llmx/codex-rs/tui/src/wrapping.rs

use ratatui::text::Line;
use ratatui::text::Span;
use std::ops::Range;
use textwrap::Options;

use crate::render::line_utils::push_owned_lines;

pub(crate) fn wrap_ranges<'a, O>(text: &str, width_or_options: O) -> Vec<Range<usize>>
where
    O: Into<Options<'a>>,
{
    let opts = width_or_options.into();
    let mut lines: Vec<Range<usize>> = Vec::new();
    for line in textwrap::wrap(text, opts).iter() {
        match line {
            std::borrow::Cow::Borrowed(slice) => {
                let start = unsafe { slice.as_ptr().offset_from(text.as_ptr()) as usize };
                let end = start + slice.len();
                let trailing_spaces = text[end..].chars().take_while(|c| *c == ' ').count();
                lines.push(start..end + trailing_spaces + 1);
            }
            std::borrow::Cow::Owned(_) => panic!("wrap_ranges: unexpected owned string"),
        }
    }
    lines
}

/// Like `wrap_ranges` but returns ranges without trailing whitespace and
/// without the sentinel extra byte. Suitable for general wrapping where
/// trailing spaces should not be preserved.
pub(crate) fn wrap_ranges_trim<'a, O>(text: &str, width_or_options: O) -> Vec<Range<usize>>
where
    O: Into<Options<'a>>,
{
    let opts = width_or_options.into();
    let mut lines: Vec<Range<usize>> = Vec::new();
    for line in textwrap::wrap(text, opts).iter() {
        match line {
            std::borrow::Cow::Borrowed(slice) => {
                let start = unsafe { slice.as_ptr().offset_from(text.as_ptr()) as usize };
                let end = start + slice.len();
                lines.push(start..end);
            }
            std::borrow::Cow::Owned(_) => panic!("wrap_ranges_trim: unexpected owned string"),
        }
    }
    lines
}

#[derive(Debug, Clone)]
pub struct RtOptions<'a> {
    /// The width in columns at which the text will be wrapped.
    pub width: usize,
    /// Line ending used for breaking lines.
    pub line_ending: textwrap::LineEnding,
    /// Indentation used for the first line of output. See the
    /// [`Options::initial_indent`] method.
    pub initial_indent: Line<'a>,
    /// Indentation used for subsequent lines of output. See the
    /// [`Options::subsequent_indent`] method.
    pub subsequent_indent: Line<'a>,
    /// Allow long words to be broken if they cannot fit on a line.
    /// When set to `false`, some lines may be longer than
    /// `self.width`. See the [`Options::break_words`] method.
    pub break_words: bool,
    /// Wrapping algorithm to use, see the implementations of the
    /// [`WrapAlgorithm`] trait for details.
    pub wrap_algorithm: textwrap::WrapAlgorithm,
    /// The line breaking algorithm to use, see the [`WordSeparator`]
    /// trait for an overview and possible implementations.
    pub word_separator: textwrap::WordSeparator,
    /// The method for splitting words. This can be used to prohibit
    /// splitting words on hyphens, or it can be used to implement
    /// language-aware machine hyphenation.
    pub word_splitter: textwrap::WordSplitter,
}
impl From<usize> for RtOptions<'_> {
    fn from(width: usize) -> Self {
        RtOptions::new(width)
    }
}

#[allow(dead_code)]
impl<'a> RtOptions<'a> {
    pub fn new(width: usize) -> Self {
        RtOptions {
            width,
            line_ending: textwrap::LineEnding::LF,
            initial_indent: Line::default(),
            subsequent_indent: Line::default(),
            break_words: true,
            word_separator: textwrap::WordSeparator::new(),
            wrap_algorithm: textwrap::WrapAlgorithm::new(),
            word_splitter: textwrap::WordSplitter::HyphenSplitter,
        }
    }

    pub fn line_ending(self, line_ending: textwrap::LineEnding) -> Self {
        RtOptions {
            line_ending,
            ..self
        }
    }

    pub fn width(self, width: usize) -> Self {
        RtOptions { width, ..self }
    }

    pub fn initial_indent(self, initial_indent: Line<'a>) -> Self {
        RtOptions {
            initial_indent,
            ..self
        }
    }

    pub fn subsequent_indent(self, subsequent_indent: Line<'a>) -> Self {
        RtOptions {
            subsequent_indent,
            ..self
        }
    }

    pub fn break_words(self, break_words: bool) -> Self {
        RtOptions {
            break_words,
            ..self
        }
    }

    pub fn word_separator(self, word_separator: textwrap::WordSeparator) -> RtOptions<'a> {
        RtOptions {
            word_separator,
            ..self
        }
    }

    pub fn wrap_algorithm(self, wrap_algorithm: textwrap::WrapAlgorithm) -> RtOptions<'a> {
        RtOptions {
            wrap_algorithm,
            ..self
        }
    }

    pub fn word_splitter(self, word_splitter: textwrap::WordSplitter) -> RtOptions<'a> {
        RtOptions {
            word_splitter,
            ..self
        }
    }
}

pub(crate) fn word_wrap_line<'a, O>(line: &'a Line<'a>, width_or_options: O) -> Vec<Line<'a>>
where
    O: Into<RtOptions<'a>>,
{
    // Flatten the line and record span byte ranges.
    let mut flat = String::new();
    let mut span_bounds = Vec::new();
    let mut acc = 0usize;
    for s in &line.spans {
        let text = s.content.as_ref();
        let start = acc;
        flat.push_str(text);
        acc += text.len();
        span_bounds.push((start..acc, s.style));
    }

    let rt_opts: RtOptions<'a> = width_or_options.into();
    let opts = Options::new(rt_opts.width)
        .line_ending(rt_opts.line_ending)
        .break_words(rt_opts.break_words)
        .wrap_algorithm(rt_opts.wrap_algorithm)
        .word_separator(rt_opts.word_separator)
        .word_splitter(rt_opts.word_splitter);

    let mut out: Vec<Line<'a>> = Vec::new();

    // Compute first line range with reduced width due to initial indent.
    let initial_width_available = opts
        .width
        .saturating_sub(rt_opts.initial_indent.width())
        .max(1);
    let initial_wrapped = wrap_ranges_trim(&flat, opts.clone().width(initial_width_available));
    let Some(first_line_range) = initial_wrapped.first() else {
        return vec![rt_opts.initial_indent.clone()];
    };

    // Build first wrapped line with initial indent.
    let mut first_line = rt_opts.initial_indent.clone().style(line.style);
    {
        let sliced = slice_line_spans(line, &span_bounds, first_line_range);
        let mut spans = first_line.spans;
        spans.append(
            &mut sliced
                .spans
                .into_iter()
                .map(|s| s.patch_style(line.style))
                .collect(),
        );
        first_line.spans = spans;
        out.push(first_line);
    }

    // Wrap the remainder using subsequent indent width and map back to original indices.
    let base = first_line_range.end;
    let skip_leading_spaces = flat[base..].chars().take_while(|c| *c == ' ').count();
    let base = base + skip_leading_spaces;
    let subsequent_width_available = opts
        .width
        .saturating_sub(rt_opts.subsequent_indent.width())
        .max(1);
    let remaining_wrapped = wrap_ranges_trim(&flat[base..], opts.width(subsequent_width_available));
    for r in &remaining_wrapped {
        if r.is_empty() {
            continue;
        }
        let mut subsequent_line = rt_opts.subsequent_indent.clone().style(line.style);
        let offset_range = (r.start + base)..(r.end + base);
        let sliced = slice_line_spans(line, &span_bounds, &offset_range);
        let mut spans = subsequent_line.spans;
        spans.append(
            &mut sliced
                .spans
                .into_iter()
                .map(|s| s.patch_style(line.style))
                .collect(),
        );
        subsequent_line.spans = spans;
        out.push(subsequent_line);
    }

    out
}

/// Wrap a sequence of lines, applying the initial indent only to the very first
/// output line, and using the subsequent indent for all later wrapped pieces.
#[allow(dead_code)]
pub(crate) fn word_wrap_lines<'a, I, O>(lines: I, width_or_options: O) -> Vec<Line<'static>>
where
    I: IntoIterator<Item = &'a Line<'a>>,
    O: Into<RtOptions<'a>>,
{
    let base_opts: RtOptions<'a> = width_or_options.into();
    let mut out: Vec<Line<'static>> = Vec::new();

    for (idx, line) in lines.into_iter().enumerate() {
        let opts = if idx == 0 {
            base_opts.clone()
        } else {
            let mut o = base_opts.clone();
            let sub = o.subsequent_indent.clone();
            o = o.initial_indent(sub);
            o
        };
        let wrapped = word_wrap_line(line, opts);
        push_owned_lines(&wrapped, &mut out);
    }

    out
}

#[allow(dead_code)]
pub(crate) fn word_wrap_lines_borrowed<'a, I, O>(lines: I, width_or_options: O) -> Vec<Line<'a>>
where
    I: IntoIterator<Item = &'a Line<'a>>,
    O: Into<RtOptions<'a>>,
{
    let base_opts: RtOptions<'a> = width_or_options.into();
    let mut out: Vec<Line<'a>> = Vec::new();
    let mut first = true;
    for line in lines.into_iter() {
        let opts = if first {
            base_opts.clone()
        } else {
            base_opts
                .clone()
                .initial_indent(base_opts.subsequent_indent.clone())
        };
        out.extend(word_wrap_line(line, opts));
        first = false;
    }
    out
}

fn slice_line_spans<'a>(
    original: &'a Line<'a>,
    span_bounds: &[(Range<usize>, ratatui::style::Style)],
    range: &Range<usize>,
) -> Line<'a> {
    let start_byte = range.start;
    let end_byte = range.end;
    let mut acc: Vec<Span<'a>> = Vec::new();
    for (i, (range, style)) in span_bounds.iter().enumerate() {
        let s = range.start;
        let e = range.end;
        if e <= start_byte {
            continue;
        }
        if s >= end_byte {
            break;
        }
        let seg_start = start_byte.max(s);
        let seg_end = end_byte.min(e);
        if seg_end > seg_start {
            let local_start = seg_start - s;
            let local_end = seg_end - s;
            let content = original.spans[i].content.as_ref();
            let slice = &content[local_start..local_end];
            acc.push(Span {
                style: *style,
                content: std::borrow::Cow::Borrowed(slice),
            });
        }
        if e >= end_byte {
            break;
        }
    }
    Line {
        style: original.style,
        alignment: original.alignment,
        spans: acc,
    }
}

#[cfg(test)]
mod tests {
    use super::*;
    use itertools::Itertools as _;
    use pretty_assertions::assert_eq;
    use ratatui::style::Color;
    use ratatui::style::Stylize;
    use std::string::ToString;

    fn concat_line(line: &Line) -> String {
        line.spans
            .iter()
            .map(|s| s.content.as_ref())
            .collect::<String>()
    }

    #[test]
    fn trivial_unstyled_no_indents_wide_width() {
        let line = Line::from("hello");
        let out = word_wrap_line(&line, 10);
        assert_eq!(out.len(), 1);
        assert_eq!(concat_line(&out[0]), "hello");
    }

    #[test]
    fn simple_unstyled_wrap_narrow_width() {
        let line = Line::from("hello world");
        let out = word_wrap_line(&line, 5);
        assert_eq!(out.len(), 2);
        assert_eq!(concat_line(&out[0]), "hello");
        assert_eq!(concat_line(&out[1]), "world");
    }

    #[test]
    fn simple_styled_wrap_preserves_styles() {
        let line = Line::from(vec!["hello ".red(), "world".into()]);
        let out = word_wrap_line(&line, 6);
        assert_eq!(out.len(), 2);
        // First line should carry the red style
        assert_eq!(concat_line(&out[0]), "hello");
        assert_eq!(out[0].spans.len(), 1);
        assert_eq!(out[0].spans[0].style.fg, Some(Color::Red));
        // Second line is unstyled
        assert_eq!(concat_line(&out[1]), "world");
        assert_eq!(out[1].spans.len(), 1);
        assert_eq!(out[1].spans[0].style.fg, None);
    }

    #[test]
    fn with_initial_and_subsequent_indents() {
        let opts = RtOptions::new(8)
            .initial_indent(Line::from("- "))
            .subsequent_indent(Line::from("  "));
        let line = Line::from("hello world foo");
        let out = word_wrap_line(&line, opts);
        // Expect three lines with proper prefixes
        assert!(concat_line(&out[0]).starts_with("- "));
        assert!(concat_line(&out[1]).starts_with("  "));
        assert!(concat_line(&out[2]).starts_with("  "));
        // And content roughly segmented
        assert_eq!(concat_line(&out[0]), "- hello");
        assert_eq!(concat_line(&out[1]), "  world");
        assert_eq!(concat_line(&out[2]), "  foo");
    }

    #[test]
    fn empty_initial_indent_subsequent_spaces() {
        let opts = RtOptions::new(8)
            .initial_indent(Line::from(""))
            .subsequent_indent(Line::from("    "));
        let line = Line::from("hello world foobar");
        let out = word_wrap_line(&line, opts);
        assert!(concat_line(&out[0]).starts_with("hello"));
        for l in &out[1..] {
            assert!(concat_line(l).starts_with("    "));
        }
    }

    #[test]
    fn empty_input_yields_single_empty_line() {
        let line = Line::from("");
        let out = word_wrap_line(&line, 10);
        assert_eq!(out.len(), 1);
        assert_eq!(concat_line(&out[0]), "");
    }

    #[test]
    fn leading_spaces_preserved_on_first_line() {
        let line = Line::from("   hello");
        let out = word_wrap_line(&line, 8);
        assert_eq!(out.len(), 1);
        assert_eq!(concat_line(&out[0]), "   hello");
    }

    #[test]
    fn multiple_spaces_between_words_dont_start_next_line_with_spaces() {
        let line = Line::from("hello   world");
        let out = word_wrap_line(&line, 8);
        assert_eq!(out.len(), 2);
        assert_eq!(concat_line(&out[0]), "hello");
        assert_eq!(concat_line(&out[1]), "world");
    }

    #[test]
    fn break_words_false_allows_overflow_for_long_word() {
        let opts = RtOptions::new(5).break_words(false);
        let line = Line::from("supercalifragilistic");
        let out = word_wrap_line(&line, opts);
        assert_eq!(out.len(), 1);
        assert_eq!(concat_line(&out[0]), "supercalifragilistic");
    }

    #[test]
    fn hyphen_splitter_breaks_at_hyphen() {
        let line = Line::from("hello-world");
        let out = word_wrap_line(&line, 7);
        assert_eq!(out.len(), 2);
        assert_eq!(concat_line(&out[0]), "hello-");
        assert_eq!(concat_line(&out[1]), "world");
    }

    #[test]
    fn indent_consumes_width_leaving_one_char_space() {
        let opts = RtOptions::new(4)
            .initial_indent(Line::from(">>>>"))
            .subsequent_indent(Line::from("--"));
        let line = Line::from("hello");
        let out = word_wrap_line(&line, opts);
        assert_eq!(out.len(), 3);
        assert_eq!(concat_line(&out[0]), ">>>>h");
        assert_eq!(concat_line(&out[1]), "--el");
        assert_eq!(concat_line(&out[2]), "--lo");
    }

    #[test]
    fn wide_unicode_wraps_by_display_width() {
        let line = Line::from("😀😀😀");
        let out = word_wrap_line(&line, 4);
        assert_eq!(out.len(), 2);
        assert_eq!(concat_line(&out[0]), "😀😀");
        assert_eq!(concat_line(&out[1]), "😀");
    }

    #[test]
    fn styled_split_within_span_preserves_style() {
        use ratatui::style::Stylize;
        let line = Line::from(vec!["abcd".red()]);
        let out = word_wrap_line(&line, 2);
        assert_eq!(out.len(), 2);
        assert_eq!(out[0].spans.len(), 1);
        assert_eq!(out[1].spans.len(), 1);
        assert_eq!(out[0].spans[0].style.fg, Some(Color::Red));
        assert_eq!(out[1].spans[0].style.fg, Some(Color::Red));
        assert_eq!(concat_line(&out[0]), "ab");
        assert_eq!(concat_line(&out[1]), "cd");
    }

    #[test]
    fn wrap_lines_applies_initial_indent_only_once() {
        let opts = RtOptions::new(8)
            .initial_indent(Line::from("- "))
            .subsequent_indent(Line::from("  "));

        let lines = vec![Line::from("hello world"), Line::from("foo bar baz")];
        let out = word_wrap_lines(&lines, opts);

        // Expect: first line prefixed with "- ", subsequent wrapped pieces with "  "
        // and for the second input line, there should be no "- " prefix on its first piece
        let rendered: Vec<String> = out.iter().map(concat_line).collect();
        assert!(rendered[0].starts_with("- "));
        for r in rendered.iter().skip(1) {
            assert!(r.starts_with("  "));
        }
    }

    #[test]
    fn wrap_lines_without_indents_is_concat_of_single_wraps() {
        let lines = vec![Line::from("hello"), Line::from("world!")];
        let out = word_wrap_lines(&lines, 10);
        let rendered: Vec<String> = out.iter().map(concat_line).collect();
        assert_eq!(rendered, vec!["hello", "world!"]);
    }

    #[test]
    fn wrap_lines_borrowed_applies_initial_indent_only_once() {
        let opts = RtOptions::new(8)
            .initial_indent(Line::from("- "))
            .subsequent_indent(Line::from("  "));

        let lines = [Line::from("hello world"), Line::from("foo bar baz")];
        let out = word_wrap_lines_borrowed(lines.iter(), opts);

        let rendered: Vec<String> = out.iter().map(concat_line).collect();
        assert!(rendered.first().unwrap().starts_with("- "));
        for r in rendered.iter().skip(1) {
            assert!(r.starts_with("  "));
        }
    }

    #[test]
    fn wrap_lines_borrowed_without_indents_is_concat_of_single_wraps() {
        let lines = [Line::from("hello"), Line::from("world!")];
        let out = word_wrap_lines_borrowed(lines.iter(), 10);
        let rendered: Vec<String> = out.iter().map(concat_line).collect();
        assert_eq!(rendered, vec!["hello", "world!"]);
    }

    #[test]
    fn line_height_counts_double_width_emoji() {
        let line = "😀😀😀".into(); // each emoji ~ width 2
        assert_eq!(word_wrap_line(&line, 4).len(), 2);
        assert_eq!(word_wrap_line(&line, 2).len(), 3);
        assert_eq!(word_wrap_line(&line, 6).len(), 1);
    }

    #[test]
    fn word_wrap_does_not_split_words_simple_english() {
        let sample = "Years passed, and Willowmere thrived in peace and friendship. Mira’s herb garden flourished with both ordinary and enchanted plants, and travelers spoke of the kindness of the woman who tended them.";
        let line = Line::from(sample);
        let lines = [line];
        // Force small width to exercise wrapping at spaces.
        let wrapped = word_wrap_lines_borrowed(&lines, 40);
        let joined: String = wrapped.iter().map(ToString::to_string).join("\n");
        assert_eq!(
            joined,
            r#"Years passed, and Willowmere thrived
in peace and friendship. Mira’s herb
garden flourished with both ordinary and
enchanted plants, and travelers spoke
of the kindness of the woman who tended
them."#
        );
    }
}