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 /// Returns the index to the end of the url if the given string includes an
162 /// URL or alike. Otherwise, returns None;
163 fn detect_url(s: &[&str], index: usize) -> Option<usize> {
164 let start = match s[..=index].iter().rposition(|g| is_whitespace(g)) {
165 Some(pos) => pos + 1,
168 if s.len() < start + 8 {
171 let prefix = s[start..start + 8].join("");
172 if prefix.starts_with("https://")
173 || prefix.starts_with("http://")
174 || prefix.starts_with("ftp://")
175 || prefix.starts_with("file://")
177 match s[index..].iter().position(|g| is_whitespace(g)) {
178 Some(pos) => Some(index + pos - 1),
179 None => Some(s.len() - 1),
186 /// Trims whitespaces to the right except for the line feed character.
187 fn trim_right_but_line_feed(trim_end: bool, result: String) -> String {
188 let whitespace_except_line_feed = |c: char| c.is_whitespace() && c != '\n';
189 if trim_end && result.ends_with(whitespace_except_line_feed) {
191 .trim_right_matches(whitespace_except_line_feed)
198 /// Result of breaking a string so it fits in a line and the state it ended in.
199 /// The state informs about what to do with the snippet and how to continue the breaking process.
200 #[derive(Debug, PartialEq)]
202 /// The input could not be broken and so rewriting the string is finished.
204 /// The input could be broken and the returned snippet should be ended with a
205 /// `[StringFormat::line_end]`. The next snippet needs to be indented.
207 /// The returned string is the line to print out and the number is the length that got read in
208 /// the text being rewritten. That length may be greater than the returned string if trailing
209 /// whitespaces got trimmed.
210 LineEnd(String, usize),
211 /// The input could be broken but a newline is present that cannot be trimmed. The next snippet
212 /// to be rewritten *could* use more width than what is specified by the given shape. For
213 /// example with a multiline string, the next snippet does not need to be indented, allowing
214 /// more characters to be fit within a line.
216 /// The returned string is the line to print out and the number is the length that got read in
217 /// the text being rewritten.
218 EndWithLineFeed(String, usize),
221 fn not_whitespace_except_line_feed(g: &str) -> bool {
222 is_line_feed(g) || !is_whitespace(g)
225 /// Break the input string at a boundary character around the offset `max_chars`. A boundary
226 /// character is either a punctuation or a whitespace.
227 fn break_string(max_chars: usize, trim_end: bool, line_end: &str, input: &[&str]) -> SnippetState {
228 let break_at = |index /* grapheme at index is included */| {
229 // Take in any whitespaces to the left/right of `input[index]` while
230 // preserving line feeds
231 let index_minus_ws = input[0..=index]
233 .rposition(|grapheme| not_whitespace_except_line_feed(grapheme))
235 // Take into account newlines occuring in input[0..=index], i.e., the possible next new
236 // line. If there is one, then text after it could be rewritten in a way that the available
237 // space is fully used.
238 for (i, grapheme) in input[0..=index].iter().enumerate() {
239 if is_line_feed(grapheme) {
240 if i <= index_minus_ws {
241 let mut line = input[0..i].join("");
243 line = line.trim_right().to_string();
245 return SnippetState::EndWithLineFeed(format!("{}\n", line), i + 1);
251 let mut index_plus_ws = index;
252 for (i, grapheme) in input[index + 1..].iter().enumerate() {
253 if !trim_end && is_line_feed(grapheme) {
254 return SnippetState::EndWithLineFeed(
255 input[0..=index + 1 + i].join("").to_string(),
258 } else if not_whitespace_except_line_feed(grapheme) {
259 index_plus_ws = index + i;
265 SnippetState::LineEnd(
266 input[0..=index_minus_ws].join("").to_string(),
270 SnippetState::LineEnd(
271 input[0..=index_plus_ws].join("").to_string(),
277 // Find the position in input for breaking the string
278 if line_end.is_empty()
280 && !is_whitespace(input[max_chars - 1])
281 && is_whitespace(input[max_chars])
283 // At a breaking point already
284 // The line won't invalidate the rewriting because:
285 // - no extra space needed for the line_end character
286 // - extra whitespaces to the right can be trimmed
287 return break_at(max_chars - 1);
289 if let Some(url_index_end) = detect_url(input, max_chars) {
290 let index_plus_ws = url_index_end
291 + input[url_index_end..]
294 .position(|grapheme| not_whitespace_except_line_feed(grapheme))
297 SnippetState::LineEnd(
298 input[..=url_index_end].join("").to_string(),
302 return SnippetState::LineEnd(
303 input[..=index_plus_ws].join("").to_string(),
308 match input[0..max_chars]
310 .rposition(|grapheme| is_whitespace(grapheme))
312 // Found a whitespace and what is on its left side is big enough.
313 Some(index) if index >= MIN_STRING => break_at(index),
314 // No whitespace found, try looking for a punctuation instead
315 _ => match input[0..max_chars]
317 .rposition(|grapheme| is_punctuation(grapheme))
319 // Found a punctuation and what is on its left side is big enough.
320 Some(index) if index >= MIN_STRING => break_at(index),
321 // Either no boundary character was found to the left of `input[max_chars]`, or the line
322 // got too small. We try searching for a boundary character to the right.
323 _ => match input[max_chars..]
325 .position(|grapheme| is_whitespace(grapheme) || is_punctuation(grapheme))
327 // A boundary was found after the line limit
328 Some(index) => break_at(max_chars + index),
329 // No boundary to the right, the input cannot be broken
330 None => SnippetState::EndOfInput(input.join("").to_string()),
336 fn is_line_feed(grapheme: &str) -> bool {
337 grapheme.as_bytes()[0] == b'\n'
340 fn is_whitespace(grapheme: &str) -> bool {
341 grapheme.chars().all(|c| c.is_whitespace())
344 fn is_punctuation(grapheme: &str) -> bool {
345 match grapheme.as_bytes()[0] {
346 b':' | b',' | b';' | b'.' => true,
353 use super::{break_string, detect_url, rewrite_string, SnippetState, StringFormat};
355 use shape::{Indent, Shape};
356 use unicode_segmentation::UnicodeSegmentation;
360 let config = Default::default();
361 let fmt = StringFormat::new(Shape::legacy(2, Indent::empty()), &config);
362 rewrite_string("eq_", &fmt);
366 fn should_break_on_whitespace() {
367 let string = "Placerat felis. Mauris porta ante sagittis purus.";
368 let graphemes = UnicodeSegmentation::graphemes(&*string, false).collect::<Vec<&str>>();
370 break_string(20, false, "", &graphemes[..]),
371 SnippetState::LineEnd("Placerat felis. ".to_string(), 16)
374 break_string(20, true, "", &graphemes[..]),
375 SnippetState::LineEnd("Placerat felis.".to_string(), 16)
380 fn should_break_on_punctuation() {
381 let string = "Placerat_felis._Mauris_porta_ante_sagittis_purus.";
382 let graphemes = UnicodeSegmentation::graphemes(&*string, false).collect::<Vec<&str>>();
384 break_string(20, false, "", &graphemes[..]),
385 SnippetState::LineEnd("Placerat_felis.".to_string(), 15)
390 fn should_break_forward() {
391 let string = "Venenatis_tellus_vel_tellus. Aliquam aliquam dolor at justo.";
392 let graphemes = UnicodeSegmentation::graphemes(&*string, false).collect::<Vec<&str>>();
394 break_string(20, false, "", &graphemes[..]),
395 SnippetState::LineEnd("Venenatis_tellus_vel_tellus. ".to_string(), 29)
398 break_string(20, true, "", &graphemes[..]),
399 SnippetState::LineEnd("Venenatis_tellus_vel_tellus.".to_string(), 29)
404 fn nothing_to_break() {
405 let string = "Venenatis_tellus_vel_tellus";
406 let graphemes = UnicodeSegmentation::graphemes(&*string, false).collect::<Vec<&str>>();
408 break_string(20, false, "", &graphemes[..]),
409 SnippetState::EndOfInput("Venenatis_tellus_vel_tellus".to_string())
414 fn significant_whitespaces() {
415 let string = "Neque in sem. \n Pellentesque tellus augue.";
416 let graphemes = UnicodeSegmentation::graphemes(&*string, false).collect::<Vec<&str>>();
418 break_string(15, false, "", &graphemes[..]),
419 SnippetState::EndWithLineFeed("Neque in sem. \n".to_string(), 20)
422 break_string(25, false, "", &graphemes[..]),
423 SnippetState::EndWithLineFeed("Neque in sem. \n".to_string(), 20)
427 break_string(15, true, "", &graphemes[..]),
428 SnippetState::LineEnd("Neque in sem.".to_string(), 19)
431 break_string(25, true, "", &graphemes[..]),
432 SnippetState::EndWithLineFeed("Neque in sem.\n".to_string(), 20)
437 fn big_whitespace() {
438 let string = "Neque in sem. Pellentesque tellus augue.";
439 let graphemes = UnicodeSegmentation::graphemes(&*string, false).collect::<Vec<&str>>();
441 break_string(20, false, "", &graphemes[..]),
442 SnippetState::LineEnd("Neque in sem. ".to_string(), 25)
445 break_string(20, true, "", &graphemes[..]),
446 SnippetState::LineEnd("Neque in sem.".to_string(), 25)
451 fn newline_in_candidate_line() {
452 let string = "Nulla\nconsequat erat at massa. Vivamus id mi.";
454 let graphemes = UnicodeSegmentation::graphemes(&*string, false).collect::<Vec<&str>>();
456 break_string(25, false, "", &graphemes[..]),
457 SnippetState::EndWithLineFeed("Nulla\n".to_string(), 6)
460 break_string(25, true, "", &graphemes[..]),
461 SnippetState::EndWithLineFeed("Nulla\n".to_string(), 6)
464 let mut config: Config = Default::default();
465 config.set().max_width(27);
466 let fmt = StringFormat::new(Shape::legacy(25, Indent::empty()), &config);
467 let rewritten_string = rewrite_string(string, &fmt);
470 Some("\"Nulla\nconsequat erat at massa. \\\n Vivamus id mi.\"".to_string())
475 fn last_line_fit_with_trailing_whitespaces() {
476 let string = "Vivamus id mi. ";
477 let config: Config = Default::default();
478 let mut fmt = StringFormat::new(Shape::legacy(25, Indent::empty()), &config);
481 let rewritten_string = rewrite_string(string, &fmt);
482 assert_eq!(rewritten_string, Some("\"Vivamus id mi.\"".to_string()));
484 fmt.trim_end = false; // default value of trim_end
485 let rewritten_string = rewrite_string(string, &fmt);
486 assert_eq!(rewritten_string, Some("\"Vivamus id mi. \"".to_string()));
490 fn last_line_fit_with_newline() {
491 let string = "Vivamus id mi.\nVivamus id mi.";
492 let config: Config = Default::default();
493 let fmt = StringFormat {
498 shape: Shape::legacy(100, Indent::from_width(&config, 4)),
503 let rewritten_string = rewrite_string(string, &fmt);
506 Some("Vivamus id mi.\n // Vivamus id mi.".to_string())
511 fn overflow_in_non_string_content() {
512 let comment = "Aenean metus.\nVestibulum ac lacus. Vivamus porttitor";
513 let config: Config = Default::default();
514 let fmt = StringFormat {
519 shape: Shape::legacy(30, Indent::from_width(&config, 8)),
525 rewrite_string(comment, &fmt),
527 "Aenean metus.\n // Vestibulum ac lacus. Vivamus\n // porttitor"
534 fn overflow_in_non_string_content_with_line_end() {
535 let comment = "Aenean metus.\nVestibulum ac lacus. Vivamus porttitor";
536 let config: Config = Default::default();
537 let fmt = StringFormat {
542 shape: Shape::legacy(30, Indent::from_width(&config, 8)),
548 rewrite_string(comment, &fmt),
550 "Aenean metus.\n // Vestibulum ac lacus. Vivamus@\n // porttitor"
557 fn blank_line_with_non_empty_line_start() {
558 let config: Config = Default::default();
559 let mut fmt = StringFormat {
564 shape: Shape::legacy(30, Indent::from_width(&config, 4)),
569 let comment = "Aenean metus. Vestibulum\n\nac lacus. Vivamus porttitor";
571 rewrite_string(comment, &fmt),
573 "Aenean metus. Vestibulum\n //\n // ac lacus. Vivamus porttitor".to_string()
577 fmt.shape = Shape::legacy(15, Indent::from_width(&config, 4));
578 let comment = "Aenean\n\nmetus. Vestibulum ac lacus. Vivamus porttitor";
580 rewrite_string(comment, &fmt),
593 fn retain_blank_lines() {
594 let config: Config = Default::default();
595 let fmt = StringFormat {
600 shape: Shape::legacy(20, Indent::from_width(&config, 4)),
605 let comment = "Aenean\n\nmetus. Vestibulum ac lacus.\n\n";
607 rewrite_string(comment, &fmt),
609 "Aenean\n //\n // metus. Vestibulum ac\n // lacus.\n //\n".to_string()
613 let comment = "Aenean\n\nmetus. Vestibulum ac lacus.\n";
615 rewrite_string(comment, &fmt),
616 Some("Aenean\n //\n // metus. Vestibulum ac\n // lacus.\n".to_string())
619 let comment = "Aenean\n \nmetus. Vestibulum ac lacus.";
621 rewrite_string(comment, &fmt),
622 Some("Aenean\n //\n // metus. Vestibulum ac\n // lacus.".to_string())
627 fn boundary_on_edge() {
628 let config: Config = Default::default();
629 let mut fmt = StringFormat {
634 shape: Shape::legacy(13, Indent::from_width(&config, 4)),
639 let comment = "Aenean metus. Vestibulum ac lacus.";
641 rewrite_string(comment, &fmt),
642 Some("Aenean metus.\n // Vestibulum ac\n // lacus.".to_string())
645 fmt.trim_end = false;
646 let comment = "Vestibulum ac lacus.";
648 rewrite_string(comment, &fmt),
649 Some("Vestibulum \n // ac lacus.".to_string())
654 let comment = "Vestibulum ac lacus.";
656 rewrite_string(comment, &fmt),
657 Some("Vestibulum\\\n // ac lacus.".to_string())
663 let string = "aaa http://example.org something";
664 let graphemes = UnicodeSegmentation::graphemes(&*string, false).collect::<Vec<&str>>();
665 assert_eq!(detect_url(&graphemes, 8), Some(21));
667 let string = "https://example.org something";
668 let graphemes = UnicodeSegmentation::graphemes(&*string, false).collect::<Vec<&str>>();
669 assert_eq!(detect_url(&graphemes, 0), Some(18));
671 let string = "aaa ftp://example.org something";
672 let graphemes = UnicodeSegmentation::graphemes(&*string, false).collect::<Vec<&str>>();
673 assert_eq!(detect_url(&graphemes, 8), Some(20));
675 let string = "aaa file://example.org something";
676 let graphemes = UnicodeSegmentation::graphemes(&*string, false).collect::<Vec<&str>>();
677 assert_eq!(detect_url(&graphemes, 8), Some(21));
679 let string = "aaa http not an url";
680 let graphemes = UnicodeSegmentation::graphemes(&*string, false).collect::<Vec<&str>>();
681 assert_eq!(detect_url(&graphemes, 6), None);
683 let string = "aaa file://example.org";
684 let graphemes = UnicodeSegmentation::graphemes(&*string, false).collect::<Vec<&str>>();
685 assert_eq!(detect_url(&graphemes, 8), Some(21));