1 // Copyright 2015 The Rust Project Developers. See the COPYRIGHT
2 // file at the top-level directory of this distribution and at
3 // http://rust-lang.org/COPYRIGHT.
5 // Licensed under the Apache License, Version 2.0 <LICENSE-APACHE or
6 // http://www.apache.org/licenses/LICENSE-2.0> or the MIT license
7 // <LICENSE-MIT or http://opensource.org/licenses/MIT>, at your
8 // option. This file may not be copied, modified, or distributed
9 // except according to those terms.
11 // Format string literals.
14 use unicode_segmentation::UnicodeSegmentation;
20 const MIN_STRING: usize = 10;
22 /// Describes the layout of a piece of text.
23 pub struct StringFormat<'a> {
24 /// The opening sequence of characters for the piece of text
26 /// The closing sequence of characters for the piece of text
28 /// The opening sequence of characters for a line
29 pub line_start: &'a str,
30 /// The closing sequence of characters for a line
31 pub line_end: &'a str,
32 /// The allocated box to fit the text into
34 /// Trim trailing whitespaces
36 pub config: &'a Config,
39 impl<'a> StringFormat<'a> {
40 pub fn new(shape: Shape, config: &'a Config) -> StringFormat<'a> {
52 /// Returns the maximum number of graphemes that is possible on a line while taking the
53 /// indentation into account.
55 /// If we cannot put at least a single character per line, the rewrite won't succeed.
56 fn max_chars_with_indent(&self) -> Option<usize> {
60 .checked_sub(self.opener.len() + self.line_end.len() + 1)?
65 /// Like max_chars_with_indent but the indentation is not substracted.
66 /// This allows to fit more graphemes from the string on a line when
67 /// SnippetState::EndWithLineFeed.
68 fn max_chars_without_indent(&self) -> Option<usize> {
69 Some(self.config.max_width().checked_sub(self.line_end.len())?)
73 pub fn rewrite_string<'a>(orig: &str, fmt: &StringFormat<'a>) -> Option<String> {
74 let max_chars_with_indent = fmt.max_chars_with_indent()?;
75 let max_chars_without_indent = fmt.max_chars_without_indent()?;
76 let indent_with_newline = fmt.shape.indent.to_string_with_newline(fmt.config);
77 let indent_without_newline = fmt.shape.indent.to_string(fmt.config);
80 // With this regex applied, all remaining whitespaces are significant
81 let strip_line_breaks_re = Regex::new(r"([^\\](\\\\)*)\\[\n\r][[:space:]]*").unwrap();
82 let stripped_str = strip_line_breaks_re.replace_all(orig, "$1");
84 let graphemes = UnicodeSegmentation::graphemes(&*stripped_str, false).collect::<Vec<&str>>();
86 // `cur_start` is the position in `orig` of the start of the current line.
87 let mut cur_start = 0;
88 let mut result = String::with_capacity(
91 .checked_next_power_of_two()
92 .unwrap_or(usize::max_value()),
94 result.push_str(fmt.opener);
96 // Snip a line at a time from `stripped_str` until it is used up. Push the snippet
98 let mut cur_max_chars = max_chars_with_indent;
99 let is_bareline_ok = fmt.line_start.is_empty() || is_whitespace(fmt.line_start);
101 // All the input starting at cur_start fits on the current line
102 if graphemes.len() - cur_start <= cur_max_chars {
103 for (i, grapheme) in graphemes[cur_start..].iter().enumerate() {
104 if is_line_feed(grapheme) {
105 // take care of blank lines
106 result = trim_right_but_line_feed(fmt.trim_end, result);
107 result.push_str("\n");
108 if !is_bareline_ok && cur_start + i + 1 < graphemes.len() {
109 result.push_str(&indent_without_newline);
110 result.push_str(fmt.line_start);
113 result.push_str(grapheme);
116 result = trim_right_but_line_feed(fmt.trim_end, result);
120 // The input starting at cur_start needs to be broken
125 &graphemes[cur_start..],
127 SnippetState::LineEnd(line, len) => {
128 result.push_str(&line);
129 result.push_str(fmt.line_end);
130 result.push_str(&indent_with_newline);
131 result.push_str(fmt.line_start);
132 cur_max_chars = max_chars_with_indent;
135 SnippetState::EndWithLineFeed(line, len) => {
136 if line == "\n" && fmt.trim_end {
137 result = result.trim_right().to_string();
139 result.push_str(&line);
141 // the next line can benefit from the full width
142 cur_max_chars = max_chars_without_indent;
144 result.push_str(&indent_without_newline);
145 result.push_str(fmt.line_start);
146 cur_max_chars = max_chars_with_indent;
150 SnippetState::EndOfInput(line) => {
151 result.push_str(&line);
157 result.push_str(fmt.closer);
158 wrap_str(result, fmt.config.max_width(), fmt.shape)
161 /// Trims whitespaces to the right except for the line feed character.
162 fn trim_right_but_line_feed(trim_end: bool, result: String) -> String {
163 let whitespace_except_line_feed = |c: char| c.is_whitespace() && c != '\n';
164 if trim_end && result.ends_with(whitespace_except_line_feed) {
166 .trim_right_matches(whitespace_except_line_feed)
173 /// Result of breaking a string so it fits in a line and the state it ended in.
174 /// The state informs about what to do with the snippet and how to continue the breaking process.
175 #[derive(Debug, PartialEq)]
177 /// The input could not be broken and so rewriting the string is finished.
179 /// The input could be broken and the returned snippet should be ended with a
180 /// `[StringFormat::line_end]`. The next snippet needs to be indented.
182 /// The returned string is the line to print out and the number is the length that got read in
183 /// the text being rewritten. That length may be greater than the returned string if trailing
184 /// whitespaces got trimmed.
185 LineEnd(String, usize),
186 /// The input could be broken but a newline is present that cannot be trimmed. The next snippet
187 /// to be rewritten *could* use more width than what is specified by the given shape. For
188 /// example with a multiline string, the next snippet does not need to be indented, allowing
189 /// more characters to be fit within a line.
191 /// The returned string is the line to print out and the number is the length that got read in
192 /// the text being rewritten.
193 EndWithLineFeed(String, usize),
196 /// Break the input string at a boundary character around the offset `max_chars`. A boundary
197 /// character is either a punctuation or a whitespace.
198 fn break_string(max_chars: usize, trim_end: bool, line_end: &str, input: &[&str]) -> SnippetState {
199 let break_at = |index /* grapheme at index is included */| {
200 // Take in any whitespaces to the left/right of `input[index]` while
201 // preserving line feeds
202 let not_whitespace_except_line_feed = |g| is_line_feed(g) || !is_whitespace(g);
203 let index_minus_ws = input[0..=index]
205 .rposition(|grapheme| not_whitespace_except_line_feed(grapheme))
207 // Take into account newlines occuring in input[0..=index], i.e., the possible next new
208 // line. If there is one, then text after it could be rewritten in a way that the available
209 // space is fully used.
210 for (i, grapheme) in input[0..=index].iter().enumerate() {
211 if is_line_feed(grapheme) {
212 if i <= index_minus_ws {
213 let mut line = input[0..i].join("");
215 line = line.trim_right().to_string();
217 return SnippetState::EndWithLineFeed(format!("{}\n", line), i + 1);
223 let mut index_plus_ws = index;
224 for (i, grapheme) in input[index + 1..].iter().enumerate() {
225 if !trim_end && is_line_feed(grapheme) {
226 return SnippetState::EndWithLineFeed(
227 input[0..=index + 1 + i].join("").to_string(),
230 } else if not_whitespace_except_line_feed(grapheme) {
231 index_plus_ws = index + i;
237 SnippetState::LineEnd(
238 input[0..=index_minus_ws].join("").to_string(),
242 SnippetState::LineEnd(
243 input[0..=index_plus_ws].join("").to_string(),
249 // Find the position in input for breaking the string
250 if line_end.is_empty()
252 && !is_whitespace(input[max_chars - 1])
253 && is_whitespace(input[max_chars])
255 // At a breaking point already
256 // The line won't invalidate the rewriting because:
257 // - no extra space needed for the line_end character
258 // - extra whitespaces to the right can be trimmed
259 return break_at(max_chars - 1);
261 match input[0..max_chars]
263 .rposition(|grapheme| is_whitespace(grapheme))
265 // Found a whitespace and what is on its left side is big enough.
266 Some(index) if index >= MIN_STRING => break_at(index),
267 // No whitespace found, try looking for a punctuation instead
268 _ => match input[0..max_chars]
270 .rposition(|grapheme| is_punctuation(grapheme))
272 // Found a punctuation and what is on its left side is big enough.
273 Some(index) if index >= MIN_STRING => break_at(index),
274 // Either no boundary character was found to the left of `input[max_chars]`, or the line
275 // got too small. We try searching for a boundary character to the right.
276 _ => match input[max_chars..]
278 .position(|grapheme| is_whitespace(grapheme) || is_punctuation(grapheme))
280 // A boundary was found after the line limit
281 Some(index) => break_at(max_chars + index),
282 // No boundary to the right, the input cannot be broken
283 None => SnippetState::EndOfInput(input.join("").to_string()),
289 fn is_line_feed(grapheme: &str) -> bool {
290 grapheme.as_bytes()[0] == b'\n'
293 fn is_whitespace(grapheme: &str) -> bool {
294 grapheme.chars().all(|c| c.is_whitespace())
297 fn is_punctuation(grapheme: &str) -> bool {
298 match grapheme.as_bytes()[0] {
299 b':' | b',' | b';' | b'.' => true,
306 use super::{break_string, rewrite_string, SnippetState, StringFormat};
308 use shape::{Indent, Shape};
309 use unicode_segmentation::UnicodeSegmentation;
313 let config = Default::default();
314 let fmt = StringFormat::new(Shape::legacy(2, Indent::empty()), &config);
315 rewrite_string("eq_", &fmt);
319 fn should_break_on_whitespace() {
320 let string = "Placerat felis. Mauris porta ante sagittis purus.";
321 let graphemes = UnicodeSegmentation::graphemes(&*string, false).collect::<Vec<&str>>();
323 break_string(20, false, "", &graphemes[..]),
324 SnippetState::LineEnd("Placerat felis. ".to_string(), 16)
327 break_string(20, true, "", &graphemes[..]),
328 SnippetState::LineEnd("Placerat felis.".to_string(), 16)
333 fn should_break_on_punctuation() {
334 let string = "Placerat_felis._Mauris_porta_ante_sagittis_purus.";
335 let graphemes = UnicodeSegmentation::graphemes(&*string, false).collect::<Vec<&str>>();
337 break_string(20, false, "", &graphemes[..]),
338 SnippetState::LineEnd("Placerat_felis.".to_string(), 15)
343 fn should_break_forward() {
344 let string = "Venenatis_tellus_vel_tellus. Aliquam aliquam dolor at justo.";
345 let graphemes = UnicodeSegmentation::graphemes(&*string, false).collect::<Vec<&str>>();
347 break_string(20, false, "", &graphemes[..]),
348 SnippetState::LineEnd("Venenatis_tellus_vel_tellus. ".to_string(), 29)
351 break_string(20, true, "", &graphemes[..]),
352 SnippetState::LineEnd("Venenatis_tellus_vel_tellus.".to_string(), 29)
357 fn nothing_to_break() {
358 let string = "Venenatis_tellus_vel_tellus";
359 let graphemes = UnicodeSegmentation::graphemes(&*string, false).collect::<Vec<&str>>();
361 break_string(20, false, "", &graphemes[..]),
362 SnippetState::EndOfInput("Venenatis_tellus_vel_tellus".to_string())
367 fn significant_whitespaces() {
368 let string = "Neque in sem. \n Pellentesque tellus augue.";
369 let graphemes = UnicodeSegmentation::graphemes(&*string, false).collect::<Vec<&str>>();
371 break_string(15, false, "", &graphemes[..]),
372 SnippetState::EndWithLineFeed("Neque in sem. \n".to_string(), 20)
375 break_string(25, false, "", &graphemes[..]),
376 SnippetState::EndWithLineFeed("Neque in sem. \n".to_string(), 20)
380 break_string(15, true, "", &graphemes[..]),
381 SnippetState::LineEnd("Neque in sem.".to_string(), 19)
384 break_string(25, true, "", &graphemes[..]),
385 SnippetState::EndWithLineFeed("Neque in sem.\n".to_string(), 20)
390 fn big_whitespace() {
391 let string = "Neque in sem. Pellentesque tellus augue.";
392 let graphemes = UnicodeSegmentation::graphemes(&*string, false).collect::<Vec<&str>>();
394 break_string(20, false, "", &graphemes[..]),
395 SnippetState::LineEnd("Neque in sem. ".to_string(), 25)
398 break_string(20, true, "", &graphemes[..]),
399 SnippetState::LineEnd("Neque in sem.".to_string(), 25)
404 fn newline_in_candidate_line() {
405 let string = "Nulla\nconsequat erat at massa. Vivamus id mi.";
407 let graphemes = UnicodeSegmentation::graphemes(&*string, false).collect::<Vec<&str>>();
409 break_string(25, false, "", &graphemes[..]),
410 SnippetState::EndWithLineFeed("Nulla\n".to_string(), 6)
413 break_string(25, true, "", &graphemes[..]),
414 SnippetState::EndWithLineFeed("Nulla\n".to_string(), 6)
417 let mut config: Config = Default::default();
418 config.set().max_width(27);
419 let fmt = StringFormat::new(Shape::legacy(25, Indent::empty()), &config);
420 let rewritten_string = rewrite_string(string, &fmt);
423 Some("\"Nulla\nconsequat erat at massa. \\\n Vivamus id mi.\"".to_string())
428 fn last_line_fit_with_trailing_whitespaces() {
429 let string = "Vivamus id mi. ";
430 let config: Config = Default::default();
431 let mut fmt = StringFormat::new(Shape::legacy(25, Indent::empty()), &config);
434 let rewritten_string = rewrite_string(string, &fmt);
435 assert_eq!(rewritten_string, Some("\"Vivamus id mi.\"".to_string()));
437 fmt.trim_end = false; // default value of trim_end
438 let rewritten_string = rewrite_string(string, &fmt);
439 assert_eq!(rewritten_string, Some("\"Vivamus id mi. \"".to_string()));
443 fn last_line_fit_with_newline() {
444 let string = "Vivamus id mi.\nVivamus id mi.";
445 let config: Config = Default::default();
446 let fmt = StringFormat {
451 shape: Shape::legacy(100, Indent::from_width(&config, 4)),
456 let rewritten_string = rewrite_string(string, &fmt);
459 Some("Vivamus id mi.\n // Vivamus id mi.".to_string())
464 fn overflow_in_non_string_content() {
465 let comment = "Aenean metus.\nVestibulum ac lacus. Vivamus porttitor";
466 let config: Config = Default::default();
467 let fmt = StringFormat {
472 shape: Shape::legacy(30, Indent::from_width(&config, 8)),
478 rewrite_string(comment, &fmt),
480 "Aenean metus.\n // Vestibulum ac lacus. Vivamus\n // porttitor"
487 fn overflow_in_non_string_content_with_line_end() {
488 let comment = "Aenean metus.\nVestibulum ac lacus. Vivamus porttitor";
489 let config: Config = Default::default();
490 let fmt = StringFormat {
495 shape: Shape::legacy(30, Indent::from_width(&config, 8)),
501 rewrite_string(comment, &fmt),
503 "Aenean metus.\n // Vestibulum ac lacus. Vivamus@\n // porttitor"
510 fn blank_line_with_non_empty_line_start() {
511 let config: Config = Default::default();
512 let mut fmt = StringFormat {
517 shape: Shape::legacy(30, Indent::from_width(&config, 4)),
522 let comment = "Aenean metus. Vestibulum\n\nac lacus. Vivamus porttitor";
524 rewrite_string(comment, &fmt),
526 "Aenean metus. Vestibulum\n //\n // ac lacus. Vivamus porttitor".to_string()
530 fmt.shape = Shape::legacy(15, Indent::from_width(&config, 4));
531 let comment = "Aenean\n\nmetus. Vestibulum ac lacus. Vivamus porttitor";
533 rewrite_string(comment, &fmt),
546 fn retain_blank_lines() {
547 let config: Config = Default::default();
548 let fmt = StringFormat {
553 shape: Shape::legacy(20, Indent::from_width(&config, 4)),
558 let comment = "Aenean\n\nmetus. Vestibulum ac lacus.\n\n";
560 rewrite_string(comment, &fmt),
562 "Aenean\n //\n // metus. Vestibulum ac\n // lacus.\n //\n".to_string()
566 let comment = "Aenean\n\nmetus. Vestibulum ac lacus.\n";
568 rewrite_string(comment, &fmt),
569 Some("Aenean\n //\n // metus. Vestibulum ac\n // lacus.\n".to_string())
572 let comment = "Aenean\n \nmetus. Vestibulum ac lacus.";
574 rewrite_string(comment, &fmt),
575 Some("Aenean\n //\n // metus. Vestibulum ac\n // lacus.".to_string())
580 fn boundary_on_edge() {
581 let config: Config = Default::default();
582 let mut fmt = StringFormat {
587 shape: Shape::legacy(13, Indent::from_width(&config, 4)),
592 let comment = "Aenean metus. Vestibulum ac lacus.";
594 rewrite_string(comment, &fmt),
595 Some("Aenean metus.\n // Vestibulum ac\n // lacus.".to_string())
598 fmt.trim_end = false;
599 let comment = "Vestibulum ac lacus.";
601 rewrite_string(comment, &fmt),
602 Some("Vestibulum \n // ac lacus.".to_string())
607 let comment = "Vestibulum ac lacus.";
609 rewrite_string(comment, &fmt),
610 Some("Vestibulum\\\n // ac lacus.".to_string())