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 size;
   9#[cfg(any(test, feature = "test-support"))]
  10pub mod test;
  11pub mod time;
  12
  13use anyhow::Result;
  14use futures::Future;
  15use itertools::Either;
  16use regex::Regex;
  17use std::sync::{LazyLock, OnceLock};
  18use std::{
  19    borrow::Cow,
  20    cmp::{self, Ordering},
  21    env,
  22    ops::{AddAssign, Range, RangeInclusive},
  23    panic::Location,
  24    pin::Pin,
  25    task::{Context, Poll},
  26    time::Instant,
  27};
  28use unicase::UniCase;
  29
  30#[cfg(unix)]
  31use anyhow::Context as _;
  32
  33pub use take_until::*;
  34#[cfg(any(test, feature = "test-support"))]
  35pub use util_macros::{line_endings, separator, uri};
  36
  37#[macro_export]
  38macro_rules! debug_panic {
  39    ( $($fmt_arg:tt)* ) => {
  40        if cfg!(debug_assertions) {
  41            panic!( $($fmt_arg)* );
  42        } else {
  43            let backtrace = std::backtrace::Backtrace::capture();
  44            log::error!("{}\n{:?}", format_args!($($fmt_arg)*), backtrace);
  45        }
  46    };
  47}
  48
  49/// A macro to add "C:" to the beginning of a path literal on Windows, and replace all
  50/// the separator from `/` to `\`.
  51/// But on non-Windows platforms, it will return the path literal as is.
  52///
  53/// # Examples
  54/// ```rust
  55/// use util::path;
  56///
  57/// let path = path!("/Users/user/file.txt");
  58/// #[cfg(target_os = "windows")]
  59/// assert_eq!(path, "C:\\Users\\user\\file.txt");
  60/// #[cfg(not(target_os = "windows"))]
  61/// assert_eq!(path, "/Users/user/file.txt");
  62/// ```
  63#[cfg(all(any(test, feature = "test-support"), target_os = "windows"))]
  64#[macro_export]
  65macro_rules! path {
  66    ($path:literal) => {
  67        concat!("C:", util::separator!($path))
  68    };
  69}
  70
  71/// A macro to add "C:" to the beginning of a path literal on Windows, and replace all
  72/// the separator from `/` to `\`.
  73/// But on non-Windows platforms, it will return the path literal as is.
  74///
  75/// # Examples
  76/// ```rust
  77/// use util::path;
  78///
  79/// let path = path!("/Users/user/file.txt");
  80/// #[cfg(target_os = "windows")]
  81/// assert_eq!(path, "C:\\Users\\user\\file.txt");
  82/// #[cfg(not(target_os = "windows"))]
  83/// assert_eq!(path, "/Users/user/file.txt");
  84/// ```
  85#[cfg(all(any(test, feature = "test-support"), not(target_os = "windows")))]
  86#[macro_export]
  87macro_rules! path {
  88    ($path:literal) => {
  89        $path
  90    };
  91}
  92
  93pub fn truncate(s: &str, max_chars: usize) -> &str {
  94    match s.char_indices().nth(max_chars) {
  95        None => s,
  96        Some((idx, _)) => &s[..idx],
  97    }
  98}
  99
 100/// Removes characters from the end of the string if its length is greater than `max_chars` and
 101/// appends "..." to the string. Returns string unchanged if its length is smaller than max_chars.
 102pub fn truncate_and_trailoff(s: &str, max_chars: usize) -> String {
 103    debug_assert!(max_chars >= 5);
 104
 105    // If the string's byte length is <= max_chars, walking the string can be skipped since the
 106    // number of chars is <= the number of bytes.
 107    if s.len() <= max_chars {
 108        return s.to_string();
 109    }
 110    let truncation_ix = s.char_indices().map(|(i, _)| i).nth(max_chars);
 111    match truncation_ix {
 112        Some(index) => s[..index].to_string() + "",
 113        _ => s.to_string(),
 114    }
 115}
 116
 117/// Removes characters from the front of the string if its length is greater than `max_chars` and
 118/// prepends the string with "...". Returns string unchanged if its length is smaller than max_chars.
 119pub fn truncate_and_remove_front(s: &str, max_chars: usize) -> String {
 120    debug_assert!(max_chars >= 5);
 121
 122    // If the string's byte length is <= max_chars, walking the string can be skipped since the
 123    // number of chars is <= the number of bytes.
 124    if s.len() <= max_chars {
 125        return s.to_string();
 126    }
 127    let suffix_char_length = max_chars.saturating_sub(1);
 128    let truncation_ix = s
 129        .char_indices()
 130        .map(|(i, _)| i)
 131        .nth_back(suffix_char_length);
 132    match truncation_ix {
 133        Some(index) if index > 0 => "".to_string() + &s[index..],
 134        _ => s.to_string(),
 135    }
 136}
 137
 138/// Takes only `max_lines` from the string and, if there were more than `max_lines-1`, appends a
 139/// a newline and "..." to the string, so that `max_lines` are returned.
 140/// Returns string unchanged if its length is smaller than max_lines.
 141pub fn truncate_lines_and_trailoff(s: &str, max_lines: usize) -> String {
 142    let mut lines = s.lines().take(max_lines).collect::<Vec<_>>();
 143    if lines.len() > max_lines - 1 {
 144        lines.pop();
 145        lines.join("\n") + "\n"
 146    } else {
 147        lines.join("\n")
 148    }
 149}
 150
 151/// Truncates the string at a character boundary, such that the result is less than `max_bytes` in
 152/// length.
 153pub fn truncate_to_byte_limit(s: &str, max_bytes: usize) -> &str {
 154    if s.len() < max_bytes {
 155        return s;
 156    }
 157
 158    for i in (0..max_bytes).rev() {
 159        if s.is_char_boundary(i) {
 160            return &s[..i];
 161        }
 162    }
 163
 164    ""
 165}
 166
 167/// Takes a prefix of complete lines which fit within the byte limit. If the first line is longer
 168/// than the limit, truncates at a character boundary.
 169pub fn truncate_lines_to_byte_limit(s: &str, max_bytes: usize) -> &str {
 170    if s.len() < max_bytes {
 171        return s;
 172    }
 173
 174    for i in (0..max_bytes).rev() {
 175        if s.is_char_boundary(i) {
 176            if s.as_bytes()[i] == b'\n' {
 177                // Since the i-th character is \n, valid to slice at i + 1.
 178                return &s[..i + 1];
 179            }
 180        }
 181    }
 182
 183    truncate_to_byte_limit(s, max_bytes)
 184}
 185
 186#[test]
 187fn test_truncate_lines_to_byte_limit() {
 188    let text = "Line 1\nLine 2\nLine 3\nLine 4";
 189
 190    // Limit that includes all lines
 191    assert_eq!(truncate_lines_to_byte_limit(text, 100), text);
 192
 193    // Exactly the first line
 194    assert_eq!(truncate_lines_to_byte_limit(text, 7), "Line 1\n");
 195
 196    // Limit between lines
 197    assert_eq!(truncate_lines_to_byte_limit(text, 13), "Line 1\n");
 198    assert_eq!(truncate_lines_to_byte_limit(text, 20), "Line 1\nLine 2\n");
 199
 200    // Limit before first newline
 201    assert_eq!(truncate_lines_to_byte_limit(text, 6), "Line ");
 202
 203    // Test with non-ASCII characters
 204    let text_utf8 = "Line 1\nLíne 2\nLine 3";
 205    assert_eq!(
 206        truncate_lines_to_byte_limit(text_utf8, 15),
 207        "Line 1\nLíne 2\n"
 208    );
 209}
 210
 211pub fn post_inc<T: From<u8> + AddAssign<T> + Copy>(value: &mut T) -> T {
 212    let prev = *value;
 213    *value += T::from(1);
 214    prev
 215}
 216
 217/// Extend a sorted vector with a sorted sequence of items, maintaining the vector's sort order and
 218/// enforcing a maximum length. This also de-duplicates items. Sort the items according to the given callback. Before calling this,
 219/// both `vec` and `new_items` should already be sorted according to the `cmp` comparator.
 220pub fn extend_sorted<T, I, F>(vec: &mut Vec<T>, new_items: I, limit: usize, mut cmp: F)
 221where
 222    I: IntoIterator<Item = T>,
 223    F: FnMut(&T, &T) -> Ordering,
 224{
 225    let mut start_index = 0;
 226    for new_item in new_items {
 227        if let Err(i) = vec[start_index..].binary_search_by(|m| cmp(m, &new_item)) {
 228            let index = start_index + i;
 229            if vec.len() < limit {
 230                vec.insert(index, new_item);
 231            } else if index < vec.len() {
 232                vec.pop();
 233                vec.insert(index, new_item);
 234            }
 235            start_index = index;
 236        }
 237    }
 238}
 239
 240pub fn truncate_to_bottom_n_sorted_by<T, F>(items: &mut Vec<T>, limit: usize, compare: &F)
 241where
 242    F: Fn(&T, &T) -> Ordering,
 243{
 244    if limit == 0 {
 245        items.truncate(0);
 246    }
 247    if items.len() <= limit {
 248        items.sort_by(compare);
 249        return;
 250    }
 251    // When limit is near to items.len() it may be more efficient to sort the whole list and
 252    // truncate, rather than always doing selection first as is done below. It's hard to analyze
 253    // where the threshold for this should be since the quickselect style algorithm used by
 254    // `select_nth_unstable_by` makes the prefix partially sorted, and so its work is not wasted -
 255    // the expected number of comparisons needed by `sort_by` is less than it is for some arbitrary
 256    // unsorted input.
 257    items.select_nth_unstable_by(limit, compare);
 258    items.truncate(limit);
 259    items.sort_by(compare);
 260}
 261
 262#[cfg(unix)]
 263fn load_shell_from_passwd() -> Result<()> {
 264    let buflen = match unsafe { libc::sysconf(libc::_SC_GETPW_R_SIZE_MAX) } {
 265        n if n < 0 => 1024,
 266        n => n as usize,
 267    };
 268    let mut buffer = Vec::with_capacity(buflen);
 269
 270    let mut pwd: std::mem::MaybeUninit<libc::passwd> = std::mem::MaybeUninit::uninit();
 271    let mut result: *mut libc::passwd = std::ptr::null_mut();
 272
 273    let uid = unsafe { libc::getuid() };
 274    let status = unsafe {
 275        libc::getpwuid_r(
 276            uid,
 277            pwd.as_mut_ptr(),
 278            buffer.as_mut_ptr() as *mut libc::c_char,
 279            buflen,
 280            &mut result,
 281        )
 282    };
 283    let entry = unsafe { pwd.assume_init() };
 284
 285    anyhow::ensure!(
 286        status == 0,
 287        "call to getpwuid_r failed. uid: {}, status: {}",
 288        uid,
 289        status
 290    );
 291    anyhow::ensure!(!result.is_null(), "passwd entry for uid {} not found", uid);
 292    anyhow::ensure!(
 293        entry.pw_uid == uid,
 294        "passwd entry has different uid ({}) than getuid ({}) returned",
 295        entry.pw_uid,
 296        uid,
 297    );
 298
 299    let shell = unsafe { std::ffi::CStr::from_ptr(entry.pw_shell).to_str().unwrap() };
 300    if env::var("SHELL").map_or(true, |shell_env| shell_env != shell) {
 301        log::info!(
 302            "updating SHELL environment variable to value from passwd entry: {:?}",
 303            shell,
 304        );
 305        unsafe { env::set_var("SHELL", shell) };
 306    }
 307
 308    Ok(())
 309}
 310
 311#[cfg(unix)]
 312pub fn load_login_shell_environment() -> Result<()> {
 313    load_shell_from_passwd().log_err();
 314
 315    let marker = "ZED_LOGIN_SHELL_START";
 316    let shell = env::var("SHELL").context(
 317        "SHELL environment variable is not assigned so we can't source login environment variables",
 318    )?;
 319
 320    // If possible, we want to `cd` in the user's `$HOME` to trigger programs
 321    // such as direnv, asdf, mise, ... to adjust the PATH. These tools often hook
 322    // into shell's `cd` command (and hooks) to manipulate env.
 323    // We do this so that we get the env a user would have when spawning a shell
 324    // in home directory.
 325    let shell_cmd_prefix = std::env::var_os("HOME")
 326        .and_then(|home| home.into_string().ok())
 327        .map(|home| format!("cd '{home}';"));
 328
 329    let shell_cmd = format!(
 330        "{}printf '%s' {marker}; /usr/bin/env;",
 331        shell_cmd_prefix.as_deref().unwrap_or("")
 332    );
 333
 334    let output = set_pre_exec_to_start_new_session(
 335        std::process::Command::new(&shell).args(["-l", "-i", "-c", &shell_cmd]),
 336    )
 337    .output()
 338    .context("failed to spawn login shell to source login environment variables")?;
 339    anyhow::ensure!(output.status.success(), "login shell exited with error");
 340
 341    let stdout = String::from_utf8_lossy(&output.stdout);
 342
 343    if let Some(env_output_start) = stdout.find(marker) {
 344        let env_output = &stdout[env_output_start + marker.len()..];
 345
 346        parse_env_output(env_output, |key, value| unsafe { env::set_var(key, value) });
 347
 348        log::info!(
 349            "set environment variables from shell:{}, path:{}",
 350            shell,
 351            env::var("PATH").unwrap_or_default(),
 352        );
 353    }
 354
 355    Ok(())
 356}
 357
 358/// Configures the process to start a new session, to prevent interactive shells from taking control
 359/// of the terminal.
 360///
 361/// For more details: https://registerspill.thorstenball.com/p/how-to-lose-control-of-your-shell
 362pub fn set_pre_exec_to_start_new_session(
 363    command: &mut std::process::Command,
 364) -> &mut std::process::Command {
 365    // safety: code in pre_exec should be signal safe.
 366    // https://man7.org/linux/man-pages/man7/signal-safety.7.html
 367    #[cfg(not(target_os = "windows"))]
 368    unsafe {
 369        use std::os::unix::process::CommandExt;
 370        command.pre_exec(|| {
 371            libc::setsid();
 372            Ok(())
 373        });
 374    };
 375    command
 376}
 377
 378/// Parse the result of calling `usr/bin/env` with no arguments
 379pub fn parse_env_output(env: &str, mut f: impl FnMut(String, String)) {
 380    let mut current_key: Option<String> = None;
 381    let mut current_value: Option<String> = None;
 382
 383    for line in env.split_terminator('\n') {
 384        if let Some(separator_index) = line.find('=') {
 385            if !line[..separator_index].is_empty() {
 386                if let Some((key, value)) = Option::zip(current_key.take(), current_value.take()) {
 387                    f(key, value)
 388                }
 389                current_key = Some(line[..separator_index].to_string());
 390                current_value = Some(line[separator_index + 1..].to_string());
 391                continue;
 392            };
 393        }
 394        if let Some(value) = current_value.as_mut() {
 395            value.push('\n');
 396            value.push_str(line);
 397        }
 398    }
 399    if let Some((key, value)) = Option::zip(current_key.take(), current_value.take()) {
 400        f(key, value)
 401    }
 402}
 403
 404pub fn merge_json_value_into(source: serde_json::Value, target: &mut serde_json::Value) {
 405    use serde_json::Value;
 406
 407    match (source, target) {
 408        (Value::Object(source), Value::Object(target)) => {
 409            for (key, value) in source {
 410                if let Some(target) = target.get_mut(&key) {
 411                    merge_json_value_into(value, target);
 412                } else {
 413                    target.insert(key, value);
 414                }
 415            }
 416        }
 417
 418        (Value::Array(source), Value::Array(target)) => {
 419            for value in source {
 420                target.push(value);
 421            }
 422        }
 423
 424        (source, target) => *target = source,
 425    }
 426}
 427
 428pub fn merge_non_null_json_value_into(source: serde_json::Value, target: &mut serde_json::Value) {
 429    use serde_json::Value;
 430    if let Value::Object(source_object) = source {
 431        let target_object = if let Value::Object(target) = target {
 432            target
 433        } else {
 434            *target = Value::Object(Default::default());
 435            target.as_object_mut().unwrap()
 436        };
 437        for (key, value) in source_object {
 438            if let Some(target) = target_object.get_mut(&key) {
 439                merge_non_null_json_value_into(value, target);
 440            } else if !value.is_null() {
 441                target_object.insert(key, value);
 442            }
 443        }
 444    } else if !source.is_null() {
 445        *target = source
 446    }
 447}
 448
 449pub fn measure<R>(label: &str, f: impl FnOnce() -> R) -> R {
 450    static ZED_MEASUREMENTS: OnceLock<bool> = OnceLock::new();
 451    let zed_measurements = ZED_MEASUREMENTS.get_or_init(|| {
 452        env::var("ZED_MEASUREMENTS")
 453            .map(|measurements| measurements == "1" || measurements == "true")
 454            .unwrap_or(false)
 455    });
 456
 457    if *zed_measurements {
 458        let start = Instant::now();
 459        let result = f();
 460        let elapsed = start.elapsed();
 461        eprintln!("{}: {:?}", label, elapsed);
 462        result
 463    } else {
 464        f()
 465    }
 466}
 467
 468pub fn iterate_expanded_and_wrapped_usize_range(
 469    range: Range<usize>,
 470    additional_before: usize,
 471    additional_after: usize,
 472    wrap_length: usize,
 473) -> impl Iterator<Item = usize> {
 474    let start_wraps = range.start < additional_before;
 475    let end_wraps = wrap_length < range.end + additional_after;
 476    if start_wraps && end_wraps {
 477        Either::Left(0..wrap_length)
 478    } else if start_wraps {
 479        let wrapped_start = (range.start + wrap_length).saturating_sub(additional_before);
 480        if wrapped_start <= range.end {
 481            Either::Left(0..wrap_length)
 482        } else {
 483            Either::Right((0..range.end + additional_after).chain(wrapped_start..wrap_length))
 484        }
 485    } else if end_wraps {
 486        let wrapped_end = range.end + additional_after - wrap_length;
 487        if range.start <= wrapped_end {
 488            Either::Left(0..wrap_length)
 489        } else {
 490            Either::Right((0..wrapped_end).chain(range.start - additional_before..wrap_length))
 491        }
 492    } else {
 493        Either::Left((range.start - additional_before)..(range.end + additional_after))
 494    }
 495}
 496
 497#[cfg(target_os = "windows")]
 498pub fn get_windows_system_shell() -> String {
 499    use std::path::PathBuf;
 500
 501    fn find_pwsh_in_programfiles(find_alternate: bool, find_preview: bool) -> Option<PathBuf> {
 502        #[cfg(target_pointer_width = "64")]
 503        let env_var = if find_alternate {
 504            "ProgramFiles(x86)"
 505        } else {
 506            "ProgramFiles"
 507        };
 508
 509        #[cfg(target_pointer_width = "32")]
 510        let env_var = if find_alternate {
 511            "ProgramW6432"
 512        } else {
 513            "ProgramFiles"
 514        };
 515
 516        let install_base_dir = PathBuf::from(std::env::var_os(env_var)?).join("PowerShell");
 517        install_base_dir
 518            .read_dir()
 519            .ok()?
 520            .filter_map(Result::ok)
 521            .filter(|entry| matches!(entry.file_type(), Ok(ft) if ft.is_dir()))
 522            .filter_map(|entry| {
 523                let dir_name = entry.file_name();
 524                let dir_name = dir_name.to_string_lossy();
 525
 526                let version = if find_preview {
 527                    let dash_index = dir_name.find('-')?;
 528                    if &dir_name[dash_index + 1..] != "preview" {
 529                        return None;
 530                    };
 531                    dir_name[..dash_index].parse::<u32>().ok()?
 532                } else {
 533                    dir_name.parse::<u32>().ok()?
 534                };
 535
 536                let exe_path = entry.path().join("pwsh.exe");
 537                if exe_path.exists() {
 538                    Some((version, exe_path))
 539                } else {
 540                    None
 541                }
 542            })
 543            .max_by_key(|(version, _)| *version)
 544            .map(|(_, path)| path)
 545    }
 546
 547    fn find_pwsh_in_msix(find_preview: bool) -> Option<PathBuf> {
 548        let msix_app_dir =
 549            PathBuf::from(std::env::var_os("LOCALAPPDATA")?).join("Microsoft\\WindowsApps");
 550        if !msix_app_dir.exists() {
 551            return None;
 552        }
 553
 554        let prefix = if find_preview {
 555            "Microsoft.PowerShellPreview_"
 556        } else {
 557            "Microsoft.PowerShell_"
 558        };
 559        msix_app_dir
 560            .read_dir()
 561            .ok()?
 562            .filter_map(|entry| {
 563                let entry = entry.ok()?;
 564                if !matches!(entry.file_type(), Ok(ft) if ft.is_dir()) {
 565                    return None;
 566                }
 567
 568                if !entry.file_name().to_string_lossy().starts_with(prefix) {
 569                    return None;
 570                }
 571
 572                let exe_path = entry.path().join("pwsh.exe");
 573                exe_path.exists().then_some(exe_path)
 574            })
 575            .next()
 576    }
 577
 578    fn find_pwsh_in_scoop() -> Option<PathBuf> {
 579        let pwsh_exe =
 580            PathBuf::from(std::env::var_os("USERPROFILE")?).join("scoop\\shims\\pwsh.exe");
 581        pwsh_exe.exists().then_some(pwsh_exe)
 582    }
 583
 584    static SYSTEM_SHELL: LazyLock<String> = LazyLock::new(|| {
 585        find_pwsh_in_programfiles(false, false)
 586            .or_else(|| find_pwsh_in_programfiles(true, false))
 587            .or_else(|| find_pwsh_in_msix(false))
 588            .or_else(|| find_pwsh_in_programfiles(false, true))
 589            .or_else(|| find_pwsh_in_msix(true))
 590            .or_else(|| find_pwsh_in_programfiles(true, true))
 591            .or_else(find_pwsh_in_scoop)
 592            .map(|p| p.to_string_lossy().to_string())
 593            .unwrap_or("powershell.exe".to_string())
 594    });
 595
 596    (*SYSTEM_SHELL).clone()
 597}
 598
 599pub trait ResultExt<E> {
 600    type Ok;
 601
 602    fn log_err(self) -> Option<Self::Ok>;
 603    /// Assert that this result should never be an error in development or tests.
 604    fn debug_assert_ok(self, reason: &str) -> Self;
 605    fn warn_on_err(self) -> Option<Self::Ok>;
 606    fn log_with_level(self, level: log::Level) -> Option<Self::Ok>;
 607    fn anyhow(self) -> anyhow::Result<Self::Ok>
 608    where
 609        E: Into<anyhow::Error>;
 610}
 611
 612impl<T, E> ResultExt<E> for Result<T, E>
 613where
 614    E: std::fmt::Debug,
 615{
 616    type Ok = T;
 617
 618    #[track_caller]
 619    fn log_err(self) -> Option<T> {
 620        self.log_with_level(log::Level::Error)
 621    }
 622
 623    #[track_caller]
 624    fn debug_assert_ok(self, reason: &str) -> Self {
 625        if let Err(error) = &self {
 626            debug_panic!("{reason} - {error:?}");
 627        }
 628        self
 629    }
 630
 631    #[track_caller]
 632    fn warn_on_err(self) -> Option<T> {
 633        self.log_with_level(log::Level::Warn)
 634    }
 635
 636    #[track_caller]
 637    fn log_with_level(self, level: log::Level) -> Option<T> {
 638        match self {
 639            Ok(value) => Some(value),
 640            Err(error) => {
 641                log_error_with_caller(*Location::caller(), error, level);
 642                None
 643            }
 644        }
 645    }
 646
 647    fn anyhow(self) -> anyhow::Result<T>
 648    where
 649        E: Into<anyhow::Error>,
 650    {
 651        self.map_err(Into::into)
 652    }
 653}
 654
 655fn log_error_with_caller<E>(caller: core::panic::Location<'_>, error: E, level: log::Level)
 656where
 657    E: std::fmt::Debug,
 658{
 659    #[cfg(not(target_os = "windows"))]
 660    let file = caller.file();
 661    #[cfg(target_os = "windows")]
 662    let file = caller.file().replace('\\', "/");
 663    // In this codebase, the first segment of the file path is
 664    // the 'crates' folder, followed by the crate name.
 665    let target = file.split('/').nth(1);
 666
 667    log::logger().log(
 668        &log::Record::builder()
 669            .target(target.unwrap_or(""))
 670            .module_path(target)
 671            .args(format_args!("{:?}", error))
 672            .file(Some(caller.file()))
 673            .line(Some(caller.line()))
 674            .level(level)
 675            .build(),
 676    );
 677}
 678
 679pub fn log_err<E: std::fmt::Debug>(error: &E) {
 680    log_error_with_caller(*Location::caller(), error, log::Level::Warn);
 681}
 682
 683pub trait TryFutureExt {
 684    fn log_err(self) -> LogErrorFuture<Self>
 685    where
 686        Self: Sized;
 687
 688    fn log_tracked_err(self, location: core::panic::Location<'static>) -> LogErrorFuture<Self>
 689    where
 690        Self: Sized;
 691
 692    fn warn_on_err(self) -> LogErrorFuture<Self>
 693    where
 694        Self: Sized;
 695    fn unwrap(self) -> UnwrapFuture<Self>
 696    where
 697        Self: Sized;
 698}
 699
 700impl<F, T, E> TryFutureExt for F
 701where
 702    F: Future<Output = Result<T, E>>,
 703    E: std::fmt::Debug,
 704{
 705    #[track_caller]
 706    fn log_err(self) -> LogErrorFuture<Self>
 707    where
 708        Self: Sized,
 709    {
 710        let location = Location::caller();
 711        LogErrorFuture(self, log::Level::Error, *location)
 712    }
 713
 714    fn log_tracked_err(self, location: core::panic::Location<'static>) -> LogErrorFuture<Self>
 715    where
 716        Self: Sized,
 717    {
 718        LogErrorFuture(self, log::Level::Error, location)
 719    }
 720
 721    #[track_caller]
 722    fn warn_on_err(self) -> LogErrorFuture<Self>
 723    where
 724        Self: Sized,
 725    {
 726        let location = Location::caller();
 727        LogErrorFuture(self, log::Level::Warn, *location)
 728    }
 729
 730    fn unwrap(self) -> UnwrapFuture<Self>
 731    where
 732        Self: Sized,
 733    {
 734        UnwrapFuture(self)
 735    }
 736}
 737
 738#[must_use]
 739pub struct LogErrorFuture<F>(F, log::Level, core::panic::Location<'static>);
 740
 741impl<F, T, E> Future for LogErrorFuture<F>
 742where
 743    F: Future<Output = Result<T, E>>,
 744    E: std::fmt::Debug,
 745{
 746    type Output = Option<T>;
 747
 748    fn poll(self: Pin<&mut Self>, cx: &mut Context) -> Poll<Self::Output> {
 749        let level = self.1;
 750        let location = self.2;
 751        let inner = unsafe { Pin::new_unchecked(&mut self.get_unchecked_mut().0) };
 752        match inner.poll(cx) {
 753            Poll::Ready(output) => Poll::Ready(match output {
 754                Ok(output) => Some(output),
 755                Err(error) => {
 756                    log_error_with_caller(location, error, level);
 757                    None
 758                }
 759            }),
 760            Poll::Pending => Poll::Pending,
 761        }
 762    }
 763}
 764
 765pub struct UnwrapFuture<F>(F);
 766
 767impl<F, T, E> Future for UnwrapFuture<F>
 768where
 769    F: Future<Output = Result<T, E>>,
 770    E: std::fmt::Debug,
 771{
 772    type Output = T;
 773
 774    fn poll(self: Pin<&mut Self>, cx: &mut Context) -> Poll<Self::Output> {
 775        let inner = unsafe { Pin::new_unchecked(&mut self.get_unchecked_mut().0) };
 776        match inner.poll(cx) {
 777            Poll::Ready(result) => Poll::Ready(result.unwrap()),
 778            Poll::Pending => Poll::Pending,
 779        }
 780    }
 781}
 782
 783pub struct Deferred<F: FnOnce()>(Option<F>);
 784
 785impl<F: FnOnce()> Deferred<F> {
 786    /// Drop without running the deferred function.
 787    pub fn abort(mut self) {
 788        self.0.take();
 789    }
 790}
 791
 792impl<F: FnOnce()> Drop for Deferred<F> {
 793    fn drop(&mut self) {
 794        if let Some(f) = self.0.take() {
 795            f()
 796        }
 797    }
 798}
 799
 800/// Run the given function when the returned value is dropped (unless it's cancelled).
 801#[must_use]
 802pub fn defer<F: FnOnce()>(f: F) -> Deferred<F> {
 803    Deferred(Some(f))
 804}
 805
 806#[cfg(any(test, feature = "test-support"))]
 807mod rng {
 808    use rand::{Rng, seq::SliceRandom};
 809    pub struct RandomCharIter<T: Rng> {
 810        rng: T,
 811        simple_text: bool,
 812    }
 813
 814    impl<T: Rng> RandomCharIter<T> {
 815        pub fn new(rng: T) -> Self {
 816            Self {
 817                rng,
 818                simple_text: std::env::var("SIMPLE_TEXT").map_or(false, |v| !v.is_empty()),
 819            }
 820        }
 821
 822        pub fn with_simple_text(mut self) -> Self {
 823            self.simple_text = true;
 824            self
 825        }
 826    }
 827
 828    impl<T: Rng> Iterator for RandomCharIter<T> {
 829        type Item = char;
 830
 831        fn next(&mut self) -> Option<Self::Item> {
 832            if self.simple_text {
 833                return if self.rng.gen_range(0..100) < 5 {
 834                    Some('\n')
 835                } else {
 836                    Some(self.rng.gen_range(b'a'..b'z' + 1).into())
 837                };
 838            }
 839
 840            match self.rng.gen_range(0..100) {
 841                // whitespace
 842                0..=19 => [' ', '\n', '\r', '\t'].choose(&mut self.rng).copied(),
 843                // two-byte greek letters
 844                20..=32 => char::from_u32(self.rng.gen_range(('α' as u32)..('ω' as u32 + 1))),
 845                // // three-byte characters
 846                33..=45 => ['✋', '✅', '❌', '❎', '⭐']
 847                    .choose(&mut self.rng)
 848                    .copied(),
 849                // // four-byte characters
 850                46..=58 => ['🍐', '🏀', '🍗', '🎉'].choose(&mut self.rng).copied(),
 851                // ascii letters
 852                _ => Some(self.rng.gen_range(b'a'..b'z' + 1).into()),
 853            }
 854        }
 855    }
 856}
 857#[cfg(any(test, feature = "test-support"))]
 858pub use rng::RandomCharIter;
 859/// Get an embedded file as a string.
 860pub fn asset_str<A: rust_embed::RustEmbed>(path: &str) -> Cow<'static, str> {
 861    match A::get(path).expect(path).data {
 862        Cow::Borrowed(bytes) => Cow::Borrowed(std::str::from_utf8(bytes).unwrap()),
 863        Cow::Owned(bytes) => Cow::Owned(String::from_utf8(bytes).unwrap()),
 864    }
 865}
 866
 867/// Expands to an immediately-invoked function expression. Good for using the ? operator
 868/// in functions which do not return an Option or Result.
 869///
 870/// Accepts a normal block, an async block, or an async move block.
 871#[macro_export]
 872macro_rules! maybe {
 873    ($block:block) => {
 874        (|| $block)()
 875    };
 876    (async $block:block) => {
 877        (|| async $block)()
 878    };
 879    (async move $block:block) => {
 880        (|| async move $block)()
 881    };
 882}
 883
 884pub trait RangeExt<T> {
 885    fn sorted(&self) -> Self;
 886    fn to_inclusive(&self) -> RangeInclusive<T>;
 887    fn overlaps(&self, other: &Range<T>) -> bool;
 888    fn contains_inclusive(&self, other: &Range<T>) -> bool;
 889}
 890
 891impl<T: Ord + Clone> RangeExt<T> for Range<T> {
 892    fn sorted(&self) -> Self {
 893        cmp::min(&self.start, &self.end).clone()..cmp::max(&self.start, &self.end).clone()
 894    }
 895
 896    fn to_inclusive(&self) -> RangeInclusive<T> {
 897        self.start.clone()..=self.end.clone()
 898    }
 899
 900    fn overlaps(&self, other: &Range<T>) -> bool {
 901        self.start < other.end && other.start < self.end
 902    }
 903
 904    fn contains_inclusive(&self, other: &Range<T>) -> bool {
 905        self.start <= other.start && other.end <= self.end
 906    }
 907}
 908
 909impl<T: Ord + Clone> RangeExt<T> for RangeInclusive<T> {
 910    fn sorted(&self) -> Self {
 911        cmp::min(self.start(), self.end()).clone()..=cmp::max(self.start(), self.end()).clone()
 912    }
 913
 914    fn to_inclusive(&self) -> RangeInclusive<T> {
 915        self.clone()
 916    }
 917
 918    fn overlaps(&self, other: &Range<T>) -> bool {
 919        self.start() < &other.end && &other.start <= self.end()
 920    }
 921
 922    fn contains_inclusive(&self, other: &Range<T>) -> bool {
 923        self.start() <= &other.start && &other.end <= self.end()
 924    }
 925}
 926
 927/// A way to sort strings with starting numbers numerically first, falling back to alphanumeric one,
 928/// case-insensitive.
 929///
 930/// This is useful for turning regular alphanumerically sorted sequences as `1-abc, 10, 11-def, .., 2, 21-abc`
 931/// into `1-abc, 2, 10, 11-def, .., 21-abc`
 932#[derive(Debug, PartialEq, Eq)]
 933pub struct NumericPrefixWithSuffix<'a>(Option<u64>, &'a str);
 934
 935impl<'a> NumericPrefixWithSuffix<'a> {
 936    pub fn from_numeric_prefixed_str(str: &'a str) -> Self {
 937        let i = str.chars().take_while(|c| c.is_ascii_digit()).count();
 938        let (prefix, remainder) = str.split_at(i);
 939
 940        let prefix = prefix.parse().ok();
 941        Self(prefix, remainder)
 942    }
 943}
 944
 945/// When dealing with equality, we need to consider the case of the strings to achieve strict equality
 946/// to handle cases like "a" < "A" instead of "a" == "A".
 947impl Ord for NumericPrefixWithSuffix<'_> {
 948    fn cmp(&self, other: &Self) -> Ordering {
 949        match (self.0, other.0) {
 950            (None, None) => UniCase::new(self.1)
 951                .cmp(&UniCase::new(other.1))
 952                .then_with(|| self.1.cmp(other.1).reverse()),
 953            (None, Some(_)) => Ordering::Greater,
 954            (Some(_), None) => Ordering::Less,
 955            (Some(a), Some(b)) => a.cmp(&b).then_with(|| {
 956                UniCase::new(self.1)
 957                    .cmp(&UniCase::new(other.1))
 958                    .then_with(|| self.1.cmp(other.1).reverse())
 959            }),
 960        }
 961    }
 962}
 963
 964impl PartialOrd for NumericPrefixWithSuffix<'_> {
 965    fn partial_cmp(&self, other: &Self) -> Option<Ordering> {
 966        Some(self.cmp(other))
 967    }
 968}
 969
 970/// Capitalizes the first character of a string.
 971///
 972/// This function takes a string slice as input and returns a new `String` with the first character
 973/// capitalized.
 974///
 975/// # Examples
 976///
 977/// ```
 978/// use util::capitalize;
 979///
 980/// assert_eq!(capitalize("hello"), "Hello");
 981/// assert_eq!(capitalize("WORLD"), "WORLD");
 982/// assert_eq!(capitalize(""), "");
 983/// ```
 984pub fn capitalize(str: &str) -> String {
 985    let mut chars = str.chars();
 986    match chars.next() {
 987        None => String::new(),
 988        Some(first_char) => first_char.to_uppercase().collect::<String>() + chars.as_str(),
 989    }
 990}
 991
 992fn emoji_regex() -> &'static Regex {
 993    static EMOJI_REGEX: LazyLock<Regex> =
 994        LazyLock::new(|| Regex::new("(\\p{Emoji}|\u{200D})").unwrap());
 995    &EMOJI_REGEX
 996}
 997
 998/// Returns true if the given string consists of emojis only.
 999/// E.g. "👨‍👩‍👧‍👧👋" will return true, but "👋!" will return false.
1000pub fn word_consists_of_emojis(s: &str) -> bool {
1001    let mut prev_end = 0;
1002    for capture in emoji_regex().find_iter(s) {
1003        if capture.start() != prev_end {
1004            return false;
1005        }
1006        prev_end = capture.end();
1007    }
1008    prev_end == s.len()
1009}
1010
1011pub fn default<D: Default>() -> D {
1012    Default::default()
1013}
1014
1015pub fn get_system_shell() -> String {
1016    #[cfg(target_os = "windows")]
1017    {
1018        get_windows_system_shell()
1019    }
1020
1021    #[cfg(not(target_os = "windows"))]
1022    {
1023        std::env::var("SHELL").unwrap_or("/bin/sh".to_string())
1024    }
1025}
1026
1027#[derive(Debug)]
1028pub enum ConnectionResult<O> {
1029    Timeout,
1030    ConnectionReset,
1031    Result(anyhow::Result<O>),
1032}
1033
1034impl<O> ConnectionResult<O> {
1035    pub fn into_response(self) -> anyhow::Result<O> {
1036        match self {
1037            ConnectionResult::Timeout => anyhow::bail!("Request timed out"),
1038            ConnectionResult::ConnectionReset => anyhow::bail!("Server reset the connection"),
1039            ConnectionResult::Result(r) => r,
1040        }
1041    }
1042}
1043
1044impl<O> From<anyhow::Result<O>> for ConnectionResult<O> {
1045    fn from(result: anyhow::Result<O>) -> Self {
1046        ConnectionResult::Result(result)
1047    }
1048}
1049
1050#[cfg(test)]
1051mod tests {
1052    use super::*;
1053
1054    #[test]
1055    fn test_extend_sorted() {
1056        let mut vec = vec![];
1057
1058        extend_sorted(&mut vec, vec![21, 17, 13, 8, 1, 0], 5, |a, b| b.cmp(a));
1059        assert_eq!(vec, &[21, 17, 13, 8, 1]);
1060
1061        extend_sorted(&mut vec, vec![101, 19, 17, 8, 2], 8, |a, b| b.cmp(a));
1062        assert_eq!(vec, &[101, 21, 19, 17, 13, 8, 2, 1]);
1063
1064        extend_sorted(&mut vec, vec![1000, 19, 17, 9, 5], 8, |a, b| b.cmp(a));
1065        assert_eq!(vec, &[1000, 101, 21, 19, 17, 13, 9, 8]);
1066    }
1067
1068    #[test]
1069    fn test_truncate_to_bottom_n_sorted_by() {
1070        let mut vec: Vec<u32> = vec![5, 2, 3, 4, 1];
1071        truncate_to_bottom_n_sorted_by(&mut vec, 10, &u32::cmp);
1072        assert_eq!(vec, &[1, 2, 3, 4, 5]);
1073
1074        vec = vec![5, 2, 3, 4, 1];
1075        truncate_to_bottom_n_sorted_by(&mut vec, 5, &u32::cmp);
1076        assert_eq!(vec, &[1, 2, 3, 4, 5]);
1077
1078        vec = vec![5, 2, 3, 4, 1];
1079        truncate_to_bottom_n_sorted_by(&mut vec, 4, &u32::cmp);
1080        assert_eq!(vec, &[1, 2, 3, 4]);
1081
1082        vec = vec![5, 2, 3, 4, 1];
1083        truncate_to_bottom_n_sorted_by(&mut vec, 1, &u32::cmp);
1084        assert_eq!(vec, &[1]);
1085
1086        vec = vec![5, 2, 3, 4, 1];
1087        truncate_to_bottom_n_sorted_by(&mut vec, 0, &u32::cmp);
1088        assert!(vec.is_empty());
1089    }
1090
1091    #[test]
1092    fn test_iife() {
1093        fn option_returning_function() -> Option<()> {
1094            None
1095        }
1096
1097        let foo = maybe!({
1098            option_returning_function()?;
1099            Some(())
1100        });
1101
1102        assert_eq!(foo, None);
1103    }
1104
1105    #[test]
1106    fn test_truncate_and_trailoff() {
1107        assert_eq!(truncate_and_trailoff("", 5), "");
1108        assert_eq!(truncate_and_trailoff("aaaaaa", 7), "aaaaaa");
1109        assert_eq!(truncate_and_trailoff("aaaaaa", 6), "aaaaaa");
1110        assert_eq!(truncate_and_trailoff("aaaaaa", 5), "aaaaa…");
1111        assert_eq!(truncate_and_trailoff("èèèèèè", 7), "èèèèèè");
1112        assert_eq!(truncate_and_trailoff("èèèèèè", 6), "èèèèèè");
1113        assert_eq!(truncate_and_trailoff("èèèèèè", 5), "èèèèè…");
1114    }
1115
1116    #[test]
1117    fn test_truncate_and_remove_front() {
1118        assert_eq!(truncate_and_remove_front("", 5), "");
1119        assert_eq!(truncate_and_remove_front("aaaaaa", 7), "aaaaaa");
1120        assert_eq!(truncate_and_remove_front("aaaaaa", 6), "aaaaaa");
1121        assert_eq!(truncate_and_remove_front("aaaaaa", 5), "…aaaaa");
1122        assert_eq!(truncate_and_remove_front("èèèèèè", 7), "èèèèèè");
1123        assert_eq!(truncate_and_remove_front("èèèèèè", 6), "èèèèèè");
1124        assert_eq!(truncate_and_remove_front("èèèèèè", 5), "…èèèèè");
1125    }
1126
1127    #[test]
1128    fn test_numeric_prefix_str_method() {
1129        let target = "1a";
1130        assert_eq!(
1131            NumericPrefixWithSuffix::from_numeric_prefixed_str(target),
1132            NumericPrefixWithSuffix(Some(1), "a")
1133        );
1134
1135        let target = "12ab";
1136        assert_eq!(
1137            NumericPrefixWithSuffix::from_numeric_prefixed_str(target),
1138            NumericPrefixWithSuffix(Some(12), "ab")
1139        );
1140
1141        let target = "12_ab";
1142        assert_eq!(
1143            NumericPrefixWithSuffix::from_numeric_prefixed_str(target),
1144            NumericPrefixWithSuffix(Some(12), "_ab")
1145        );
1146
1147        let target = "1_2ab";
1148        assert_eq!(
1149            NumericPrefixWithSuffix::from_numeric_prefixed_str(target),
1150            NumericPrefixWithSuffix(Some(1), "_2ab")
1151        );
1152
1153        let target = "1.2";
1154        assert_eq!(
1155            NumericPrefixWithSuffix::from_numeric_prefixed_str(target),
1156            NumericPrefixWithSuffix(Some(1), ".2")
1157        );
1158
1159        let target = "1.2_a";
1160        assert_eq!(
1161            NumericPrefixWithSuffix::from_numeric_prefixed_str(target),
1162            NumericPrefixWithSuffix(Some(1), ".2_a")
1163        );
1164
1165        let target = "12.2_a";
1166        assert_eq!(
1167            NumericPrefixWithSuffix::from_numeric_prefixed_str(target),
1168            NumericPrefixWithSuffix(Some(12), ".2_a")
1169        );
1170
1171        let target = "12a.2_a";
1172        assert_eq!(
1173            NumericPrefixWithSuffix::from_numeric_prefixed_str(target),
1174            NumericPrefixWithSuffix(Some(12), "a.2_a")
1175        );
1176    }
1177
1178    #[test]
1179    fn test_numeric_prefix_with_suffix() {
1180        let mut sorted = vec!["1-abc", "10", "11def", "2", "21-abc"];
1181        sorted.sort_by_key(|s| NumericPrefixWithSuffix::from_numeric_prefixed_str(s));
1182        assert_eq!(sorted, ["1-abc", "2", "10", "11def", "21-abc"]);
1183
1184        for numeric_prefix_less in ["numeric_prefix_less", "aaa", "~™£"] {
1185            assert_eq!(
1186                NumericPrefixWithSuffix::from_numeric_prefixed_str(numeric_prefix_less),
1187                NumericPrefixWithSuffix(None, numeric_prefix_less),
1188                "String without numeric prefix `{numeric_prefix_less}` should not be converted into NumericPrefixWithSuffix"
1189            )
1190        }
1191    }
1192
1193    #[test]
1194    fn test_word_consists_of_emojis() {
1195        let words_to_test = vec![
1196            ("👨‍👩‍👧‍👧👋🥒", true),
1197            ("👋", true),
1198            ("!👋", false),
1199            ("👋!", false),
1200            ("👋 ", false),
1201            (" 👋", false),
1202            ("Test", false),
1203        ];
1204
1205        for (text, expected_result) in words_to_test {
1206            assert_eq!(word_consists_of_emojis(text), expected_result);
1207        }
1208    }
1209
1210    #[test]
1211    fn test_truncate_lines_and_trailoff() {
1212        let text = r#"Line 1
1213Line 2
1214Line 3"#;
1215
1216        assert_eq!(
1217            truncate_lines_and_trailoff(text, 2),
1218            r#"Line 1
1219…"#
1220        );
1221
1222        assert_eq!(
1223            truncate_lines_and_trailoff(text, 3),
1224            r#"Line 1
1225Line 2
1226…"#
1227        );
1228
1229        assert_eq!(
1230            truncate_lines_and_trailoff(text, 4),
1231            r#"Line 1
1232Line 2
1233Line 3"#
1234        );
1235    }
1236
1237    #[test]
1238    fn test_iterate_expanded_and_wrapped_usize_range() {
1239        // Neither wrap
1240        assert_eq!(
1241            iterate_expanded_and_wrapped_usize_range(2..4, 1, 1, 8).collect::<Vec<usize>>(),
1242            (1..5).collect::<Vec<usize>>()
1243        );
1244        // Start wraps
1245        assert_eq!(
1246            iterate_expanded_and_wrapped_usize_range(2..4, 3, 1, 8).collect::<Vec<usize>>(),
1247            ((0..5).chain(7..8)).collect::<Vec<usize>>()
1248        );
1249        // Start wraps all the way around
1250        assert_eq!(
1251            iterate_expanded_and_wrapped_usize_range(2..4, 5, 1, 8).collect::<Vec<usize>>(),
1252            (0..8).collect::<Vec<usize>>()
1253        );
1254        // Start wraps all the way around and past 0
1255        assert_eq!(
1256            iterate_expanded_and_wrapped_usize_range(2..4, 10, 1, 8).collect::<Vec<usize>>(),
1257            (0..8).collect::<Vec<usize>>()
1258        );
1259        // End wraps
1260        assert_eq!(
1261            iterate_expanded_and_wrapped_usize_range(3..5, 1, 4, 8).collect::<Vec<usize>>(),
1262            (0..1).chain(2..8).collect::<Vec<usize>>()
1263        );
1264        // End wraps all the way around
1265        assert_eq!(
1266            iterate_expanded_and_wrapped_usize_range(3..5, 1, 5, 8).collect::<Vec<usize>>(),
1267            (0..8).collect::<Vec<usize>>()
1268        );
1269        // End wraps all the way around and past the end
1270        assert_eq!(
1271            iterate_expanded_and_wrapped_usize_range(3..5, 1, 10, 8).collect::<Vec<usize>>(),
1272            (0..8).collect::<Vec<usize>>()
1273        );
1274        // Both start and end wrap
1275        assert_eq!(
1276            iterate_expanded_and_wrapped_usize_range(3..5, 4, 4, 8).collect::<Vec<usize>>(),
1277            (0..8).collect::<Vec<usize>>()
1278        );
1279    }
1280}