ochrepad-core 0.3.0

//! `Document`: the rope, addressed only by [`ByteOffset`].
//!
//! Every Ropey call in the crate lives in this module, so a later move to Ropey 2
//! (byte-indexed) stays local. Ropey is built without `unicode_lines`/`cr_lines`,
//! so only `\n` ends a line; line endings are normalized by `fileio`.

use std::fmt;
use std::ops::Range;

use ropey::Rope;

use crate::position::{ByteOffset, CharIdx};

#[derive(Clone, Default)]
pub struct Document {
    rope: Rope,
}

impl Document {
    pub fn new(text: &str) -> Self {
        Self {
            rope: Rope::from_str(text),
        }
    }

    /// The text in order, in the rope's pieces (for hashing without a copy).
    pub fn chunks(&self) -> impl Iterator<Item = &str> + '_ {
        self.rope.chunks()
    }

    /// Length in bytes; also the offset of the end of the document.
    pub fn len(&self) -> ByteOffset {
        ByteOffset(self.rope.len_bytes())
    }

    pub fn is_empty(&self) -> bool {
        self.rope.len_bytes() == 0
    }

    /// Number of lines. An empty document, or one ending in `\n`, still has a
    /// (possibly empty) last line.
    pub fn line_count(&self) -> usize {
        self.rope.len_lines()
    }

    /// Line containing `pos`. A `\n` belongs to the line it ends.
    pub fn line_of(&self, pos: ByteOffset) -> usize {
        self.rope.byte_to_line(pos.0)
    }

    pub fn line_start(&self, line: usize) -> ByteOffset {
        ByteOffset(self.rope.line_to_byte(line))
    }

    /// End of the line's content, excluding its `\n`.
    pub fn line_end(&self, line: usize) -> ByteOffset {
        if line + 1 < self.line_count() {
            ByteOffset(self.rope.line_to_byte(line + 1) - 1)
        } else {
            self.len()
        }
    }

    /// The line's content, excluding its `\n`.
    pub fn line_text(&self, line: usize) -> String {
        self.slice(self.line_start(line)..self.line_end(line))
    }

    pub fn slice(&self, range: Range<ByteOffset>) -> String {
        self.rope.byte_slice(range.start.0..range.end.0).to_string()
    }

    pub fn is_char_boundary(&self, pos: ByteOffset) -> bool {
        if pos.0 > self.rope.len_bytes() {
            return false;
        }
        let (chunk, chunk_start, _, _) = self.rope.chunk_at_byte(pos.0);
        chunk.is_char_boundary(pos.0 - chunk_start)
    }

    /// The chunk of text containing `pos` and the offset at which it starts.
    /// At the end of the document this is the last chunk. Grapheme segmentation
    /// in `position` walks the rope with this.
    pub fn chunk_at(&self, pos: ByteOffset) -> (&str, ByteOffset) {
        let (chunk, start, _, _) = self.rope.chunk_at_byte(pos.0);
        (chunk, ByteOffset(start))
    }

    /// The char starting at `pos`, if any.
    pub fn char_at(&self, pos: ByteOffset) -> Option<char> {
        if pos >= self.len() {
            return None;
        }
        let (chunk, start, _, _) = self.rope.chunk_at_byte(pos.0);
        chunk[pos.0 - start..].chars().next()
    }

    pub fn to_char_idx(&self, pos: ByteOffset) -> CharIdx {
        CharIdx(self.rope.byte_to_char(pos.0))
    }

    /// Replaces `range` with `text`. Both ends must be char boundaries.
    pub fn replace(&mut self, range: Range<ByteOffset>, text: &str) {
        debug_assert!(self.is_char_boundary(range.start) && self.is_char_boundary(range.end));
        let start = self.to_char_idx(range.start);
        if range.start < range.end {
            let end = self.to_char_idx(range.end);
            self.rope.remove(start.0..end.0);
        }
        if !text.is_empty() {
            self.rope.insert(start.0, text);
        }
    }
}

impl fmt::Display for Document {
    fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
        for chunk in self.rope.chunks() {
            f.write_str(chunk)?;
        }
        Ok(())
    }
}

impl fmt::Debug for Document {
    fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
        write!(f, "Document({:?})", self.to_string())
    }
}

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

    #[test]
    fn lines_split_only_on_lf() {
        let d = Document::new("a\r\nb\rc\u{2028}d\n");
        assert_eq!(d.line_count(), 3);
        assert_eq!(d.line_text(0), "a\r");
        assert_eq!(d.line_text(1), "b\rc\u{2028}d");
        assert_eq!(d.line_text(2), "");
        assert_eq!(d.line_end(0), ByteOffset(2));
    }

    #[test]
    fn replace_by_bytes() {
        let mut d = Document::new("héllo wörld");
        d.replace(ByteOffset(1)..ByteOffset(3), "e");
        assert_eq!(d.to_string(), "hello wörld");
        d.replace(ByteOffset(5)..ByteOffset(5), ",");
        assert_eq!(d.to_string(), "hello, wörld");
        assert!(!d.is_char_boundary(ByteOffset(9)));
    }
}