util.rs

   1pub mod arc_cow;
   2pub mod archive;
   3pub mod command;
   4pub mod fs;
   5pub mod markdown;
   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::{Context as _, Result};
  21use futures::Future;
  22use itertools::Either;
  23use paths::PathExt;
  24use regex::Regex;
  25use std::path::PathBuf;
  26use std::sync::{LazyLock, OnceLock};
  27use std::{
  28    borrow::Cow,
  29    cmp::{self, Ordering},
  30    env,
  31    ops::{AddAssign, Range, RangeInclusive},
  32    panic::Location,
  33    pin::Pin,
  34    task::{Context, Poll},
  35    time::Instant,
  36};
  37use unicase::UniCase;
  38
  39pub use take_until::*;
  40#[cfg(any(test, feature = "test-support"))]
  41pub use util_macros::{line_endings, path, uri};
  42
  43#[macro_export]
  44macro_rules! debug_panic {
  45    ( $($fmt_arg:tt)* ) => {
  46        if cfg!(debug_assertions) {
  47            panic!( $($fmt_arg)* );
  48        } else {
  49            let backtrace = std::backtrace::Backtrace::capture();
  50            log::error!("{}\n{:?}", format_args!($($fmt_arg)*), backtrace);
  51        }
  52    };
  53}
  54
  55#[inline]
  56pub const fn is_utf8_char_boundary(u8: u8) -> bool {
  57    // This is bit magic equivalent to: b < 128 || b >= 192
  58    (u8 as i8) >= -0x40
  59}
  60
  61pub fn truncate(s: &str, max_chars: usize) -> &str {
  62    match s.char_indices().nth(max_chars) {
  63        None => s,
  64        Some((idx, _)) => &s[..idx],
  65    }
  66}
  67
  68/// Removes characters from the end of the string if its length is greater than `max_chars` and
  69/// appends "..." to the string. Returns string unchanged if its length is smaller than max_chars.
  70pub fn truncate_and_trailoff(s: &str, max_chars: usize) -> String {
  71    debug_assert!(max_chars >= 5);
  72
  73    // If the string's byte length is <= max_chars, walking the string can be skipped since the
  74    // number of chars is <= the number of bytes.
  75    if s.len() <= max_chars {
  76        return s.to_string();
  77    }
  78    let truncation_ix = s.char_indices().map(|(i, _)| i).nth(max_chars);
  79    match truncation_ix {
  80        Some(index) => s[..index].to_string() + "…",
  81        _ => s.to_string(),
  82    }
  83}
  84
  85/// Removes characters from the front of the string if its length is greater than `max_chars` and
  86/// prepends the string with "...". Returns string unchanged if its length is smaller than max_chars.
  87pub fn truncate_and_remove_front(s: &str, max_chars: usize) -> String {
  88    debug_assert!(max_chars >= 5);
  89
  90    // If the string's byte length is <= max_chars, walking the string can be skipped since the
  91    // number of chars is <= the number of bytes.
  92    if s.len() <= max_chars {
  93        return s.to_string();
  94    }
  95    let suffix_char_length = max_chars.saturating_sub(1);
  96    let truncation_ix = s
  97        .char_indices()
  98        .map(|(i, _)| i)
  99        .nth_back(suffix_char_length);
 100    match truncation_ix {
 101        Some(index) if index > 0 => "…".to_string() + &s[index..],
 102        _ => s.to_string(),
 103    }
 104}
 105
 106/// Takes only `max_lines` from the string and, if there were more than `max_lines-1`, appends a
 107/// a newline and "..." to the string, so that `max_lines` are returned.
 108/// Returns string unchanged if its length is smaller than max_lines.
 109pub fn truncate_lines_and_trailoff(s: &str, max_lines: usize) -> String {
 110    let mut lines = s.lines().take(max_lines).collect::<Vec<_>>();
 111    if lines.len() > max_lines - 1 {
 112        lines.pop();
 113        lines.join("\n") + "\n…"
 114    } else {
 115        lines.join("\n")
 116    }
 117}
 118
 119/// Truncates the string at a character boundary, such that the result is less than `max_bytes` in
 120/// length.
 121pub fn truncate_to_byte_limit(s: &str, max_bytes: usize) -> &str {
 122    if s.len() < max_bytes {
 123        return s;
 124    }
 125
 126    for i in (0..max_bytes).rev() {
 127        if s.is_char_boundary(i) {
 128            return &s[..i];
 129        }
 130    }
 131
 132    ""
 133}
 134
 135/// Takes a prefix of complete lines which fit within the byte limit. If the first line is longer
 136/// than the limit, truncates at a character boundary.
 137pub fn truncate_lines_to_byte_limit(s: &str, max_bytes: usize) -> &str {
 138    if s.len() < max_bytes {
 139        return s;
 140    }
 141
 142    for i in (0..max_bytes).rev() {
 143        if s.is_char_boundary(i) && s.as_bytes()[i] == b'\n' {
 144            // Since the i-th character is \n, valid to slice at i + 1.
 145            return &s[..i + 1];
 146        }
 147    }
 148
 149    truncate_to_byte_limit(s, max_bytes)
 150}
 151
 152#[test]
 153fn test_truncate_lines_to_byte_limit() {
 154    let text = "Line 1\nLine 2\nLine 3\nLine 4";
 155
 156    // Limit that includes all lines
 157    assert_eq!(truncate_lines_to_byte_limit(text, 100), text);
 158
 159    // Exactly the first line
 160    assert_eq!(truncate_lines_to_byte_limit(text, 7), "Line 1\n");
 161
 162    // Limit between lines
 163    assert_eq!(truncate_lines_to_byte_limit(text, 13), "Line 1\n");
 164    assert_eq!(truncate_lines_to_byte_limit(text, 20), "Line 1\nLine 2\n");
 165
 166    // Limit before first newline
 167    assert_eq!(truncate_lines_to_byte_limit(text, 6), "Line ");
 168
 169    // Test with non-ASCII characters
 170    let text_utf8 = "Line 1\nLΓ­ne 2\nLine 3";
 171    assert_eq!(
 172        truncate_lines_to_byte_limit(text_utf8, 15),
 173        "Line 1\nLΓ­ne 2\n"
 174    );
 175}
 176
 177pub fn post_inc<T: From<u8> + AddAssign<T> + Copy>(value: &mut T) -> T {
 178    let prev = *value;
 179    *value += T::from(1);
 180    prev
 181}
 182
 183/// Extend a sorted vector with a sorted sequence of items, maintaining the vector's sort order and
 184/// enforcing a maximum length. This also de-duplicates items. Sort the items according to the given callback. Before calling this,
 185/// both `vec` and `new_items` should already be sorted according to the `cmp` comparator.
 186pub fn extend_sorted<T, I, F>(vec: &mut Vec<T>, new_items: I, limit: usize, mut cmp: F)
 187where
 188    I: IntoIterator<Item = T>,
 189    F: FnMut(&T, &T) -> Ordering,
 190{
 191    let mut start_index = 0;
 192    for new_item in new_items {
 193        if let Err(i) = vec[start_index..].binary_search_by(|m| cmp(m, &new_item)) {
 194            let index = start_index + i;
 195            if vec.len() < limit {
 196                vec.insert(index, new_item);
 197            } else if index < vec.len() {
 198                vec.pop();
 199                vec.insert(index, new_item);
 200            }
 201            start_index = index;
 202        }
 203    }
 204}
 205
 206pub fn truncate_to_bottom_n_sorted_by<T, F>(items: &mut Vec<T>, limit: usize, compare: &F)
 207where
 208    F: Fn(&T, &T) -> Ordering,
 209{
 210    if limit == 0 {
 211        items.truncate(0);
 212    }
 213    if items.len() <= limit {
 214        items.sort_by(compare);
 215        return;
 216    }
 217    // When limit is near to items.len() it may be more efficient to sort the whole list and
 218    // truncate, rather than always doing selection first as is done below. It's hard to analyze
 219    // where the threshold for this should be since the quickselect style algorithm used by
 220    // `select_nth_unstable_by` makes the prefix partially sorted, and so its work is not wasted -
 221    // the expected number of comparisons needed by `sort_by` is less than it is for some arbitrary
 222    // unsorted input.
 223    items.select_nth_unstable_by(limit, compare);
 224    items.truncate(limit);
 225    items.sort_by(compare);
 226}
 227
 228/// Prevents execution of the application with root privileges on Unix systems.
 229///
 230/// This function checks if the current process is running with root privileges
 231/// and terminates the program with an error message unless explicitly allowed via the
 232/// `ZED_ALLOW_ROOT` environment variable.
 233#[cfg(unix)]
 234pub fn prevent_root_execution() {
 235    let is_root = nix::unistd::geteuid().is_root();
 236    let allow_root = std::env::var("ZED_ALLOW_ROOT").is_ok_and(|val| val == "true");
 237
 238    if is_root && !allow_root {
 239        eprintln!(
 240            "\
 241Error: Running Zed as root or via sudo is unsupported.
 242       Doing so (even once) may subtly break things for all subsequent non-root usage of Zed.
 243       It is untested and not recommended, don't complain when things break.
 244       If you wish to proceed anyways, set `ZED_ALLOW_ROOT=true` in your environment."
 245        );
 246        std::process::exit(1);
 247    }
 248}
 249
 250#[cfg(unix)]
 251fn load_shell_from_passwd() -> Result<()> {
 252    let buflen = match unsafe { libc::sysconf(libc::_SC_GETPW_R_SIZE_MAX) } {
 253        n if n < 0 => 1024,
 254        n => n as usize,
 255    };
 256    let mut buffer = Vec::with_capacity(buflen);
 257
 258    let mut pwd: std::mem::MaybeUninit<libc::passwd> = std::mem::MaybeUninit::uninit();
 259    let mut result: *mut libc::passwd = std::ptr::null_mut();
 260
 261    let uid = unsafe { libc::getuid() };
 262    let status = unsafe {
 263        libc::getpwuid_r(
 264            uid,
 265            pwd.as_mut_ptr(),
 266            buffer.as_mut_ptr() as *mut libc::c_char,
 267            buflen,
 268            &mut result,
 269        )
 270    };
 271    anyhow::ensure!(!result.is_null(), "passwd entry for uid {} not found", uid);
 272
 273    // SAFETY: If `getpwuid_r` doesn't error, we have the entry here.
 274    let entry = unsafe { pwd.assume_init() };
 275
 276    anyhow::ensure!(
 277        status == 0,
 278        "call to getpwuid_r failed. uid: {}, status: {}",
 279        uid,
 280        status
 281    );
 282    anyhow::ensure!(
 283        entry.pw_uid == uid,
 284        "passwd entry has different uid ({}) than getuid ({}) returned",
 285        entry.pw_uid,
 286        uid,
 287    );
 288
 289    let shell = unsafe { std::ffi::CStr::from_ptr(entry.pw_shell).to_str().unwrap() };
 290    let should_set_shell = env::var("SHELL").map_or(true, |shell_env| {
 291        shell_env != shell && !std::path::Path::new(&shell_env).exists()
 292    });
 293
 294    if should_set_shell {
 295        log::info!(
 296            "updating SHELL environment variable to value from passwd entry: {:?}",
 297            shell,
 298        );
 299        unsafe { env::set_var("SHELL", shell) };
 300    }
 301
 302    Ok(())
 303}
 304
 305/// Returns a shell escaped path for the current zed executable
 306pub fn get_shell_safe_zed_path(shell_kind: shell::ShellKind) -> anyhow::Result<String> {
 307    let zed_path =
 308        std::env::current_exe().context("Failed to determine current zed executable path.")?;
 309
 310    zed_path
 311        .try_shell_safe(shell_kind)
 312        .context("Failed to shell-escape Zed executable path.")
 313}
 314
 315/// Returns a path for the zed cli executable, this function
 316/// should be called from the zed executable, not zed-cli.
 317pub fn get_zed_cli_path() -> Result<PathBuf> {
 318    let zed_path =
 319        std::env::current_exe().context("Failed to determine current zed executable path.")?;
 320    let parent = zed_path
 321        .parent()
 322        .context("Failed to determine parent directory of zed executable path.")?;
 323
 324    let possible_locations: &[&str] = if cfg!(target_os = "macos") {
 325        // On macOS, the zed executable and zed-cli are inside the app bundle,
 326        // so here ./cli is for both installed and development builds.
 327        &["./cli"]
 328    } else if cfg!(target_os = "windows") {
 329        // bin/zed.exe is for installed builds, ./cli.exe is for development builds.
 330        &["bin/zed.exe", "./cli.exe"]
 331    } else if cfg!(target_os = "linux") || cfg!(target_os = "freebsd") {
 332        // bin is the standard, ./cli is for the target directory in development builds.
 333        &["../bin/zed", "./cli"]
 334    } else {
 335        anyhow::bail!("unsupported platform for determining zed-cli path");
 336    };
 337
 338    possible_locations
 339        .iter()
 340        .find_map(|p| {
 341            parent
 342                .join(p)
 343                .canonicalize()
 344                .ok()
 345                .filter(|p| p != &zed_path)
 346        })
 347        .with_context(|| {
 348            format!(
 349                "could not find zed-cli from any of: {}",
 350                possible_locations.join(", ")
 351            )
 352        })
 353}
 354
 355#[cfg(unix)]
 356pub async fn load_login_shell_environment() -> Result<()> {
 357    load_shell_from_passwd().log_err();
 358
 359    // If possible, we want to `cd` in the user's `$HOME` to trigger programs
 360    // such as direnv, asdf, mise, ... to adjust the PATH. These tools often hook
 361    // into shell's `cd` command (and hooks) to manipulate env.
 362    // We do this so that we get the env a user would have when spawning a shell
 363    // in home directory.
 364    for (name, value) in shell_env::capture(get_system_shell(), &[], paths::home_dir())
 365        .await
 366        .with_context(|| format!("capturing environment with {:?}", get_system_shell()))?
 367    {
 368        // Skip SHLVL to prevent it from polluting Zed's process environment.
 369        // The login shell used for env capture increments SHLVL, and if we propagate it,
 370        // terminals spawned by Zed will inherit it and increment again, causing SHLVL
 371        // to start at 2 instead of 1 (and increase by 2 on each reload).
 372        if name == "SHLVL" {
 373            continue;
 374        }
 375        unsafe { env::set_var(&name, &value) };
 376    }
 377
 378    log::info!(
 379        "set environment variables from shell:{}, path:{}",
 380        std::env::var("SHELL").unwrap_or_default(),
 381        std::env::var("PATH").unwrap_or_default(),
 382    );
 383
 384    Ok(())
 385}
 386
 387/// Configures the process to start a new session, to prevent interactive shells from taking control
 388/// of the terminal.
 389///
 390/// For more details: <https://registerspill.thorstenball.com/p/how-to-lose-control-of-your-shell>
 391pub fn set_pre_exec_to_start_new_session(
 392    command: &mut std::process::Command,
 393) -> &mut std::process::Command {
 394    // safety: code in pre_exec should be signal safe.
 395    // https://man7.org/linux/man-pages/man7/signal-safety.7.html
 396    #[cfg(not(target_os = "windows"))]
 397    unsafe {
 398        use std::os::unix::process::CommandExt;
 399        command.pre_exec(|| {
 400            libc::setsid();
 401            #[cfg(target_os = "macos")]
 402            crate::command::reset_exception_ports();
 403            Ok(())
 404        });
 405    };
 406    command
 407}
 408
 409pub fn merge_json_lenient_value_into(
 410    source: serde_json_lenient::Value,
 411    target: &mut serde_json_lenient::Value,
 412) {
 413    match (source, target) {
 414        (serde_json_lenient::Value::Object(source), serde_json_lenient::Value::Object(target)) => {
 415            for (key, value) in source {
 416                if let Some(target) = target.get_mut(&key) {
 417                    merge_json_lenient_value_into(value, target);
 418                } else {
 419                    target.insert(key, value);
 420                }
 421            }
 422        }
 423
 424        (serde_json_lenient::Value::Array(source), serde_json_lenient::Value::Array(target)) => {
 425            for value in source {
 426                target.push(value);
 427            }
 428        }
 429
 430        (source, target) => *target = source,
 431    }
 432}
 433
 434pub fn merge_json_value_into(source: serde_json::Value, target: &mut serde_json::Value) {
 435    use serde_json::Value;
 436
 437    match (source, target) {
 438        (Value::Object(source), Value::Object(target)) => {
 439            for (key, value) in source {
 440                if let Some(target) = target.get_mut(&key) {
 441                    merge_json_value_into(value, target);
 442                } else {
 443                    target.insert(key, value);
 444                }
 445            }
 446        }
 447
 448        (Value::Array(source), Value::Array(target)) => {
 449            for value in source {
 450                target.push(value);
 451            }
 452        }
 453
 454        (source, target) => *target = source,
 455    }
 456}
 457
 458pub fn merge_non_null_json_value_into(source: serde_json::Value, target: &mut serde_json::Value) {
 459    use serde_json::Value;
 460    if let Value::Object(source_object) = source {
 461        let target_object = if let Value::Object(target) = target {
 462            target
 463        } else {
 464            *target = Value::Object(Default::default());
 465            target.as_object_mut().unwrap()
 466        };
 467        for (key, value) in source_object {
 468            if let Some(target) = target_object.get_mut(&key) {
 469                merge_non_null_json_value_into(value, target);
 470            } else if !value.is_null() {
 471                target_object.insert(key, value);
 472            }
 473        }
 474    } else if !source.is_null() {
 475        *target = source
 476    }
 477}
 478
 479pub fn measure<R>(label: &str, f: impl FnOnce() -> R) -> R {
 480    static ZED_MEASUREMENTS: OnceLock<bool> = OnceLock::new();
 481    let zed_measurements = ZED_MEASUREMENTS.get_or_init(|| {
 482        env::var("ZED_MEASUREMENTS")
 483            .map(|measurements| measurements == "1" || measurements == "true")
 484            .unwrap_or(false)
 485    });
 486
 487    if *zed_measurements {
 488        let start = Instant::now();
 489        let result = f();
 490        let elapsed = start.elapsed();
 491        eprintln!("{}: {:?}", label, elapsed);
 492        result
 493    } else {
 494        f()
 495    }
 496}
 497
 498pub fn expanded_and_wrapped_usize_range(
 499    range: Range<usize>,
 500    additional_before: usize,
 501    additional_after: usize,
 502    wrap_length: usize,
 503) -> impl Iterator<Item = usize> {
 504    let start_wraps = range.start < additional_before;
 505    let end_wraps = wrap_length < range.end + additional_after;
 506    if start_wraps && end_wraps {
 507        Either::Left(0..wrap_length)
 508    } else if start_wraps {
 509        let wrapped_start = (range.start + wrap_length).saturating_sub(additional_before);
 510        if wrapped_start <= range.end {
 511            Either::Left(0..wrap_length)
 512        } else {
 513            Either::Right((0..range.end + additional_after).chain(wrapped_start..wrap_length))
 514        }
 515    } else if end_wraps {
 516        let wrapped_end = range.end + additional_after - wrap_length;
 517        if range.start <= wrapped_end {
 518            Either::Left(0..wrap_length)
 519        } else {
 520            Either::Right((0..wrapped_end).chain(range.start - additional_before..wrap_length))
 521        }
 522    } else {
 523        Either::Left((range.start - additional_before)..(range.end + additional_after))
 524    }
 525}
 526
 527/// Yields `[i, i + 1, i - 1, i + 2, ..]`, each modulo `wrap_length` and bounded by
 528/// `additional_before` and `additional_after`. If the wrapping causes overlap, duplicates are not
 529/// emitted. If wrap_length is 0, nothing is yielded.
 530pub fn wrapped_usize_outward_from(
 531    start: usize,
 532    additional_before: usize,
 533    additional_after: usize,
 534    wrap_length: usize,
 535) -> impl Iterator<Item = usize> {
 536    let mut count = 0;
 537    let mut after_offset = 1;
 538    let mut before_offset = 1;
 539
 540    std::iter::from_fn(move || {
 541        count += 1;
 542        if count > wrap_length {
 543            None
 544        } else if count == 1 {
 545            Some(start % wrap_length)
 546        } else if after_offset <= additional_after && after_offset <= before_offset {
 547            let value = (start + after_offset) % wrap_length;
 548            after_offset += 1;
 549            Some(value)
 550        } else if before_offset <= additional_before {
 551            let value = (start + wrap_length - before_offset) % wrap_length;
 552            before_offset += 1;
 553            Some(value)
 554        } else if after_offset <= additional_after {
 555            let value = (start + after_offset) % wrap_length;
 556            after_offset += 1;
 557            Some(value)
 558        } else {
 559            None
 560        }
 561    })
 562}
 563
 564pub trait ResultExt<E> {
 565    type Ok;
 566
 567    fn log_err(self) -> Option<Self::Ok>;
 568    /// Assert that this result should never be an error in development or tests.
 569    fn debug_assert_ok(self, reason: &str) -> Self;
 570    fn warn_on_err(self) -> Option<Self::Ok>;
 571    fn log_with_level(self, level: log::Level) -> Option<Self::Ok>;
 572    fn anyhow(self) -> anyhow::Result<Self::Ok>
 573    where
 574        E: Into<anyhow::Error>;
 575}
 576
 577impl<T, E> ResultExt<E> for Result<T, E>
 578where
 579    E: std::fmt::Debug,
 580{
 581    type Ok = T;
 582
 583    #[track_caller]
 584    fn log_err(self) -> Option<T> {
 585        self.log_with_level(log::Level::Error)
 586    }
 587
 588    #[track_caller]
 589    fn debug_assert_ok(self, reason: &str) -> Self {
 590        if let Err(error) = &self {
 591            debug_panic!("{reason} - {error:?}");
 592        }
 593        self
 594    }
 595
 596    #[track_caller]
 597    fn warn_on_err(self) -> Option<T> {
 598        self.log_with_level(log::Level::Warn)
 599    }
 600
 601    #[track_caller]
 602    fn log_with_level(self, level: log::Level) -> Option<T> {
 603        match self {
 604            Ok(value) => Some(value),
 605            Err(error) => {
 606                log_error_with_caller(*Location::caller(), error, level);
 607                None
 608            }
 609        }
 610    }
 611
 612    fn anyhow(self) -> anyhow::Result<T>
 613    where
 614        E: Into<anyhow::Error>,
 615    {
 616        self.map_err(Into::into)
 617    }
 618}
 619
 620fn log_error_with_caller<E>(caller: core::panic::Location<'_>, error: E, level: log::Level)
 621where
 622    E: std::fmt::Debug,
 623{
 624    #[cfg(not(target_os = "windows"))]
 625    let file = caller.file();
 626    #[cfg(target_os = "windows")]
 627    let file = caller.file().replace('\\', "/");
 628    // In this codebase all crates reside in a `crates` directory,
 629    // so discard the prefix up to that segment to find the crate name
 630    let file = file.split_once("crates/");
 631    let target = file.as_ref().and_then(|(_, s)| s.split_once("/src/"));
 632
 633    let module_path = target.map(|(krate, module)| {
 634        if module.starts_with(krate) {
 635            module.trim_end_matches(".rs").replace('/', "::")
 636        } else {
 637            krate.to_owned() + "::" + &module.trim_end_matches(".rs").replace('/', "::")
 638        }
 639    });
 640    let file = file.map(|(_, file)| format!("crates/{file}"));
 641    log::logger().log(
 642        &log::Record::builder()
 643            .target(module_path.as_deref().unwrap_or(""))
 644            .module_path(file.as_deref())
 645            .args(format_args!("{:?}", error))
 646            .file(Some(caller.file()))
 647            .line(Some(caller.line()))
 648            .level(level)
 649            .build(),
 650    );
 651}
 652
 653pub fn log_err<E: std::fmt::Debug>(error: &E) {
 654    log_error_with_caller(*Location::caller(), error, log::Level::Error);
 655}
 656
 657pub trait TryFutureExt {
 658    fn log_err(self) -> LogErrorFuture<Self>
 659    where
 660        Self: Sized;
 661
 662    fn log_tracked_err(self, location: core::panic::Location<'static>) -> LogErrorFuture<Self>
 663    where
 664        Self: Sized;
 665
 666    fn warn_on_err(self) -> LogErrorFuture<Self>
 667    where
 668        Self: Sized;
 669    fn unwrap(self) -> UnwrapFuture<Self>
 670    where
 671        Self: Sized;
 672}
 673
 674impl<F, T, E> TryFutureExt for F
 675where
 676    F: Future<Output = Result<T, E>>,
 677    E: std::fmt::Debug,
 678{
 679    #[track_caller]
 680    fn log_err(self) -> LogErrorFuture<Self>
 681    where
 682        Self: Sized,
 683    {
 684        let location = Location::caller();
 685        LogErrorFuture(self, log::Level::Error, *location)
 686    }
 687
 688    fn log_tracked_err(self, location: core::panic::Location<'static>) -> LogErrorFuture<Self>
 689    where
 690        Self: Sized,
 691    {
 692        LogErrorFuture(self, log::Level::Error, location)
 693    }
 694
 695    #[track_caller]
 696    fn warn_on_err(self) -> LogErrorFuture<Self>
 697    where
 698        Self: Sized,
 699    {
 700        let location = Location::caller();
 701        LogErrorFuture(self, log::Level::Warn, *location)
 702    }
 703
 704    fn unwrap(self) -> UnwrapFuture<Self>
 705    where
 706        Self: Sized,
 707    {
 708        UnwrapFuture(self)
 709    }
 710}
 711
 712#[must_use]
 713pub struct LogErrorFuture<F>(F, log::Level, core::panic::Location<'static>);
 714
 715impl<F, T, E> Future for LogErrorFuture<F>
 716where
 717    F: Future<Output = Result<T, E>>,
 718    E: std::fmt::Debug,
 719{
 720    type Output = Option<T>;
 721
 722    fn poll(self: Pin<&mut Self>, cx: &mut Context) -> Poll<Self::Output> {
 723        let level = self.1;
 724        let location = self.2;
 725        let inner = unsafe { Pin::new_unchecked(&mut self.get_unchecked_mut().0) };
 726        match inner.poll(cx) {
 727            Poll::Ready(output) => Poll::Ready(match output {
 728                Ok(output) => Some(output),
 729                Err(error) => {
 730                    log_error_with_caller(location, error, level);
 731                    None
 732                }
 733            }),
 734            Poll::Pending => Poll::Pending,
 735        }
 736    }
 737}
 738
 739pub struct UnwrapFuture<F>(F);
 740
 741impl<F, T, E> Future for UnwrapFuture<F>
 742where
 743    F: Future<Output = Result<T, E>>,
 744    E: std::fmt::Debug,
 745{
 746    type Output = T;
 747
 748    fn poll(self: Pin<&mut Self>, cx: &mut Context) -> Poll<Self::Output> {
 749        let inner = unsafe { Pin::new_unchecked(&mut self.get_unchecked_mut().0) };
 750        match inner.poll(cx) {
 751            Poll::Ready(result) => Poll::Ready(result.unwrap()),
 752            Poll::Pending => Poll::Pending,
 753        }
 754    }
 755}
 756
 757pub struct Deferred<F: FnOnce()>(Option<F>);
 758
 759impl<F: FnOnce()> Deferred<F> {
 760    /// Drop without running the deferred function.
 761    pub fn abort(mut self) {
 762        self.0.take();
 763    }
 764}
 765
 766impl<F: FnOnce()> Drop for Deferred<F> {
 767    fn drop(&mut self) {
 768        if let Some(f) = self.0.take() {
 769            f()
 770        }
 771    }
 772}
 773
 774/// Run the given function when the returned value is dropped (unless it's cancelled).
 775#[must_use]
 776pub fn defer<F: FnOnce()>(f: F) -> Deferred<F> {
 777    Deferred(Some(f))
 778}
 779
 780#[cfg(any(test, feature = "test-support"))]
 781mod rng {
 782    use rand::prelude::*;
 783
 784    pub struct RandomCharIter<T: Rng> {
 785        rng: T,
 786        simple_text: bool,
 787    }
 788
 789    impl<T: Rng> RandomCharIter<T> {
 790        pub fn new(rng: T) -> Self {
 791            Self {
 792                rng,
 793                simple_text: std::env::var("SIMPLE_TEXT").is_ok_and(|v| !v.is_empty()),
 794            }
 795        }
 796
 797        pub fn with_simple_text(mut self) -> Self {
 798            self.simple_text = true;
 799            self
 800        }
 801    }
 802
 803    impl<T: Rng> Iterator for RandomCharIter<T> {
 804        type Item = char;
 805
 806        fn next(&mut self) -> Option<Self::Item> {
 807            if self.simple_text {
 808                return if self.rng.random_range(0..100) < 5 {
 809                    Some('\n')
 810                } else {
 811                    Some(self.rng.random_range(b'a'..b'z' + 1).into())
 812                };
 813            }
 814
 815            match self.rng.random_range(0..100) {
 816                // whitespace
 817                0..=19 => [' ', '\n', '\r', '\t'].choose(&mut self.rng).copied(),
 818                // two-byte greek letters
 819                20..=32 => char::from_u32(self.rng.random_range(('Ξ±' as u32)..('Ο‰' as u32 + 1))),
 820                // // three-byte characters
 821                33..=45 => ['βœ‹', 'βœ…', '❌', '❎', '⭐']
 822                    .choose(&mut self.rng)
 823                    .copied(),
 824                // // four-byte characters
 825                46..=58 => ['🍐', 'πŸ€', 'πŸ—', 'πŸŽ‰'].choose(&mut self.rng).copied(),
 826                // ascii letters
 827                _ => Some(self.rng.random_range(b'a'..b'z' + 1).into()),
 828            }
 829        }
 830    }
 831}
 832#[cfg(any(test, feature = "test-support"))]
 833pub use rng::RandomCharIter;
 834
 835/// Get an embedded file as a string.
 836pub fn asset_str<A: rust_embed::RustEmbed>(path: &str) -> Cow<'static, str> {
 837    match A::get(path).expect(path).data {
 838        Cow::Borrowed(bytes) => Cow::Borrowed(std::str::from_utf8(bytes).unwrap()),
 839        Cow::Owned(bytes) => Cow::Owned(String::from_utf8(bytes).unwrap()),
 840    }
 841}
 842
 843/// Expands to an immediately-invoked function expression. Good for using the ? operator
 844/// in functions which do not return an Option or Result.
 845///
 846/// Accepts a normal block, an async block, or an async move block.
 847#[macro_export]
 848macro_rules! maybe {
 849    ($block:block) => {
 850        (|| $block)()
 851    };
 852    (async $block:block) => {
 853        (async || $block)()
 854    };
 855    (async move $block:block) => {
 856        (async move || $block)()
 857    };
 858}
 859
 860pub trait RangeExt<T> {
 861    fn sorted(&self) -> Self;
 862    fn to_inclusive(&self) -> RangeInclusive<T>;
 863    fn overlaps(&self, other: &Range<T>) -> bool;
 864    fn contains_inclusive(&self, other: &Range<T>) -> bool;
 865}
 866
 867impl<T: Ord + Clone> RangeExt<T> for Range<T> {
 868    fn sorted(&self) -> Self {
 869        cmp::min(&self.start, &self.end).clone()..cmp::max(&self.start, &self.end).clone()
 870    }
 871
 872    fn to_inclusive(&self) -> RangeInclusive<T> {
 873        self.start.clone()..=self.end.clone()
 874    }
 875
 876    fn overlaps(&self, other: &Range<T>) -> bool {
 877        self.start < other.end && other.start < self.end
 878    }
 879
 880    fn contains_inclusive(&self, other: &Range<T>) -> bool {
 881        self.start <= other.start && other.end <= self.end
 882    }
 883}
 884
 885impl<T: Ord + Clone> RangeExt<T> for RangeInclusive<T> {
 886    fn sorted(&self) -> Self {
 887        cmp::min(self.start(), self.end()).clone()..=cmp::max(self.start(), self.end()).clone()
 888    }
 889
 890    fn to_inclusive(&self) -> RangeInclusive<T> {
 891        self.clone()
 892    }
 893
 894    fn overlaps(&self, other: &Range<T>) -> bool {
 895        self.start() < &other.end && &other.start <= self.end()
 896    }
 897
 898    fn contains_inclusive(&self, other: &Range<T>) -> bool {
 899        self.start() <= &other.start && &other.end <= self.end()
 900    }
 901}
 902
 903/// A way to sort strings with starting numbers numerically first, falling back to alphanumeric one,
 904/// case-insensitive.
 905///
 906/// This is useful for turning regular alphanumerically sorted sequences as `1-abc, 10, 11-def, .., 2, 21-abc`
 907/// into `1-abc, 2, 10, 11-def, .., 21-abc`
 908#[derive(Debug, PartialEq, Eq)]
 909pub struct NumericPrefixWithSuffix<'a>(Option<u64>, &'a str);
 910
 911impl<'a> NumericPrefixWithSuffix<'a> {
 912    pub fn from_numeric_prefixed_str(str: &'a str) -> Self {
 913        let i = str.chars().take_while(|c| c.is_ascii_digit()).count();
 914        let (prefix, remainder) = str.split_at(i);
 915
 916        let prefix = prefix.parse().ok();
 917        Self(prefix, remainder)
 918    }
 919}
 920
 921/// When dealing with equality, we need to consider the case of the strings to achieve strict equality
 922/// to handle cases like "a" < "A" instead of "a" == "A".
 923impl Ord for NumericPrefixWithSuffix<'_> {
 924    fn cmp(&self, other: &Self) -> Ordering {
 925        match (self.0, other.0) {
 926            (None, None) => UniCase::new(self.1)
 927                .cmp(&UniCase::new(other.1))
 928                .then_with(|| self.1.cmp(other.1).reverse()),
 929            (None, Some(_)) => Ordering::Greater,
 930            (Some(_), None) => Ordering::Less,
 931            (Some(a), Some(b)) => a.cmp(&b).then_with(|| {
 932                UniCase::new(self.1)
 933                    .cmp(&UniCase::new(other.1))
 934                    .then_with(|| self.1.cmp(other.1).reverse())
 935            }),
 936        }
 937    }
 938}
 939
 940impl PartialOrd for NumericPrefixWithSuffix<'_> {
 941    fn partial_cmp(&self, other: &Self) -> Option<Ordering> {
 942        Some(self.cmp(other))
 943    }
 944}
 945
 946/// Capitalizes the first character of a string.
 947///
 948/// This function takes a string slice as input and returns a new `String` with the first character
 949/// capitalized.
 950///
 951/// # Examples
 952///
 953/// ```
 954/// use util::capitalize;
 955///
 956/// assert_eq!(capitalize("hello"), "Hello");
 957/// assert_eq!(capitalize("WORLD"), "WORLD");
 958/// assert_eq!(capitalize(""), "");
 959/// ```
 960pub fn capitalize(str: &str) -> String {
 961    let mut chars = str.chars();
 962    match chars.next() {
 963        None => String::new(),
 964        Some(first_char) => first_char.to_uppercase().collect::<String>() + chars.as_str(),
 965    }
 966}
 967
 968fn emoji_regex() -> &'static Regex {
 969    static EMOJI_REGEX: LazyLock<Regex> =
 970        LazyLock::new(|| Regex::new("(\\p{Emoji}|\u{200D})").unwrap());
 971    &EMOJI_REGEX
 972}
 973
 974/// Returns true if the given string consists of emojis only.
 975/// E.g. "πŸ‘¨β€πŸ‘©β€πŸ‘§β€πŸ‘§πŸ‘‹" will return true, but "πŸ‘‹!" will return false.
 976pub fn word_consists_of_emojis(s: &str) -> bool {
 977    let mut prev_end = 0;
 978    for capture in emoji_regex().find_iter(s) {
 979        if capture.start() != prev_end {
 980            return false;
 981        }
 982        prev_end = capture.end();
 983    }
 984    prev_end == s.len()
 985}
 986
 987/// Similar to `str::split`, but also provides byte-offset ranges of the results. Unlike
 988/// `str::split`, this is not generic on pattern types and does not return an `Iterator`.
 989pub fn split_str_with_ranges(s: &str, pat: impl Fn(char) -> bool) -> Vec<(Range<usize>, &str)> {
 990    let mut result = Vec::new();
 991    let mut start = 0;
 992
 993    for (i, ch) in s.char_indices() {
 994        if pat(ch) {
 995            if i > start {
 996                result.push((start..i, &s[start..i]));
 997            }
 998            start = i + ch.len_utf8();
 999        }
1000    }
1001
1002    if s.len() > start {
1003        result.push((start..s.len(), &s[start..s.len()]));
1004    }
1005
1006    result
1007}
1008
1009pub fn default<D: Default>() -> D {
1010    Default::default()
1011}
1012
1013pub use self::shell::{
1014    get_default_system_shell, get_default_system_shell_preferring_bash, get_system_shell,
1015};
1016
1017#[derive(Debug)]
1018pub enum ConnectionResult<O> {
1019    Timeout,
1020    ConnectionReset,
1021    Result(anyhow::Result<O>),
1022}
1023
1024impl<O> ConnectionResult<O> {
1025    pub fn into_response(self) -> anyhow::Result<O> {
1026        match self {
1027            ConnectionResult::Timeout => anyhow::bail!("Request timed out"),
1028            ConnectionResult::ConnectionReset => anyhow::bail!("Server reset the connection"),
1029            ConnectionResult::Result(r) => r,
1030        }
1031    }
1032}
1033
1034impl<O> From<anyhow::Result<O>> for ConnectionResult<O> {
1035    fn from(result: anyhow::Result<O>) -> Self {
1036        ConnectionResult::Result(result)
1037    }
1038}
1039
1040#[track_caller]
1041pub fn some_or_debug_panic<T>(option: Option<T>) -> Option<T> {
1042    #[cfg(debug_assertions)]
1043    if option.is_none() {
1044        panic!("Unexpected None");
1045    }
1046    option
1047}
1048
1049#[cfg(test)]
1050mod tests {
1051    use super::*;
1052
1053    #[test]
1054    fn test_extend_sorted() {
1055        let mut vec = vec![];
1056
1057        extend_sorted(&mut vec, vec![21, 17, 13, 8, 1, 0], 5, |a, b| b.cmp(a));
1058        assert_eq!(vec, &[21, 17, 13, 8, 1]);
1059
1060        extend_sorted(&mut vec, vec![101, 19, 17, 8, 2], 8, |a, b| b.cmp(a));
1061        assert_eq!(vec, &[101, 21, 19, 17, 13, 8, 2, 1]);
1062
1063        extend_sorted(&mut vec, vec![1000, 19, 17, 9, 5], 8, |a, b| b.cmp(a));
1064        assert_eq!(vec, &[1000, 101, 21, 19, 17, 13, 9, 8]);
1065    }
1066
1067    #[test]
1068    fn test_truncate_to_bottom_n_sorted_by() {
1069        let mut vec: Vec<u32> = vec![5, 2, 3, 4, 1];
1070        truncate_to_bottom_n_sorted_by(&mut vec, 10, &u32::cmp);
1071        assert_eq!(vec, &[1, 2, 3, 4, 5]);
1072
1073        vec = vec![5, 2, 3, 4, 1];
1074        truncate_to_bottom_n_sorted_by(&mut vec, 5, &u32::cmp);
1075        assert_eq!(vec, &[1, 2, 3, 4, 5]);
1076
1077        vec = vec![5, 2, 3, 4, 1];
1078        truncate_to_bottom_n_sorted_by(&mut vec, 4, &u32::cmp);
1079        assert_eq!(vec, &[1, 2, 3, 4]);
1080
1081        vec = vec![5, 2, 3, 4, 1];
1082        truncate_to_bottom_n_sorted_by(&mut vec, 1, &u32::cmp);
1083        assert_eq!(vec, &[1]);
1084
1085        vec = vec![5, 2, 3, 4, 1];
1086        truncate_to_bottom_n_sorted_by(&mut vec, 0, &u32::cmp);
1087        assert!(vec.is_empty());
1088    }
1089
1090    #[test]
1091    fn test_iife() {
1092        fn option_returning_function() -> Option<()> {
1093            None
1094        }
1095
1096        let foo = maybe!({
1097            option_returning_function()?;
1098            Some(())
1099        });
1100
1101        assert_eq!(foo, None);
1102    }
1103
1104    #[test]
1105    fn test_truncate_and_trailoff() {
1106        assert_eq!(truncate_and_trailoff("", 5), "");
1107        assert_eq!(truncate_and_trailoff("aaaaaa", 7), "aaaaaa");
1108        assert_eq!(truncate_and_trailoff("aaaaaa", 6), "aaaaaa");
1109        assert_eq!(truncate_and_trailoff("aaaaaa", 5), "aaaaa…");
1110        assert_eq!(truncate_and_trailoff("èèèèèè", 7), "èèèèèè");
1111        assert_eq!(truncate_and_trailoff("èèèèèè", 6), "èèèèèè");
1112        assert_eq!(truncate_and_trailoff("èèèèèè", 5), "èèèèè…");
1113    }
1114
1115    #[test]
1116    fn test_truncate_and_remove_front() {
1117        assert_eq!(truncate_and_remove_front("", 5), "");
1118        assert_eq!(truncate_and_remove_front("aaaaaa", 7), "aaaaaa");
1119        assert_eq!(truncate_and_remove_front("aaaaaa", 6), "aaaaaa");
1120        assert_eq!(truncate_and_remove_front("aaaaaa", 5), "…aaaaa");
1121        assert_eq!(truncate_and_remove_front("èèèèèè", 7), "èèèèèè");
1122        assert_eq!(truncate_and_remove_front("èèèèèè", 6), "èèèèèè");
1123        assert_eq!(truncate_and_remove_front("èèèèèè", 5), "…èèèèè");
1124    }
1125
1126    #[test]
1127    fn test_numeric_prefix_str_method() {
1128        let target = "1a";
1129        assert_eq!(
1130            NumericPrefixWithSuffix::from_numeric_prefixed_str(target),
1131            NumericPrefixWithSuffix(Some(1), "a")
1132        );
1133
1134        let target = "12ab";
1135        assert_eq!(
1136            NumericPrefixWithSuffix::from_numeric_prefixed_str(target),
1137            NumericPrefixWithSuffix(Some(12), "ab")
1138        );
1139
1140        let target = "12_ab";
1141        assert_eq!(
1142            NumericPrefixWithSuffix::from_numeric_prefixed_str(target),
1143            NumericPrefixWithSuffix(Some(12), "_ab")
1144        );
1145
1146        let target = "1_2ab";
1147        assert_eq!(
1148            NumericPrefixWithSuffix::from_numeric_prefixed_str(target),
1149            NumericPrefixWithSuffix(Some(1), "_2ab")
1150        );
1151
1152        let target = "1.2";
1153        assert_eq!(
1154            NumericPrefixWithSuffix::from_numeric_prefixed_str(target),
1155            NumericPrefixWithSuffix(Some(1), ".2")
1156        );
1157
1158        let target = "1.2_a";
1159        assert_eq!(
1160            NumericPrefixWithSuffix::from_numeric_prefixed_str(target),
1161            NumericPrefixWithSuffix(Some(1), ".2_a")
1162        );
1163
1164        let target = "12.2_a";
1165        assert_eq!(
1166            NumericPrefixWithSuffix::from_numeric_prefixed_str(target),
1167            NumericPrefixWithSuffix(Some(12), ".2_a")
1168        );
1169
1170        let target = "12a.2_a";
1171        assert_eq!(
1172            NumericPrefixWithSuffix::from_numeric_prefixed_str(target),
1173            NumericPrefixWithSuffix(Some(12), "a.2_a")
1174        );
1175    }
1176
1177    #[test]
1178    fn test_numeric_prefix_with_suffix() {
1179        let mut sorted = vec!["1-abc", "10", "11def", "2", "21-abc"];
1180        sorted.sort_by_key(|s| NumericPrefixWithSuffix::from_numeric_prefixed_str(s));
1181        assert_eq!(sorted, ["1-abc", "2", "10", "11def", "21-abc"]);
1182
1183        for numeric_prefix_less in ["numeric_prefix_less", "aaa", "~β„’Β£"] {
1184            assert_eq!(
1185                NumericPrefixWithSuffix::from_numeric_prefixed_str(numeric_prefix_less),
1186                NumericPrefixWithSuffix(None, numeric_prefix_less),
1187                "String without numeric prefix `{numeric_prefix_less}` should not be converted into NumericPrefixWithSuffix"
1188            )
1189        }
1190    }
1191
1192    #[test]
1193    fn test_word_consists_of_emojis() {
1194        let words_to_test = vec![
1195            ("πŸ‘¨β€πŸ‘©β€πŸ‘§β€πŸ‘§πŸ‘‹πŸ₯’", true),
1196            ("πŸ‘‹", true),
1197            ("!πŸ‘‹", false),
1198            ("πŸ‘‹!", false),
1199            ("πŸ‘‹ ", false),
1200            (" πŸ‘‹", false),
1201            ("Test", false),
1202        ];
1203
1204        for (text, expected_result) in words_to_test {
1205            assert_eq!(word_consists_of_emojis(text), expected_result);
1206        }
1207    }
1208
1209    #[test]
1210    fn test_truncate_lines_and_trailoff() {
1211        let text = r#"Line 1
1212Line 2
1213Line 3"#;
1214
1215        assert_eq!(
1216            truncate_lines_and_trailoff(text, 2),
1217            r#"Line 1
1218…"#
1219        );
1220
1221        assert_eq!(
1222            truncate_lines_and_trailoff(text, 3),
1223            r#"Line 1
1224Line 2
1225…"#
1226        );
1227
1228        assert_eq!(
1229            truncate_lines_and_trailoff(text, 4),
1230            r#"Line 1
1231Line 2
1232Line 3"#
1233        );
1234    }
1235
1236    #[test]
1237    fn test_expanded_and_wrapped_usize_range() {
1238        // Neither wrap
1239        assert_eq!(
1240            expanded_and_wrapped_usize_range(2..4, 1, 1, 8).collect::<Vec<usize>>(),
1241            (1..5).collect::<Vec<usize>>()
1242        );
1243        // Start wraps
1244        assert_eq!(
1245            expanded_and_wrapped_usize_range(2..4, 3, 1, 8).collect::<Vec<usize>>(),
1246            ((0..5).chain(7..8)).collect::<Vec<usize>>()
1247        );
1248        // Start wraps all the way around
1249        assert_eq!(
1250            expanded_and_wrapped_usize_range(2..4, 5, 1, 8).collect::<Vec<usize>>(),
1251            (0..8).collect::<Vec<usize>>()
1252        );
1253        // Start wraps all the way around and past 0
1254        assert_eq!(
1255            expanded_and_wrapped_usize_range(2..4, 10, 1, 8).collect::<Vec<usize>>(),
1256            (0..8).collect::<Vec<usize>>()
1257        );
1258        // End wraps
1259        assert_eq!(
1260            expanded_and_wrapped_usize_range(3..5, 1, 4, 8).collect::<Vec<usize>>(),
1261            (0..1).chain(2..8).collect::<Vec<usize>>()
1262        );
1263        // End wraps all the way around
1264        assert_eq!(
1265            expanded_and_wrapped_usize_range(3..5, 1, 5, 8).collect::<Vec<usize>>(),
1266            (0..8).collect::<Vec<usize>>()
1267        );
1268        // End wraps all the way around and past the end
1269        assert_eq!(
1270            expanded_and_wrapped_usize_range(3..5, 1, 10, 8).collect::<Vec<usize>>(),
1271            (0..8).collect::<Vec<usize>>()
1272        );
1273        // Both start and end wrap
1274        assert_eq!(
1275            expanded_and_wrapped_usize_range(3..5, 4, 4, 8).collect::<Vec<usize>>(),
1276            (0..8).collect::<Vec<usize>>()
1277        );
1278    }
1279
1280    #[test]
1281    fn test_wrapped_usize_outward_from() {
1282        // No wrapping
1283        assert_eq!(
1284            wrapped_usize_outward_from(4, 2, 2, 10).collect::<Vec<usize>>(),
1285            vec![4, 5, 3, 6, 2]
1286        );
1287        // Wrapping at end
1288        assert_eq!(
1289            wrapped_usize_outward_from(8, 2, 3, 10).collect::<Vec<usize>>(),
1290            vec![8, 9, 7, 0, 6, 1]
1291        );
1292        // Wrapping at start
1293        assert_eq!(
1294            wrapped_usize_outward_from(1, 3, 2, 10).collect::<Vec<usize>>(),
1295            vec![1, 2, 0, 3, 9, 8]
1296        );
1297        // All values wrap around
1298        assert_eq!(
1299            wrapped_usize_outward_from(5, 10, 10, 8).collect::<Vec<usize>>(),
1300            vec![5, 6, 4, 7, 3, 0, 2, 1]
1301        );
1302        // None before / after
1303        assert_eq!(
1304            wrapped_usize_outward_from(3, 0, 0, 8).collect::<Vec<usize>>(),
1305            vec![3]
1306        );
1307        // Starting point already wrapped
1308        assert_eq!(
1309            wrapped_usize_outward_from(15, 2, 2, 10).collect::<Vec<usize>>(),
1310            vec![5, 6, 4, 7, 3]
1311        );
1312        // wrap_length of 0
1313        assert_eq!(
1314            wrapped_usize_outward_from(4, 2, 2, 0).collect::<Vec<usize>>(),
1315            Vec::<usize>::new()
1316        );
1317    }
1318
1319    #[test]
1320    fn test_split_with_ranges() {
1321        let input = "hi";
1322        let result = split_str_with_ranges(input, |c| c == ' ');
1323
1324        assert_eq!(result.len(), 1);
1325        assert_eq!(result[0], (0..2, "hi"));
1326
1327        let input = "hΓ©lloπŸ¦€world";
1328        let result = split_str_with_ranges(input, |c| c == 'πŸ¦€');
1329
1330        assert_eq!(result.len(), 2);
1331        assert_eq!(result[0], (0..6, "hΓ©llo")); // 'Γ©' is 2 bytes
1332        assert_eq!(result[1], (10..15, "world")); // 'πŸ¦€' is 4 bytes
1333    }
1334}