util.rs

   1pub mod arc_cow;
   2pub mod archive;
   3pub mod command;
   4pub mod fs;
   5pub mod markdown;
   6pub mod paths;
   7pub mod serde;
   8pub mod shell_env;
   9pub mod size;
  10#[cfg(any(test, feature = "test-support"))]
  11pub mod test;
  12pub mod time;
  13
  14use anyhow::Result;
  15use futures::Future;
  16use itertools::Either;
  17use regex::Regex;
  18use std::sync::{LazyLock, OnceLock};
  19use std::{
  20    borrow::Cow,
  21    cmp::{self, Ordering},
  22    env,
  23    ops::{AddAssign, Range, RangeInclusive},
  24    panic::Location,
  25    pin::Pin,
  26    task::{Context, Poll},
  27    time::Instant,
  28};
  29use unicase::UniCase;
  30
  31pub use take_until::*;
  32#[cfg(any(test, feature = "test-support"))]
  33pub use util_macros::{line_endings, path, uri};
  34
  35#[macro_export]
  36macro_rules! debug_panic {
  37    ( $($fmt_arg:tt)* ) => {
  38        if cfg!(debug_assertions) {
  39            panic!( $($fmt_arg)* );
  40        } else {
  41            let backtrace = std::backtrace::Backtrace::capture();
  42            log::error!("{}\n{:?}", format_args!($($fmt_arg)*), backtrace);
  43        }
  44    };
  45}
  46
  47pub fn truncate(s: &str, max_chars: usize) -> &str {
  48    match s.char_indices().nth(max_chars) {
  49        None => s,
  50        Some((idx, _)) => &s[..idx],
  51    }
  52}
  53
  54/// Removes characters from the end of the string if its length is greater than `max_chars` and
  55/// appends "..." to the string. Returns string unchanged if its length is smaller than max_chars.
  56pub fn truncate_and_trailoff(s: &str, max_chars: usize) -> String {
  57    debug_assert!(max_chars >= 5);
  58
  59    // If the string's byte length is <= max_chars, walking the string can be skipped since the
  60    // number of chars is <= the number of bytes.
  61    if s.len() <= max_chars {
  62        return s.to_string();
  63    }
  64    let truncation_ix = s.char_indices().map(|(i, _)| i).nth(max_chars);
  65    match truncation_ix {
  66        Some(index) => s[..index].to_string() + "…",
  67        _ => s.to_string(),
  68    }
  69}
  70
  71/// Removes characters from the front of the string if its length is greater than `max_chars` and
  72/// prepends the string with "...". Returns string unchanged if its length is smaller than max_chars.
  73pub fn truncate_and_remove_front(s: &str, max_chars: usize) -> String {
  74    debug_assert!(max_chars >= 5);
  75
  76    // If the string's byte length is <= max_chars, walking the string can be skipped since the
  77    // number of chars is <= the number of bytes.
  78    if s.len() <= max_chars {
  79        return s.to_string();
  80    }
  81    let suffix_char_length = max_chars.saturating_sub(1);
  82    let truncation_ix = s
  83        .char_indices()
  84        .map(|(i, _)| i)
  85        .nth_back(suffix_char_length);
  86    match truncation_ix {
  87        Some(index) if index > 0 => "…".to_string() + &s[index..],
  88        _ => s.to_string(),
  89    }
  90}
  91
  92/// Takes only `max_lines` from the string and, if there were more than `max_lines-1`, appends a
  93/// a newline and "..." to the string, so that `max_lines` are returned.
  94/// Returns string unchanged if its length is smaller than max_lines.
  95pub fn truncate_lines_and_trailoff(s: &str, max_lines: usize) -> String {
  96    let mut lines = s.lines().take(max_lines).collect::<Vec<_>>();
  97    if lines.len() > max_lines - 1 {
  98        lines.pop();
  99        lines.join("\n") + "\n…"
 100    } else {
 101        lines.join("\n")
 102    }
 103}
 104
 105/// Truncates the string at a character boundary, such that the result is less than `max_bytes` in
 106/// length.
 107pub fn truncate_to_byte_limit(s: &str, max_bytes: usize) -> &str {
 108    if s.len() < max_bytes {
 109        return s;
 110    }
 111
 112    for i in (0..max_bytes).rev() {
 113        if s.is_char_boundary(i) {
 114            return &s[..i];
 115        }
 116    }
 117
 118    ""
 119}
 120
 121/// Takes a prefix of complete lines which fit within the byte limit. If the first line is longer
 122/// than the limit, truncates at a character boundary.
 123pub fn truncate_lines_to_byte_limit(s: &str, max_bytes: usize) -> &str {
 124    if s.len() < max_bytes {
 125        return s;
 126    }
 127
 128    for i in (0..max_bytes).rev() {
 129        if s.is_char_boundary(i) {
 130            if s.as_bytes()[i] == b'\n' {
 131                // Since the i-th character is \n, valid to slice at i + 1.
 132                return &s[..i + 1];
 133            }
 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
 165pub fn post_inc<T: From<u8> + AddAssign<T> + Copy>(value: &mut T) -> T {
 166    let prev = *value;
 167    *value += T::from(1);
 168    prev
 169}
 170
 171/// Extend a sorted vector with a sorted sequence of items, maintaining the vector's sort order and
 172/// enforcing a maximum length. This also de-duplicates items. Sort the items according to the given callback. Before calling this,
 173/// both `vec` and `new_items` should already be sorted according to the `cmp` comparator.
 174pub fn extend_sorted<T, I, F>(vec: &mut Vec<T>, new_items: I, limit: usize, mut cmp: F)
 175where
 176    I: IntoIterator<Item = T>,
 177    F: FnMut(&T, &T) -> Ordering,
 178{
 179    let mut start_index = 0;
 180    for new_item in new_items {
 181        if let Err(i) = vec[start_index..].binary_search_by(|m| cmp(m, &new_item)) {
 182            let index = start_index + i;
 183            if vec.len() < limit {
 184                vec.insert(index, new_item);
 185            } else if index < vec.len() {
 186                vec.pop();
 187                vec.insert(index, new_item);
 188            }
 189            start_index = index;
 190        }
 191    }
 192}
 193
 194pub fn truncate_to_bottom_n_sorted_by<T, F>(items: &mut Vec<T>, limit: usize, compare: &F)
 195where
 196    F: Fn(&T, &T) -> Ordering,
 197{
 198    if limit == 0 {
 199        items.truncate(0);
 200    }
 201    if items.len() <= limit {
 202        items.sort_by(compare);
 203        return;
 204    }
 205    // When limit is near to items.len() it may be more efficient to sort the whole list and
 206    // truncate, rather than always doing selection first as is done below. It's hard to analyze
 207    // where the threshold for this should be since the quickselect style algorithm used by
 208    // `select_nth_unstable_by` makes the prefix partially sorted, and so its work is not wasted -
 209    // the expected number of comparisons needed by `sort_by` is less than it is for some arbitrary
 210    // unsorted input.
 211    items.select_nth_unstable_by(limit, compare);
 212    items.truncate(limit);
 213    items.sort_by(compare);
 214}
 215
 216#[cfg(unix)]
 217fn load_shell_from_passwd() -> Result<()> {
 218    let buflen = match unsafe { libc::sysconf(libc::_SC_GETPW_R_SIZE_MAX) } {
 219        n if n < 0 => 1024,
 220        n => n as usize,
 221    };
 222    let mut buffer = Vec::with_capacity(buflen);
 223
 224    let mut pwd: std::mem::MaybeUninit<libc::passwd> = std::mem::MaybeUninit::uninit();
 225    let mut result: *mut libc::passwd = std::ptr::null_mut();
 226
 227    let uid = unsafe { libc::getuid() };
 228    let status = unsafe {
 229        libc::getpwuid_r(
 230            uid,
 231            pwd.as_mut_ptr(),
 232            buffer.as_mut_ptr() as *mut libc::c_char,
 233            buflen,
 234            &mut result,
 235        )
 236    };
 237    let entry = unsafe { pwd.assume_init() };
 238
 239    anyhow::ensure!(
 240        status == 0,
 241        "call to getpwuid_r failed. uid: {}, status: {}",
 242        uid,
 243        status
 244    );
 245    anyhow::ensure!(!result.is_null(), "passwd entry for uid {} not found", uid);
 246    anyhow::ensure!(
 247        entry.pw_uid == uid,
 248        "passwd entry has different uid ({}) than getuid ({}) returned",
 249        entry.pw_uid,
 250        uid,
 251    );
 252
 253    let shell = unsafe { std::ffi::CStr::from_ptr(entry.pw_shell).to_str().unwrap() };
 254    if env::var("SHELL").map_or(true, |shell_env| shell_env != shell) {
 255        log::info!(
 256            "updating SHELL environment variable to value from passwd entry: {:?}",
 257            shell,
 258        );
 259        unsafe { env::set_var("SHELL", shell) };
 260    }
 261
 262    Ok(())
 263}
 264
 265#[cfg(unix)]
 266/// Returns a shell escaped path for the current zed executable
 267pub fn get_shell_safe_zed_path() -> anyhow::Result<String> {
 268    use anyhow::Context;
 269
 270    let zed_path = std::env::current_exe()
 271        .context("Failed to determine current zed executable path.")?
 272        .to_string_lossy()
 273        .trim_end_matches(" (deleted)") // see https://github.com/rust-lang/rust/issues/69343
 274        .to_string();
 275
 276    // As of writing, this can only be fail if the path contains a null byte, which shouldn't be possible
 277    // but shlex has annotated the error as #[non_exhaustive] so we can't make it a compile error if other
 278    // errors are introduced in the future :(
 279    let zed_path_escaped =
 280        shlex::try_quote(&zed_path).context("Failed to shell-escape Zed executable path.")?;
 281
 282    return Ok(zed_path_escaped.to_string());
 283}
 284
 285#[cfg(unix)]
 286pub fn load_login_shell_environment() -> Result<()> {
 287    load_shell_from_passwd().log_err();
 288
 289    // If possible, we want to `cd` in the user's `$HOME` to trigger programs
 290    // such as direnv, asdf, mise, ... to adjust the PATH. These tools often hook
 291    // into shell's `cd` command (and hooks) to manipulate env.
 292    // We do this so that we get the env a user would have when spawning a shell
 293    // in home directory.
 294    for (name, value) in shell_env::capture(paths::home_dir())? {
 295        unsafe { env::set_var(&name, &value) };
 296    }
 297
 298    log::info!(
 299        "set environment variables from shell:{}, path:{}",
 300        std::env::var("SHELL").unwrap_or_default(),
 301        std::env::var("PATH").unwrap_or_default(),
 302    );
 303
 304    Ok(())
 305}
 306
 307/// Configures the process to start a new session, to prevent interactive shells from taking control
 308/// of the terminal.
 309///
 310/// For more details: https://registerspill.thorstenball.com/p/how-to-lose-control-of-your-shell
 311pub fn set_pre_exec_to_start_new_session(
 312    command: &mut std::process::Command,
 313) -> &mut std::process::Command {
 314    // safety: code in pre_exec should be signal safe.
 315    // https://man7.org/linux/man-pages/man7/signal-safety.7.html
 316    #[cfg(not(target_os = "windows"))]
 317    unsafe {
 318        use std::os::unix::process::CommandExt;
 319        command.pre_exec(|| {
 320            libc::setsid();
 321            Ok(())
 322        });
 323    };
 324    command
 325}
 326
 327pub fn merge_json_lenient_value_into(
 328    source: serde_json_lenient::Value,
 329    target: &mut serde_json_lenient::Value,
 330) {
 331    match (source, target) {
 332        (serde_json_lenient::Value::Object(source), serde_json_lenient::Value::Object(target)) => {
 333            for (key, value) in source {
 334                if let Some(target) = target.get_mut(&key) {
 335                    merge_json_lenient_value_into(value, target);
 336                } else {
 337                    target.insert(key, value);
 338                }
 339            }
 340        }
 341
 342        (serde_json_lenient::Value::Array(source), serde_json_lenient::Value::Array(target)) => {
 343            for value in source {
 344                target.push(value);
 345            }
 346        }
 347
 348        (source, target) => *target = source,
 349    }
 350}
 351
 352pub fn merge_json_value_into(source: serde_json::Value, target: &mut serde_json::Value) {
 353    use serde_json::Value;
 354
 355    match (source, target) {
 356        (Value::Object(source), Value::Object(target)) => {
 357            for (key, value) in source {
 358                if let Some(target) = target.get_mut(&key) {
 359                    merge_json_value_into(value, target);
 360                } else {
 361                    target.insert(key, value);
 362                }
 363            }
 364        }
 365
 366        (Value::Array(source), Value::Array(target)) => {
 367            for value in source {
 368                target.push(value);
 369            }
 370        }
 371
 372        (source, target) => *target = source,
 373    }
 374}
 375
 376pub fn merge_non_null_json_value_into(source: serde_json::Value, target: &mut serde_json::Value) {
 377    use serde_json::Value;
 378    if let Value::Object(source_object) = source {
 379        let target_object = if let Value::Object(target) = target {
 380            target
 381        } else {
 382            *target = Value::Object(Default::default());
 383            target.as_object_mut().unwrap()
 384        };
 385        for (key, value) in source_object {
 386            if let Some(target) = target_object.get_mut(&key) {
 387                merge_non_null_json_value_into(value, target);
 388            } else if !value.is_null() {
 389                target_object.insert(key, value);
 390            }
 391        }
 392    } else if !source.is_null() {
 393        *target = source
 394    }
 395}
 396
 397pub fn measure<R>(label: &str, f: impl FnOnce() -> R) -> R {
 398    static ZED_MEASUREMENTS: OnceLock<bool> = OnceLock::new();
 399    let zed_measurements = ZED_MEASUREMENTS.get_or_init(|| {
 400        env::var("ZED_MEASUREMENTS")
 401            .map(|measurements| measurements == "1" || measurements == "true")
 402            .unwrap_or(false)
 403    });
 404
 405    if *zed_measurements {
 406        let start = Instant::now();
 407        let result = f();
 408        let elapsed = start.elapsed();
 409        eprintln!("{}: {:?}", label, elapsed);
 410        result
 411    } else {
 412        f()
 413    }
 414}
 415
 416pub fn expanded_and_wrapped_usize_range(
 417    range: Range<usize>,
 418    additional_before: usize,
 419    additional_after: usize,
 420    wrap_length: usize,
 421) -> impl Iterator<Item = usize> {
 422    let start_wraps = range.start < additional_before;
 423    let end_wraps = wrap_length < range.end + additional_after;
 424    if start_wraps && end_wraps {
 425        Either::Left(0..wrap_length)
 426    } else if start_wraps {
 427        let wrapped_start = (range.start + wrap_length).saturating_sub(additional_before);
 428        if wrapped_start <= range.end {
 429            Either::Left(0..wrap_length)
 430        } else {
 431            Either::Right((0..range.end + additional_after).chain(wrapped_start..wrap_length))
 432        }
 433    } else if end_wraps {
 434        let wrapped_end = range.end + additional_after - wrap_length;
 435        if range.start <= wrapped_end {
 436            Either::Left(0..wrap_length)
 437        } else {
 438            Either::Right((0..wrapped_end).chain(range.start - additional_before..wrap_length))
 439        }
 440    } else {
 441        Either::Left((range.start - additional_before)..(range.end + additional_after))
 442    }
 443}
 444
 445/// Yields `[i, i + 1, i - 1, i + 2, ..]`, each modulo `wrap_length` and bounded by
 446/// `additional_before` and `additional_after`. If the wrapping causes overlap, duplicates are not
 447/// emitted. If wrap_length is 0, nothing is yielded.
 448pub fn wrapped_usize_outward_from(
 449    start: usize,
 450    additional_before: usize,
 451    additional_after: usize,
 452    wrap_length: usize,
 453) -> impl Iterator<Item = usize> {
 454    let mut count = 0;
 455    let mut after_offset = 1;
 456    let mut before_offset = 1;
 457
 458    std::iter::from_fn(move || {
 459        count += 1;
 460        if count > wrap_length {
 461            None
 462        } else if count == 1 {
 463            Some(start % wrap_length)
 464        } else if after_offset <= additional_after && after_offset <= before_offset {
 465            let value = (start + after_offset) % wrap_length;
 466            after_offset += 1;
 467            Some(value)
 468        } else if before_offset <= additional_before {
 469            let value = (start + wrap_length - before_offset) % wrap_length;
 470            before_offset += 1;
 471            Some(value)
 472        } else if after_offset <= additional_after {
 473            let value = (start + after_offset) % wrap_length;
 474            after_offset += 1;
 475            Some(value)
 476        } else {
 477            None
 478        }
 479    })
 480}
 481
 482#[cfg(target_os = "windows")]
 483pub fn get_windows_system_shell() -> String {
 484    use std::path::PathBuf;
 485
 486    fn find_pwsh_in_programfiles(find_alternate: bool, find_preview: bool) -> Option<PathBuf> {
 487        #[cfg(target_pointer_width = "64")]
 488        let env_var = if find_alternate {
 489            "ProgramFiles(x86)"
 490        } else {
 491            "ProgramFiles"
 492        };
 493
 494        #[cfg(target_pointer_width = "32")]
 495        let env_var = if find_alternate {
 496            "ProgramW6432"
 497        } else {
 498            "ProgramFiles"
 499        };
 500
 501        let install_base_dir = PathBuf::from(std::env::var_os(env_var)?).join("PowerShell");
 502        install_base_dir
 503            .read_dir()
 504            .ok()?
 505            .filter_map(Result::ok)
 506            .filter(|entry| matches!(entry.file_type(), Ok(ft) if ft.is_dir()))
 507            .filter_map(|entry| {
 508                let dir_name = entry.file_name();
 509                let dir_name = dir_name.to_string_lossy();
 510
 511                let version = if find_preview {
 512                    let dash_index = dir_name.find('-')?;
 513                    if &dir_name[dash_index + 1..] != "preview" {
 514                        return None;
 515                    };
 516                    dir_name[..dash_index].parse::<u32>().ok()?
 517                } else {
 518                    dir_name.parse::<u32>().ok()?
 519                };
 520
 521                let exe_path = entry.path().join("pwsh.exe");
 522                if exe_path.exists() {
 523                    Some((version, exe_path))
 524                } else {
 525                    None
 526                }
 527            })
 528            .max_by_key(|(version, _)| *version)
 529            .map(|(_, path)| path)
 530    }
 531
 532    fn find_pwsh_in_msix(find_preview: bool) -> Option<PathBuf> {
 533        let msix_app_dir =
 534            PathBuf::from(std::env::var_os("LOCALAPPDATA")?).join("Microsoft\\WindowsApps");
 535        if !msix_app_dir.exists() {
 536            return None;
 537        }
 538
 539        let prefix = if find_preview {
 540            "Microsoft.PowerShellPreview_"
 541        } else {
 542            "Microsoft.PowerShell_"
 543        };
 544        msix_app_dir
 545            .read_dir()
 546            .ok()?
 547            .filter_map(|entry| {
 548                let entry = entry.ok()?;
 549                if !matches!(entry.file_type(), Ok(ft) if ft.is_dir()) {
 550                    return None;
 551                }
 552
 553                if !entry.file_name().to_string_lossy().starts_with(prefix) {
 554                    return None;
 555                }
 556
 557                let exe_path = entry.path().join("pwsh.exe");
 558                exe_path.exists().then_some(exe_path)
 559            })
 560            .next()
 561    }
 562
 563    fn find_pwsh_in_scoop() -> Option<PathBuf> {
 564        let pwsh_exe =
 565            PathBuf::from(std::env::var_os("USERPROFILE")?).join("scoop\\shims\\pwsh.exe");
 566        pwsh_exe.exists().then_some(pwsh_exe)
 567    }
 568
 569    static SYSTEM_SHELL: LazyLock<String> = LazyLock::new(|| {
 570        find_pwsh_in_programfiles(false, false)
 571            .or_else(|| find_pwsh_in_programfiles(true, false))
 572            .or_else(|| find_pwsh_in_msix(false))
 573            .or_else(|| find_pwsh_in_programfiles(false, true))
 574            .or_else(|| find_pwsh_in_msix(true))
 575            .or_else(|| find_pwsh_in_programfiles(true, true))
 576            .or_else(find_pwsh_in_scoop)
 577            .map(|p| p.to_string_lossy().to_string())
 578            .unwrap_or("powershell.exe".to_string())
 579    });
 580
 581    (*SYSTEM_SHELL).clone()
 582}
 583
 584pub trait ResultExt<E> {
 585    type Ok;
 586
 587    fn log_err(self) -> Option<Self::Ok>;
 588    /// Assert that this result should never be an error in development or tests.
 589    fn debug_assert_ok(self, reason: &str) -> Self;
 590    fn warn_on_err(self) -> Option<Self::Ok>;
 591    fn log_with_level(self, level: log::Level) -> Option<Self::Ok>;
 592    fn anyhow(self) -> anyhow::Result<Self::Ok>
 593    where
 594        E: Into<anyhow::Error>;
 595}
 596
 597impl<T, E> ResultExt<E> for Result<T, E>
 598where
 599    E: std::fmt::Debug,
 600{
 601    type Ok = T;
 602
 603    #[track_caller]
 604    fn log_err(self) -> Option<T> {
 605        self.log_with_level(log::Level::Error)
 606    }
 607
 608    #[track_caller]
 609    fn debug_assert_ok(self, reason: &str) -> Self {
 610        if let Err(error) = &self {
 611            debug_panic!("{reason} - {error:?}");
 612        }
 613        self
 614    }
 615
 616    #[track_caller]
 617    fn warn_on_err(self) -> Option<T> {
 618        self.log_with_level(log::Level::Warn)
 619    }
 620
 621    #[track_caller]
 622    fn log_with_level(self, level: log::Level) -> Option<T> {
 623        match self {
 624            Ok(value) => Some(value),
 625            Err(error) => {
 626                log_error_with_caller(*Location::caller(), error, level);
 627                None
 628            }
 629        }
 630    }
 631
 632    fn anyhow(self) -> anyhow::Result<T>
 633    where
 634        E: Into<anyhow::Error>,
 635    {
 636        self.map_err(Into::into)
 637    }
 638}
 639
 640fn log_error_with_caller<E>(caller: core::panic::Location<'_>, error: E, level: log::Level)
 641where
 642    E: std::fmt::Debug,
 643{
 644    #[cfg(not(target_os = "windows"))]
 645    let file = caller.file();
 646    #[cfg(target_os = "windows")]
 647    let file = caller.file().replace('\\', "/");
 648    // In this codebase, the first segment of the file path is
 649    // the 'crates' folder, followed by the crate name.
 650    let target = file.split('/').nth(1);
 651
 652    log::logger().log(
 653        &log::Record::builder()
 654            .target(target.unwrap_or(""))
 655            .module_path(target)
 656            .args(format_args!("{:?}", error))
 657            .file(Some(caller.file()))
 658            .line(Some(caller.line()))
 659            .level(level)
 660            .build(),
 661    );
 662}
 663
 664pub fn log_err<E: std::fmt::Debug>(error: &E) {
 665    log_error_with_caller(*Location::caller(), error, log::Level::Warn);
 666}
 667
 668pub trait TryFutureExt {
 669    fn log_err(self) -> LogErrorFuture<Self>
 670    where
 671        Self: Sized;
 672
 673    fn log_tracked_err(self, location: core::panic::Location<'static>) -> LogErrorFuture<Self>
 674    where
 675        Self: Sized;
 676
 677    fn warn_on_err(self) -> LogErrorFuture<Self>
 678    where
 679        Self: Sized;
 680    fn unwrap(self) -> UnwrapFuture<Self>
 681    where
 682        Self: Sized;
 683}
 684
 685impl<F, T, E> TryFutureExt for F
 686where
 687    F: Future<Output = Result<T, E>>,
 688    E: std::fmt::Debug,
 689{
 690    #[track_caller]
 691    fn log_err(self) -> LogErrorFuture<Self>
 692    where
 693        Self: Sized,
 694    {
 695        let location = Location::caller();
 696        LogErrorFuture(self, log::Level::Error, *location)
 697    }
 698
 699    fn log_tracked_err(self, location: core::panic::Location<'static>) -> LogErrorFuture<Self>
 700    where
 701        Self: Sized,
 702    {
 703        LogErrorFuture(self, log::Level::Error, location)
 704    }
 705
 706    #[track_caller]
 707    fn warn_on_err(self) -> LogErrorFuture<Self>
 708    where
 709        Self: Sized,
 710    {
 711        let location = Location::caller();
 712        LogErrorFuture(self, log::Level::Warn, *location)
 713    }
 714
 715    fn unwrap(self) -> UnwrapFuture<Self>
 716    where
 717        Self: Sized,
 718    {
 719        UnwrapFuture(self)
 720    }
 721}
 722
 723#[must_use]
 724pub struct LogErrorFuture<F>(F, log::Level, core::panic::Location<'static>);
 725
 726impl<F, T, E> Future for LogErrorFuture<F>
 727where
 728    F: Future<Output = Result<T, E>>,
 729    E: std::fmt::Debug,
 730{
 731    type Output = Option<T>;
 732
 733    fn poll(self: Pin<&mut Self>, cx: &mut Context) -> Poll<Self::Output> {
 734        let level = self.1;
 735        let location = self.2;
 736        let inner = unsafe { Pin::new_unchecked(&mut self.get_unchecked_mut().0) };
 737        match inner.poll(cx) {
 738            Poll::Ready(output) => Poll::Ready(match output {
 739                Ok(output) => Some(output),
 740                Err(error) => {
 741                    log_error_with_caller(location, error, level);
 742                    None
 743                }
 744            }),
 745            Poll::Pending => Poll::Pending,
 746        }
 747    }
 748}
 749
 750pub struct UnwrapFuture<F>(F);
 751
 752impl<F, T, E> Future for UnwrapFuture<F>
 753where
 754    F: Future<Output = Result<T, E>>,
 755    E: std::fmt::Debug,
 756{
 757    type Output = T;
 758
 759    fn poll(self: Pin<&mut Self>, cx: &mut Context) -> Poll<Self::Output> {
 760        let inner = unsafe { Pin::new_unchecked(&mut self.get_unchecked_mut().0) };
 761        match inner.poll(cx) {
 762            Poll::Ready(result) => Poll::Ready(result.unwrap()),
 763            Poll::Pending => Poll::Pending,
 764        }
 765    }
 766}
 767
 768pub struct Deferred<F: FnOnce()>(Option<F>);
 769
 770impl<F: FnOnce()> Deferred<F> {
 771    /// Drop without running the deferred function.
 772    pub fn abort(mut self) {
 773        self.0.take();
 774    }
 775}
 776
 777impl<F: FnOnce()> Drop for Deferred<F> {
 778    fn drop(&mut self) {
 779        if let Some(f) = self.0.take() {
 780            f()
 781        }
 782    }
 783}
 784
 785/// Run the given function when the returned value is dropped (unless it's cancelled).
 786#[must_use]
 787pub fn defer<F: FnOnce()>(f: F) -> Deferred<F> {
 788    Deferred(Some(f))
 789}
 790
 791#[cfg(any(test, feature = "test-support"))]
 792mod rng {
 793    use rand::{Rng, seq::SliceRandom};
 794    pub struct RandomCharIter<T: Rng> {
 795        rng: T,
 796        simple_text: bool,
 797    }
 798
 799    impl<T: Rng> RandomCharIter<T> {
 800        pub fn new(rng: T) -> Self {
 801            Self {
 802                rng,
 803                simple_text: std::env::var("SIMPLE_TEXT").map_or(false, |v| !v.is_empty()),
 804            }
 805        }
 806
 807        pub fn with_simple_text(mut self) -> Self {
 808            self.simple_text = true;
 809            self
 810        }
 811    }
 812
 813    impl<T: Rng> Iterator for RandomCharIter<T> {
 814        type Item = char;
 815
 816        fn next(&mut self) -> Option<Self::Item> {
 817            if self.simple_text {
 818                return if self.rng.gen_range(0..100) < 5 {
 819                    Some('\n')
 820                } else {
 821                    Some(self.rng.gen_range(b'a'..b'z' + 1).into())
 822                };
 823            }
 824
 825            match self.rng.gen_range(0..100) {
 826                // whitespace
 827                0..=19 => [' ', '\n', '\r', '\t'].choose(&mut self.rng).copied(),
 828                // two-byte greek letters
 829                20..=32 => char::from_u32(self.rng.gen_range(('Ξ±' as u32)..('Ο‰' as u32 + 1))),
 830                // // three-byte characters
 831                33..=45 => ['βœ‹', 'βœ…', '❌', '❎', '⭐']
 832                    .choose(&mut self.rng)
 833                    .copied(),
 834                // // four-byte characters
 835                46..=58 => ['🍐', 'πŸ€', 'πŸ—', 'πŸŽ‰'].choose(&mut self.rng).copied(),
 836                // ascii letters
 837                _ => Some(self.rng.gen_range(b'a'..b'z' + 1).into()),
 838            }
 839        }
 840    }
 841}
 842#[cfg(any(test, feature = "test-support"))]
 843pub use rng::RandomCharIter;
 844/// Get an embedded file as a string.
 845pub fn asset_str<A: rust_embed::RustEmbed>(path: &str) -> Cow<'static, str> {
 846    match A::get(path).expect(path).data {
 847        Cow::Borrowed(bytes) => Cow::Borrowed(std::str::from_utf8(bytes).unwrap()),
 848        Cow::Owned(bytes) => Cow::Owned(String::from_utf8(bytes).unwrap()),
 849    }
 850}
 851
 852/// Expands to an immediately-invoked function expression. Good for using the ? operator
 853/// in functions which do not return an Option or Result.
 854///
 855/// Accepts a normal block, an async block, or an async move block.
 856#[macro_export]
 857macro_rules! maybe {
 858    ($block:block) => {
 859        (|| $block)()
 860    };
 861    (async $block:block) => {
 862        (|| async $block)()
 863    };
 864    (async move $block:block) => {
 865        (|| async move $block)()
 866    };
 867}
 868
 869pub trait RangeExt<T> {
 870    fn sorted(&self) -> Self;
 871    fn to_inclusive(&self) -> RangeInclusive<T>;
 872    fn overlaps(&self, other: &Range<T>) -> bool;
 873    fn contains_inclusive(&self, other: &Range<T>) -> bool;
 874}
 875
 876impl<T: Ord + Clone> RangeExt<T> for Range<T> {
 877    fn sorted(&self) -> Self {
 878        cmp::min(&self.start, &self.end).clone()..cmp::max(&self.start, &self.end).clone()
 879    }
 880
 881    fn to_inclusive(&self) -> RangeInclusive<T> {
 882        self.start.clone()..=self.end.clone()
 883    }
 884
 885    fn overlaps(&self, other: &Range<T>) -> bool {
 886        self.start < other.end && other.start < self.end
 887    }
 888
 889    fn contains_inclusive(&self, other: &Range<T>) -> bool {
 890        self.start <= other.start && other.end <= self.end
 891    }
 892}
 893
 894impl<T: Ord + Clone> RangeExt<T> for RangeInclusive<T> {
 895    fn sorted(&self) -> Self {
 896        cmp::min(self.start(), self.end()).clone()..=cmp::max(self.start(), self.end()).clone()
 897    }
 898
 899    fn to_inclusive(&self) -> RangeInclusive<T> {
 900        self.clone()
 901    }
 902
 903    fn overlaps(&self, other: &Range<T>) -> bool {
 904        self.start() < &other.end && &other.start <= self.end()
 905    }
 906
 907    fn contains_inclusive(&self, other: &Range<T>) -> bool {
 908        self.start() <= &other.start && &other.end <= self.end()
 909    }
 910}
 911
 912/// A way to sort strings with starting numbers numerically first, falling back to alphanumeric one,
 913/// case-insensitive.
 914///
 915/// This is useful for turning regular alphanumerically sorted sequences as `1-abc, 10, 11-def, .., 2, 21-abc`
 916/// into `1-abc, 2, 10, 11-def, .., 21-abc`
 917#[derive(Debug, PartialEq, Eq)]
 918pub struct NumericPrefixWithSuffix<'a>(Option<u64>, &'a str);
 919
 920impl<'a> NumericPrefixWithSuffix<'a> {
 921    pub fn from_numeric_prefixed_str(str: &'a str) -> Self {
 922        let i = str.chars().take_while(|c| c.is_ascii_digit()).count();
 923        let (prefix, remainder) = str.split_at(i);
 924
 925        let prefix = prefix.parse().ok();
 926        Self(prefix, remainder)
 927    }
 928}
 929
 930/// When dealing with equality, we need to consider the case of the strings to achieve strict equality
 931/// to handle cases like "a" < "A" instead of "a" == "A".
 932impl Ord for NumericPrefixWithSuffix<'_> {
 933    fn cmp(&self, other: &Self) -> Ordering {
 934        match (self.0, other.0) {
 935            (None, None) => UniCase::new(self.1)
 936                .cmp(&UniCase::new(other.1))
 937                .then_with(|| self.1.cmp(other.1).reverse()),
 938            (None, Some(_)) => Ordering::Greater,
 939            (Some(_), None) => Ordering::Less,
 940            (Some(a), Some(b)) => a.cmp(&b).then_with(|| {
 941                UniCase::new(self.1)
 942                    .cmp(&UniCase::new(other.1))
 943                    .then_with(|| self.1.cmp(other.1).reverse())
 944            }),
 945        }
 946    }
 947}
 948
 949impl PartialOrd for NumericPrefixWithSuffix<'_> {
 950    fn partial_cmp(&self, other: &Self) -> Option<Ordering> {
 951        Some(self.cmp(other))
 952    }
 953}
 954
 955/// Capitalizes the first character of a string.
 956///
 957/// This function takes a string slice as input and returns a new `String` with the first character
 958/// capitalized.
 959///
 960/// # Examples
 961///
 962/// ```
 963/// use util::capitalize;
 964///
 965/// assert_eq!(capitalize("hello"), "Hello");
 966/// assert_eq!(capitalize("WORLD"), "WORLD");
 967/// assert_eq!(capitalize(""), "");
 968/// ```
 969pub fn capitalize(str: &str) -> String {
 970    let mut chars = str.chars();
 971    match chars.next() {
 972        None => String::new(),
 973        Some(first_char) => first_char.to_uppercase().collect::<String>() + chars.as_str(),
 974    }
 975}
 976
 977fn emoji_regex() -> &'static Regex {
 978    static EMOJI_REGEX: LazyLock<Regex> =
 979        LazyLock::new(|| Regex::new("(\\p{Emoji}|\u{200D})").unwrap());
 980    &EMOJI_REGEX
 981}
 982
 983/// Returns true if the given string consists of emojis only.
 984/// E.g. "πŸ‘¨β€πŸ‘©β€πŸ‘§β€πŸ‘§πŸ‘‹" will return true, but "πŸ‘‹!" will return false.
 985pub fn word_consists_of_emojis(s: &str) -> bool {
 986    let mut prev_end = 0;
 987    for capture in emoji_regex().find_iter(s) {
 988        if capture.start() != prev_end {
 989            return false;
 990        }
 991        prev_end = capture.end();
 992    }
 993    prev_end == s.len()
 994}
 995
 996/// Similar to `str::split`, but also provides byte-offset ranges of the results. Unlike
 997/// `str::split`, this is not generic on pattern types and does not return an `Iterator`.
 998pub fn split_str_with_ranges(s: &str, pat: impl Fn(char) -> bool) -> Vec<(Range<usize>, &str)> {
 999    let mut result = Vec::new();
1000    let mut start = 0;
1001
1002    for (i, ch) in s.char_indices() {
1003        if pat(ch) {
1004            if i > start {
1005                result.push((start..i, &s[start..i]));
1006            }
1007            start = i + ch.len_utf8();
1008        }
1009    }
1010
1011    if s.len() > start {
1012        result.push((start..s.len(), &s[start..s.len()]));
1013    }
1014
1015    result
1016}
1017
1018pub fn default<D: Default>() -> D {
1019    Default::default()
1020}
1021
1022pub fn get_system_shell() -> String {
1023    #[cfg(target_os = "windows")]
1024    {
1025        get_windows_system_shell()
1026    }
1027
1028    #[cfg(not(target_os = "windows"))]
1029    {
1030        std::env::var("SHELL").unwrap_or("/bin/sh".to_string())
1031    }
1032}
1033
1034#[derive(Debug)]
1035pub enum ConnectionResult<O> {
1036    Timeout,
1037    ConnectionReset,
1038    Result(anyhow::Result<O>),
1039}
1040
1041impl<O> ConnectionResult<O> {
1042    pub fn into_response(self) -> anyhow::Result<O> {
1043        match self {
1044            ConnectionResult::Timeout => anyhow::bail!("Request timed out"),
1045            ConnectionResult::ConnectionReset => anyhow::bail!("Server reset the connection"),
1046            ConnectionResult::Result(r) => r,
1047        }
1048    }
1049}
1050
1051impl<O> From<anyhow::Result<O>> for ConnectionResult<O> {
1052    fn from(result: anyhow::Result<O>) -> Self {
1053        ConnectionResult::Result(result)
1054    }
1055}
1056
1057#[cfg(test)]
1058mod tests {
1059    use super::*;
1060
1061    #[test]
1062    fn test_extend_sorted() {
1063        let mut vec = vec![];
1064
1065        extend_sorted(&mut vec, vec![21, 17, 13, 8, 1, 0], 5, |a, b| b.cmp(a));
1066        assert_eq!(vec, &[21, 17, 13, 8, 1]);
1067
1068        extend_sorted(&mut vec, vec![101, 19, 17, 8, 2], 8, |a, b| b.cmp(a));
1069        assert_eq!(vec, &[101, 21, 19, 17, 13, 8, 2, 1]);
1070
1071        extend_sorted(&mut vec, vec![1000, 19, 17, 9, 5], 8, |a, b| b.cmp(a));
1072        assert_eq!(vec, &[1000, 101, 21, 19, 17, 13, 9, 8]);
1073    }
1074
1075    #[test]
1076    fn test_truncate_to_bottom_n_sorted_by() {
1077        let mut vec: Vec<u32> = vec![5, 2, 3, 4, 1];
1078        truncate_to_bottom_n_sorted_by(&mut vec, 10, &u32::cmp);
1079        assert_eq!(vec, &[1, 2, 3, 4, 5]);
1080
1081        vec = vec![5, 2, 3, 4, 1];
1082        truncate_to_bottom_n_sorted_by(&mut vec, 5, &u32::cmp);
1083        assert_eq!(vec, &[1, 2, 3, 4, 5]);
1084
1085        vec = vec![5, 2, 3, 4, 1];
1086        truncate_to_bottom_n_sorted_by(&mut vec, 4, &u32::cmp);
1087        assert_eq!(vec, &[1, 2, 3, 4]);
1088
1089        vec = vec![5, 2, 3, 4, 1];
1090        truncate_to_bottom_n_sorted_by(&mut vec, 1, &u32::cmp);
1091        assert_eq!(vec, &[1]);
1092
1093        vec = vec![5, 2, 3, 4, 1];
1094        truncate_to_bottom_n_sorted_by(&mut vec, 0, &u32::cmp);
1095        assert!(vec.is_empty());
1096    }
1097
1098    #[test]
1099    fn test_iife() {
1100        fn option_returning_function() -> Option<()> {
1101            None
1102        }
1103
1104        let foo = maybe!({
1105            option_returning_function()?;
1106            Some(())
1107        });
1108
1109        assert_eq!(foo, None);
1110    }
1111
1112    #[test]
1113    fn test_truncate_and_trailoff() {
1114        assert_eq!(truncate_and_trailoff("", 5), "");
1115        assert_eq!(truncate_and_trailoff("aaaaaa", 7), "aaaaaa");
1116        assert_eq!(truncate_and_trailoff("aaaaaa", 6), "aaaaaa");
1117        assert_eq!(truncate_and_trailoff("aaaaaa", 5), "aaaaa…");
1118        assert_eq!(truncate_and_trailoff("èèèèèè", 7), "èèèèèè");
1119        assert_eq!(truncate_and_trailoff("èèèèèè", 6), "èèèèèè");
1120        assert_eq!(truncate_and_trailoff("èèèèèè", 5), "èèèèè…");
1121    }
1122
1123    #[test]
1124    fn test_truncate_and_remove_front() {
1125        assert_eq!(truncate_and_remove_front("", 5), "");
1126        assert_eq!(truncate_and_remove_front("aaaaaa", 7), "aaaaaa");
1127        assert_eq!(truncate_and_remove_front("aaaaaa", 6), "aaaaaa");
1128        assert_eq!(truncate_and_remove_front("aaaaaa", 5), "…aaaaa");
1129        assert_eq!(truncate_and_remove_front("èèèèèè", 7), "èèèèèè");
1130        assert_eq!(truncate_and_remove_front("èèèèèè", 6), "èèèèèè");
1131        assert_eq!(truncate_and_remove_front("èèèèèè", 5), "…èèèèè");
1132    }
1133
1134    #[test]
1135    fn test_numeric_prefix_str_method() {
1136        let target = "1a";
1137        assert_eq!(
1138            NumericPrefixWithSuffix::from_numeric_prefixed_str(target),
1139            NumericPrefixWithSuffix(Some(1), "a")
1140        );
1141
1142        let target = "12ab";
1143        assert_eq!(
1144            NumericPrefixWithSuffix::from_numeric_prefixed_str(target),
1145            NumericPrefixWithSuffix(Some(12), "ab")
1146        );
1147
1148        let target = "12_ab";
1149        assert_eq!(
1150            NumericPrefixWithSuffix::from_numeric_prefixed_str(target),
1151            NumericPrefixWithSuffix(Some(12), "_ab")
1152        );
1153
1154        let target = "1_2ab";
1155        assert_eq!(
1156            NumericPrefixWithSuffix::from_numeric_prefixed_str(target),
1157            NumericPrefixWithSuffix(Some(1), "_2ab")
1158        );
1159
1160        let target = "1.2";
1161        assert_eq!(
1162            NumericPrefixWithSuffix::from_numeric_prefixed_str(target),
1163            NumericPrefixWithSuffix(Some(1), ".2")
1164        );
1165
1166        let target = "1.2_a";
1167        assert_eq!(
1168            NumericPrefixWithSuffix::from_numeric_prefixed_str(target),
1169            NumericPrefixWithSuffix(Some(1), ".2_a")
1170        );
1171
1172        let target = "12.2_a";
1173        assert_eq!(
1174            NumericPrefixWithSuffix::from_numeric_prefixed_str(target),
1175            NumericPrefixWithSuffix(Some(12), ".2_a")
1176        );
1177
1178        let target = "12a.2_a";
1179        assert_eq!(
1180            NumericPrefixWithSuffix::from_numeric_prefixed_str(target),
1181            NumericPrefixWithSuffix(Some(12), "a.2_a")
1182        );
1183    }
1184
1185    #[test]
1186    fn test_numeric_prefix_with_suffix() {
1187        let mut sorted = vec!["1-abc", "10", "11def", "2", "21-abc"];
1188        sorted.sort_by_key(|s| NumericPrefixWithSuffix::from_numeric_prefixed_str(s));
1189        assert_eq!(sorted, ["1-abc", "2", "10", "11def", "21-abc"]);
1190
1191        for numeric_prefix_less in ["numeric_prefix_less", "aaa", "~β„’Β£"] {
1192            assert_eq!(
1193                NumericPrefixWithSuffix::from_numeric_prefixed_str(numeric_prefix_less),
1194                NumericPrefixWithSuffix(None, numeric_prefix_less),
1195                "String without numeric prefix `{numeric_prefix_less}` should not be converted into NumericPrefixWithSuffix"
1196            )
1197        }
1198    }
1199
1200    #[test]
1201    fn test_word_consists_of_emojis() {
1202        let words_to_test = vec![
1203            ("πŸ‘¨β€πŸ‘©β€πŸ‘§β€πŸ‘§πŸ‘‹πŸ₯’", true),
1204            ("πŸ‘‹", true),
1205            ("!πŸ‘‹", false),
1206            ("πŸ‘‹!", false),
1207            ("πŸ‘‹ ", false),
1208            (" πŸ‘‹", false),
1209            ("Test", false),
1210        ];
1211
1212        for (text, expected_result) in words_to_test {
1213            assert_eq!(word_consists_of_emojis(text), expected_result);
1214        }
1215    }
1216
1217    #[test]
1218    fn test_truncate_lines_and_trailoff() {
1219        let text = r#"Line 1
1220Line 2
1221Line 3"#;
1222
1223        assert_eq!(
1224            truncate_lines_and_trailoff(text, 2),
1225            r#"Line 1
1226…"#
1227        );
1228
1229        assert_eq!(
1230            truncate_lines_and_trailoff(text, 3),
1231            r#"Line 1
1232Line 2
1233…"#
1234        );
1235
1236        assert_eq!(
1237            truncate_lines_and_trailoff(text, 4),
1238            r#"Line 1
1239Line 2
1240Line 3"#
1241        );
1242    }
1243
1244    #[test]
1245    fn test_expanded_and_wrapped_usize_range() {
1246        // Neither wrap
1247        assert_eq!(
1248            expanded_and_wrapped_usize_range(2..4, 1, 1, 8).collect::<Vec<usize>>(),
1249            (1..5).collect::<Vec<usize>>()
1250        );
1251        // Start wraps
1252        assert_eq!(
1253            expanded_and_wrapped_usize_range(2..4, 3, 1, 8).collect::<Vec<usize>>(),
1254            ((0..5).chain(7..8)).collect::<Vec<usize>>()
1255        );
1256        // Start wraps all the way around
1257        assert_eq!(
1258            expanded_and_wrapped_usize_range(2..4, 5, 1, 8).collect::<Vec<usize>>(),
1259            (0..8).collect::<Vec<usize>>()
1260        );
1261        // Start wraps all the way around and past 0
1262        assert_eq!(
1263            expanded_and_wrapped_usize_range(2..4, 10, 1, 8).collect::<Vec<usize>>(),
1264            (0..8).collect::<Vec<usize>>()
1265        );
1266        // End wraps
1267        assert_eq!(
1268            expanded_and_wrapped_usize_range(3..5, 1, 4, 8).collect::<Vec<usize>>(),
1269            (0..1).chain(2..8).collect::<Vec<usize>>()
1270        );
1271        // End wraps all the way around
1272        assert_eq!(
1273            expanded_and_wrapped_usize_range(3..5, 1, 5, 8).collect::<Vec<usize>>(),
1274            (0..8).collect::<Vec<usize>>()
1275        );
1276        // End wraps all the way around and past the end
1277        assert_eq!(
1278            expanded_and_wrapped_usize_range(3..5, 1, 10, 8).collect::<Vec<usize>>(),
1279            (0..8).collect::<Vec<usize>>()
1280        );
1281        // Both start and end wrap
1282        assert_eq!(
1283            expanded_and_wrapped_usize_range(3..5, 4, 4, 8).collect::<Vec<usize>>(),
1284            (0..8).collect::<Vec<usize>>()
1285        );
1286    }
1287
1288    #[test]
1289    fn test_wrapped_usize_outward_from() {
1290        // No wrapping
1291        assert_eq!(
1292            wrapped_usize_outward_from(4, 2, 2, 10).collect::<Vec<usize>>(),
1293            vec![4, 5, 3, 6, 2]
1294        );
1295        // Wrapping at end
1296        assert_eq!(
1297            wrapped_usize_outward_from(8, 2, 3, 10).collect::<Vec<usize>>(),
1298            vec![8, 9, 7, 0, 6, 1]
1299        );
1300        // Wrapping at start
1301        assert_eq!(
1302            wrapped_usize_outward_from(1, 3, 2, 10).collect::<Vec<usize>>(),
1303            vec![1, 2, 0, 3, 9, 8]
1304        );
1305        // All values wrap around
1306        assert_eq!(
1307            wrapped_usize_outward_from(5, 10, 10, 8).collect::<Vec<usize>>(),
1308            vec![5, 6, 4, 7, 3, 0, 2, 1]
1309        );
1310        // None before / after
1311        assert_eq!(
1312            wrapped_usize_outward_from(3, 0, 0, 8).collect::<Vec<usize>>(),
1313            vec![3]
1314        );
1315        // Starting point already wrapped
1316        assert_eq!(
1317            wrapped_usize_outward_from(15, 2, 2, 10).collect::<Vec<usize>>(),
1318            vec![5, 6, 4, 7, 3]
1319        );
1320        // wrap_length of 0
1321        assert_eq!(
1322            wrapped_usize_outward_from(4, 2, 2, 0).collect::<Vec<usize>>(),
1323            Vec::<usize>::new()
1324        );
1325    }
1326
1327    #[test]
1328    fn test_split_with_ranges() {
1329        let input = "hi";
1330        let result = split_str_with_ranges(input, |c| c == ' ');
1331
1332        assert_eq!(result.len(), 1);
1333        assert_eq!(result[0], (0..2, "hi"));
1334
1335        let input = "hΓ©lloπŸ¦€world";
1336        let result = split_str_with_ranges(input, |c| c == 'πŸ¦€');
1337
1338        assert_eq!(result.len(), 2);
1339        assert_eq!(result[0], (0..6, "hΓ©llo")); // 'Γ©' is 2 bytes
1340        assert_eq!(result[1], (10..15, "world")); // 'πŸ¦€' is 4 bytes
1341    }
1342}