1 // Format string literals.
4 use unicode_categories::UnicodeCategories;
5 use unicode_segmentation::UnicodeSegmentation;
7 use crate::config::Config;
8 use crate::shape::Shape;
9 use crate::utils::{unicode_str_width, wrap_str};
11 const MIN_STRING: usize = 10;
13 /// Describes the layout of a piece of text.
14 pub(crate) struct StringFormat<'a> {
15 /// The opening sequence of characters for the piece of text
16 pub(crate) opener: &'a str,
17 /// The closing sequence of characters for the piece of text
18 pub(crate) closer: &'a str,
19 /// The opening sequence of characters for a line
20 pub(crate) line_start: &'a str,
21 /// The closing sequence of characters for a line
22 pub(crate) line_end: &'a str,
23 /// The allocated box to fit the text into
24 pub(crate) shape: Shape,
25 /// Trim trailing whitespaces
26 pub(crate) trim_end: bool,
27 pub(crate) config: &'a Config,
30 impl<'a> StringFormat<'a> {
31 pub(crate) fn new(shape: Shape, config: &'a Config) -> StringFormat<'a> {
43 /// Returns the maximum number of graphemes that is possible on a line while taking the
44 /// indentation into account.
46 /// If we cannot put at least a single character per line, the rewrite won't succeed.
47 fn max_width_with_indent(&self) -> Option<usize> {
51 .checked_sub(self.opener.len() + self.line_end.len() + 1)?
56 /// Like max_width_with_indent but the indentation is not subtracted.
57 /// This allows to fit more graphemes from the string on a line when
58 /// SnippetState::EndWithLineFeed.
59 fn max_width_without_indent(&self) -> Option<usize> {
60 self.config.max_width().checked_sub(self.line_end.len())
64 pub(crate) fn rewrite_string<'a>(
66 fmt: &StringFormat<'a>,
67 newline_max_chars: usize,
69 let max_width_with_indent = fmt.max_width_with_indent()?;
70 let max_width_without_indent = fmt.max_width_without_indent()?;
71 let indent_with_newline = fmt.shape.indent.to_string_with_newline(fmt.config);
72 let indent_without_newline = fmt.shape.indent.to_string(fmt.config);
75 // With this regex applied, all remaining whitespaces are significant
76 let strip_line_breaks_re = Regex::new(r"([^\\](\\\\)*)\\[\n\r][[:space:]]*").unwrap();
77 let stripped_str = strip_line_breaks_re.replace_all(orig, "$1");
79 let graphemes = UnicodeSegmentation::graphemes(&*stripped_str, false).collect::<Vec<&str>>();
81 // `cur_start` is the position in `orig` of the start of the current line.
82 let mut cur_start = 0;
83 let mut result = String::with_capacity(
86 .checked_next_power_of_two()
87 .unwrap_or(usize::max_value()),
89 result.push_str(fmt.opener);
91 // Snip a line at a time from `stripped_str` until it is used up. Push the snippet
93 let mut cur_max_width = max_width_with_indent;
94 let is_bareline_ok = fmt.line_start.is_empty() || is_whitespace(fmt.line_start);
96 // All the input starting at cur_start fits on the current line
97 if graphemes_width(&graphemes[cur_start..]) <= cur_max_width {
98 for (i, grapheme) in graphemes[cur_start..].iter().enumerate() {
99 if is_new_line(grapheme) {
100 // take care of blank lines
101 result = trim_end_but_line_feed(fmt.trim_end, result);
103 if !is_bareline_ok && cur_start + i + 1 < graphemes.len() {
104 result.push_str(&indent_without_newline);
105 result.push_str(fmt.line_start);
108 result.push_str(grapheme);
111 result = trim_end_but_line_feed(fmt.trim_end, result);
115 // The input starting at cur_start needs to be broken
120 &graphemes[cur_start..],
122 SnippetState::LineEnd(line, len) => {
123 result.push_str(&line);
124 result.push_str(fmt.line_end);
125 result.push_str(&indent_with_newline);
126 result.push_str(fmt.line_start);
127 cur_max_width = newline_max_chars;
130 SnippetState::EndWithLineFeed(line, len) => {
131 if line == "\n" && fmt.trim_end {
132 result = result.trim_end().to_string();
134 result.push_str(&line);
136 // the next line can benefit from the full width
137 cur_max_width = max_width_without_indent;
139 result.push_str(&indent_without_newline);
140 result.push_str(fmt.line_start);
141 cur_max_width = max_width_with_indent;
145 SnippetState::EndOfInput(line) => {
146 result.push_str(&line);
152 result.push_str(fmt.closer);
153 wrap_str(result, fmt.config.max_width(), fmt.shape)
156 /// Returns the index to the end of the URL if the split at index of the given string includes a
157 /// URL or alike. Otherwise, returns `None`.
158 fn detect_url(s: &[&str], index: usize) -> Option<usize> {
159 let start = match s[..=index].iter().rposition(|g| is_whitespace(g)) {
160 Some(pos) => pos + 1,
163 // 8 = minimum length for a string to contain a URL
164 if s.len() < start + 8 {
167 let split = s[start..].concat();
168 if split.contains("https://")
169 || split.contains("http://")
170 || split.contains("ftp://")
171 || split.contains("file://")
173 match s[index..].iter().position(|g| is_whitespace(g)) {
174 Some(pos) => Some(index + pos - 1),
175 None => Some(s.len() - 1),
182 /// Trims whitespaces to the right except for the line feed character.
183 fn trim_end_but_line_feed(trim_end: bool, result: String) -> String {
184 let whitespace_except_line_feed = |c: char| c.is_whitespace() && c != '\n';
185 if trim_end && result.ends_with(whitespace_except_line_feed) {
187 .trim_end_matches(whitespace_except_line_feed)
194 /// Result of breaking a string so it fits in a line and the state it ended in.
195 /// The state informs about what to do with the snippet and how to continue the breaking process.
196 #[derive(Debug, PartialEq)]
198 /// The input could not be broken and so rewriting the string is finished.
200 /// The input could be broken and the returned snippet should be ended with a
201 /// `[StringFormat::line_end]`. The next snippet needs to be indented.
203 /// The returned string is the line to print out and the number is the length that got read in
204 /// the text being rewritten. That length may be greater than the returned string if trailing
205 /// whitespaces got trimmed.
206 LineEnd(String, usize),
207 /// The input could be broken but a newline is present that cannot be trimmed. The next snippet
208 /// to be rewritten *could* use more width than what is specified by the given shape. For
209 /// example with a multiline string, the next snippet does not need to be indented, allowing
210 /// more characters to be fit within a line.
212 /// The returned string is the line to print out and the number is the length that got read in
213 /// the text being rewritten.
214 EndWithLineFeed(String, usize),
217 fn not_whitespace_except_line_feed(g: &str) -> bool {
218 is_new_line(g) || !is_whitespace(g)
221 /// Break the input string at a boundary character around the offset `max_width`. A boundary
222 /// character is either a punctuation or a whitespace.
223 /// FIXME(issue#3281): We must follow UAX#14 algorithm instead of this.
224 fn break_string(max_width: usize, trim_end: bool, line_end: &str, input: &[&str]) -> SnippetState {
225 let break_at = |index /* grapheme at index is included */| {
226 // Take in any whitespaces to the left/right of `input[index]` while
227 // preserving line feeds
228 let index_minus_ws = input[0..=index]
230 .rposition(|grapheme| not_whitespace_except_line_feed(grapheme))
232 // Take into account newlines occurring in input[0..=index], i.e., the possible next new
233 // line. If there is one, then text after it could be rewritten in a way that the available
234 // space is fully used.
235 for (i, grapheme) in input[0..=index].iter().enumerate() {
236 if is_new_line(grapheme) {
237 if i <= index_minus_ws {
238 let mut line = &input[0..i].concat()[..];
240 line = line.trim_end();
242 return SnippetState::EndWithLineFeed(format!("{}\n", line), i + 1);
248 let mut index_plus_ws = index;
249 for (i, grapheme) in input[index + 1..].iter().enumerate() {
250 if !trim_end && is_new_line(grapheme) {
251 return SnippetState::EndWithLineFeed(
252 input[0..=index + 1 + i].concat(),
255 } else if not_whitespace_except_line_feed(grapheme) {
256 index_plus_ws = index + i;
262 SnippetState::LineEnd(input[0..=index_minus_ws].concat(), index_plus_ws + 1)
264 SnippetState::LineEnd(input[0..=index_plus_ws].concat(), index_plus_ws + 1)
268 // find a first index where the unicode width of input[0..x] become > max_width
269 let max_width_index_in_input = {
270 let mut cur_width = 0;
271 let mut cur_index = 0;
272 for (i, grapheme) in input.iter().enumerate() {
273 cur_width += unicode_str_width(grapheme);
275 if cur_width > max_width {
281 if max_width_index_in_input == 0 {
282 return SnippetState::EndOfInput(input.concat());
285 // Find the position in input for breaking the string
286 if line_end.is_empty()
288 && !is_whitespace(input[max_width_index_in_input - 1])
289 && is_whitespace(input[max_width_index_in_input])
291 // At a breaking point already
292 // The line won't invalidate the rewriting because:
293 // - no extra space needed for the line_end character
294 // - extra whitespaces to the right can be trimmed
295 return break_at(max_width_index_in_input - 1);
297 if let Some(url_index_end) = detect_url(input, max_width_index_in_input) {
298 let index_plus_ws = url_index_end
299 + input[url_index_end..]
302 .position(|grapheme| not_whitespace_except_line_feed(grapheme))
305 SnippetState::LineEnd(input[..=url_index_end].concat(), index_plus_ws + 1)
307 SnippetState::LineEnd(input[..=index_plus_ws].concat(), index_plus_ws + 1)
311 match input[0..max_width_index_in_input]
313 .rposition(|grapheme| is_whitespace(grapheme))
315 // Found a whitespace and what is on its left side is big enough.
316 Some(index) if index >= MIN_STRING => break_at(index),
317 // No whitespace found, try looking for a punctuation instead
318 _ => match (0..max_width_index_in_input)
320 .skip_while(|pos| !is_valid_linebreak(input, *pos))
323 // Found a punctuation and what is on its left side is big enough.
324 Some(index) if index >= MIN_STRING => break_at(index),
325 // Either no boundary character was found to the left of `input[max_chars]`, or the line
326 // got too small. We try searching for a boundary character to the right.
327 _ => match (max_width_index_in_input..input.len())
328 .skip_while(|pos| !is_valid_linebreak(input, *pos))
331 // A boundary was found after the line limit
332 Some(index) => break_at(index),
333 // No boundary to the right, the input cannot be broken
334 None => SnippetState::EndOfInput(input.concat()),
340 fn is_valid_linebreak(input: &[&str], pos: usize) -> bool {
341 let is_whitespace = is_whitespace(input[pos]);
345 let is_punctuation = is_punctuation(input[pos]);
346 if is_punctuation && !is_part_of_type(input, pos) {
352 fn is_part_of_type(input: &[&str], pos: usize) -> bool {
353 input.get(pos..=pos + 1) == Some(&[":", ":"])
354 || input.get(pos.saturating_sub(1)..=pos) == Some(&[":", ":"])
357 fn is_new_line(grapheme: &str) -> bool {
358 let bytes = grapheme.as_bytes();
359 bytes.starts_with(b"\n") || bytes.starts_with(b"\r\n")
362 fn is_whitespace(grapheme: &str) -> bool {
363 grapheme.chars().all(char::is_whitespace)
366 fn is_punctuation(grapheme: &str) -> bool {
369 .all(UnicodeCategories::is_punctuation_other)
372 fn graphemes_width(graphemes: &[&str]) -> usize {
373 graphemes.iter().map(|s| unicode_str_width(s)).sum()
378 use super::{break_string, detect_url, rewrite_string, SnippetState, StringFormat};
379 use crate::config::Config;
380 use crate::shape::{Indent, Shape};
381 use unicode_segmentation::UnicodeSegmentation;
385 let config = Default::default();
386 let fmt = StringFormat::new(Shape::legacy(2, Indent::empty()), &config);
387 rewrite_string("eq_", &fmt, 2);
391 fn line_break_at_valid_points_test() {
392 let string = "[TheName](Dont::break::my::type::That::would::be::very::nice) break here";
393 let graphemes = UnicodeSegmentation::graphemes(&*string, false).collect::<Vec<&str>>();
395 break_string(20, false, "", &graphemes[..]),
396 SnippetState::LineEnd(
397 "[TheName](Dont::break::my::type::That::would::be::very::nice) ".to_string(),
404 fn should_break_on_whitespace() {
405 let string = "Placerat felis. Mauris porta ante sagittis purus.";
406 let graphemes = UnicodeSegmentation::graphemes(&*string, false).collect::<Vec<&str>>();
408 break_string(20, false, "", &graphemes[..]),
409 SnippetState::LineEnd("Placerat felis. ".to_string(), 16)
412 break_string(20, true, "", &graphemes[..]),
413 SnippetState::LineEnd("Placerat felis.".to_string(), 16)
418 fn should_break_on_punctuation() {
419 let string = "Placerat_felis._Mauris_porta_ante_sagittis_purus.";
420 let graphemes = UnicodeSegmentation::graphemes(&*string, false).collect::<Vec<&str>>();
422 break_string(20, false, "", &graphemes[..]),
423 SnippetState::LineEnd("Placerat_felis.".to_string(), 15)
428 fn should_break_forward() {
429 let string = "Venenatis_tellus_vel_tellus. Aliquam aliquam dolor at justo.";
430 let graphemes = UnicodeSegmentation::graphemes(&*string, false).collect::<Vec<&str>>();
432 break_string(20, false, "", &graphemes[..]),
433 SnippetState::LineEnd("Venenatis_tellus_vel_tellus. ".to_string(), 29)
436 break_string(20, true, "", &graphemes[..]),
437 SnippetState::LineEnd("Venenatis_tellus_vel_tellus.".to_string(), 29)
442 fn nothing_to_break() {
443 let string = "Venenatis_tellus_vel_tellus";
444 let graphemes = UnicodeSegmentation::graphemes(&*string, false).collect::<Vec<&str>>();
446 break_string(20, false, "", &graphemes[..]),
447 SnippetState::EndOfInput("Venenatis_tellus_vel_tellus".to_string())
452 fn significant_whitespaces() {
453 let string = "Neque in sem. \n Pellentesque tellus augue.";
454 let graphemes = UnicodeSegmentation::graphemes(&*string, false).collect::<Vec<&str>>();
456 break_string(15, false, "", &graphemes[..]),
457 SnippetState::EndWithLineFeed("Neque in sem. \n".to_string(), 20)
460 break_string(25, false, "", &graphemes[..]),
461 SnippetState::EndWithLineFeed("Neque in sem. \n".to_string(), 20)
465 break_string(15, true, "", &graphemes[..]),
466 SnippetState::LineEnd("Neque in sem.".to_string(), 19)
469 break_string(25, true, "", &graphemes[..]),
470 SnippetState::EndWithLineFeed("Neque in sem.\n".to_string(), 20)
475 fn big_whitespace() {
476 let string = "Neque in sem. Pellentesque tellus augue.";
477 let graphemes = UnicodeSegmentation::graphemes(&*string, false).collect::<Vec<&str>>();
479 break_string(20, false, "", &graphemes[..]),
480 SnippetState::LineEnd("Neque in sem. ".to_string(), 25)
483 break_string(20, true, "", &graphemes[..]),
484 SnippetState::LineEnd("Neque in sem.".to_string(), 25)
489 fn newline_in_candidate_line() {
490 let string = "Nulla\nconsequat erat at massa. Vivamus id mi.";
492 let graphemes = UnicodeSegmentation::graphemes(&*string, false).collect::<Vec<&str>>();
494 break_string(25, false, "", &graphemes[..]),
495 SnippetState::EndWithLineFeed("Nulla\n".to_string(), 6)
498 break_string(25, true, "", &graphemes[..]),
499 SnippetState::EndWithLineFeed("Nulla\n".to_string(), 6)
502 let mut config: Config = Default::default();
503 config.set().max_width(27);
504 let fmt = StringFormat::new(Shape::legacy(25, Indent::empty()), &config);
505 let rewritten_string = rewrite_string(string, &fmt, 27);
508 Some("\"Nulla\nconsequat erat at massa. \\\n Vivamus id mi.\"".to_string())
513 fn last_line_fit_with_trailing_whitespaces() {
514 let string = "Vivamus id mi. ";
515 let config: Config = Default::default();
516 let mut fmt = StringFormat::new(Shape::legacy(25, Indent::empty()), &config);
519 let rewritten_string = rewrite_string(string, &fmt, 25);
520 assert_eq!(rewritten_string, Some("\"Vivamus id mi.\"".to_string()));
522 fmt.trim_end = false; // default value of trim_end
523 let rewritten_string = rewrite_string(string, &fmt, 25);
524 assert_eq!(rewritten_string, Some("\"Vivamus id mi. \"".to_string()));
528 fn last_line_fit_with_newline() {
529 let string = "Vivamus id mi.\nVivamus id mi.";
530 let config: Config = Default::default();
531 let fmt = StringFormat {
536 shape: Shape::legacy(100, Indent::from_width(&config, 4)),
541 let rewritten_string = rewrite_string(string, &fmt, 100);
544 Some("Vivamus id mi.\n // Vivamus id mi.".to_string())
549 fn overflow_in_non_string_content() {
550 let comment = "Aenean metus.\nVestibulum ac lacus. Vivamus porttitor";
551 let config: Config = Default::default();
552 let fmt = StringFormat {
557 shape: Shape::legacy(30, Indent::from_width(&config, 8)),
563 rewrite_string(comment, &fmt, 30),
565 "Aenean metus.\n // Vestibulum ac lacus. Vivamus\n // porttitor"
572 fn overflow_in_non_string_content_with_line_end() {
573 let comment = "Aenean metus.\nVestibulum ac lacus. Vivamus porttitor";
574 let config: Config = Default::default();
575 let fmt = StringFormat {
580 shape: Shape::legacy(30, Indent::from_width(&config, 8)),
586 rewrite_string(comment, &fmt, 30),
588 "Aenean metus.\n // Vestibulum ac lacus. Vivamus@\n // porttitor"
595 fn blank_line_with_non_empty_line_start() {
596 let config: Config = Default::default();
597 let mut fmt = StringFormat {
602 shape: Shape::legacy(30, Indent::from_width(&config, 4)),
607 let comment = "Aenean metus. Vestibulum\n\nac lacus. Vivamus porttitor";
609 rewrite_string(comment, &fmt, 30),
611 "Aenean metus. Vestibulum\n //\n // ac lacus. Vivamus porttitor".to_string()
615 fmt.shape = Shape::legacy(15, Indent::from_width(&config, 4));
616 let comment = "Aenean\n\nmetus. Vestibulum ac lacus. Vivamus porttitor";
618 rewrite_string(comment, &fmt, 15),
631 fn retain_blank_lines() {
632 let config: Config = Default::default();
633 let fmt = StringFormat {
638 shape: Shape::legacy(20, Indent::from_width(&config, 4)),
643 let comment = "Aenean\n\nmetus. Vestibulum ac lacus.\n\n";
645 rewrite_string(comment, &fmt, 20),
647 "Aenean\n //\n // metus. Vestibulum ac\n // lacus.\n //\n".to_string()
651 let comment = "Aenean\n\nmetus. Vestibulum ac lacus.\n";
653 rewrite_string(comment, &fmt, 20),
654 Some("Aenean\n //\n // metus. Vestibulum ac\n // lacus.\n".to_string())
657 let comment = "Aenean\n \nmetus. Vestibulum ac lacus.";
659 rewrite_string(comment, &fmt, 20),
660 Some("Aenean\n //\n // metus. Vestibulum ac\n // lacus.".to_string())
665 fn boundary_on_edge() {
666 let config: Config = Default::default();
667 let mut fmt = StringFormat {
672 shape: Shape::legacy(13, Indent::from_width(&config, 4)),
677 let comment = "Aenean metus. Vestibulum ac lacus.";
679 rewrite_string(comment, &fmt, 13),
680 Some("Aenean metus.\n // Vestibulum ac\n // lacus.".to_string())
683 fmt.trim_end = false;
684 let comment = "Vestibulum ac lacus.";
686 rewrite_string(comment, &fmt, 13),
687 Some("Vestibulum \n // ac lacus.".to_string())
692 let comment = "Vestibulum ac lacus.";
694 rewrite_string(comment, &fmt, 13),
695 Some("Vestibulum\\\n // ac lacus.".to_string())
701 let string = "aaa http://example.org something";
702 let graphemes = UnicodeSegmentation::graphemes(&*string, false).collect::<Vec<&str>>();
703 assert_eq!(detect_url(&graphemes, 8), Some(21));
705 let string = "https://example.org something";
706 let graphemes = UnicodeSegmentation::graphemes(&*string, false).collect::<Vec<&str>>();
707 assert_eq!(detect_url(&graphemes, 0), Some(18));
709 let string = "aaa ftp://example.org something";
710 let graphemes = UnicodeSegmentation::graphemes(&*string, false).collect::<Vec<&str>>();
711 assert_eq!(detect_url(&graphemes, 8), Some(20));
713 let string = "aaa file://example.org something";
714 let graphemes = UnicodeSegmentation::graphemes(&*string, false).collect::<Vec<&str>>();
715 assert_eq!(detect_url(&graphemes, 8), Some(21));
717 let string = "aaa http not an url";
718 let graphemes = UnicodeSegmentation::graphemes(&*string, false).collect::<Vec<&str>>();
719 assert_eq!(detect_url(&graphemes, 6), None);
721 let string = "aaa file://example.org";
722 let graphemes = UnicodeSegmentation::graphemes(&*string, false).collect::<Vec<&str>>();
723 assert_eq!(detect_url(&graphemes, 8), Some(21));