1pub mod archive;
2pub mod command;
3pub mod fs;
4pub mod markdown;
5pub mod path_list;
6pub mod paths;
7pub mod process;
8pub mod redact;
9pub mod rel_path;
10pub mod schemars;
11pub mod serde;
12pub mod shell;
13pub mod shell_builder;
14pub mod shell_env;
15pub mod size;
16#[cfg(any(test, feature = "test-support"))]
17pub mod test;
18pub mod time;
19
20use anyhow::Result;
21use itertools::Either;
22use regex::Regex;
23use rust_embed::EmbeddedFile;
24use std::path::{Path, PathBuf};
25use std::sync::LazyLock;
26use std::{
27 borrow::Cow,
28 cmp::{self, Ordering},
29 ops::{Range, RangeInclusive},
30};
31use unicase::UniCase;
32
33pub use gpui_util::*;
34
35pub use take_until::*;
36#[cfg(any(test, feature = "test-support"))]
37pub use util_macros::{line_endings, path, uri};
38
39pub use self::shell::{
40 get_default_system_shell, get_default_system_shell_preferring_bash, get_system_shell,
41};
42
43#[inline]
44pub const fn is_utf8_char_boundary(u8: u8) -> bool {
45 // This is bit magic equivalent to: b < 128 || b >= 192
46 (u8 as i8) >= -0x40
47}
48
49pub fn truncate(s: &str, max_chars: usize) -> &str {
50 match s.char_indices().nth(max_chars) {
51 None => s,
52 Some((idx, _)) => &s[..idx],
53 }
54}
55
56/// Removes characters from the end of the string if its length is greater than `max_chars` and
57/// appends "..." to the string. Returns string unchanged if its length is smaller than max_chars.
58pub fn truncate_and_trailoff(s: &str, max_chars: usize) -> String {
59 debug_assert!(max_chars >= 5);
60
61 // If the string's byte length is <= max_chars, walking the string can be skipped since the
62 // number of chars is <= the number of bytes.
63 if s.len() <= max_chars {
64 return s.to_string();
65 }
66 let truncation_ix = s.char_indices().map(|(i, _)| i).nth(max_chars);
67 match truncation_ix {
68 Some(index) => s[..index].to_string() + "β¦",
69 _ => s.to_string(),
70 }
71}
72
73/// Removes characters from the front of the string if its length is greater than `max_chars` and
74/// prepends the string with "...". Returns string unchanged if its length is smaller than max_chars.
75pub fn truncate_and_remove_front(s: &str, max_chars: usize) -> String {
76 debug_assert!(max_chars >= 5);
77
78 // If the string's byte length is <= max_chars, walking the string can be skipped since the
79 // number of chars is <= the number of bytes.
80 if s.len() <= max_chars {
81 return s.to_string();
82 }
83 let suffix_char_length = max_chars.saturating_sub(1);
84 let truncation_ix = s
85 .char_indices()
86 .map(|(i, _)| i)
87 .nth_back(suffix_char_length);
88 match truncation_ix {
89 Some(index) if index > 0 => "β¦".to_string() + &s[index..],
90 _ => s.to_string(),
91 }
92}
93
94/// Takes only `max_lines` from the string and, if there were more than `max_lines-1`, appends a
95/// a newline and "..." to the string, so that `max_lines` are returned.
96/// Returns string unchanged if its length is smaller than max_lines.
97pub fn truncate_lines_and_trailoff(s: &str, max_lines: usize) -> String {
98 let mut lines = s.lines().take(max_lines).collect::<Vec<_>>();
99 if lines.len() > max_lines - 1 {
100 lines.pop();
101 lines.join("\n") + "\nβ¦"
102 } else {
103 lines.join("\n")
104 }
105}
106
107/// Truncates the string at a character boundary, such that the result is less than `max_bytes` in
108/// length.
109pub fn truncate_to_byte_limit(s: &str, max_bytes: usize) -> &str {
110 if s.len() < max_bytes {
111 return s;
112 }
113
114 for i in (0..max_bytes).rev() {
115 if s.is_char_boundary(i) {
116 return &s[..i];
117 }
118 }
119
120 ""
121}
122
123/// Takes a prefix of complete lines which fit within the byte limit. If the first line is longer
124/// than the limit, truncates at a character boundary.
125pub fn truncate_lines_to_byte_limit(s: &str, max_bytes: usize) -> &str {
126 if s.len() < max_bytes {
127 return s;
128 }
129
130 for i in (0..max_bytes).rev() {
131 if s.is_char_boundary(i) && s.as_bytes()[i] == b'\n' {
132 // Since the i-th character is \n, valid to slice at i + 1.
133 return &s[..i + 1];
134 }
135 }
136
137 truncate_to_byte_limit(s, max_bytes)
138}
139
140#[test]
141fn test_truncate_lines_to_byte_limit() {
142 let text = "Line 1\nLine 2\nLine 3\nLine 4";
143
144 // Limit that includes all lines
145 assert_eq!(truncate_lines_to_byte_limit(text, 100), text);
146
147 // Exactly the first line
148 assert_eq!(truncate_lines_to_byte_limit(text, 7), "Line 1\n");
149
150 // Limit between lines
151 assert_eq!(truncate_lines_to_byte_limit(text, 13), "Line 1\n");
152 assert_eq!(truncate_lines_to_byte_limit(text, 20), "Line 1\nLine 2\n");
153
154 // Limit before first newline
155 assert_eq!(truncate_lines_to_byte_limit(text, 6), "Line ");
156
157 // Test with non-ASCII characters
158 let text_utf8 = "Line 1\nLΓne 2\nLine 3";
159 assert_eq!(
160 truncate_lines_to_byte_limit(text_utf8, 15),
161 "Line 1\nLΓne 2\n"
162 );
163}
164
165/// Extend a sorted vector with a sorted sequence of items, maintaining the vector's sort order and
166/// enforcing a maximum length. This also de-duplicates items. Sort the items according to the given callback. Before calling this,
167/// both `vec` and `new_items` should already be sorted according to the `cmp` comparator.
168pub fn extend_sorted<T, I, F>(vec: &mut Vec<T>, new_items: I, limit: usize, mut cmp: F)
169where
170 I: IntoIterator<Item = T>,
171 F: FnMut(&T, &T) -> Ordering,
172{
173 let mut start_index = 0;
174 for new_item in new_items {
175 if let Err(i) = vec[start_index..].binary_search_by(|m| cmp(m, &new_item)) {
176 let index = start_index + i;
177 if vec.len() < limit {
178 vec.insert(index, new_item);
179 } else if index < vec.len() {
180 vec.pop();
181 vec.insert(index, new_item);
182 }
183 start_index = index;
184 }
185 }
186}
187
188pub fn truncate_to_bottom_n_sorted_by<T, F>(items: &mut Vec<T>, limit: usize, compare: &F)
189where
190 F: Fn(&T, &T) -> Ordering,
191{
192 if limit == 0 {
193 items.truncate(0);
194 }
195 if items.len() <= limit {
196 items.sort_by(compare);
197 return;
198 }
199 // When limit is near to items.len() it may be more efficient to sort the whole list and
200 // truncate, rather than always doing selection first as is done below. It's hard to analyze
201 // where the threshold for this should be since the quickselect style algorithm used by
202 // `select_nth_unstable_by` makes the prefix partially sorted, and so its work is not wasted -
203 // the expected number of comparisons needed by `sort_by` is less than it is for some arbitrary
204 // unsorted input.
205 items.select_nth_unstable_by(limit, compare);
206 items.truncate(limit);
207 items.sort_by(compare);
208}
209
210/// Prevents execution of the application with root privileges on Unix systems.
211///
212/// This function checks if the current process is running with root privileges
213/// and terminates the program with an error message unless explicitly allowed via the
214/// `ZED_ALLOW_ROOT` environment variable.
215#[cfg(unix)]
216pub fn prevent_root_execution() {
217 let is_root = nix::unistd::geteuid().is_root();
218 let allow_root = std::env::var("ZED_ALLOW_ROOT").is_ok_and(|val| val == "true");
219
220 if is_root && !allow_root {
221 eprintln!(
222 "\
223Error: Running Zed as root or via sudo is unsupported.
224 Doing so (even once) may subtly break things for all subsequent non-root usage of Zed.
225 It is untested and not recommended, don't complain when things break.
226 If you wish to proceed anyways, set `ZED_ALLOW_ROOT=true` in your environment."
227 );
228 std::process::exit(1);
229 }
230}
231
232#[cfg(unix)]
233fn load_shell_from_passwd() -> Result<()> {
234 let buflen = match unsafe { libc::sysconf(libc::_SC_GETPW_R_SIZE_MAX) } {
235 n if n < 0 => 1024,
236 n => n as usize,
237 };
238 let mut buffer = Vec::with_capacity(buflen);
239
240 let mut pwd: std::mem::MaybeUninit<libc::passwd> = std::mem::MaybeUninit::uninit();
241 let mut result: *mut libc::passwd = std::ptr::null_mut();
242
243 let uid = unsafe { libc::getuid() };
244 let status = unsafe {
245 libc::getpwuid_r(
246 uid,
247 pwd.as_mut_ptr(),
248 buffer.as_mut_ptr() as *mut libc::c_char,
249 buflen,
250 &mut result,
251 )
252 };
253 anyhow::ensure!(!result.is_null(), "passwd entry for uid {} not found", uid);
254
255 // SAFETY: If `getpwuid_r` doesn't error, we have the entry here.
256 let entry = unsafe { pwd.assume_init() };
257
258 anyhow::ensure!(
259 status == 0,
260 "call to getpwuid_r failed. uid: {}, status: {}",
261 uid,
262 status
263 );
264 anyhow::ensure!(
265 entry.pw_uid == uid,
266 "passwd entry has different uid ({}) than getuid ({}) returned",
267 entry.pw_uid,
268 uid,
269 );
270
271 let shell = unsafe { std::ffi::CStr::from_ptr(entry.pw_shell).to_str().unwrap() };
272 let should_set_shell = std::env::var("SHELL").map_or(true, |shell_env| {
273 shell_env != shell && !std::path::Path::new(&shell_env).exists()
274 });
275
276 if should_set_shell {
277 log::info!(
278 "updating SHELL environment variable to value from passwd entry: {:?}",
279 shell,
280 );
281 unsafe { std::env::set_var("SHELL", shell) };
282 }
283
284 Ok(())
285}
286
287/// Returns a shell escaped path for the current zed executable
288pub fn get_shell_safe_zed_path(shell_kind: shell::ShellKind) -> anyhow::Result<String> {
289 use anyhow::Context as _;
290 use paths::PathExt;
291 let mut zed_path =
292 std::env::current_exe().context("Failed to determine current zed executable path.")?;
293 if cfg!(target_os = "linux")
294 && !zed_path.is_file()
295 && let Some(truncated) = zed_path
296 .clone()
297 .file_name()
298 .and_then(|s| s.to_str())
299 .and_then(|n| n.strip_suffix(" (deleted)"))
300 {
301 // Might have been deleted during update; let's use the new binary if there is one.
302 zed_path.set_file_name(truncated);
303 }
304
305 zed_path
306 .try_shell_safe(shell_kind)
307 .context("Failed to shell-escape Zed executable path.")
308}
309
310/// Returns a path for the zed cli executable, this function
311/// should be called from the zed executable, not zed-cli.
312pub fn get_zed_cli_path() -> Result<PathBuf> {
313 use anyhow::Context as _;
314 let zed_path =
315 std::env::current_exe().context("Failed to determine current zed executable path.")?;
316 let parent = zed_path
317 .parent()
318 .context("Failed to determine parent directory of zed executable path.")?;
319
320 let possible_locations: &[&str] = if cfg!(target_os = "macos") {
321 // On macOS, the zed executable and zed-cli are inside the app bundle,
322 // so here ./cli is for both installed and development builds.
323 &["./cli"]
324 } else if cfg!(target_os = "windows") {
325 // bin/zed.exe is for installed builds, ./cli.exe is for development builds.
326 &["bin/zed.exe", "./cli.exe"]
327 } else if cfg!(target_os = "linux") || cfg!(target_os = "freebsd") {
328 // bin is the standard, ./cli is for the target directory in development builds.
329 &["../bin/zed", "./cli"]
330 } else {
331 anyhow::bail!("unsupported platform for determining zed-cli path");
332 };
333
334 possible_locations
335 .iter()
336 .find_map(|p| {
337 parent
338 .join(p)
339 .canonicalize()
340 .ok()
341 .filter(|p| p != &zed_path)
342 })
343 .with_context(|| {
344 format!(
345 "could not find zed-cli from any of: {}",
346 possible_locations.join(", ")
347 )
348 })
349}
350
351#[cfg(unix)]
352pub async fn load_login_shell_environment() -> Result<()> {
353 use anyhow::Context as _;
354
355 load_shell_from_passwd().log_err();
356
357 // If possible, we want to `cd` in the user's `$HOME` to trigger programs
358 // such as direnv, asdf, mise, ... to adjust the PATH. These tools often hook
359 // into shell's `cd` command (and hooks) to manipulate env.
360 // We do this so that we get the env a user would have when spawning a shell
361 // in home directory.
362 for (name, value) in shell_env::capture(get_system_shell(), &[], paths::home_dir())
363 .await
364 .with_context(|| format!("capturing environment with {:?}", get_system_shell()))?
365 {
366 // Skip SHLVL to prevent it from polluting Zed's process environment.
367 // The login shell used for env capture increments SHLVL, and if we propagate it,
368 // terminals spawned by Zed will inherit it and increment again, causing SHLVL
369 // to start at 2 instead of 1 (and increase by 2 on each reload).
370 if name == "SHLVL" {
371 continue;
372 }
373 unsafe { std::env::set_var(&name, &value) };
374 }
375
376 log::info!(
377 "set environment variables from shell:{}, path:{}",
378 std::env::var("SHELL").unwrap_or_default(),
379 std::env::var("PATH").unwrap_or_default(),
380 );
381
382 Ok(())
383}
384
385/// Configures the process to start a new session, to prevent interactive shells from taking control
386/// of the terminal.
387///
388/// For more details: <https://registerspill.thorstenball.com/p/how-to-lose-control-of-your-shell>
389pub fn set_pre_exec_to_start_new_session(
390 command: &mut std::process::Command,
391) -> &mut std::process::Command {
392 // safety: code in pre_exec should be signal safe.
393 // https://man7.org/linux/man-pages/man7/signal-safety.7.html
394 #[cfg(unix)]
395 unsafe {
396 use std::os::unix::process::CommandExt;
397 command.pre_exec(|| {
398 libc::setsid();
399 Ok(())
400 });
401 };
402 command
403}
404
405pub fn merge_json_lenient_value_into(
406 source: serde_json_lenient::Value,
407 target: &mut serde_json_lenient::Value,
408) {
409 match (source, target) {
410 (serde_json_lenient::Value::Object(source), serde_json_lenient::Value::Object(target)) => {
411 for (key, value) in source {
412 if let Some(target) = target.get_mut(&key) {
413 merge_json_lenient_value_into(value, target);
414 } else {
415 target.insert(key, value);
416 }
417 }
418 }
419
420 (serde_json_lenient::Value::Array(source), serde_json_lenient::Value::Array(target)) => {
421 for value in source {
422 target.push(value);
423 }
424 }
425
426 (source, target) => *target = source,
427 }
428}
429
430pub fn merge_json_value_into(source: serde_json::Value, target: &mut serde_json::Value) {
431 use serde_json::Value;
432
433 match (source, target) {
434 (Value::Object(source), Value::Object(target)) => {
435 for (key, value) in source {
436 if let Some(target) = target.get_mut(&key) {
437 merge_json_value_into(value, target);
438 } else {
439 target.insert(key, value);
440 }
441 }
442 }
443
444 (Value::Array(source), Value::Array(target)) => {
445 for value in source {
446 target.push(value);
447 }
448 }
449
450 (source, target) => *target = source,
451 }
452}
453
454pub fn merge_non_null_json_value_into(source: serde_json::Value, target: &mut serde_json::Value) {
455 use serde_json::Value;
456 if let Value::Object(source_object) = source {
457 let target_object = if let Value::Object(target) = target {
458 target
459 } else {
460 *target = Value::Object(Default::default());
461 target.as_object_mut().unwrap()
462 };
463 for (key, value) in source_object {
464 if let Some(target) = target_object.get_mut(&key) {
465 merge_non_null_json_value_into(value, target);
466 } else if !value.is_null() {
467 target_object.insert(key, value);
468 }
469 }
470 } else if !source.is_null() {
471 *target = source
472 }
473}
474
475pub fn expanded_and_wrapped_usize_range(
476 range: Range<usize>,
477 additional_before: usize,
478 additional_after: usize,
479 wrap_length: usize,
480) -> impl Iterator<Item = usize> {
481 let start_wraps = range.start < additional_before;
482 let end_wraps = wrap_length < range.end + additional_after;
483 if start_wraps && end_wraps {
484 Either::Left(0..wrap_length)
485 } else if start_wraps {
486 let wrapped_start = (range.start + wrap_length).saturating_sub(additional_before);
487 if wrapped_start <= range.end {
488 Either::Left(0..wrap_length)
489 } else {
490 Either::Right((0..range.end + additional_after).chain(wrapped_start..wrap_length))
491 }
492 } else if end_wraps {
493 let wrapped_end = range.end + additional_after - wrap_length;
494 if range.start <= wrapped_end {
495 Either::Left(0..wrap_length)
496 } else {
497 Either::Right((0..wrapped_end).chain(range.start - additional_before..wrap_length))
498 }
499 } else {
500 Either::Left((range.start - additional_before)..(range.end + additional_after))
501 }
502}
503
504/// Yields `[i, i + 1, i - 1, i + 2, ..]`, each modulo `wrap_length` and bounded by
505/// `additional_before` and `additional_after`. If the wrapping causes overlap, duplicates are not
506/// emitted. If wrap_length is 0, nothing is yielded.
507pub fn wrapped_usize_outward_from(
508 start: usize,
509 additional_before: usize,
510 additional_after: usize,
511 wrap_length: usize,
512) -> impl Iterator<Item = usize> {
513 let mut count = 0;
514 let mut after_offset = 1;
515 let mut before_offset = 1;
516
517 std::iter::from_fn(move || {
518 count += 1;
519 if count > wrap_length {
520 None
521 } else if count == 1 {
522 Some(start % wrap_length)
523 } else if after_offset <= additional_after && after_offset <= before_offset {
524 let value = (start + after_offset) % wrap_length;
525 after_offset += 1;
526 Some(value)
527 } else if before_offset <= additional_before {
528 let value = (start + wrap_length - before_offset) % wrap_length;
529 before_offset += 1;
530 Some(value)
531 } else if after_offset <= additional_after {
532 let value = (start + after_offset) % wrap_length;
533 after_offset += 1;
534 Some(value)
535 } else {
536 None
537 }
538 })
539}
540
541#[cfg(any(test, feature = "test-support"))]
542mod rng {
543 use rand::prelude::*;
544
545 pub struct RandomCharIter<T: Rng> {
546 rng: T,
547 simple_text: bool,
548 }
549
550 impl<T: Rng> RandomCharIter<T> {
551 pub fn new(rng: T) -> Self {
552 Self {
553 rng,
554 simple_text: std::env::var("SIMPLE_TEXT").is_ok_and(|v| !v.is_empty()),
555 }
556 }
557
558 pub fn with_simple_text(mut self) -> Self {
559 self.simple_text = true;
560 self
561 }
562 }
563
564 impl<T: Rng> Iterator for RandomCharIter<T> {
565 type Item = char;
566
567 fn next(&mut self) -> Option<Self::Item> {
568 if self.simple_text {
569 return if self.rng.random_range(0..100) < 5 {
570 Some('\n')
571 } else {
572 Some(self.rng.random_range(b'a'..b'z' + 1).into())
573 };
574 }
575
576 match self.rng.random_range(0..100) {
577 // whitespace
578 0..=19 => [' ', '\n', '\r', '\t'].choose(&mut self.rng).copied(),
579 // two-byte greek letters
580 20..=32 => char::from_u32(self.rng.random_range(('Ξ±' as u32)..('Ο' as u32 + 1))),
581 // // three-byte characters
582 33..=45 => ['β', 'β
', 'β', 'β', 'β']
583 .choose(&mut self.rng)
584 .copied(),
585 // // four-byte characters
586 46..=58 => ['π', 'π', 'π', 'π'].choose(&mut self.rng).copied(),
587 // ascii letters
588 _ => Some(self.rng.random_range(b'a'..b'z' + 1).into()),
589 }
590 }
591 }
592}
593#[cfg(any(test, feature = "test-support"))]
594pub use rng::RandomCharIter;
595
596/// Get an embedded file as a string.
597pub fn asset_str<A: rust_embed::RustEmbed>(path: &str) -> Cow<'static, str> {
598 let asset = A::get(path).expect(path);
599 embedded_file_to_str(asset)
600}
601
602pub fn embedded_file_to_str(file: EmbeddedFile) -> Cow<'static, str> {
603 match file.data {
604 Cow::Borrowed(bytes) => Cow::Borrowed(std::str::from_utf8(bytes).unwrap()),
605 Cow::Owned(bytes) => Cow::Owned(String::from_utf8(bytes).unwrap()),
606 }
607}
608
609pub trait RangeExt<T> {
610 fn sorted(&self) -> Self;
611 fn to_inclusive(&self) -> RangeInclusive<T>;
612 fn overlaps(&self, other: &Range<T>) -> bool;
613 fn contains_inclusive(&self, other: &Range<T>) -> bool;
614}
615
616impl<T: Ord + Clone> RangeExt<T> for Range<T> {
617 fn sorted(&self) -> Self {
618 cmp::min(&self.start, &self.end).clone()..cmp::max(&self.start, &self.end).clone()
619 }
620
621 fn to_inclusive(&self) -> RangeInclusive<T> {
622 self.start.clone()..=self.end.clone()
623 }
624
625 fn overlaps(&self, other: &Range<T>) -> bool {
626 self.start < other.end && other.start < self.end
627 }
628
629 fn contains_inclusive(&self, other: &Range<T>) -> bool {
630 self.start <= other.start && other.end <= self.end
631 }
632}
633
634impl<T: Ord + Clone> RangeExt<T> for RangeInclusive<T> {
635 fn sorted(&self) -> Self {
636 cmp::min(self.start(), self.end()).clone()..=cmp::max(self.start(), self.end()).clone()
637 }
638
639 fn to_inclusive(&self) -> RangeInclusive<T> {
640 self.clone()
641 }
642
643 fn overlaps(&self, other: &Range<T>) -> bool {
644 self.start() < &other.end && &other.start <= self.end()
645 }
646
647 fn contains_inclusive(&self, other: &Range<T>) -> bool {
648 self.start() <= &other.start && &other.end <= self.end()
649 }
650}
651
652/// A way to sort strings with starting numbers numerically first, falling back to alphanumeric one,
653/// case-insensitive.
654///
655/// This is useful for turning regular alphanumerically sorted sequences as `1-abc, 10, 11-def, .., 2, 21-abc`
656/// into `1-abc, 2, 10, 11-def, .., 21-abc`
657#[derive(Debug, PartialEq, Eq)]
658pub struct NumericPrefixWithSuffix<'a>(Option<u64>, &'a str);
659
660impl<'a> NumericPrefixWithSuffix<'a> {
661 pub fn from_numeric_prefixed_str(str: &'a str) -> Self {
662 let i = str.chars().take_while(|c| c.is_ascii_digit()).count();
663 let (prefix, remainder) = str.split_at(i);
664
665 let prefix = prefix.parse().ok();
666 Self(prefix, remainder)
667 }
668}
669
670/// When dealing with equality, we need to consider the case of the strings to achieve strict equality
671/// to handle cases like "a" < "A" instead of "a" == "A".
672impl Ord for NumericPrefixWithSuffix<'_> {
673 fn cmp(&self, other: &Self) -> Ordering {
674 match (self.0, other.0) {
675 (None, None) => UniCase::new(self.1)
676 .cmp(&UniCase::new(other.1))
677 .then_with(|| self.1.cmp(other.1).reverse()),
678 (None, Some(_)) => Ordering::Greater,
679 (Some(_), None) => Ordering::Less,
680 (Some(a), Some(b)) => a.cmp(&b).then_with(|| {
681 UniCase::new(self.1)
682 .cmp(&UniCase::new(other.1))
683 .then_with(|| self.1.cmp(other.1).reverse())
684 }),
685 }
686 }
687}
688
689impl PartialOrd for NumericPrefixWithSuffix<'_> {
690 fn partial_cmp(&self, other: &Self) -> Option<Ordering> {
691 Some(self.cmp(other))
692 }
693}
694
695/// Capitalizes the first character of a string.
696///
697/// This function takes a string slice as input and returns a new `String` with the first character
698/// capitalized.
699///
700/// # Examples
701///
702/// ```
703/// use util::capitalize;
704///
705/// assert_eq!(capitalize("hello"), "Hello");
706/// assert_eq!(capitalize("WORLD"), "WORLD");
707/// assert_eq!(capitalize(""), "");
708/// ```
709pub fn capitalize(str: &str) -> String {
710 let mut chars = str.chars();
711 match chars.next() {
712 None => String::new(),
713 Some(first_char) => first_char.to_uppercase().collect::<String>() + chars.as_str(),
714 }
715}
716
717fn emoji_regex() -> &'static Regex {
718 static EMOJI_REGEX: LazyLock<Regex> =
719 LazyLock::new(|| Regex::new("(\\p{Emoji}|\u{200D})").unwrap());
720 &EMOJI_REGEX
721}
722
723/// Returns true if the given string consists of emojis only.
724/// E.g. "π¨βπ©βπ§βπ§π" will return true, but "π!" will return false.
725pub fn word_consists_of_emojis(s: &str) -> bool {
726 let mut prev_end = 0;
727 for capture in emoji_regex().find_iter(s) {
728 if capture.start() != prev_end {
729 return false;
730 }
731 prev_end = capture.end();
732 }
733 prev_end == s.len()
734}
735
736/// Similar to `str::split`, but also provides byte-offset ranges of the results. Unlike
737/// `str::split`, this is not generic on pattern types and does not return an `Iterator`.
738pub fn split_str_with_ranges<'s>(
739 s: &'s str,
740 pat: &dyn Fn(char) -> bool,
741) -> Vec<(Range<usize>, &'s str)> {
742 let mut result = Vec::new();
743 let mut start = 0;
744
745 for (i, ch) in s.char_indices() {
746 if pat(ch) {
747 if i > start {
748 result.push((start..i, &s[start..i]));
749 }
750 start = i + ch.len_utf8();
751 }
752 }
753
754 if s.len() > start {
755 result.push((start..s.len(), &s[start..s.len()]));
756 }
757
758 result
759}
760
761pub fn default<D: Default>() -> D {
762 Default::default()
763}
764
765#[derive(Debug)]
766pub enum ConnectionResult<O> {
767 Timeout,
768 ConnectionReset,
769 Result(anyhow::Result<O>),
770}
771
772impl<O> ConnectionResult<O> {
773 pub fn into_response(self) -> anyhow::Result<O> {
774 match self {
775 ConnectionResult::Timeout => anyhow::bail!("Request timed out"),
776 ConnectionResult::ConnectionReset => anyhow::bail!("Server reset the connection"),
777 ConnectionResult::Result(r) => r,
778 }
779 }
780}
781
782impl<O> From<anyhow::Result<O>> for ConnectionResult<O> {
783 fn from(result: anyhow::Result<O>) -> Self {
784 ConnectionResult::Result(result)
785 }
786}
787
788/// Normalizes a path by resolving `.` and `..` components without
789/// requiring the path to exist on disk (unlike `canonicalize`).
790pub fn normalize_path(path: &Path) -> PathBuf {
791 use std::path::Component;
792 let mut components = path.components().peekable();
793 let mut ret = if let Some(c @ Component::Prefix(..)) = components.peek().cloned() {
794 components.next();
795 PathBuf::from(c.as_os_str())
796 } else {
797 PathBuf::new()
798 };
799
800 for component in components {
801 match component {
802 Component::Prefix(..) => unreachable!(),
803 Component::RootDir => {
804 ret.push(component.as_os_str());
805 }
806 Component::CurDir => {}
807 Component::ParentDir => {
808 ret.pop();
809 }
810 Component::Normal(c) => {
811 ret.push(c);
812 }
813 }
814 }
815 ret
816}
817
818#[cfg(test)]
819mod tests {
820 use super::*;
821
822 #[test]
823 fn test_extend_sorted() {
824 let mut vec = vec![];
825
826 extend_sorted(&mut vec, vec![21, 17, 13, 8, 1, 0], 5, |a, b| b.cmp(a));
827 assert_eq!(vec, &[21, 17, 13, 8, 1]);
828
829 extend_sorted(&mut vec, vec![101, 19, 17, 8, 2], 8, |a, b| b.cmp(a));
830 assert_eq!(vec, &[101, 21, 19, 17, 13, 8, 2, 1]);
831
832 extend_sorted(&mut vec, vec![1000, 19, 17, 9, 5], 8, |a, b| b.cmp(a));
833 assert_eq!(vec, &[1000, 101, 21, 19, 17, 13, 9, 8]);
834 }
835
836 #[test]
837 fn test_truncate_to_bottom_n_sorted_by() {
838 let mut vec: Vec<u32> = vec![5, 2, 3, 4, 1];
839 truncate_to_bottom_n_sorted_by(&mut vec, 10, &u32::cmp);
840 assert_eq!(vec, &[1, 2, 3, 4, 5]);
841
842 vec = vec![5, 2, 3, 4, 1];
843 truncate_to_bottom_n_sorted_by(&mut vec, 5, &u32::cmp);
844 assert_eq!(vec, &[1, 2, 3, 4, 5]);
845
846 vec = vec![5, 2, 3, 4, 1];
847 truncate_to_bottom_n_sorted_by(&mut vec, 4, &u32::cmp);
848 assert_eq!(vec, &[1, 2, 3, 4]);
849
850 vec = vec![5, 2, 3, 4, 1];
851 truncate_to_bottom_n_sorted_by(&mut vec, 1, &u32::cmp);
852 assert_eq!(vec, &[1]);
853
854 vec = vec![5, 2, 3, 4, 1];
855 truncate_to_bottom_n_sorted_by(&mut vec, 0, &u32::cmp);
856 assert!(vec.is_empty());
857 }
858
859 #[test]
860 fn test_iife() {
861 fn option_returning_function() -> Option<()> {
862 None
863 }
864
865 let foo = maybe!({
866 option_returning_function()?;
867 Some(())
868 });
869
870 assert_eq!(foo, None);
871 }
872
873 #[test]
874 fn test_truncate_and_trailoff() {
875 assert_eq!(truncate_and_trailoff("", 5), "");
876 assert_eq!(truncate_and_trailoff("aaaaaa", 7), "aaaaaa");
877 assert_eq!(truncate_and_trailoff("aaaaaa", 6), "aaaaaa");
878 assert_eq!(truncate_and_trailoff("aaaaaa", 5), "aaaaaβ¦");
879 assert_eq!(truncate_and_trailoff("èèèèèè", 7), "èèèèèè");
880 assert_eq!(truncate_and_trailoff("èèèèèè", 6), "èèèèèè");
881 assert_eq!(truncate_and_trailoff("èèèèèè", 5), "èèèèèβ¦");
882 }
883
884 #[test]
885 fn test_truncate_and_remove_front() {
886 assert_eq!(truncate_and_remove_front("", 5), "");
887 assert_eq!(truncate_and_remove_front("aaaaaa", 7), "aaaaaa");
888 assert_eq!(truncate_and_remove_front("aaaaaa", 6), "aaaaaa");
889 assert_eq!(truncate_and_remove_front("aaaaaa", 5), "β¦aaaaa");
890 assert_eq!(truncate_and_remove_front("èèèèèè", 7), "èèèèèè");
891 assert_eq!(truncate_and_remove_front("èèèèèè", 6), "èèèèèè");
892 assert_eq!(truncate_and_remove_front("èèèèèè", 5), "β¦Γ¨Γ¨Γ¨Γ¨Γ¨");
893 }
894
895 #[test]
896 fn test_numeric_prefix_str_method() {
897 let target = "1a";
898 assert_eq!(
899 NumericPrefixWithSuffix::from_numeric_prefixed_str(target),
900 NumericPrefixWithSuffix(Some(1), "a")
901 );
902
903 let target = "12ab";
904 assert_eq!(
905 NumericPrefixWithSuffix::from_numeric_prefixed_str(target),
906 NumericPrefixWithSuffix(Some(12), "ab")
907 );
908
909 let target = "12_ab";
910 assert_eq!(
911 NumericPrefixWithSuffix::from_numeric_prefixed_str(target),
912 NumericPrefixWithSuffix(Some(12), "_ab")
913 );
914
915 let target = "1_2ab";
916 assert_eq!(
917 NumericPrefixWithSuffix::from_numeric_prefixed_str(target),
918 NumericPrefixWithSuffix(Some(1), "_2ab")
919 );
920
921 let target = "1.2";
922 assert_eq!(
923 NumericPrefixWithSuffix::from_numeric_prefixed_str(target),
924 NumericPrefixWithSuffix(Some(1), ".2")
925 );
926
927 let target = "1.2_a";
928 assert_eq!(
929 NumericPrefixWithSuffix::from_numeric_prefixed_str(target),
930 NumericPrefixWithSuffix(Some(1), ".2_a")
931 );
932
933 let target = "12.2_a";
934 assert_eq!(
935 NumericPrefixWithSuffix::from_numeric_prefixed_str(target),
936 NumericPrefixWithSuffix(Some(12), ".2_a")
937 );
938
939 let target = "12a.2_a";
940 assert_eq!(
941 NumericPrefixWithSuffix::from_numeric_prefixed_str(target),
942 NumericPrefixWithSuffix(Some(12), "a.2_a")
943 );
944 }
945
946 #[test]
947 fn test_numeric_prefix_with_suffix() {
948 let mut sorted = vec!["1-abc", "10", "11def", "2", "21-abc"];
949 sorted.sort_by_key(|s| NumericPrefixWithSuffix::from_numeric_prefixed_str(s));
950 assert_eq!(sorted, ["1-abc", "2", "10", "11def", "21-abc"]);
951
952 for numeric_prefix_less in ["numeric_prefix_less", "aaa", "~β’Β£"] {
953 assert_eq!(
954 NumericPrefixWithSuffix::from_numeric_prefixed_str(numeric_prefix_less),
955 NumericPrefixWithSuffix(None, numeric_prefix_less),
956 "String without numeric prefix `{numeric_prefix_less}` should not be converted into NumericPrefixWithSuffix"
957 )
958 }
959 }
960
961 #[test]
962 fn test_word_consists_of_emojis() {
963 let words_to_test = vec![
964 ("π¨βπ©βπ§βπ§ππ₯", true),
965 ("π", true),
966 ("!π", false),
967 ("π!", false),
968 ("π ", false),
969 (" π", false),
970 ("Test", false),
971 ];
972
973 for (text, expected_result) in words_to_test {
974 assert_eq!(word_consists_of_emojis(text), expected_result);
975 }
976 }
977
978 #[test]
979 fn test_truncate_lines_and_trailoff() {
980 let text = r#"Line 1
981Line 2
982Line 3"#;
983
984 assert_eq!(
985 truncate_lines_and_trailoff(text, 2),
986 r#"Line 1
987β¦"#
988 );
989
990 assert_eq!(
991 truncate_lines_and_trailoff(text, 3),
992 r#"Line 1
993Line 2
994β¦"#
995 );
996
997 assert_eq!(
998 truncate_lines_and_trailoff(text, 4),
999 r#"Line 1
1000Line 2
1001Line 3"#
1002 );
1003 }
1004
1005 #[test]
1006 fn test_expanded_and_wrapped_usize_range() {
1007 // Neither wrap
1008 assert_eq!(
1009 expanded_and_wrapped_usize_range(2..4, 1, 1, 8).collect::<Vec<usize>>(),
1010 (1..5).collect::<Vec<usize>>()
1011 );
1012 // Start wraps
1013 assert_eq!(
1014 expanded_and_wrapped_usize_range(2..4, 3, 1, 8).collect::<Vec<usize>>(),
1015 ((0..5).chain(7..8)).collect::<Vec<usize>>()
1016 );
1017 // Start wraps all the way around
1018 assert_eq!(
1019 expanded_and_wrapped_usize_range(2..4, 5, 1, 8).collect::<Vec<usize>>(),
1020 (0..8).collect::<Vec<usize>>()
1021 );
1022 // Start wraps all the way around and past 0
1023 assert_eq!(
1024 expanded_and_wrapped_usize_range(2..4, 10, 1, 8).collect::<Vec<usize>>(),
1025 (0..8).collect::<Vec<usize>>()
1026 );
1027 // End wraps
1028 assert_eq!(
1029 expanded_and_wrapped_usize_range(3..5, 1, 4, 8).collect::<Vec<usize>>(),
1030 (0..1).chain(2..8).collect::<Vec<usize>>()
1031 );
1032 // End wraps all the way around
1033 assert_eq!(
1034 expanded_and_wrapped_usize_range(3..5, 1, 5, 8).collect::<Vec<usize>>(),
1035 (0..8).collect::<Vec<usize>>()
1036 );
1037 // End wraps all the way around and past the end
1038 assert_eq!(
1039 expanded_and_wrapped_usize_range(3..5, 1, 10, 8).collect::<Vec<usize>>(),
1040 (0..8).collect::<Vec<usize>>()
1041 );
1042 // Both start and end wrap
1043 assert_eq!(
1044 expanded_and_wrapped_usize_range(3..5, 4, 4, 8).collect::<Vec<usize>>(),
1045 (0..8).collect::<Vec<usize>>()
1046 );
1047 }
1048
1049 #[test]
1050 fn test_wrapped_usize_outward_from() {
1051 // No wrapping
1052 assert_eq!(
1053 wrapped_usize_outward_from(4, 2, 2, 10).collect::<Vec<usize>>(),
1054 vec![4, 5, 3, 6, 2]
1055 );
1056 // Wrapping at end
1057 assert_eq!(
1058 wrapped_usize_outward_from(8, 2, 3, 10).collect::<Vec<usize>>(),
1059 vec![8, 9, 7, 0, 6, 1]
1060 );
1061 // Wrapping at start
1062 assert_eq!(
1063 wrapped_usize_outward_from(1, 3, 2, 10).collect::<Vec<usize>>(),
1064 vec![1, 2, 0, 3, 9, 8]
1065 );
1066 // All values wrap around
1067 assert_eq!(
1068 wrapped_usize_outward_from(5, 10, 10, 8).collect::<Vec<usize>>(),
1069 vec![5, 6, 4, 7, 3, 0, 2, 1]
1070 );
1071 // None before / after
1072 assert_eq!(
1073 wrapped_usize_outward_from(3, 0, 0, 8).collect::<Vec<usize>>(),
1074 vec![3]
1075 );
1076 // Starting point already wrapped
1077 assert_eq!(
1078 wrapped_usize_outward_from(15, 2, 2, 10).collect::<Vec<usize>>(),
1079 vec![5, 6, 4, 7, 3]
1080 );
1081 // wrap_length of 0
1082 assert_eq!(
1083 wrapped_usize_outward_from(4, 2, 2, 0).collect::<Vec<usize>>(),
1084 Vec::<usize>::new()
1085 );
1086 }
1087
1088 #[test]
1089 fn test_split_with_ranges() {
1090 let input = "hi";
1091 let result = split_str_with_ranges(input, &|c| c == ' ');
1092
1093 assert_eq!(result.len(), 1);
1094 assert_eq!(result[0], (0..2, "hi"));
1095
1096 let input = "hΓ©lloπ¦world";
1097 let result = split_str_with_ranges(input, &|c| c == 'π¦');
1098
1099 assert_eq!(result.len(), 2);
1100 assert_eq!(result[0], (0..6, "hΓ©llo")); // 'Γ©' is 2 bytes
1101 assert_eq!(result[1], (10..15, "world")); // 'π¦' is 4 bytes
1102 }
1103}