rich/text.rs
1//! Styled text with spans.
2//!
3//! Port of upstream `rich/text.py` (core subset). [`Text`] is a plain string
4//! plus a list of [`Span`]s, each applying a [`Style`] to a byte range. Spans
5//! may overlap and nest; [`Text::render`] flattens them into non-overlapping
6//! [`Segment`]s by combining every span covering each run.
7
8use crate::cells::{cell_len, set_cell_size};
9use crate::console::{Justify, Overflow};
10use crate::errors::Result;
11use crate::markup;
12use crate::segment::Segment;
13use crate::style::{Style, StyleType};
14use crate::theme::Theme;
15
16/// The control codes upstream drops in `Text.__init__` (`strip_control_codes`):
17/// BEL, backspace, vertical tab, form feed and carriage return. Tab and newline
18/// are layout, not control, and are kept.
19///
20/// Crate-visible because **every** producer of a `Text` plus its spans must agree
21/// on this set. `markup::render` computes span byte-offsets as it builds the
22/// plain string; if it kept a code that `Text::new` later removed, the content
23/// would shift left while the offsets stayed put, and a boundary landing inside
24/// a multi-byte character panics on slicing.
25pub(crate) fn is_control_code(c: char) -> bool {
26 matches!(c, '\u{7}' | '\u{8}' | '\u{b}' | '\u{c}' | '\r')
27}
28
29/// Cell width of a tab stop: upstream's `Console.tab_size` default, and the
30/// fallback when neither a text nor its console sets one (`tab_size or 8`).
31pub const DEFAULT_TAB_SIZE: usize = 8;
32
33/// A style applied to a byte range `[start, end)` of a [`Text`]'s plain string.
34/// Mirrors `rich.text.Span`.
35///
36/// The style may be a *name* rather than a resolved [`Style`]; see [`StyleType`].
37/// Names are resolved when the text is rendered, against the theme of whichever
38/// console renders it.
39#[derive(Debug, Clone, PartialEq, Eq)]
40pub struct Span {
41 pub start: usize,
42 pub end: usize,
43 pub style: StyleType,
44}
45
46/// Styled text. Mirrors `rich.text.Text`.
47#[derive(Debug, Clone, Default)]
48pub struct Text {
49 plain: String,
50 spans: Vec<Span>,
51 /// A base style applied to the whole text. May be an unresolved name.
52 style: StyleType,
53 /// How lines are justified within the render width. `None` defers to
54 /// the console options (upstream's `justify=None`); `Some(Default)` is an
55 /// explicit `justify="default"`, which a container's justify does not
56 /// override.
57 justify: Option<Justify>,
58 /// What to do with lines wider than the render width. `None` defers to the
59 /// console options, then to [`Overflow::Fold`].
60 overflow: Option<Overflow>,
61 /// Whether to skip wrapping. `None` defers to the console options, then to
62 /// `false`.
63 no_wrap: Option<bool>,
64 /// Tab stop width. `None` defers to the console's `tab_size` when
65 /// rendered, then to [`DEFAULT_TAB_SIZE`]. Upstream's `Text.tab_size`.
66 tab_size: Option<usize>,
67}
68
69impl Text {
70 /// Strip the control codes upstream removes in `Text.__init__`
71 /// (`strip_control_codes`): BEL, backspace, vertical tab, form feed and
72 /// carriage return. Tab and newline are deliberately kept — they are layout,
73 /// not control.
74 fn strip_control_codes(text: &str) -> String {
75 if text.chars().any(is_control_code) {
76 text.chars().filter(|c| !is_control_code(*c)).collect()
77 } else {
78 text.to_string()
79 }
80 }
81
82 /// Plain, unstyled text.
83 pub fn new(plain: impl Into<String>) -> Self {
84 Text {
85 plain: Text::strip_control_codes(&plain.into()),
86 spans: Vec::new(),
87 style: StyleType::default(),
88 justify: None,
89 overflow: None,
90 no_wrap: None,
91 tab_size: None,
92 }
93 }
94
95 /// Text with a base style, which may be a style *name* resolved at render
96 /// time (`Text::styled("hi", "repr.number")`) or a resolved [`Style`].
97 pub fn styled(plain: impl Into<String>, style: impl Into<StyleType>) -> Self {
98 Text {
99 // Strips too: upstream's `Text.__init__` does this regardless of
100 // style, and a constructor that skipped it would reintroduce the
101 // offset divergence the moment a caller added spans.
102 plain: Text::strip_control_codes(&plain.into()),
103 spans: Vec::new(),
104 style: style.into(),
105 justify: None,
106 overflow: None,
107 no_wrap: None,
108 tab_size: None,
109 }
110 }
111
112 /// Set how lines are justified within the render width (builder form).
113 pub fn justify(mut self, justify: Justify) -> Self {
114 self.set_justify(justify);
115 self
116 }
117
118 /// Set how lines are justified within the render width.
119 ///
120 /// [`Justify::Default`] means *unset* here (upstream's `justify=None`),
121 /// deferring to the console options; use
122 /// [`set_justify_option`](Self::set_justify_option) for an explicit
123 /// `justify="default"`.
124 pub fn set_justify(&mut self, justify: Justify) {
125 self.justify = (justify != Justify::Default).then_some(justify);
126 }
127
128 /// This text's own justify method ([`Justify::Default`] when unset).
129 pub fn get_justify(&self) -> Justify {
130 self.justify.unwrap_or_default()
131 }
132
133 /// Set this text's justify, distinguishing an explicit
134 /// `Some(Justify::Default)` (upstream's `justify="default"`, which a
135 /// table column's or the print call's justify does not override) from
136 /// `None` (upstream's `justify=None`, which defers to them).
137 pub fn set_justify_option(&mut self, justify: Option<Justify>) {
138 self.justify = justify;
139 }
140
141 /// This text's own justify, `None` when unset. See
142 /// [`set_justify_option`](Self::set_justify_option).
143 pub fn get_justify_option(&self) -> Option<Justify> {
144 self.justify
145 }
146
147 /// Set what happens to lines wider than the render width (builder form).
148 pub fn overflow(mut self, overflow: Overflow) -> Self {
149 self.overflow = Some(overflow);
150 self
151 }
152
153 /// Set what happens to lines wider than the render width. Pass `None` to
154 /// defer to the console options.
155 pub fn set_overflow(&mut self, overflow: Option<Overflow>) {
156 self.overflow = overflow;
157 }
158
159 /// This text's own overflow method, if it set one.
160 pub fn get_overflow(&self) -> Option<Overflow> {
161 self.overflow
162 }
163
164 /// Disable (or re-enable) wrapping for this text (builder form).
165 pub fn no_wrap(mut self, no_wrap: bool) -> Self {
166 self.no_wrap = Some(no_wrap);
167 self
168 }
169
170 /// Disable (or re-enable) wrapping. Pass `None` to defer to the console
171 /// options.
172 pub fn set_no_wrap(&mut self, no_wrap: Option<bool>) {
173 self.no_wrap = no_wrap;
174 }
175
176 /// This text's own no-wrap setting, if it set one.
177 pub fn get_no_wrap(&self) -> Option<bool> {
178 self.no_wrap
179 }
180
181 /// Set the tab stop width (builder form). Port of `Text(tab_size=…)`.
182 pub fn tab_size(mut self, tab_size: usize) -> Self {
183 self.tab_size = Some(tab_size);
184 self
185 }
186
187 /// Set the tab stop width. Pass `None` to defer to the console's
188 /// `tab_size`. Upstream's `Text.tab_size` attribute.
189 pub fn set_tab_size(&mut self, tab_size: Option<usize>) {
190 self.tab_size = tab_size;
191 }
192
193 /// This text's own tab stop width, if it set one.
194 pub fn get_tab_size(&self) -> Option<usize> {
195 self.tab_size
196 }
197
198 /// Shorten this text to at most `max_width` cells, optionally padding it out
199 /// to exactly `max_width` when it is shorter. Port of `Text.truncate`.
200 ///
201 /// `overflow` defaults to this text's own method, then to [`Overflow::Fold`];
202 /// [`Overflow::Ignore`] leaves the text alone entirely. Note that `Fold` and
203 /// `Crop` behave identically here — folding is a property of *wrapping*, and
204 /// a line that has already been wrapped can only be cut.
205 pub fn truncate(&mut self, max_width: usize, overflow: Option<Overflow>, pad: bool) {
206 let overflow = overflow.or(self.overflow).unwrap_or(Overflow::Fold);
207 if overflow == Overflow::Ignore {
208 return;
209 }
210 let length = cell_len(&self.plain);
211 if length > max_width {
212 let plain = if overflow == Overflow::Ellipsis {
213 // `…` is one cell wide, so cut one short and add it back.
214 format!(
215 "{}…",
216 set_cell_size(&self.plain, max_width.saturating_sub(1))
217 )
218 } else {
219 set_cell_size(&self.plain, max_width)
220 };
221 self.set_plain(plain);
222 } else if pad {
223 let plain = set_cell_size(&self.plain, max_width);
224 self.set_plain(plain);
225 }
226 }
227
228 /// Replace the plain string, clamping every span into the new length so no
229 /// span can dangle past the end. Upstream's `Text.plain` setter does the
230 /// same via `_trim_spans`.
231 fn set_plain(&mut self, plain: String) {
232 // Upstream spans use character offsets. A truncation may replace a
233 // wide glyph with padding or ASCII with a multibyte ellipsis, so byte
234 // offsets alone cannot be clamped into the replacement string.
235 if !self.spans.is_empty()
236 && !self.plain.starts_with(&plain)
237 && !plain.starts_with(&self.plain)
238 {
239 let old_offsets: Vec<usize> = self
240 .plain
241 .char_indices()
242 .map(|(i, _)| i)
243 .chain(std::iter::once(self.plain.len()))
244 .collect();
245 let new_offsets: Vec<usize> = plain
246 .char_indices()
247 .map(|(i, _)| i)
248 .chain(std::iter::once(plain.len()))
249 .collect();
250 let map_offset = |offset: usize| {
251 let character = old_offsets.partition_point(|old| *old < offset);
252 new_offsets[character.min(new_offsets.len() - 1)]
253 };
254 for span in &mut self.spans {
255 span.start = map_offset(span.start);
256 span.end = map_offset(span.end);
257 }
258 }
259 let length = plain.len();
260 self.plain = plain;
261 self.spans.retain(|span| span.start < length);
262 for span in &mut self.spans {
263 span.end = span.end.min(length);
264 }
265 }
266
267 /// An empty `Text` carrying this one's style, justify, overflow and no-wrap.
268 /// Port of `Text.blank_copy`.
269 pub fn blank_copy(&self) -> Text {
270 Text {
271 plain: String::new(),
272 spans: Vec::new(),
273 style: self.style.clone(),
274 justify: self.justify,
275 overflow: self.overflow,
276 no_wrap: self.no_wrap,
277 tab_size: self.tab_size,
278 }
279 }
280
281 /// Cut this text at each byte offset in `offsets`, returning the pieces.
282 /// Port of `Text.divide`.
283 ///
284 /// Every piece inherits the base style, justify, overflow and no-wrap, and
285 /// each span is re-based onto the pieces it covers. Spans that would come out
286 /// empty are dropped, matching upstream's `new_end > new_start`.
287 ///
288 /// Offsets are **byte** offsets (as everywhere else in this port's span
289 /// arithmetic) and must fall on `char` boundaries.
290 pub fn divide(&self, offsets: &[usize]) -> Vec<Text> {
291 if offsets.is_empty() {
292 return vec![self.clone()];
293 }
294 let mut bounds = Vec::with_capacity(offsets.len() + 2);
295 bounds.push(0);
296 bounds.extend(offsets.iter().copied());
297 bounds.push(self.plain.len());
298
299 let mut lines: Vec<Text> = bounds
300 .windows(2)
301 .map(|w| {
302 let (start, end) = (w[0].min(self.plain.len()), w[1].min(self.plain.len()));
303 let mut line = self.blank_copy();
304 if start < end {
305 line.plain = self.plain[start..end].to_string();
306 }
307 line
308 })
309 .collect();
310
311 // Upstream binary-searches for the first and last line a span touches
312 // rather than visiting every line for every span, which made
313 // splitting a highlighted file quadratic in its line count. Offsets
314 // from `split` are sorted; for anything else keep the full scan, whose
315 // result the search would not reproduce.
316 let sorted = bounds.windows(2).all(|w| w[0] <= w[1]);
317 for span in &self.spans {
318 let lines_touched = if sorted {
319 // Lines ending at or before the span starts cannot hold it,
320 // nor can lines starting at or after it ends.
321 let first = bounds[1..].partition_point(|&end| end <= span.start);
322 let last = bounds[..bounds.len() - 1].partition_point(|&start| start < span.end);
323 first..last.max(first)
324 } else {
325 0..bounds.len() - 1
326 };
327 for index in lines_touched {
328 let (line_start, line_end) = (bounds[index], bounds[index + 1]);
329 let new_start = span.start.max(line_start) - line_start;
330 let new_end = span.end.min(line_end).saturating_sub(line_start);
331 if new_end > new_start {
332 lines[index].spans.push(Span {
333 start: new_start,
334 end: new_end,
335 style: span.style.clone(),
336 });
337 }
338 }
339 }
340 lines
341 }
342
343 /// Split on `separator`. Port of `Text.split`.
344 ///
345 /// `include_separator` keeps the separator at the end of each piece.
346 /// `allow_blank` keeps the trailing empty piece that a text ending in the
347 /// separator would otherwise produce.
348 ///
349 /// # Panics
350 /// If `separator` is empty, which upstream asserts against.
351 pub fn split(&self, separator: &str, include_separator: bool, allow_blank: bool) -> Vec<Text> {
352 assert!(!separator.is_empty(), "separator must not be empty");
353 if !self.plain.contains(separator) {
354 return vec![self.clone()];
355 }
356 let matches: Vec<usize> = self
357 .plain
358 .match_indices(separator)
359 .map(|(i, _)| i)
360 .collect();
361 let mut lines = if include_separator {
362 let offsets: Vec<usize> = matches.iter().map(|s| s + separator.len()).collect();
363 self.divide(&offsets)
364 } else {
365 // Cut on both sides of every separator, then drop the separators.
366 let mut offsets = Vec::with_capacity(matches.len() * 2);
367 for start in &matches {
368 offsets.push(*start);
369 offsets.push(start + separator.len());
370 }
371 self.divide(&offsets)
372 .into_iter()
373 .filter(|line| line.plain != separator)
374 .collect()
375 };
376 if !allow_blank && self.plain.ends_with(separator) {
377 lines.pop();
378 }
379 lines
380 }
381
382 /// Pad both sides with `count` copies of `character`. Port of `Text.pad`.
383 pub fn pad(&mut self, count: usize, character: char) {
384 self.pad_left(count, character);
385 self.pad_right(count, character);
386 }
387
388 /// Pad the left with `count` copies of `character`, shifting every span to
389 /// follow the text. Port of `Text.pad_left`.
390 pub fn pad_left(&mut self, count: usize, character: char) {
391 if count == 0 {
392 return;
393 }
394 let padding: String = std::iter::repeat_n(character, count).collect();
395 let offset = padding.len();
396 self.plain.insert_str(0, &padding);
397 for span in &mut self.spans {
398 span.start += offset;
399 span.end += offset;
400 }
401 }
402
403 /// Pad the right with `count` copies of `character`. Port of
404 /// `Text.pad_right`. Spans are untouched, so the padding is unstyled.
405 pub fn pad_right(&mut self, count: usize, character: char) {
406 if count == 0 {
407 return;
408 }
409 self.plain.extend(std::iter::repeat_n(character, count));
410 }
411
412 /// Drop the last `amount` bytes, clipping any span that reached into them.
413 /// Port of `Text.right_crop`.
414 pub fn right_crop(&mut self, amount: usize) {
415 if amount == 0 {
416 return;
417 }
418 let max_offset = self.plain.len().saturating_sub(amount);
419 let plain = self.plain[..max_offset].to_string();
420 self.set_plain(plain);
421 }
422
423 /// Remove trailing whitespace. Port of `Text.rstrip`.
424 pub fn rstrip(&mut self) {
425 // Python's `str.rstrip()` whitespace, which includes U+001C..U+001F.
426 let plain = self.plain.trim_end_matches(is_python_space).to_string();
427 self.set_plain(plain);
428 }
429
430 /// Remove *only as much* trailing whitespace as it takes to get down to
431 /// `size` characters, leaving the rest. Port of `Text.rstrip_end`.
432 ///
433 /// This is what lets a wrapped line keep the space that ended it while a
434 /// line that overshot the width gives its padding back. As upstream, the
435 /// length is `len(self)` — characters, not cells.
436 pub fn rstrip_end(&mut self, size: usize) {
437 let length = self.plain.chars().count();
438 if length <= size {
439 return;
440 }
441 let excess = length - size;
442 let trimmed = self.plain.trim_end_matches(is_python_space);
443 let whitespace = self.plain[trimmed.len()..].chars().count();
444 if whitespace > 0 {
445 let keep = length - whitespace.min(excess);
446 let bytes = self.plain.len() - char_to_byte(&self.plain, keep);
447 self.right_crop(bytes);
448 }
449 }
450
451 /// Replace tabs with spaces up to the next `tab_size` stop. Port of
452 /// `Text.expand_tabs`.
453 ///
454 /// Styles extend over the inserted spaces, so a styled tab pads in its own
455 /// style rather than punching an unstyled hole (upstream reaches the same
456 /// result via `extend_style`).
457 /// Append `count` spaces, extending any span that reached the end so the
458 /// padding takes its style. Port of `Text.extend_style`.
459 fn extend_style(&mut self, count: usize) {
460 if count == 0 {
461 return;
462 }
463 let length = self.plain.len();
464 self.plain.extend(std::iter::repeat_n(' ', count));
465 for span in &mut self.spans {
466 if span.end >= length {
467 span.end += count;
468 }
469 }
470 }
471
472 pub fn expand_tabs(&mut self, tab_size: usize) {
473 if !self.plain.contains('\t') || tab_size == 0 {
474 return;
475 }
476 // Rebuilt part-by-part rather than by remapping offsets, because the
477 // *split* is observable: upstream turns each tab-terminated run into its
478 // own piece, so a span crossing several tabs comes back as several spans
479 // and renders as several segments. Remapping offsets keeps one span and
480 // emits one segment — same colours, different bytes.
481 let mut result = Text::new("");
482 for line in self.split("\n", true, false) {
483 if !line.plain.contains('\t') {
484 result = result.append_text(&line);
485 continue;
486 }
487 let mut cell_position = 0usize;
488 for mut part in line.split("\t", true, false) {
489 if part.plain.ends_with('\t') {
490 // The tab becomes one space, then the run is padded out to
491 // the next stop — so a tab always advances at least one cell.
492 part.plain.pop();
493 part.plain.push(' ');
494 cell_position += part.cell_len();
495 let remainder = cell_position % tab_size;
496 if remainder != 0 {
497 let spaces = tab_size - remainder;
498 part.extend_style(spaces);
499 cell_position += spaces;
500 }
501 } else {
502 cell_position += part.cell_len();
503 }
504 result = result.append_text(&part);
505 }
506 }
507 self.plain = result.plain;
508 self.spans = result.spans;
509 }
510
511 /// Join `lines` with this text as the separator, carrying each piece's base
512 /// style across as a covering span. Port of `Text.join`.
513 pub fn join(&self, lines: &[Text]) -> Text {
514 let mut joined = self.blank_copy();
515 let last = lines.len().saturating_sub(1);
516 for (index, line) in lines.iter().enumerate() {
517 joined = joined.append_text(line);
518 if !self.plain.is_empty() && index != last {
519 joined = joined.append_text(self);
520 }
521 }
522 joined
523 }
524
525 /// A `Text` from a string containing ANSI escape codes, one decoded line
526 /// per line joined by newlines, with base `style`. Port of
527 /// `Text.from_ansi` (its `tab_size` defaults to 8; set `justify`,
528 /// `overflow` and `no_wrap` with the builders).
529 pub fn from_ansi(text: &str, style: impl Into<StyleType>) -> Text {
530 let mut joiner = Text::styled("\n", style);
531 joiner.tab_size = Some(DEFAULT_TAB_SIZE);
532 let lines = crate::ansi::AnsiDecoder::new().decode_split_newlines(text);
533 joiner.join(&lines)
534 }
535
536 /// The indentation unit of the text: the greatest common divisor of its
537 /// even line indents, or 1. Port of `Text.detect_indentation`.
538 pub fn detect_indentation(&self) -> usize {
539 fn gcd(a: usize, b: usize) -> usize {
540 if b == 0 {
541 a
542 } else {
543 gcd(b, a % b)
544 }
545 }
546 let mut indents: Vec<usize> = self
547 .plain
548 .split('\n')
549 .map(|line| line.len() - line.trim_start_matches(' ').len())
550 .filter(|indent| indent % 2 == 0)
551 .collect();
552 indents.sort_unstable();
553 indents.dedup();
554 match indents.into_iter().reduce(gcd) {
555 None | Some(0) => 1,
556 Some(indentation) => indentation,
557 }
558 }
559
560 /// A copy with indent guides drawn in each line's leading spaces: every
561 /// `indent_size` spaces (auto-detected for `None`) become `character`
562 /// then padding, styled `style`. Port of `Text.with_indent_guides`
563 /// (upstream's defaults are `"│"` and `"dim green"`).
564 pub fn with_indent_guides(
565 &self,
566 indent_size: Option<usize>,
567 character: &str,
568 style: impl Into<StyleType>,
569 ) -> Text {
570 let style = style.into();
571 let indent_size = indent_size.unwrap_or_else(|| self.detect_indentation());
572 let mut text = self.clone();
573 text.expand_tabs(match self.tab_size {
574 None => DEFAULT_TAB_SIZE,
575 Some(tab_size) => tab_size,
576 });
577 let indent_line = format!("{character}{}", " ".repeat(indent_size.saturating_sub(1)));
578 let mut new_lines: Vec<Text> = Vec::new();
579 let mut blank_lines = 0usize;
580 for mut line in text.split("\n", false, true) {
581 let indent = line.plain.len() - line.plain.trim_start_matches(' ').len();
582 // `re.match(r"^( *)(.*)$")`: `.` stops at a line break, so a
583 // line with nothing after its indent (or a `\r`) counts as blank.
584 let rest = &line.plain[indent..];
585 if rest.is_empty() || rest.starts_with('\r') {
586 blank_lines += 1;
587 continue;
588 }
589 let (full_indents, remaining_space) = (
590 indent.checked_div(indent_size).unwrap_or(0),
591 indent.checked_rem(indent_size).unwrap_or(indent),
592 );
593 let new_indent = format!(
594 "{}{}",
595 indent_line.repeat(full_indents),
596 " ".repeat(remaining_space)
597 );
598 // `line.plain = new_indent + line.plain[len(new_indent):]`: the
599 // same number of characters, so upstream's (character) spans stay
600 // put; here the byte offsets past the indent shift.
601 let old_bytes = char_to_byte(&line.plain, new_indent.chars().count());
602 let plain = format!("{new_indent}{}", &line.plain[old_bytes..]);
603 let remap = |position: usize| {
604 if position >= old_bytes {
605 position - old_bytes + new_indent.len()
606 } else {
607 char_to_byte(&new_indent, position)
608 }
609 };
610 for span in &mut line.spans {
611 span.start = remap(span.start);
612 span.end = remap(span.end);
613 }
614 line.plain = plain;
615 line.stylize(style.clone(), 0, new_indent.len());
616 for _ in 0..std::mem::take(&mut blank_lines) {
617 new_lines.push(Text::styled(new_indent.clone(), style.clone()));
618 }
619 new_lines.push(line);
620 }
621 for _ in 0..blank_lines {
622 new_lines.push(Text::styled("", style.clone()));
623 }
624 let mut separator = text.blank_copy();
625 separator.plain.push('\n');
626 separator.join(&new_lines)
627 }
628
629 /// Style every occurrence of any of `words`. Port of `Text.highlight_words`,
630 /// returning the number of matches.
631 pub fn highlight_words(
632 &mut self,
633 words: &[&str],
634 style: impl Into<StyleType>,
635 case_sensitive: bool,
636 ) -> Result<usize> {
637 let alternation = words
638 .iter()
639 .map(|word| fancy_regex::escape(word).into_owned())
640 .collect::<Vec<_>>()
641 .join("|");
642 if alternation.is_empty() {
643 return Ok(0);
644 }
645 let pattern = if case_sensitive {
646 alternation
647 } else {
648 format!("(?i){alternation}")
649 };
650 self.highlight_regex(&pattern, Some(style.into()), "")
651 }
652
653 /// Style every match of `pattern`, returning the number of matches. Full port
654 /// of `Text.highlight_regex`.
655 ///
656 /// `style`, when given, styles the whole match. Each **named group** is then
657 /// styled with `{style_prefix}{name}` as a style *name*, left for the theme
658 /// to resolve at render time — which is how a highlighter colours its groups
659 /// without ever seeing a console.
660 ///
661 /// Groups that did not participate in the match, and zero-width ones, are
662 /// skipped.
663 pub fn highlight_regex(
664 &mut self,
665 pattern: &str,
666 style: Option<StyleType>,
667 style_prefix: &str,
668 ) -> Result<usize> {
669 let regex = fancy_regex::Regex::new(pattern)
670 .map_err(|e| crate::errors::RichError::Regex(format!("invalid pattern: {e}")))?;
671 Ok(self.highlight_with_regex(®ex, style, style_prefix))
672 }
673
674 /// As [`highlight_regex`](Self::highlight_regex) with an already-compiled
675 /// pattern, for callers that apply the same patterns repeatedly.
676 ///
677 /// A match that errors mid-scan (a `fancy-regex` backtrack-limit hit) stops
678 /// the scan and keeps the spans found so far, rather than discarding them.
679 pub(crate) fn highlight_with_regex(
680 &mut self,
681 regex: &fancy_regex::Regex,
682 style: Option<StyleType>,
683 style_prefix: &str,
684 ) -> usize {
685 // Capture-definition order, matching upstream's `match.groupdict()`.
686 let names: Vec<(usize, String)> = regex
687 .capture_names()
688 .enumerate()
689 .filter_map(|(index, name)| name.map(|name| (index, name.to_string())))
690 .collect();
691
692 // Scanning borrows the plain string while the spans are pushed, so move
693 // it out and put it back — no copy, and no fighting the borrow checker.
694 let plain = std::mem::take(&mut self.plain);
695 let mut count = 0;
696 for captures in regex.captures_iter(&plain) {
697 let Ok(captures) = captures else { break };
698 if let (Some(style), Some(whole)) = (style.as_ref(), captures.get(0)) {
699 if whole.end() > whole.start() {
700 self.spans.push(Span {
701 start: whole.start(),
702 end: whole.end(),
703 style: style.clone(),
704 });
705 }
706 }
707 count += 1;
708 for (index, name) in &names {
709 if let Some(group) = captures.get(*index) {
710 if group.end() > group.start() {
711 self.spans.push(Span {
712 start: group.start(),
713 end: group.end(),
714 style: StyleType::Name(format!("{style_prefix}{name}")),
715 });
716 }
717 }
718 }
719 }
720 self.plain = plain;
721 count
722 }
723
724 /// Build styled text from console markup. Port of `Text.from_markup`.
725 ///
726 /// Tag names are stored on the spans and resolved when the text is rendered,
727 /// so no theme is needed here.
728 pub fn from_markup(markup_text: &str) -> Result<Text> {
729 markup::render(markup_text)
730 }
731
732 /// The unstyled string content.
733 pub fn plain(&self) -> &str {
734 &self.plain
735 }
736
737 /// The spans currently applied.
738 pub fn spans(&self) -> &[Span] {
739 &self.spans
740 }
741
742 /// Length in terminal cells.
743 pub fn cell_len(&self) -> usize {
744 cell_len(&self.plain)
745 }
746
747 /// True when there is no content.
748 pub fn is_empty(&self) -> bool {
749 self.plain.is_empty()
750 }
751
752 /// Append more text, optionally under `style` (a resolved [`Style`] or a
753 /// style name).
754 pub fn append(&mut self, text: &str, style: Option<StyleType>) {
755 let start = self.plain.len();
756 // Strip here as well as in `new`: upstream's `Text.append` runs the same
757 // `strip_control_codes`, and skipping it let BEL, backspace, vertical
758 // tab and form feed reach the terminal through every path that builds
759 // text incrementally — Markdown, Syntax and plain files. A backspace run
760 // is a spoofing tool: `FAILED\u{8}\u{8}\u{8}\u{8}\u{8}\u{8}PASSED`
761 // displays as `PASSED`.
762 self.plain.push_str(&Text::strip_control_codes(text));
763 let end = self.plain.len();
764 // `if style:` — a null style or an empty name adds no span, so it
765 // leaves no segment boundary behind either.
766 if let Some(style) = style.filter(|style| !is_falsy_style(style)) {
767 self.spans.push(Span { start, end, style });
768 }
769 }
770
771 /// Append another `Text`, carrying over its base style (as a covering span)
772 /// and all of its spans, shifted to their new offsets. Port of
773 /// `Text.append_text`. Consumes `self` and returns it for chaining.
774 pub fn append_text(mut self, other: &Text) -> Text {
775 let offset = self.plain.len();
776 self.plain.push_str(&other.plain);
777 let end = self.plain.len();
778 if !other.style.is_null_style() {
779 self.spans.push(Span {
780 start: offset,
781 end,
782 style: other.style.clone(),
783 });
784 }
785 for span in &other.spans {
786 self.spans.push(Span {
787 start: span.start + offset,
788 end: span.end + offset,
789 style: span.style.clone(),
790 });
791 }
792 self
793 }
794
795 /// Apply `style` to the byte range `[start, end)`. Port of `Text.stylize`,
796 /// including its argument order.
797 ///
798 /// `style` may be a resolved [`Style`] or a name (`"repr.number"`) left for
799 /// the renderer to look up. Byte offsets, not char offsets; ASCII-only
800 /// callers such as highlighters are unaffected by the distinction.
801 ///
802 /// A range that is empty or inverted is ignored, which is what gives us
803 /// upstream's `end > start` skip for non-participating regex groups.
804 pub fn stylize(&mut self, style: impl Into<StyleType>, start: usize, end: usize) {
805 let style = style.into();
806 let end = end.min(self.plain.len());
807 // `if style:` — a falsy style adds no span.
808 if start >= end || is_falsy_style(&style) {
809 return;
810 }
811 self.spans.push(Span { start, end, style });
812 }
813
814 /// As [`stylize`](Self::stylize), but the span goes *first*, beneath every
815 /// existing span. Port of `Text.stylize_before`.
816 pub fn stylize_before(&mut self, style: impl Into<StyleType>, start: usize, end: usize) {
817 let style = style.into();
818 let end = end.min(self.plain.len());
819 if start >= end || is_falsy_style(&style) {
820 return;
821 }
822 self.spans.insert(0, Span { start, end, style });
823 }
824
825 /// Apply metadata to the byte range `start..end` (the rest of the text
826 /// when `end` is `None`). Port of `Text.apply_meta`: a span of
827 /// `Style.from_meta(meta)`, which an empty map (a null style) skips.
828 pub fn apply_meta(&mut self, meta: crate::style::Meta, start: usize, end: Option<usize>) {
829 let end = end.unwrap_or(self.plain.len());
830 self.stylize(Style::from_meta(meta), start, end);
831 }
832
833 /// Apply event-handler metadata to the whole text. Port of `Text.on`:
834 /// handlers are stored as `"@name"` keys over `meta`.
835 pub fn on(
836 &mut self,
837 meta: Option<crate::style::Meta>,
838 handlers: &[(&str, crate::style::MetaValue)],
839 ) -> &mut Self {
840 let len = self.plain.len();
841 self.stylize(Style::on(meta, handlers), 0, len);
842 self
843 }
844
845 /// Append a raw span, as upstream's `text.spans.append(span)` does: no
846 /// clamping and no falsy-style check.
847 ///
848 /// Offsets are **byte** offsets into [`plain`](Self::plain) and must fall
849 /// on `char` boundaries (upstream's are character offsets; convert first).
850 /// A span may extend past the end of the text, as upstream allows.
851 pub fn push_span(&mut self, span: Span) {
852 self.spans.push(span);
853 }
854
855 /// The spans, mutably. Upstream's `Text.spans` is a public, mutable list;
856 /// the same byte-offset rules as [`push_span`](Self::push_span) apply.
857 pub fn spans_mut(&mut self) -> &mut Vec<Span> {
858 &mut self.spans
859 }
860
861 /// Replace every span. Port of the `Text.spans` setter.
862 pub fn set_spans(&mut self, spans: Vec<Span>) {
863 self.spans = spans;
864 }
865
866 /// Drop this text's own `justify`, `overflow` and `no_wrap`, so they defer to
867 /// the console options as upstream's `Text.join` result does.
868 pub(crate) fn clear_layout_options(&mut self) {
869 self.justify = None;
870 self.overflow = None;
871 self.no_wrap = None;
872 self.tab_size = None;
873 }
874
875 /// Move the base style into a span over the whole text, ahead of the
876 /// existing spans — what `Text.join` does to each joined text. A falsy
877 /// style (null, or an empty name) adds no span, as `if text.style:` skips
878 /// it upstream.
879 pub(crate) fn base_style_to_span(&mut self) {
880 let style = std::mem::take(&mut self.style);
881 let falsy =
882 style.is_null_style() || matches!(&style, StyleType::Name(name) if name.is_empty());
883 if !falsy {
884 self.spans.insert(
885 0,
886 Span {
887 start: 0,
888 end: self.plain.len(),
889 style,
890 },
891 );
892 }
893 }
894
895 /// The whole-text base style, resolved or named. Port of the `Text.style`
896 /// attribute.
897 pub fn base_style(&self) -> &StyleType {
898 &self.style
899 }
900
901 /// Set the whole-text base style, resolved or named.
902 pub fn set_base_style(&mut self, style: impl Into<StyleType>) {
903 self.style = style.into();
904 }
905
906 /// Flatten into non-overlapping segments (newlines become [`Segment::line`]),
907 /// combining `base_style`, this text's base style, and every covering span.
908 /// Does **not** wrap. Port of the core of `Text.render`.
909 ///
910 /// Named span styles are resolved against `theme`.
911 pub fn render(&self, theme: &Theme, base_style: &Style) -> Vec<Segment> {
912 self.render_joined(theme, base_style, None)
913 }
914
915 /// The `(minimum, maximum)` cell width of this text: `maximum` is the widest
916 /// hard line, `minimum` the widest word. Port of `Text.__rich_measure__`.
917 pub fn measurement(&self) -> (usize, usize) {
918 // Measured against the raw string, as upstream does: `cell_len` counts
919 // a tab as zero cells, and tabs expand only at render time. A printed
920 // `Text` renders at the full width (see `Renderable::printed_text`), so
921 // this measurement never becomes its own render width (#447).
922 measure_plain(&self.plain)
923 }
924
925 /// Render into visual lines, wrapping each hard line to `width` cells when
926 /// `Some`, and justifying per this text's own justify.
927 pub fn render_lines(
928 &self,
929 theme: &Theme,
930 base_style: &Style,
931 width: Option<usize>,
932 ) -> Vec<Vec<Segment>> {
933 self.render_lines_justified(theme, base_style, width, self.get_justify())
934 }
935
936 /// Like [`render_lines`](Self::render_lines) but with an explicit `justify`
937 /// (used by the console to apply `options.justify`).
938 pub fn render_lines_justified(
939 &self,
940 theme: &Theme,
941 base_style: &Style,
942 width: Option<usize>,
943 justify: Justify,
944 ) -> Vec<Vec<Segment>> {
945 self.render_lines_wrapped(
946 theme,
947 base_style,
948 width,
949 justify,
950 self.overflow.unwrap_or(Overflow::Fold),
951 self.no_wrap.unwrap_or(false),
952 )
953 }
954
955 /// The full wrap-justify-truncate pipeline, with every knob resolved by the
956 /// caller. Port of `Text.wrap`.
957 ///
958 /// Lines are split on `\n`, wrapped to `width` (folding over-long words only
959 /// when `overflow` is [`Overflow::Fold`]), justified, and finally truncated
960 /// to `width`. [`Overflow::Ignore`] skips wrapping and truncation both, so
961 /// lines may come back wider than `width`.
962 ///
963 /// Tabs expand to this text's own [`tab_size`](Self::get_tab_size), else
964 /// [`DEFAULT_TAB_SIZE`]; [`render_lines_wrapped_tabs`](Self::render_lines_wrapped_tabs)
965 /// takes the width explicitly (upstream's `wrap(tab_size=…)`).
966 pub fn render_lines_wrapped(
967 &self,
968 theme: &Theme,
969 base_style: &Style,
970 width: Option<usize>,
971 justify: Justify,
972 overflow: Overflow,
973 no_wrap: bool,
974 ) -> Vec<Vec<Segment>> {
975 self.render_lines_wrapped_tabs(
976 theme,
977 base_style,
978 width,
979 justify,
980 overflow,
981 no_wrap,
982 match self.tab_size {
983 None | Some(0) => DEFAULT_TAB_SIZE,
984 Some(tab_size) => tab_size,
985 },
986 )
987 }
988
989 /// The tab stop width `Text.__rich_console__` wraps with on `console`:
990 /// this text's own, else the console's, and `8` for zero
991 /// (`tab_size or 8`).
992 pub fn console_tab_size(&self, console: &crate::console::Console) -> usize {
993 match self.tab_size.unwrap_or_else(|| console.tab_size()) {
994 0 => DEFAULT_TAB_SIZE,
995 tab_size => tab_size,
996 }
997 }
998
999 /// [`render_lines_wrapped`](Self::render_lines_wrapped) with an explicit
1000 /// tab stop width. Port of `Text.wrap(…, tab_size=…)`.
1001 #[allow(clippy::too_many_arguments)]
1002 pub fn render_lines_wrapped_tabs(
1003 &self,
1004 theme: &Theme,
1005 base_style: &Style,
1006 width: Option<usize>,
1007 justify: Justify,
1008 overflow: Overflow,
1009 no_wrap: bool,
1010 tab_size: usize,
1011 ) -> Vec<Vec<Segment>> {
1012 // Tabs are expanded before anything measures or wraps the text, as
1013 // upstream's `Text.wrap` does per line. Without this a tab occupies one
1014 // cell everywhere in the layout and then eight on the terminal, so every
1015 // width calculation downstream is wrong.
1016 if self.plain.contains('\t') {
1017 let mut expanded = self.clone();
1018 expanded.expand_tabs(tab_size);
1019 return expanded.render_lines_wrapped_tabs(
1020 theme, base_style, width, justify, overflow, no_wrap, tab_size,
1021 );
1022 }
1023
1024 // Resolve every span's style once, up front, into a vector parallel to
1025 // `self.spans` — upstream's `style_map`. Resolving inside the per-line
1026 // loop would re-parse the same names for every visual line.
1027 let resolved: Vec<Style> = self
1028 .spans
1029 .iter()
1030 .map(|span| theme.get_style_or_null(&span.style))
1031 .collect();
1032 let effective_base = base_style.combine(&theme.get_style_or_null(&self.style));
1033 // Upstream folds `overflow == "ignore"` into no_wrap before splitting.
1034 let no_wrap = no_wrap || overflow == Overflow::Ignore;
1035 let groups = self.wrapped_ranges(width, overflow, no_wrap);
1036 // Which spans touch which visual line, found once by binary search as
1037 // upstream's `divide` does. Scanning every span for every line (and
1038 // every cut within it) made printing a long highlighted repr quadratic.
1039 let line_spans = self.spans_by_line(&groups);
1040 let mut line_spans = line_spans.iter();
1041 let Some(width) = width else {
1042 return groups
1043 .into_iter()
1044 .flatten()
1045 .map(|(start, end)| {
1046 let ids = line_spans.next().map(Vec::as_slice).unwrap_or(&[]);
1047 self.line_segments(&resolved, start, end, &effective_base, ids)
1048 })
1049 .collect();
1050 };
1051
1052 let mut lines: Vec<Vec<Segment>> = Vec::new();
1053 // One hard line at a time, as upstream's `for line in self.split(...)`
1054 // does — the paragraph boundary is what full justification treats as
1055 // ragged, so the groups cannot be flattened first.
1056 for group in groups {
1057 let group_spans: Vec<&[usize]> = group
1058 .iter()
1059 .map(|_| line_spans.next().map(Vec::as_slice).unwrap_or(&[]))
1060 .collect();
1061 let mut new_lines: Vec<Vec<Segment>> = group
1062 .iter()
1063 .zip(&group_spans)
1064 .map(|(&(start, end), ids)| {
1065 self.line_segments(&resolved, start, end, &effective_base, ids)
1066 })
1067 .collect();
1068
1069 // `overflow == "ignore"` is a hard stop upstream: the line is
1070 // appended verbatim and the loop `continue`s, so it is neither
1071 // justified nor truncated. Padding it out to the width here was
1072 // adding trailing spaces to text upstream returns untouched.
1073 if overflow == Overflow::Ignore {
1074 lines.append(&mut new_lines);
1075 continue;
1076 }
1077
1078 // Give each wrapped line back the padding it overshot by, exactly
1079 // where upstream's `Text.wrap` does it — after dividing, before
1080 // justifying. `divide_line` counts a word *including* its trailing
1081 // space, so a line whose last word ends flush with the width comes
1082 // back one cell too long; without this the ellipsis overflow then
1083 // chops a real character to make room for a `…` that upstream never
1084 // emits ("abcdefghij more" at width 10 became "abcdefghi…", not
1085 // "abcdefghij").
1086 //
1087 // Only in the wrapping branch: upstream's `rstrip_end` loop sits
1088 // inside `Text.wrap`'s `else`, which `no_wrap` skips entirely.
1089 if !no_wrap {
1090 for line in &mut new_lines {
1091 rstrip_end_line(line, width);
1092 }
1093 }
1094 if justify != Justify::Default {
1095 let last = new_lines.len().saturating_sub(1);
1096 for (index, line) in new_lines.iter_mut().enumerate() {
1097 // Full justification leaves the final line of the paragraph
1098 // ragged, so it needs to know where it is in the group.
1099 let (start, end) = group[index];
1100 *line = justify_line(
1101 line,
1102 width,
1103 justify,
1104 overflow,
1105 &effective_base,
1106 index == last,
1107 &|char_index| self.covered_at(start, end, char_index, group_spans[index]),
1108 );
1109 }
1110 }
1111 for line in &mut new_lines {
1112 *line = truncate_line(line, width, overflow);
1113 }
1114 lines.append(&mut new_lines);
1115 }
1116 lines
1117 }
1118
1119 /// Whether any span covers the `char_index`-th character of the line
1120 /// `[start, end)` — i.e. whether upstream's `Text.render` has a span
1121 /// boundary at that character's edge of the line. Justify padding joins
1122 /// the neighbouring run only where it does not.
1123 ///
1124 /// `ids` are the spans touching that line (see [`spans_by_line`](Self::spans_by_line)).
1125 fn covered_at(&self, start: usize, end: usize, char_index: usize, ids: &[usize]) -> bool {
1126 let Some((offset, _)) = self.plain[start..end].char_indices().nth(char_index) else {
1127 return false;
1128 };
1129 let position = start + offset;
1130 ids.iter().any(|&id| {
1131 let span = &self.spans[id];
1132 span.start <= position && position < span.end
1133 })
1134 }
1135
1136 /// For each visual line of `groups` (flattened, in order), the indices of
1137 /// the spans that overlap it, in span order. Lines are ordered and
1138 /// disjoint, so each span binary-searches its first and last line, as
1139 /// upstream's `Text.divide` does; the cost is the number of (span, line)
1140 /// pairs that actually overlap rather than spans × lines.
1141 fn spans_by_line(&self, groups: &[Vec<(usize, usize)>]) -> Vec<Vec<usize>> {
1142 let lines: Vec<(usize, usize)> = groups.iter().flatten().copied().collect();
1143 let mut by_line: Vec<Vec<usize>> = vec![Vec::new(); lines.len()];
1144 // Upstream only divides a text that has a newline or a wrap cut, and
1145 // `divide` drops spans that come out empty. An undivided text keeps
1146 // its empty spans, whose start/end events still split the run
1147 // (`[b]a[i][/i]b[/b]` renders `a` and `b` as two segments).
1148 let undivided = lines.len() == 1;
1149 for (id, span) in self.spans.iter().enumerate() {
1150 if span.start >= span.end {
1151 if undivided && span.start == span.end {
1152 by_line[0].push(id);
1153 }
1154 continue;
1155 }
1156 let first = lines.partition_point(|&(_, end)| end <= span.start);
1157 let last = lines.partition_point(|&(start, _)| start < span.end);
1158 for index in first..last.max(first) {
1159 let (start, end) = lines[index];
1160 if span.start.max(start) < span.end.min(end) {
1161 by_line[index].push(id);
1162 }
1163 }
1164 }
1165 by_line
1166 }
1167
1168 /// As [`render_lines_wrapped`](Self::render_lines_wrapped), flattened into a
1169 /// single segment stream with [`Segment::line`] between visual lines.
1170 pub fn render_joined_wrapped(
1171 &self,
1172 theme: &Theme,
1173 base_style: &Style,
1174 width: usize,
1175 justify: Justify,
1176 overflow: Overflow,
1177 no_wrap: bool,
1178 ) -> Vec<Segment> {
1179 self.render_joined_wrapped_tabs(
1180 theme,
1181 base_style,
1182 width,
1183 justify,
1184 overflow,
1185 no_wrap,
1186 match self.tab_size {
1187 None | Some(0) => DEFAULT_TAB_SIZE,
1188 Some(tab_size) => tab_size,
1189 },
1190 )
1191 }
1192
1193 /// [`render_joined_wrapped`](Self::render_joined_wrapped) with an explicit
1194 /// tab stop width (see [`console_tab_size`](Self::console_tab_size)).
1195 #[allow(clippy::too_many_arguments)]
1196 pub fn render_joined_wrapped_tabs(
1197 &self,
1198 theme: &Theme,
1199 base_style: &Style,
1200 width: usize,
1201 justify: Justify,
1202 overflow: Overflow,
1203 no_wrap: bool,
1204 tab_size: usize,
1205 ) -> Vec<Segment> {
1206 let lines = self.render_lines_wrapped_tabs(
1207 theme,
1208 base_style,
1209 Some(width),
1210 justify,
1211 overflow,
1212 no_wrap,
1213 tab_size,
1214 );
1215 let mut segments = Vec::new();
1216 let last = lines.len().saturating_sub(1);
1217 for (index, line) in lines.into_iter().enumerate() {
1218 // A final empty line (a trailing newline, or an empty text left
1219 // unpadded, as with `overflow="ignore"`) is still a line upstream:
1220 // `Text.render` yields `Segment("")` before the `end`.
1221 if index == last && line.is_empty() {
1222 segments.push(Segment::new("", None));
1223 }
1224 segments.extend(line);
1225 if index != last {
1226 segments.push(Segment::line());
1227 }
1228 }
1229 segments
1230 }
1231
1232 /// Render into a flat segment stream with [`Segment::line`] between visual
1233 /// lines (wrapping when `width` is `Some`), using this text's own justify.
1234 fn render_joined(
1235 &self,
1236 theme: &Theme,
1237 base_style: &Style,
1238 width: Option<usize>,
1239 ) -> Vec<Segment> {
1240 let lines = self.render_lines(theme, base_style, width);
1241 let mut segments = Vec::new();
1242 let last = lines.len().saturating_sub(1);
1243 for (index, line) in lines.into_iter().enumerate() {
1244 segments.extend(line);
1245 if index != last {
1246 segments.push(Segment::line());
1247 }
1248 }
1249 segments
1250 }
1251
1252 /// The `(start_byte, end_byte)` range of each visual line, **grouped by the
1253 /// hard line it came from**: hard lines split on `\n`, then each wrapped to
1254 /// `width` cells when `Some`.
1255 ///
1256 /// The grouping is not cosmetic. Upstream wraps and justifies one hard line
1257 /// at a time (`for line in self.split(...)`), so full justification leaves
1258 /// the last visual line of *each paragraph* ragged. Flattening first makes
1259 /// every paragraph but the final one get stretched, which turned
1260 /// `"line here"` into `"line here"`.
1261 fn wrapped_ranges(
1262 &self,
1263 width: Option<usize>,
1264 overflow: Overflow,
1265 no_wrap: bool,
1266 ) -> Vec<Vec<(usize, usize)>> {
1267 let mut hard: Vec<(usize, usize)> = Vec::new();
1268 let mut start = 0;
1269 for (i, byte) in self.plain.bytes().enumerate() {
1270 if byte == b'\n' {
1271 hard.push((start, i));
1272 start = i + 1;
1273 }
1274 }
1275 hard.push((start, self.plain.len()));
1276
1277 let Some(width) = width else {
1278 return hard.into_iter().map(|range| vec![range]).collect();
1279 };
1280 if no_wrap {
1281 return hard.into_iter().map(|range| vec![range]).collect();
1282 }
1283
1284 let mut groups: Vec<Vec<(usize, usize)>> = Vec::with_capacity(hard.len());
1285 for (a, b) in hard {
1286 let sub = &self.plain[a..b];
1287 // Only `fold` breaks a word that is wider than the whole line; the
1288 // cropping methods leave it long and let truncation cut it.
1289 let breaks = crate::wrap::divide_line(sub, width, overflow == Overflow::Fold);
1290 let mut cuts = vec![a];
1291 let mut previous_char = 0;
1292 let mut previous_byte = 0;
1293 for char_offset in breaks {
1294 // Breaks are ordered char offsets. Scan only the next slice;
1295 // rescanning the prefix for every line is quadratic.
1296 previous_byte += char_to_byte(&sub[previous_byte..], char_offset - previous_char);
1297 previous_char = char_offset;
1298 cuts.push(a + previous_byte);
1299 }
1300 cuts.push(b);
1301 groups.push(cuts.windows(2).map(|w| (w[0], w[1])).collect());
1302 }
1303 groups
1304 }
1305
1306 /// Combine `effective_base` with every span covering `[start, end)`,
1307 /// producing non-overlapping segments for that byte range. Port of the
1308 /// sweep in upstream's `Text.render`: span start/end events are sorted
1309 /// once and walked with a stack of active spans.
1310 ///
1311 /// `resolved` is the per-render style map, index-parallel to `self.spans`;
1312 /// `ids` are the spans overlapping this line, in span order. Active spans
1313 /// are combined in span order (upstream's `sorted(stack)`), and spans that
1314 /// resolved to nothing are **not** skipped — they still contribute a
1315 /// boundary. The highlighter fixtures depend on it: an ISO-8601 date emits
1316 /// separate segments per sub-field even where the field styles are
1317 /// identical.
1318 fn line_segments(
1319 &self,
1320 resolved: &[Style],
1321 start: usize,
1322 end: usize,
1323 effective_base: &Style,
1324 ids: &[usize],
1325 ) -> Vec<Segment> {
1326 if start >= end {
1327 return Vec::new();
1328 }
1329 // (offset, leaving, span id); entering sorts before leaving.
1330 let mut events: Vec<(usize, bool, usize)> = Vec::with_capacity(ids.len() * 2 + 1);
1331 for &id in ids {
1332 let span = &self.spans[id];
1333 let (span_start, span_end) = (span.start.max(start), span.end.min(end));
1334 // An empty span (kept only on an undivided line) is a boundary.
1335 let empty = span.start == span.end && start < span.start && span.end < end;
1336 if span_start < span_end || empty {
1337 events.push((span_start, false, id));
1338 events.push((span_end, true, id));
1339 }
1340 }
1341 events.sort_unstable();
1342
1343 let mut segments = Vec::new();
1344 let mut stack: Vec<usize> = Vec::new();
1345 let mut offset = start;
1346 let mut events = events.into_iter().peekable();
1347 loop {
1348 // Apply every event at `offset`.
1349 while let Some(&(position, leaving, id)) = events.peek() {
1350 if position > offset {
1351 break;
1352 }
1353 events.next();
1354 match stack.binary_search(&id) {
1355 Ok(index) if leaving => {
1356 stack.remove(index);
1357 }
1358 Err(index) if !leaving => stack.insert(index, id),
1359 _ => {}
1360 }
1361 }
1362 let next = events
1363 .peek()
1364 .map_or(end, |&(position, _, _)| position.min(end));
1365 if next <= offset {
1366 break;
1367 }
1368 let mut style = effective_base.clone();
1369 for &id in &stack {
1370 style = style.combine(&resolved[id]);
1371 }
1372 segments.push(Segment::new(&self.plain[offset..next], Some(style)));
1373 offset = next;
1374 }
1375 segments
1376 }
1377}
1378
1379/// Python truthiness of a style argument: a null `Style` and an empty style
1380/// name are falsy, and upstream's `append`/`stylize` skip them.
1381fn is_falsy_style(style: &StyleType) -> bool {
1382 style.is_null_style() || matches!(style, StyleType::Name(name) if name.is_empty())
1383}
1384
1385/// Byte offset of the `char_idx`-th char in `text` (clamped to `text.len()`).
1386fn char_to_byte(text: &str, char_idx: usize) -> usize {
1387 text.char_indices()
1388 .nth(char_idx)
1389 .map(|(byte, _)| byte)
1390 .unwrap_or(text.len())
1391}
1392
1393/// Cut a rendered line down to `width` cells, applying `overflow`. Segment-level
1394/// counterpart of [`Text::truncate`], used once per line at the end of the wrap
1395/// pipeline.
1396///
1397/// [`Overflow::Fold`] and [`Overflow::Crop`] both plain-cut: by this point the
1398/// line has already been wrapped, so anything still over-long is an unbreakable
1399/// run that folding cannot help with.
1400///
1401/// [`Overflow::Ellipsis`] cuts one cell short and appends `…`. The marker takes
1402/// the style of the first segment the cut did *not* keep whole — upstream writes
1403/// the ellipsis into the plain string and lets span-trimming decide, which works
1404/// out to the same rule, including when the cut lands exactly on a boundary.
1405fn truncate_line(line: &[Segment], width: usize, overflow: Overflow) -> Vec<Segment> {
1406 if overflow == Overflow::Ignore {
1407 return line.to_vec();
1408 }
1409 // Measured and cut over the WHOLE line, exactly as upstream's `Text.truncate`
1410 // works on `self.plain`. Summing the segments instead is wrong wherever a
1411 // grapheme spans a segment boundary — a zero-width joiner at the end of one
1412 // segment swallows the first character of the next, so the per-segment sum
1413 // reads one cell too wide and cuts text upstream keeps.
1414 let plain: String = line.iter().map(|segment| segment.text.as_str()).collect();
1415 if cell_len(&plain) <= width {
1416 return line.to_vec();
1417 }
1418 let ellipsis = overflow == Overflow::Ellipsis;
1419 // `…` occupies one cell, so the kept text must stop one cell early.
1420 let keep = if ellipsis {
1421 width.saturating_sub(1)
1422 } else {
1423 width
1424 };
1425 // Never longer than `plain`, and only ever differs from a byte prefix of it
1426 // in its final byte (a wide grapheme straddling the cut becomes a space), so
1427 // slicing it at the original segment boundaries stays on char boundaries.
1428 let kept = set_cell_size(&plain, keep);
1429
1430 let mut result: Vec<Segment> = Vec::new();
1431 // Style the ellipsis inherits: that of the first segment the cut did not
1432 // keep whole, falling back to the last segment's when the cut lands exactly
1433 // on the end of the line's bytes.
1434 let mut cut_style: Option<Style> = line.last().and_then(|segment| segment.style.clone());
1435 let mut offset = 0usize;
1436 for segment in line {
1437 if offset >= kept.len() {
1438 cut_style = segment.style.clone();
1439 break;
1440 }
1441 let end = (offset + segment.text.len()).min(kept.len());
1442 if end > offset {
1443 result.push(Segment::new(&kept[offset..end], segment.style.clone()));
1444 }
1445 if offset + segment.text.len() > kept.len() {
1446 cut_style = segment.style.clone();
1447 break;
1448 }
1449 offset = end;
1450 }
1451 if ellipsis {
1452 // Upstream appends the marker to the plain string and re-renders, so it
1453 // lands inside the preceding run rather than beside it. Merging keeps
1454 // the byte stream identical — a separate segment would re-emit the style.
1455 match result.last_mut() {
1456 Some(last) if !last.control && last.style == cut_style => last.text.push('…'),
1457 _ => result.push(Segment::new("…", cut_style)),
1458 }
1459 }
1460 result
1461}
1462
1463/// Split a rendered line into whitespace-separated words, each word keeping its
1464/// own styled segments. Separator spaces are dropped — [`full_justify`] decides
1465/// the new gaps. Port of the `line.split(" ")` in upstream's `full` branch.
1466fn split_words(line: &[Segment]) -> Vec<Vec<Segment>> {
1467 let mut words: Vec<Vec<Segment>> = Vec::new();
1468 let mut current: Vec<Segment> = Vec::new();
1469 for segment in line {
1470 // A segment can straddle a space, so split within it and keep the style.
1471 for (index, piece) in segment.text.split(' ').enumerate() {
1472 if index > 0 {
1473 words.push(std::mem::take(&mut current));
1474 }
1475 if !piece.is_empty() {
1476 current.push(Segment::new(piece, segment.style.clone()));
1477 }
1478 }
1479 }
1480 words.push(current);
1481 // Wrapping leaves a trailing space on every line but the last, so the naive
1482 // split ends with an empty word. Upstream's `Text.split` drops it, and the
1483 // count matters: it decides how many gaps share the slack.
1484 if words.last().is_some_and(|w| w.is_empty()) {
1485 words.pop();
1486 }
1487 words
1488}
1489
1490/// Distribute `width` across `line`'s words by widening the gaps between them.
1491/// Direct port of the `justify == "full"` branch of upstream's `Lines.justify`:
1492/// every gap starts at one space, and the extra columns are handed out from the
1493/// rightmost gap backwards, cycling.
1494fn full_justify(line: &[Segment], width: usize, style: &Style) -> Vec<Segment> {
1495 let words = split_words(line);
1496 let words_size: usize = words
1497 .iter()
1498 .map(|word| word.iter().map(Segment::cell_length).sum::<usize>())
1499 .sum();
1500 let mut num_spaces = words.len().saturating_sub(1);
1501 let mut spaces = vec![1usize; num_spaces];
1502 if !spaces.is_empty() {
1503 let mut index = 0;
1504 while words_size + num_spaces < width {
1505 let slot = spaces.len() - index - 1;
1506 spaces[slot] += 1;
1507 num_spaces += 1;
1508 index = (index + 1) % spaces.len();
1509 }
1510 }
1511
1512 let mut out: Vec<Segment> = Vec::new();
1513 for (index, word) in words.iter().enumerate() {
1514 out.extend(word.iter().cloned());
1515 if let Some(&gap) = spaces.get(index) {
1516 // Upstream styles the gap with the surrounding style when the two
1517 // neighbours agree, else with the line's base style.
1518 let before = word.last().and_then(|s| s.style.clone());
1519 let after = words
1520 .get(index + 1)
1521 .and_then(|w| w.first())
1522 .and_then(|s| s.style.clone());
1523 let gap_style = if before == after {
1524 before.unwrap_or_else(|| style.clone())
1525 } else {
1526 style.clone()
1527 };
1528 out.push(Segment::new(" ".repeat(gap), Some(gap_style)));
1529 }
1530 }
1531 out
1532}
1533
1534/// Pad `line` to `width` cells according to `justify`, using `style` for the
1535/// pad (so e.g. a styled table cell fills with its own style).
1536///
1537/// `is_last` marks the final line of the paragraph, which full justification
1538/// leaves ragged rather than stretching.
1539///
1540/// `covered(i)` says whether a span covers the line's `i`-th character.
1541/// Upstream pads the plain string, so its padding joins the adjacent run
1542/// unless a span ends (or starts) at the text's edge; a separate segment
1543/// would re-emit the style as a second SGR run.
1544fn justify_line(
1545 line: &[Segment],
1546 width: usize,
1547 justify: Justify,
1548 overflow: Overflow,
1549 style: &Style,
1550 is_last: bool,
1551 covered: &dyn Fn(usize) -> bool,
1552) -> Vec<Segment> {
1553 // Full justification rewrites the interior gaps instead of padding an edge.
1554 if justify == Justify::Full {
1555 // Upstream `break`s before the final line, so it is left exactly as
1556 // wrapped — not even padded out to the width, unlike every other mode.
1557 return if is_last {
1558 line.to_vec()
1559 } else {
1560 full_justify(line, width, style)
1561 };
1562 }
1563 let mut content = line.to_vec();
1564 // Upstream's `Lines.justify` calls `line.rstrip()` in its `center` and
1565 // `right` branches — and only there — so the space wrapping left at the end
1566 // of a line is *not* content to be positioned. Keeping it shifts the visible
1567 // text half a space left when centring (`" abcd efgh ijklmnop "` became
1568 // `"abcd efgh ijklmnop "`) and a whole column left when right-aligning.
1569 // `left`/`full` deliberately keep it: upstream pads them without stripping.
1570 if matches!(justify, Justify::Center | Justify::Right) {
1571 rstrip_line(&mut content);
1572 // …and then TRUNCATES, before it pads. The order is load-bearing: cell
1573 // width is not additive across a cut, so a line whose over-long tail is
1574 // chopped can measure *less* than the width afterwards and still want
1575 // padding. A leading zero-width joiner is the clearest case — it eats the
1576 // character after it, so cutting the line hands one of its cells back —
1577 // and padding first computes the gap from the pre-cut measurement, which
1578 // is zero, and leaves the line short.
1579 content = truncate_line(&content, width, overflow);
1580 }
1581
1582 // Whether padding may join the content's first / last run: no span
1583 // covers the character at that edge, so upstream's run there is the base
1584 // style alone and the padding extends it.
1585 let content_chars: usize = content.iter().map(|s| s.text.chars().count()).sum();
1586 // With no content at all, the two pads of a centred line are one run.
1587 let join_left = content_chars > 0 && !covered(0);
1588 let join_right = content_chars == 0 || !covered(content_chars - 1);
1589
1590 let mut out = Vec::with_capacity(content.len() + 2);
1591 match justify {
1592 Justify::Right => {
1593 // `line.pad_left(width - cell_len(line.plain))`.
1594 let excess = width.saturating_sub(line_cell_len(&content));
1595 if excess > 0 {
1596 out.push(Segment::new(" ".repeat(excess), Some(style.clone())));
1597 }
1598 append_joined(&mut out, content, join_left);
1599 }
1600 Justify::Center => {
1601 // `pad_left((width - cell_len) // 2)` and then `pad_right(width -
1602 // cell_len)` — the second `cell_len` is re-measured *after* the left
1603 // pad, so the two halves are not simply `excess / 2` and the rest.
1604 let left = width.saturating_sub(line_cell_len(&content)) / 2;
1605 if left > 0 {
1606 out.push(Segment::new(" ".repeat(left), Some(style.clone())));
1607 }
1608 append_joined(&mut out, content, join_left);
1609 let right = width.saturating_sub(line_cell_len(&out));
1610 if right > 0 {
1611 push_joined(
1612 &mut out,
1613 Segment::new(" ".repeat(right), Some(style.clone())),
1614 join_right,
1615 );
1616 }
1617 }
1618 // Left, Default, and full justification's ragged last line pad right.
1619 // Upstream reaches this through `truncate(width, pad=True)`, whose pad
1620 // is driven by the *pre*-truncate length — so padding and truncating are
1621 // mutually exclusive here and the order does not matter.
1622 Justify::Left | Justify::Full | Justify::Default => {
1623 let excess = width.saturating_sub(line_cell_len(&content));
1624 out.append(&mut content);
1625 if excess > 0 {
1626 push_joined(
1627 &mut out,
1628 Segment::new(" ".repeat(excess), Some(style.clone())),
1629 join_right,
1630 );
1631 }
1632 }
1633 }
1634 out
1635}
1636
1637/// Append `segment`, extending the last segment instead when `join` allows it
1638/// and the two share a style.
1639fn push_joined(out: &mut Vec<Segment>, segment: Segment, join: bool) {
1640 match out.last_mut() {
1641 Some(last) if join && !last.control && last.style == segment.style => {
1642 last.text.push_str(&segment.text);
1643 }
1644 _ => out.push(segment),
1645 }
1646}
1647
1648/// Append `content`, joining its first segment onto the last of `out` when
1649/// `join` allows it and the two share a style.
1650fn append_joined(out: &mut Vec<Segment>, content: Vec<Segment>, join: bool) {
1651 let mut content = content.into_iter();
1652 if let Some(first) = content.next() {
1653 push_joined(out, first, join);
1654 }
1655 out.extend(content);
1656}
1657
1658/// The cell width of a rendered line.
1659fn line_cell_len(line: &[Segment]) -> usize {
1660 line.iter().map(Segment::cell_length).sum()
1661}
1662
1663/// The number of trailing whitespace *characters* on a rendered line.
1664///
1665/// Segment-level, because by the time the wrap pipeline justifies a line the
1666/// spans have already been flattened into [`Segment`]s and there is no `Text`
1667/// left to call `rstrip` on.
1668fn trailing_whitespace(line: &[Segment]) -> usize {
1669 let mut count = 0usize;
1670 for segment in line.iter().rev() {
1671 let trimmed = segment.text.trim_end_matches(is_python_space);
1672 count += segment.text[trimmed.len()..].chars().count();
1673 if !trimmed.is_empty() {
1674 break;
1675 }
1676 }
1677 count
1678}
1679
1680/// Drop the last `count` characters, discarding segments that empty out.
1681/// Segment-level counterpart of `Text.right_crop`.
1682fn right_crop_line(line: &mut Vec<Segment>, count: usize) {
1683 let mut remaining = count;
1684 while remaining > 0 {
1685 let Some(last) = line.last_mut() else { break };
1686 let length = last.text.chars().count();
1687 if length <= remaining {
1688 remaining -= length;
1689 line.pop();
1690 } else {
1691 let keep = char_to_byte(&last.text, length - remaining);
1692 last.text.truncate(keep);
1693 remaining = 0;
1694 }
1695 }
1696}
1697
1698/// Remove all trailing whitespace. Segment-level counterpart of `Text.rstrip`.
1699fn rstrip_line(line: &mut Vec<Segment>) {
1700 right_crop_line(line, trailing_whitespace(line));
1701}
1702
1703/// Remove *only as much* trailing whitespace as it takes to get the line down to
1704/// `size`, leaving the rest. Segment-level counterpart of `Text.rstrip_end`.
1705///
1706/// The length compared against `size` is a **character** count, not a cell
1707/// count: upstream's `Text.rstrip_end` uses `len(self)`, which is
1708/// `len(self.plain)`. The two only diverge on wide characters, and copying the
1709/// quirk is cheaper than explaining a one-column difference later.
1710fn rstrip_end_line(line: &mut Vec<Segment>, size: usize) {
1711 let length: usize = line.iter().map(|s| s.text.chars().count()).sum();
1712 let Some(excess) = length.checked_sub(size).filter(|excess| *excess > 0) else {
1713 return;
1714 };
1715 let whitespace = trailing_whitespace(line);
1716 if whitespace > 0 {
1717 right_crop_line(line, whitespace.min(excess));
1718 }
1719}
1720
1721/// `Text.__rich_measure__` on a plain string: `(widest word, widest line)`.
1722pub(crate) fn measure_plain(plain: &str) -> (usize, usize) {
1723 // `text.splitlines()` and `text.split()`: Python's line boundaries
1724 // (U+2028, U+0085, …) and whitespace (including U+001C..U+001F), not
1725 // just `\n` and Rust's `White_Space`.
1726 let max_line = python_splitlines(plain)
1727 .into_iter()
1728 .map(cell_len)
1729 .max()
1730 .unwrap_or(0);
1731 let min_word = plain
1732 .split(is_python_space)
1733 .filter(|word| !word.is_empty())
1734 .map(cell_len)
1735 .max()
1736 .unwrap_or(max_line);
1737 (min_word, max_line)
1738}
1739
1740/// Python's `str.isspace()`: Rust's `White_Space` plus the four ASCII
1741/// information separators U+001C..=U+001F.
1742pub(crate) fn is_python_space(c: char) -> bool {
1743 c.is_whitespace() || ('\u{1c}'..='\u{1f}').contains(&c)
1744}
1745
1746/// Port of Python's `str.splitlines()`: every Unicode line boundary ends a
1747/// line, `\r\n` counts once, and a trailing boundary adds no empty line.
1748pub(crate) fn python_splitlines(text: &str) -> Vec<&str> {
1749 let mut lines = Vec::new();
1750 let mut start = 0;
1751 let mut chars = text.char_indices().peekable();
1752 while let Some((i, c)) = chars.next() {
1753 if matches!(
1754 c,
1755 '\n' | '\r'
1756 | '\x0b'
1757 | '\x0c'
1758 | '\x1c'
1759 | '\x1d'
1760 | '\x1e'
1761 | '\u{85}'
1762 | '\u{2028}'
1763 | '\u{2029}'
1764 ) {
1765 lines.push(&text[start..i]);
1766 start = i + c.len_utf8();
1767 if c == '\r' && chars.peek().map(|&(_, n)| n) == Some('\n') {
1768 chars.next();
1769 start += 1;
1770 }
1771 }
1772 }
1773 if start < text.len() {
1774 lines.push(&text[start..]);
1775 }
1776 lines
1777}
1778
1779#[cfg(test)]
1780mod tests {
1781 use super::*;
1782
1783 /// Each divided line's text and its spans as (start, end, style).
1784 type Divided = Vec<(String, Vec<(usize, usize, String)>)>;
1785
1786 /// `divide` as it was before the per-span line search: every line for
1787 /// every span. The search must give the same lines, spans and span order.
1788 fn divide_by_scanning(text: &Text, offsets: &[usize]) -> Divided {
1789 if offsets.is_empty() {
1790 let spans = text
1791 .spans
1792 .iter()
1793 .map(|s| (s.start, s.end, format!("{:?}", s.style)))
1794 .collect();
1795 return vec![(text.plain.clone(), spans)];
1796 }
1797 let mut bounds = vec![0];
1798 bounds.extend(offsets.iter().copied());
1799 bounds.push(text.plain.len());
1800 let mut lines: Divided = bounds
1801 .windows(2)
1802 .map(|w| {
1803 let (start, end) = (w[0].min(text.plain.len()), w[1].min(text.plain.len()));
1804 let plain = if start < end {
1805 text.plain[start..end].to_string()
1806 } else {
1807 String::new()
1808 };
1809 (plain, Vec::new())
1810 })
1811 .collect();
1812 for span in &text.spans {
1813 for (index, window) in bounds.windows(2).enumerate() {
1814 let (line_start, line_end) = (window[0], window[1]);
1815 let new_start = span.start.max(line_start) - line_start;
1816 let new_end = span.end.min(line_end).saturating_sub(line_start);
1817 if new_end > new_start {
1818 lines[index]
1819 .1
1820 .push((new_start, new_end, format!("{:?}", span.style)));
1821 }
1822 }
1823 }
1824 lines
1825 }
1826
1827 #[test]
1828 fn divide_matches_the_full_scan_for_sorted_offsets() {
1829 let mut state = 0x5eed_u64;
1830 let mut next = |bound: usize| {
1831 state = state
1832 .wrapping_mul(6364136223846793005)
1833 .wrapping_add(1442695040888963407);
1834 (state >> 33) as usize % bound.max(1)
1835 };
1836 for case in 0..500 {
1837 let len = next(40);
1838 let plain: String = (0..len)
1839 .map(|i| if i % 7 == 3 { '\n' } else { 'a' })
1840 .collect();
1841 let mut text = Text::new(plain.clone());
1842 for i in 0..next(12) {
1843 // Empty, overlapping and past-the-end spans included.
1844 let start = next(len + 3);
1845 let end = start + next(len + 3);
1846 text.spans.push(Span {
1847 start,
1848 end,
1849 style: format!("s{i}").as_str().into(),
1850 });
1851 }
1852 let mut offsets: Vec<usize> = (0..next(10)).map(|_| next(len + 4)).collect();
1853 offsets.sort_unstable();
1854 let divided: Divided = text
1855 .divide(&offsets)
1856 .into_iter()
1857 .map(|line| {
1858 let spans = line
1859 .spans
1860 .iter()
1861 .map(|s| (s.start, s.end, format!("{:?}", s.style)))
1862 .collect();
1863 (line.plain, spans)
1864 })
1865 .collect();
1866 assert_eq!(
1867 divided,
1868 divide_by_scanning(&text, &offsets),
1869 "case {case}: {plain:?} {offsets:?}"
1870 );
1871 }
1872 }
1873
1874 /// An unbroken run of VS16 emoji must fold at the width like anything else.
1875 /// Measured per code point it did not: the heart reads one cell and the
1876 /// variation selector zero, so twenty hearts "fit" in thirty cells and came
1877 /// back as a single forty-cell row — wide enough to punch through the panel
1878 /// or table border drawn around it.
1879 ///
1880 /// Real rich 15.0.0, `[cell_len(l.plain) for l in Text("❤️"*20).wrap(c, 30)]`
1881 /// is `[30, 10]`.
1882 #[test]
1883 fn an_emoji_run_folds_at_the_width_instead_of_overflowing() {
1884 let hearts = "\u{2764}\u{fe0f}".repeat(20);
1885 let widths: Vec<usize> = wrapped_plain(&Text::new(&hearts), 30)
1886 .iter()
1887 .map(|line| cell_len(line))
1888 .collect();
1889 assert_eq!(widths, vec![30, 10]);
1890 }
1891
1892 /// Upstream wraps and justifies **one hard line at a time**, so the line
1893 /// full justification leaves ragged is the last of each paragraph — not just
1894 /// the last of the whole text. Flattening first stretched every paragraph
1895 /// but the final one.
1896 ///
1897 /// Real rich 15.0.0, `Text(case, justify="full").wrap(console, width)`:
1898 ///
1899 /// ```text
1900 /// width 10 -> ['word', ' indented', 'line here', 'last']
1901 /// width 30 -> ['word', ' indented line here', 'last'] (no_wrap)
1902 /// ```
1903 ///
1904 /// `line here` is the giveaway: it ends its paragraph, so upstream leaves
1905 /// the single gap alone where we widened it to `line here`.
1906 #[test]
1907 fn full_justify_leaves_each_paragraphs_last_line_ragged() {
1908 let text = Text::new("word\n indented line here\nlast").justify(Justify::Full);
1909 assert_eq!(
1910 wrapped_plain(&text, 10),
1911 vec!["word", " indented", "line here", "last"]
1912 );
1913 let no_wrap = Text::new("word\n indented line here\nlast")
1914 .justify(Justify::Full)
1915 .no_wrap(true);
1916 assert_eq!(
1917 wrapped_plain(&no_wrap, 30),
1918 vec!["word", " indented line here", "last"]
1919 );
1920 }
1921
1922 /// `overflow="ignore"` is a hard stop in upstream's `Text.wrap`: the line is
1923 /// appended verbatim and the loop `continue`s, so it is neither justified nor
1924 /// truncated. Padding it out to the width added trailing spaces to content
1925 /// upstream returns byte-for-byte.
1926 ///
1927 /// Real rich 15.0.0, `Text(case, justify=…, overflow="ignore").wrap(c, w)`:
1928 ///
1929 /// ```text
1930 /// left 'hello' @12 -> ['hello']
1931 /// center 'hello' @12 -> ['hello']
1932 /// right 'trailing ' @3 -> ['trailing ']
1933 /// ```
1934 #[test]
1935 fn overflow_ignore_is_neither_justified_nor_truncated() {
1936 for justify in [Justify::Left, Justify::Center, Justify::Right] {
1937 let text = Text::new("hello")
1938 .justify(justify)
1939 .overflow(Overflow::Ignore);
1940 assert_eq!(wrapped_plain(&text, 12), vec!["hello"], "{justify:?}");
1941 }
1942 let text = Text::new("trailing ")
1943 .justify(Justify::Right)
1944 .overflow(Overflow::Ignore);
1945 assert_eq!(wrapped_plain(&text, 3), vec!["trailing "]);
1946 }
1947
1948 /// Upstream's `Lines.justify` truncates *inside* its center and right
1949 /// branches, before it pads. The order matters because cell width is not
1950 /// additive across a cut: a leading zero-width joiner eats the character
1951 /// after it, so chopping the line's tail hands a cell back and the line then
1952 /// wants padding it did not want before. Padding first measures the un-cut
1953 /// line, finds no slack, and leaves the line a column short.
1954 ///
1955 /// Real rich 15.0.0:
1956 ///
1957 /// ```text
1958 /// right '┬┴⠁├╰⠃⡁⠃╯⠆┴' @9, ellipsis -> [' ┬┴⠁├╰⠃⡁⠃…']
1959 /// right '⠃╯⠆┴' @2, crop, no_wrap -> [' ⠃╯']
1960 /// center 's 8-o🧠e' @4, ellipsis -> [' s ', '8-o… ']
1961 /// ```
1962 #[test]
1963 fn center_and_right_truncate_before_they_pad() {
1964 let text = Text::new("\u{200d}┬┴⠁├╰⠃⡁⠃╯⠆┴")
1965 .justify(Justify::Right)
1966 .overflow(Overflow::Ellipsis);
1967 assert_eq!(wrapped_plain(&text, 9), vec![" \u{200d}┬┴⠁├╰⠃⡁⠃…"]);
1968
1969 let cropped = Text::new("\u{200d}⠃╯⠆┴")
1970 .justify(Justify::Right)
1971 .overflow(Overflow::Crop)
1972 .no_wrap(true);
1973 assert_eq!(wrapped_plain(&cropped, 2), vec![" \u{200d}⠃╯"]);
1974
1975 let centered = Text::new("s \u{200d}8-o\u{1f9e0}e")
1976 .justify(Justify::Center)
1977 .overflow(Overflow::Ellipsis);
1978 assert_eq!(wrapped_plain(¢ered, 4), vec![" s ", "\u{200d}8-o… "]);
1979 }
1980
1981 /// A line is measured and cut as one string, the way upstream's
1982 /// `Text.truncate` works on `self.plain` — not segment by segment. Cell
1983 /// width is not additive across a segment boundary: full justification
1984 /// splits the line into one segment per word, which strands the zero-width
1985 /// joiner at the end of `π` away from the space it swallows, so the
1986 /// per-segment sum reads eight cells for a seven-cell line and an ellipsis
1987 /// eats a character upstream keeps.
1988 ///
1989 /// Real rich 15.0.0,
1990 /// `Text("⚠1️;& π ψ\u{a0}τ\u{a0}γ ⡀", justify="full", overflow="ellipsis").wrap(c, 7)`:
1991 ///
1992 /// ```text
1993 /// ['⚠1️;& ', 'π ψ τ γ', '⡀'] with cell widths [7, 7, 1]
1994 /// ```
1995 #[test]
1996 fn a_line_is_measured_whole_not_segment_by_segment() {
1997 let text = Text::new("\u{26a0}1\u{fe0f};&\u{3000}\u{3c0}\u{200d} \u{3c8}\u{a0}\u{3c4}\u{a0}\u{3b3} \u{2840}")
1998 .justify(Justify::Full)
1999 .overflow(Overflow::Ellipsis);
2000 assert_eq!(
2001 wrapped_plain(&text, 7),
2002 vec![
2003 "\u{26a0}1\u{fe0f};&\u{3000}",
2004 "\u{3c0}\u{200d} \u{3c8}\u{a0}\u{3c4}\u{a0}\u{3b3}",
2005 "\u{2840}"
2006 ]
2007 );
2008 }
2009
2010 /// Full justification widens the gaps between words so every line but the
2011 /// last fills the width exactly.
2012 ///
2013 /// Captured verbatim from real rich 15.0.0 —
2014 /// `Lines.justify(console, 20, justify="full")` on
2015 /// `"aaa bbb ccc ddddddddddddddddddd ee ff"` yields:
2016 ///
2017 /// ```text
2018 /// 'aaa bbb ccc' <- stretched to exactly 20
2019 /// 'ddddddddddddddddddd' <- one word: nothing to widen, and the
2020 /// trailing space wrapping left is dropped
2021 /// 'ee ff' <- final line untouched: NOT padded to width
2022 /// ```
2023 ///
2024 /// Two details worth pinning: the slack is handed out from the rightmost
2025 /// gap backwards (so the gaps are 5 then 6, not 6 then 5), and the last
2026 /// line is the one case where a justified line is left short of the width.
2027 #[test]
2028 fn full_justify_matches_upstream() {
2029 let text = Text::new("aaa bbb ccc ddddddddddddddddddd ee ff").justify(Justify::Full);
2030 let plain: Vec<String> = text
2031 .render_lines(&Theme::default_theme(), &Style::new(), Some(20))
2032 .iter()
2033 .map(|line| line.iter().map(|s| s.text.as_str()).collect())
2034 .collect();
2035 assert_eq!(
2036 plain,
2037 vec!["aaa bbb ccc", "ddddddddddddddddddd", "ee ff"]
2038 );
2039 assert_eq!(plain[0].chars().count(), 20);
2040 }
2041
2042 fn wrapped_plain(text: &Text, width: usize) -> Vec<String> {
2043 text.render_lines(&Theme::default_theme(), &Style::new(), Some(width))
2044 .iter()
2045 .map(|line| line.iter().map(|s| s.text.as_str()).collect())
2046 .collect()
2047 }
2048
2049 /// Wrapping hands each line the space that ended it, and upstream's
2050 /// `Lines.justify` throws that space away (`line.rstrip()`) before centring
2051 /// or right-aligning — but *not* before left-aligning or full-justifying.
2052 ///
2053 /// Captured verbatim from real rich 15.0.0,
2054 /// `Text(case, justify=…).wrap(console, 20)`:
2055 ///
2056 /// ```text
2057 /// center 'abcd efgh ijklmnop qrst' -> ' abcd efgh ijklmnop '
2058 /// right 'abcd efgh ijklmnop qrst' -> ' abcd efgh ijklmnop'
2059 /// right 'aaaa bbbb cccc dddd eeee' -> ' aaaa bbbb cccc dddd'
2060 /// left 'abcd efgh ijklmnop qrst' -> 'abcd efgh ijklmnop '
2061 /// ```
2062 ///
2063 /// Counting the wrap space as content puts the centred line one column too
2064 /// far left and the right-aligned line a whole column short of the edge.
2065 #[test]
2066 fn center_and_right_rstrip_the_wrap_space() {
2067 let wrapped = "abcd efgh ijklmnop qrst";
2068 assert_eq!(
2069 wrapped_plain(&Text::new(wrapped).justify(Justify::Center), 20)[0],
2070 " abcd efgh ijklmnop "
2071 );
2072 assert_eq!(
2073 wrapped_plain(&Text::new(wrapped).justify(Justify::Right), 20)[0],
2074 " abcd efgh ijklmnop"
2075 );
2076 assert_eq!(
2077 wrapped_plain(
2078 &Text::new("aaaa bbbb cccc dddd eeee").justify(Justify::Right),
2079 20
2080 )[0],
2081 " aaaa bbbb cccc dddd"
2082 );
2083 // Left is the control: upstream pads it without stripping, so the
2084 // trailing space stays part of the line and nothing shifts.
2085 assert_eq!(
2086 wrapped_plain(&Text::new(wrapped).justify(Justify::Left), 20)[0],
2087 "abcd efgh ijklmnop "
2088 );
2089 }
2090
2091 /// `divide_line` measures a word *with* its trailing space, so a line whose
2092 /// last word ends flush with the width comes back one character too long.
2093 /// Upstream's `Text.wrap` calls `rstrip_end(width)` on every divided line to
2094 /// hand that back before overflow is applied.
2095 ///
2096 /// Real rich 15.0.0, `Text(case, overflow="ellipsis").wrap(console, 10)`:
2097 ///
2098 /// ```text
2099 /// 'abcdefghij more' -> ['abcdefghij', 'more'] <- no ellipsis
2100 /// 'abcdefghijkl more' -> ['abcdefghi…', 'more'] <- genuinely too long
2101 /// ```
2102 ///
2103 /// Skipping the rstrip makes the first case measure 11 cells, so the
2104 /// ellipsis fires and eats the `j` that upstream keeps.
2105 #[test]
2106 fn rstrip_end_stops_the_wrap_space_from_triggering_an_ellipsis() {
2107 assert_eq!(
2108 wrapped_plain(
2109 &Text::new("abcdefghij more").overflow(Overflow::Ellipsis),
2110 10
2111 ),
2112 vec!["abcdefghij", "more"]
2113 );
2114 assert_eq!(
2115 wrapped_plain(
2116 &Text::new("abcdefghijkl more").overflow(Overflow::Ellipsis),
2117 10
2118 ),
2119 vec!["abcdefghi…", "more"]
2120 );
2121 }
2122
2123 /// `Text.apply_meta` / `Text.on`, as rich 15.0.0 records them.
2124 #[test]
2125 fn apply_meta_and_on_add_meta_spans() {
2126 use crate::style::{Meta, MetaValue};
2127 let mut text = Text::new("hello");
2128 text.apply_meta([("a", MetaValue::Int(1))].into_iter().collect(), 1, Some(3));
2129 text.on(None, &[("click", MetaValue::Str("x".into()))]);
2130 text.apply_meta(Meta::new(), 0, None);
2131 let spans = text.spans();
2132 assert_eq!(spans.len(), 2);
2133 assert_eq!((spans[0].start, spans[0].end), (1, 3));
2134 let meta = |span: &Span| match &span.style {
2135 StyleType::Style(style) => style.meta(),
2136 StyleType::Name(_) => Meta::new(),
2137 };
2138 assert_eq!(
2139 meta(&spans[0]),
2140 [("a", MetaValue::Int(1))].into_iter().collect()
2141 );
2142 assert_eq!(
2143 meta(&spans[1]),
2144 [("@click", MetaValue::Str("x".into()))]
2145 .into_iter()
2146 .collect()
2147 );
2148 }
2149
2150 #[test]
2151 fn append_creates_spans() {
2152 let mut text = Text::new("");
2153 text.append("hello", Some(Style::parse("bold").unwrap().into()));
2154 text.append(" world", None);
2155 assert_eq!(text.plain(), "hello world");
2156 assert_eq!(text.spans().len(), 1);
2157 }
2158
2159 #[test]
2160 fn render_flattens_overlapping_spans() {
2161 let mut text = Text::new("abcdef");
2162 text.stylize(Style::parse("bold").unwrap(), 0, 4);
2163 text.stylize(Style::parse("red").unwrap(), 2, 6);
2164 let segments = text.render(&Theme::default_theme(), &Style::new());
2165 // Boundaries at 0,2,4,6 -> "ab"(bold) "cd"(bold+red) "ef"(red)
2166 let rendered: Vec<_> = segments.iter().map(|s| s.text.clone()).collect();
2167 assert_eq!(rendered, vec!["ab", "cd", "ef"]);
2168 }
2169
2170 /// `Text::truncate` on its own, against real rich 15.0.0. `fold` and `crop`
2171 /// deliberately agree: folding is a wrapping behaviour, and truncation has
2172 /// no line to fold onto.
2173 #[test]
2174 fn truncate_matches_upstream() {
2175 for (overflow, expected) in [
2176 (Overflow::Fold, "hello"),
2177 (Overflow::Crop, "hello"),
2178 (Overflow::Ellipsis, "hell…"),
2179 (Overflow::Ignore, "hello world"),
2180 ] {
2181 let mut text = Text::new("hello world");
2182 text.truncate(5, Some(overflow), false);
2183 assert_eq!(text.plain(), expected, "overflow {overflow:?}");
2184 }
2185 }
2186
2187 /// `pad` fills out to the width, but only when the text is short — a text
2188 /// that is already too long is cut, never padded.
2189 #[test]
2190 fn truncate_pads_only_when_short() {
2191 let mut short = Text::new("hi");
2192 short.truncate(6, Some(Overflow::Crop), true);
2193 assert_eq!(short.plain(), "hi ");
2194
2195 let mut exact = Text::new("hi");
2196 exact.truncate(2, Some(Overflow::Crop), true);
2197 assert_eq!(exact.plain(), "hi");
2198 }
2199
2200 /// Truncating must not leave a span pointing past the end of the string.
2201 #[test]
2202 fn truncate_trims_dangling_spans() {
2203 let mut text = Text::new("hello world");
2204 text.stylize(Style::parse("bold").unwrap(), 6, 11);
2205 text.stylize(Style::parse("red").unwrap(), 0, 5);
2206 text.truncate(3, Some(Overflow::Crop), false);
2207 assert_eq!(text.plain(), "hel");
2208 // The "world" span starts past the new end and is dropped entirely; the
2209 // "hello" span survives, clamped.
2210 assert_eq!(text.spans().len(), 1);
2211 assert!(text.spans().iter().all(|s| s.end <= text.plain().len()));
2212 }
2213
2214 use crate::protocol::Renderable;
2215
2216 /// The overflow method may come from the text or from the console options,
2217 /// and the text's own setting wins — mirroring upstream's
2218 /// `self.overflow or options.overflow or DEFAULT_OVERFLOW`.
2219 #[test]
2220 fn text_overflow_beats_console_options() {
2221 let console = crate::Console::builder().width(8).build();
2222 let mut options = console.options();
2223 options.overflow = Some(Overflow::Ellipsis);
2224 options.no_wrap = Some(true);
2225
2226 // Nothing set on the text: the options decide.
2227 let from_options = Text::new("the quick brown fox");
2228 assert_eq!(
2229 plain_of(&from_options.rich_render(&console, &options)),
2230 "the qui…"
2231 );
2232
2233 // Set on the text: the text decides, and the options are ignored.
2234 let from_text = Text::new("the quick brown fox").overflow(Overflow::Crop);
2235 assert_eq!(
2236 plain_of(&from_text.rich_render(&console, &options)),
2237 "the quic"
2238 );
2239 }
2240
2241 /// With no overflow anywhere, upstream's default applies: fold.
2242 #[test]
2243 fn overflow_defaults_to_fold() {
2244 let console = crate::Console::builder().width(8).build();
2245 let text = Text::new("supercalifragilistic");
2246 let rendered = plain_of(&text.rich_render(&console, &console.options()));
2247 assert_eq!(rendered, "supercal\nifragili\nstic");
2248 }
2249
2250 /// Concatenate the visible text of a segment stream, for assertions that
2251 /// care about layout rather than styling.
2252 fn plain_of(segments: &[Segment]) -> String {
2253 segments
2254 .iter()
2255 .filter(|s| !s.control)
2256 .map(|s| s.text.as_str())
2257 .collect()
2258 }
2259
2260 /// `Text::new` stripped control codes but `append` did not, so every path
2261 /// that builds text incrementally — Markdown, Syntax, plain files — leaked
2262 /// them to the terminal. A backspace run is a spoofing tool: the reader sees
2263 /// the overwritten text, not what the file says.
2264 #[test]
2265 fn append_strips_control_codes_like_new() {
2266 let mut text = Text::new("");
2267 text.append("FAILED\u{8}\u{8}\u{8}\u{8}\u{8}\u{8}PASSED", None);
2268 assert_eq!(text.plain(), "FAILEDPASSED");
2269
2270 for code in ['\u{7}', '\u{8}', '\u{b}', '\u{c}', '\u{d}'] {
2271 let mut text = Text::new("");
2272 text.append(&format!("a{code}b"), None);
2273 assert_eq!(text.plain(), "ab", "control code {code:?} survived append");
2274 }
2275 }
2276
2277 /// Upstream's `strip_control_codes` keeps NUL and ESC; only BEL, backspace,
2278 /// vertical tab, form feed and carriage return go.
2279 #[test]
2280 fn append_keeps_the_codes_upstream_keeps() {
2281 let mut text = Text::new("");
2282 text.append("a\u{0}b\u{1b}c", None);
2283 assert_eq!(text.plain(), "a\u{0}b\u{1b}c");
2284 }
2285}