organic/src/main.rs

use std::ops::RangeBounds;

use nom::{bytes::complete::tag, Compare, InputTake, Slice};

fn main() {
    let input = "this is my test input".to_owned();
    let wrapped_input = WrappedInput::new(input.as_str());

    let output = tag::<_, _, (_, nom::error::ErrorKind)>("this")(wrapped_input).unwrap();

    println!("{:#?}", output);
}

#[derive(Debug)]
struct WrappedInput<'s> {
    full_source: &'s str,
    start: usize,
    end: usize, //exclusive
    preceding_line_break: Option<usize>,
}

impl<'s> WrappedInput<'s> {
    /// Returns a wrapped string that keeps track of values we need for parsing org-mode.
    ///
    /// Only call this on the full original string. Calling this on a substring can result in invalid values.
    pub fn new(input: &'s str) -> Self {
        WrappedInput {
            full_source: input,
            start: 0,
            end: input.len(),
            preceding_line_break: None,
        }
    }

    /// Get the text since the line break preceding the start of this WrappedInput.
    pub fn text_since_line_break(&self) -> &'s str {
        let start = self.preceding_line_break.unwrap_or(0);
        &self.full_source[start..self.start]
    }
}

impl<'s> InputTake for WrappedInput<'s> {
    fn take(&self, count: usize) -> Self {
        self.slice(..count)
    }

    fn take_split(&self, count: usize) -> (Self, Self) {
        (self.slice(count..), self.slice(..count))
    }
}

impl<'s, O: Into<&'s str>> Compare<O> for WrappedInput<'s> {
    fn compare(&self, t: O) -> nom::CompareResult {
        (&self.full_source[self.start..self.end]).compare(t.into())
    }

    fn compare_no_case(&self, t: O) -> nom::CompareResult {
        (&self.full_source[self.start..self.end]).compare_no_case(t.into())
    }
}

impl<'s> From<&'s str> for WrappedInput<'s> {
    fn from(value: &'s str) -> Self {
        WrappedInput::new(value)
    }
}

impl<'s> From<&WrappedInput<'s>> for &'s str {
    fn from(value: &WrappedInput<'s>) -> Self {
        &value.full_source[value.start..value.end]
    }
}

impl<'s, R> Slice<R> for WrappedInput<'s>
where
    R: RangeBounds<usize>,
{
    fn slice(&self, range: R) -> Self {
        let new_start = match range.start_bound() {
            std::ops::Bound::Included(idx) => self.start + idx,
            std::ops::Bound::Excluded(idx) => self.start + idx - 1,
            std::ops::Bound::Unbounded => self.start,
        };
        let new_end = match range.end_bound() {
            std::ops::Bound::Included(idx) => self.start + idx + 1,
            std::ops::Bound::Excluded(idx) => self.start + idx,
            std::ops::Bound::Unbounded => self.end,
        };
        if new_start < self.start {
            panic!("Attempted to extend before the start of the WrappedInput.")
        }
        if new_end > self.end {
            panic!("Attempted to extend past the end of the WrappedInput.")
        }

        let skipped_text = &self.full_source[self.start..new_start];
        let last_line_feed = skipped_text
            .rfind('\n')
            .map(|idx| idx + self.preceding_line_break.unwrap_or(0) + 1);

        // TODO: calculate updated values for WrappedInput
        WrappedInput {
            full_source: self.full_source,
            start: new_start,
            end: new_end,
            preceding_line_break: last_line_feed,
        }
    }
}

impl<'s> std::fmt::Display for WrappedInput<'s> {
    fn fmt(&self, f: &mut std::fmt::Formatter) -> std::fmt::Result {
        Into::<&str>::into(self).fmt(f)
    }
}

#[cfg(test)]
mod tests {
    use super::*;

    #[test]
    fn range() {
        let input = WrappedInput::new("foo bar baz");
        let output = input.slice(4..7);
        assert_eq!(output.to_string(), "bar");
    }

    #[test]
    fn range_to() {
        let input = WrappedInput::new("foo bar baz");
        let output = input.slice(..7);
        assert_eq!(output.to_string(), "foo bar");
    }

    #[test]
    fn range_from() {
        let input = WrappedInput::new("foo bar baz");
        let output = input.slice(4..);
        assert_eq!(output.to_string(), "bar baz");
    }

    #[test]
    fn full_range() {
        let input = WrappedInput::new("foo bar baz");
        let output = input.slice(..);
        assert_eq!(output.to_string(), "foo bar baz");
    }

    #[test]
    fn nested_range() {
        let input = WrappedInput::new("lorem foo bar baz ipsum");
        let first_cut = input.slice(6..17);
        let output = first_cut.slice(4..7);
        assert_eq!(first_cut.to_string(), "foo bar baz");
        assert_eq!(output.to_string(), "bar");
    }

    #[test]
    #[should_panic]
    fn out_of_bounds() {
        let input = WrappedInput::new("lorem foo bar baz ipsum");
        input.slice(6..30);
    }

    #[test]
    #[should_panic]
    fn out_of_nested_bounds() {
        let input = WrappedInput::new("lorem foo bar baz ipsum");
        let first_cut = input.slice(6..17);
        first_cut.slice(4..14);
    }

    #[test]
    fn line_break() {
        let input = WrappedInput::new("lorem\nfoo\nbar\nbaz\nipsum");
        assert_eq!(input.slice(5..).preceding_line_break, None);
        assert_eq!(input.slice(6..).preceding_line_break, Some(6));
        assert_eq!(input.slice(6..).slice(10..).preceding_line_break, Some(14));
    }

    #[test]
    fn text_since_line_break() {
        let input = WrappedInput::new("lorem\nfoo\nbar\nbaz\nipsum");
        assert_eq!(input.text_since_line_break(), "");
        assert_eq!(input.slice(5..).text_since_line_break(), "lorem");
        assert_eq!(input.slice(6..).text_since_line_break(), "");
        assert_eq!(input.slice(6..).slice(10..).text_since_line_break(), "ba");
    }
}