feels like I should be able to do this with
    /// clever `clap` macros but I don't know how
    pub fn post_process(mut self) -> Self {
        if self.base_two {
            self.unit = Unit::Kibi;
        } else if self.si {
            self.unit = Unit::Kilo;
        }

        if self.show_hidden {
            self.exclude_print = Vec::new();
        }

        if self.path.is_empty() {
            self.path = vec![ ".".to_owned() ];
        }
        
        self
    }

    pub fn should_exclude(&self, path: &Path, file: &Metadata) -> bool {
        if !self.follow_links
        && file.is_symlink()
        // `.` counts as a symlink. excluding . is usually not
        // useful, so even if this is a symlink when we're not
        // supposed to be following them, if this opens with a
        // CurDir component then we shouldn't exclude
        && !matches!(path.components().nth(0), Some(Component::CurDir))
        {
            return true
        }

        any_pattern_matches_any_component(&self.exclude_search, path)
    }
    
    pub fn should_print(&self, path: &Path) -> bool {
        ! any_pattern_matches_any_component(&self.exclude_print, path)
        // TODO: this exists because when a file matches an exclude pattern
        // is it still returned, just with no size or children, so in order
        // to not accidentally print things that we said we were excluding,
        // we also have to check that it's not excluded by search.
        // `self.exclude_print.extend(&self.exclude_search)` is wasteful,
        // but until I find a better way this is what it's gotta be`
        && ! any_pattern_matches_any_component(&self.exclude_search, path)
    }

    pub const fn persistant(&self) -> bool {
        self.persistant
    }

    pub const fn minimal(&self) -> bool {
        self.minimal
    }
    
    pub const fn tree(&self) -> bool {
        self.tree
    }
    
    pub const fn total(&self) -> bool {
        self.total
    }
    
    pub const fn unit(&self) -> Unit {
        self.unit
    }

    pub fn iter(&self) -> Iter<'_, String> {
        self.path.iter()
    }
}

fn validate_path(s: &str) -> Result<String, String> {
    // try to access it's metadata, since that is what is used
    // to get its length. using symlink because that's what's
    // used in the actual program
    symlink_metadata(s)
        .map(|_| s.to_string())
        .map_err(|e| e.to_string())
}

fn parse_glob(s: &str) -> Result<Pattern, String> {
    Pattern::new(s).map_err(|globerr| format!("invalid glob \"{s}\": {}", globerr.msg))
}

fn any_pattern_matches_any_component(patterns: &[Pattern], path: &Path) -> bool {
    for pat in patterns {
        for cmp in path.components() {
            let Component::Normal(cmp) = cmp else {
                // this seems wacky, but only normal components
                // are useful
                continue
            };
            let Some(s) = cmp.to_str() else {
                // this is a code smell
                // I don't believe it, but I can't think
                // of anything worthwhile to do
                continue
            };
            if pat.matches(s) {
                return true
            }
        }
    }

    false
}