genmarkov/src/sanitise/filter.rs

//! Filter out characters and such
use smallmap::Map as SmallMap;
use std::{
    borrow::Cow,
    fmt,
    iter::{
	self,
	FromIterator,
    },
    str,
};
use once_cell::sync::OnceCell;

#[derive(Debug, Clone, PartialEq, Eq)]
pub struct Filter(SmallMap<char, ()>);

/*
impl<const N: usize> From<[char; N]> for Filter
{
    fn from(from: [char; N]) -> Self
    {
	let mut map = SmallMap::with_capacity(1 + (N / 256));
	for &chr in from.iter()
	{
	    map.insert(chr, ());
	}
	Self(map)
    }
}*/

impl<'a> From<&'a [char]> for Filter
{
    fn from(from: &'a [char]) -> Self
    {
	let mut map = SmallMap::new();
	for &chr in from.iter()
	{
	    map.insert(chr, ());
	}
	Self(map)
    }
}
impl<'a> From<&'a str> for Filter
{
    fn from(from: &'a str) -> Self
    {
	let mut output = Self::new();
	output.insert(from.chars());
	output
    }
}

impl str::FromStr for Filter
{
    type Err = std::convert::Infallible;
    fn from_str(s: &str) -> Result<Self, Self::Err> {
	Ok(Self::from(s))
    }
}

impl fmt::Display for Filter
{
    fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result
    {
	use std::fmt::Write;
	for chr in self.iter()
	{
	    f.write_char(chr)?;
	}
	Ok(())
    }
}

pub struct FilterKeyIter<'a>(smallmap::iter::Iter<'a, char, ()>, usize);

impl<'a> Iterator for FilterKeyIter<'a>
{
    type Item = char;
    fn next(&mut self) -> Option<Self::Item>
    {
	self.0.next().map(|&(x, _)| x)
    }
    fn size_hint(&self) -> (usize, Option<usize>) {
	(self.1, Some(self.1))
    }
}
impl<'a> iter::FusedIterator for FilterKeyIter<'a>{}
impl<'a> iter::ExactSizeIterator for FilterKeyIter<'a>{}

impl Filter
{
    pub fn new() -> Self
    {
	Self(SmallMap::new())
    }
    pub fn insert<I: IntoIterator<Item=char>>(&mut self, from: I)
    {
	for from in from.into_iter()
	{
	    self.0.insert(from, ());
	}
    }

    pub fn remove<I: IntoIterator<Item=char>>(&mut self, from: I)
    {
	for from in from.into_iter()
	{
	    self.0.remove(&from);
	}
    }

    pub fn len(&self) -> usize
    {
	self.0.len()
    }

    pub fn is_empty(&self) -> bool
    {
	//TODO: impl this in smallmap itself
	self.len() == 0
    }

    pub fn iter(&self) -> FilterKeyIter<'_> //impl Iterator<Item=char> + '_
    {
	//self.0.iter()
	//    .copied()
	//    .map(|(x, _)| x)
	FilterKeyIter(self.0.iter(), self.0.len())
    }

    /// Should this character be filtered?
    #[inline] pub fn check(&self, chr: char) -> bool
    {
	self.0.get(&chr).is_some()
    }

    pub fn filter_owned(&self, input: impl Into<String>) -> String
    {
	let mut input = input.into();
	self.filter(&mut input);
	input
    }
    
    pub fn filter<'a>(&self, output: &'a mut String) -> &'a mut String
    {
	if self.is_empty() {
	    return output;
	}
	output.retain(|chr| !self.check(chr));
	output
    }
    
    pub fn filter_iter<'a, I: IntoIterator<Item=char>>(&'a self, from_iter: I) -> FilterIter<'a, I::IntoIter>
    where I::IntoIter: 'a
    {
	FilterIter(&self, from_iter.into_iter().fuse())
    }

    pub fn filter_cow<'a>(&self, string: &'a (impl AsRef<str> + 'a + ?Sized)) -> Cow<'a, str>
    {
	let string = string.as_ref();
	
	if self.is_empty() {
	    return Cow::Borrowed(string);
	}
	
	let mut output = Cow::Borrowed(string);
	let mut i=0;
	for chr in string.chars()
	{
	    if self.check(chr) {
		output.to_mut().remove(i);
	    } else {
		i+=1;
	    }
	}

	output
    }

    pub fn filter_str<'a, T: AsRef<str>+'a +?Sized>(&'a self, string: &'a T) -> FilterStr<'a>
    {
	FilterStr(string.as_ref(), self, OnceCell::new())
    }
}

impl FromIterator<char> for Filter
{
    fn from_iter<I: IntoIterator<Item=char>>(iter: I) -> Self
    {
	let mut output= Self::new();
	output.insert(iter);
	output
    }
}

impl<'a> FilterStr<'a>
{
    pub fn as_str(&self) -> &str
    {	
	fn fmt(this: &FilterStr<'_>) -> String
	{
	    let chars = this.0.chars();
	    let mut f: String = crate::util::hint_cap(&chars);
	    for chr in  chars {
		if !this.1.check(chr) {
		    f.push(chr);
		}
	    }
	    f
	}
	&self.2.get_or_init(|| fmt(&self))[..]
    }
}

pub struct FilterStr<'a>(&'a str, &'a Filter, OnceCell<String>);
impl<'a> fmt::Display for FilterStr<'a>
{
    fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result
    {
	write!(f, "{}", self.as_str())
    }
}
impl<'a> FilterStr<'a>
{
    pub fn filter(&self) -> &Filter
    {
	&self.1
    }
}

pub struct FilterIter<'a, I>(&'a Filter, iter::Fuse<I>);

impl<'a, I: Iterator<Item=char>> Iterator for FilterIter<'a, I>
{
    type Item = char;
    fn next(&mut self) -> Option<Self::Item>
    {
	loop {
	    break match self.1.next() {
		Some(chr) if !self.0.check(chr) => Some(chr),
		None => None,
		_ => continue,
	    }
	}
    }

    fn size_hint(&self) -> (usize, Option<usize>) {
	let (_, high) = self.1.size_hint();
	(0, high)
    }
}
impl<'a, I> FilterIter<'a, I>
{
    pub fn filter(&self) -> &Filter
    {
	self.0
    }
}

impl<'a, I: Iterator<Item=char>> iter::FusedIterator for FilterIter<'a, I>{}

#[cfg(test)]
mod tests
{
    use super::*;
    #[test]
    fn filter_cow()
    {
	let filter: Filter = " hi".chars().collect();

	let string = "abcdef  ghi  jk1\nhian";

	assert_eq!(filter.filter_str(&string).to_string(), filter.filter_cow(&string).to_string());
	assert_eq!(filter.filter_cow(&string).to_string(), filter.filter(string.chars()).collect::<String>());
    }
}