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)]
 266pub fn load_login_shell_environment() -> Result<()> {
 267    load_shell_from_passwd().log_err();
 268
 269    // If possible, we want to `cd` in the user's `$HOME` to trigger programs
 270    // such as direnv, asdf, mise, ... to adjust the PATH. These tools often hook
 271    // into shell's `cd` command (and hooks) to manipulate env.
 272    // We do this so that we get the env a user would have when spawning a shell
 273    // in home directory.
 274    for (name, value) in shell_env::capture(paths::home_dir())? {
 275        unsafe { env::set_var(&name, &value) };
 276    }
 277
 278    log::info!(
 279        "set environment variables from shell:{}, path:{}",
 280        std::env::var("SHELL").unwrap_or_default(),
 281        std::env::var("PATH").unwrap_or_default(),
 282    );
 283
 284    Ok(())
 285}
 286
 287/// Configures the process to start a new session, to prevent interactive shells from taking control
 288/// of the terminal.
 289///
 290/// For more details: https://registerspill.thorstenball.com/p/how-to-lose-control-of-your-shell
 291pub fn set_pre_exec_to_start_new_session(
 292    command: &mut std::process::Command,
 293) -> &mut std::process::Command {
 294    // safety: code in pre_exec should be signal safe.
 295    // https://man7.org/linux/man-pages/man7/signal-safety.7.html
 296    #[cfg(not(target_os = "windows"))]
 297    unsafe {
 298        use std::os::unix::process::CommandExt;
 299        command.pre_exec(|| {
 300            libc::setsid();
 301            Ok(())
 302        });
 303    };
 304    command
 305}
 306
 307pub fn merge_json_lenient_value_into(
 308    source: serde_json_lenient::Value,
 309    target: &mut serde_json_lenient::Value,
 310) {
 311    match (source, target) {
 312        (serde_json_lenient::Value::Object(source), serde_json_lenient::Value::Object(target)) => {
 313            for (key, value) in source {
 314                if let Some(target) = target.get_mut(&key) {
 315                    merge_json_lenient_value_into(value, target);
 316                } else {
 317                    target.insert(key, value);
 318                }
 319            }
 320        }
 321
 322        (serde_json_lenient::Value::Array(source), serde_json_lenient::Value::Array(target)) => {
 323            for value in source {
 324                target.push(value);
 325            }
 326        }
 327
 328        (source, target) => *target = source,
 329    }
 330}
 331
 332pub fn merge_json_value_into(source: serde_json::Value, target: &mut serde_json::Value) {
 333    use serde_json::Value;
 334
 335    match (source, target) {
 336        (Value::Object(source), Value::Object(target)) => {
 337            for (key, value) in source {
 338                if let Some(target) = target.get_mut(&key) {
 339                    merge_json_value_into(value, target);
 340                } else {
 341                    target.insert(key, value);
 342                }
 343            }
 344        }
 345
 346        (Value::Array(source), Value::Array(target)) => {
 347            for value in source {
 348                target.push(value);
 349            }
 350        }
 351
 352        (source, target) => *target = source,
 353    }
 354}
 355
 356pub fn merge_non_null_json_value_into(source: serde_json::Value, target: &mut serde_json::Value) {
 357    use serde_json::Value;
 358    if let Value::Object(source_object) = source {
 359        let target_object = if let Value::Object(target) = target {
 360            target
 361        } else {
 362            *target = Value::Object(Default::default());
 363            target.as_object_mut().unwrap()
 364        };
 365        for (key, value) in source_object {
 366            if let Some(target) = target_object.get_mut(&key) {
 367                merge_non_null_json_value_into(value, target);
 368            } else if !value.is_null() {
 369                target_object.insert(key, value);
 370            }
 371        }
 372    } else if !source.is_null() {
 373        *target = source
 374    }
 375}
 376
 377pub fn measure<R>(label: &str, f: impl FnOnce() -> R) -> R {
 378    static ZED_MEASUREMENTS: OnceLock<bool> = OnceLock::new();
 379    let zed_measurements = ZED_MEASUREMENTS.get_or_init(|| {
 380        env::var("ZED_MEASUREMENTS")
 381            .map(|measurements| measurements == "1" || measurements == "true")
 382            .unwrap_or(false)
 383    });
 384
 385    if *zed_measurements {
 386        let start = Instant::now();
 387        let result = f();
 388        let elapsed = start.elapsed();
 389        eprintln!("{}: {:?}", label, elapsed);
 390        result
 391    } else {
 392        f()
 393    }
 394}
 395
 396pub fn expanded_and_wrapped_usize_range(
 397    range: Range<usize>,
 398    additional_before: usize,
 399    additional_after: usize,
 400    wrap_length: usize,
 401) -> impl Iterator<Item = usize> {
 402    let start_wraps = range.start < additional_before;
 403    let end_wraps = wrap_length < range.end + additional_after;
 404    if start_wraps && end_wraps {
 405        Either::Left(0..wrap_length)
 406    } else if start_wraps {
 407        let wrapped_start = (range.start + wrap_length).saturating_sub(additional_before);
 408        if wrapped_start <= range.end {
 409            Either::Left(0..wrap_length)
 410        } else {
 411            Either::Right((0..range.end + additional_after).chain(wrapped_start..wrap_length))
 412        }
 413    } else if end_wraps {
 414        let wrapped_end = range.end + additional_after - wrap_length;
 415        if range.start <= wrapped_end {
 416            Either::Left(0..wrap_length)
 417        } else {
 418            Either::Right((0..wrapped_end).chain(range.start - additional_before..wrap_length))
 419        }
 420    } else {
 421        Either::Left((range.start - additional_before)..(range.end + additional_after))
 422    }
 423}
 424
 425/// Yields `[i, i + 1, i - 1, i + 2, ..]`, each modulo `wrap_length` and bounded by
 426/// `additional_before` and `additional_after`. If the wrapping causes overlap, duplicates are not
 427/// emitted. If wrap_length is 0, nothing is yielded.
 428pub fn wrapped_usize_outward_from(
 429    start: usize,
 430    additional_before: usize,
 431    additional_after: usize,
 432    wrap_length: usize,
 433) -> impl Iterator<Item = usize> {
 434    let mut count = 0;
 435    let mut after_offset = 1;
 436    let mut before_offset = 1;
 437
 438    std::iter::from_fn(move || {
 439        count += 1;
 440        if count > wrap_length {
 441            None
 442        } else if count == 1 {
 443            Some(start % wrap_length)
 444        } else if after_offset <= additional_after && after_offset <= before_offset {
 445            let value = (start + after_offset) % wrap_length;
 446            after_offset += 1;
 447            Some(value)
 448        } else if before_offset <= additional_before {
 449            let value = (start + wrap_length - before_offset) % wrap_length;
 450            before_offset += 1;
 451            Some(value)
 452        } else if after_offset <= additional_after {
 453            let value = (start + after_offset) % wrap_length;
 454            after_offset += 1;
 455            Some(value)
 456        } else {
 457            None
 458        }
 459    })
 460}
 461
 462#[cfg(target_os = "windows")]
 463pub fn get_windows_system_shell() -> String {
 464    use std::path::PathBuf;
 465
 466    fn find_pwsh_in_programfiles(find_alternate: bool, find_preview: bool) -> Option<PathBuf> {
 467        #[cfg(target_pointer_width = "64")]
 468        let env_var = if find_alternate {
 469            "ProgramFiles(x86)"
 470        } else {
 471            "ProgramFiles"
 472        };
 473
 474        #[cfg(target_pointer_width = "32")]
 475        let env_var = if find_alternate {
 476            "ProgramW6432"
 477        } else {
 478            "ProgramFiles"
 479        };
 480
 481        let install_base_dir = PathBuf::from(std::env::var_os(env_var)?).join("PowerShell");
 482        install_base_dir
 483            .read_dir()
 484            .ok()?
 485            .filter_map(Result::ok)
 486            .filter(|entry| matches!(entry.file_type(), Ok(ft) if ft.is_dir()))
 487            .filter_map(|entry| {
 488                let dir_name = entry.file_name();
 489                let dir_name = dir_name.to_string_lossy();
 490
 491                let version = if find_preview {
 492                    let dash_index = dir_name.find('-')?;
 493                    if &dir_name[dash_index + 1..] != "preview" {
 494                        return None;
 495                    };
 496                    dir_name[..dash_index].parse::<u32>().ok()?
 497                } else {
 498                    dir_name.parse::<u32>().ok()?
 499                };
 500
 501                let exe_path = entry.path().join("pwsh.exe");
 502                if exe_path.exists() {
 503                    Some((version, exe_path))
 504                } else {
 505                    None
 506                }
 507            })
 508            .max_by_key(|(version, _)| *version)
 509            .map(|(_, path)| path)
 510    }
 511
 512    fn find_pwsh_in_msix(find_preview: bool) -> Option<PathBuf> {
 513        let msix_app_dir =
 514            PathBuf::from(std::env::var_os("LOCALAPPDATA")?).join("Microsoft\\WindowsApps");
 515        if !msix_app_dir.exists() {
 516            return None;
 517        }
 518
 519        let prefix = if find_preview {
 520            "Microsoft.PowerShellPreview_"
 521        } else {
 522            "Microsoft.PowerShell_"
 523        };
 524        msix_app_dir
 525            .read_dir()
 526            .ok()?
 527            .filter_map(|entry| {
 528                let entry = entry.ok()?;
 529                if !matches!(entry.file_type(), Ok(ft) if ft.is_dir()) {
 530                    return None;
 531                }
 532
 533                if !entry.file_name().to_string_lossy().starts_with(prefix) {
 534                    return None;
 535                }
 536
 537                let exe_path = entry.path().join("pwsh.exe");
 538                exe_path.exists().then_some(exe_path)
 539            })
 540            .next()
 541    }
 542
 543    fn find_pwsh_in_scoop() -> Option<PathBuf> {
 544        let pwsh_exe =
 545            PathBuf::from(std::env::var_os("USERPROFILE")?).join("scoop\\shims\\pwsh.exe");
 546        pwsh_exe.exists().then_some(pwsh_exe)
 547    }
 548
 549    static SYSTEM_SHELL: LazyLock<String> = LazyLock::new(|| {
 550        find_pwsh_in_programfiles(false, false)
 551            .or_else(|| find_pwsh_in_programfiles(true, false))
 552            .or_else(|| find_pwsh_in_msix(false))
 553            .or_else(|| find_pwsh_in_programfiles(false, true))
 554            .or_else(|| find_pwsh_in_msix(true))
 555            .or_else(|| find_pwsh_in_programfiles(true, true))
 556            .or_else(find_pwsh_in_scoop)
 557            .map(|p| p.to_string_lossy().to_string())
 558            .unwrap_or("powershell.exe".to_string())
 559    });
 560
 561    (*SYSTEM_SHELL).clone()
 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, the first segment of the file path is
 629    // the 'crates' folder, followed by the crate name.
 630    let target = file.split('/').nth(1);
 631
 632    log::logger().log(
 633        &log::Record::builder()
 634            .target(target.unwrap_or(""))
 635            .module_path(target)
 636            .args(format_args!("{:?}", error))
 637            .file(Some(caller.file()))
 638            .line(Some(caller.line()))
 639            .level(level)
 640            .build(),
 641    );
 642}
 643
 644pub fn log_err<E: std::fmt::Debug>(error: &E) {
 645    log_error_with_caller(*Location::caller(), error, log::Level::Warn);
 646}
 647
 648pub trait TryFutureExt {
 649    fn log_err(self) -> LogErrorFuture<Self>
 650    where
 651        Self: Sized;
 652
 653    fn log_tracked_err(self, location: core::panic::Location<'static>) -> LogErrorFuture<Self>
 654    where
 655        Self: Sized;
 656
 657    fn warn_on_err(self) -> LogErrorFuture<Self>
 658    where
 659        Self: Sized;
 660    fn unwrap(self) -> UnwrapFuture<Self>
 661    where
 662        Self: Sized;
 663}
 664
 665impl<F, T, E> TryFutureExt for F
 666where
 667    F: Future<Output = Result<T, E>>,
 668    E: std::fmt::Debug,
 669{
 670    #[track_caller]
 671    fn log_err(self) -> LogErrorFuture<Self>
 672    where
 673        Self: Sized,
 674    {
 675        let location = Location::caller();
 676        LogErrorFuture(self, log::Level::Error, *location)
 677    }
 678
 679    fn log_tracked_err(self, location: core::panic::Location<'static>) -> LogErrorFuture<Self>
 680    where
 681        Self: Sized,
 682    {
 683        LogErrorFuture(self, log::Level::Error, location)
 684    }
 685
 686    #[track_caller]
 687    fn warn_on_err(self) -> LogErrorFuture<Self>
 688    where
 689        Self: Sized,
 690    {
 691        let location = Location::caller();
 692        LogErrorFuture(self, log::Level::Warn, *location)
 693    }
 694
 695    fn unwrap(self) -> UnwrapFuture<Self>
 696    where
 697        Self: Sized,
 698    {
 699        UnwrapFuture(self)
 700    }
 701}
 702
 703#[must_use]
 704pub struct LogErrorFuture<F>(F, log::Level, core::panic::Location<'static>);
 705
 706impl<F, T, E> Future for LogErrorFuture<F>
 707where
 708    F: Future<Output = Result<T, E>>,
 709    E: std::fmt::Debug,
 710{
 711    type Output = Option<T>;
 712
 713    fn poll(self: Pin<&mut Self>, cx: &mut Context) -> Poll<Self::Output> {
 714        let level = self.1;
 715        let location = self.2;
 716        let inner = unsafe { Pin::new_unchecked(&mut self.get_unchecked_mut().0) };
 717        match inner.poll(cx) {
 718            Poll::Ready(output) => Poll::Ready(match output {
 719                Ok(output) => Some(output),
 720                Err(error) => {
 721                    log_error_with_caller(location, error, level);
 722                    None
 723                }
 724            }),
 725            Poll::Pending => Poll::Pending,
 726        }
 727    }
 728}
 729
 730pub struct UnwrapFuture<F>(F);
 731
 732impl<F, T, E> Future for UnwrapFuture<F>
 733where
 734    F: Future<Output = Result<T, E>>,
 735    E: std::fmt::Debug,
 736{
 737    type Output = T;
 738
 739    fn poll(self: Pin<&mut Self>, cx: &mut Context) -> Poll<Self::Output> {
 740        let inner = unsafe { Pin::new_unchecked(&mut self.get_unchecked_mut().0) };
 741        match inner.poll(cx) {
 742            Poll::Ready(result) => Poll::Ready(result.unwrap()),
 743            Poll::Pending => Poll::Pending,
 744        }
 745    }
 746}
 747
 748pub struct Deferred<F: FnOnce()>(Option<F>);
 749
 750impl<F: FnOnce()> Deferred<F> {
 751    /// Drop without running the deferred function.
 752    pub fn abort(mut self) {
 753        self.0.take();
 754    }
 755}
 756
 757impl<F: FnOnce()> Drop for Deferred<F> {
 758    fn drop(&mut self) {
 759        if let Some(f) = self.0.take() {
 760            f()
 761        }
 762    }
 763}
 764
 765/// Run the given function when the returned value is dropped (unless it's cancelled).
 766#[must_use]
 767pub fn defer<F: FnOnce()>(f: F) -> Deferred<F> {
 768    Deferred(Some(f))
 769}
 770
 771#[cfg(any(test, feature = "test-support"))]
 772mod rng {
 773    use rand::{Rng, seq::SliceRandom};
 774    pub struct RandomCharIter<T: Rng> {
 775        rng: T,
 776        simple_text: bool,
 777    }
 778
 779    impl<T: Rng> RandomCharIter<T> {
 780        pub fn new(rng: T) -> Self {
 781            Self {
 782                rng,
 783                simple_text: std::env::var("SIMPLE_TEXT").map_or(false, |v| !v.is_empty()),
 784            }
 785        }
 786
 787        pub fn with_simple_text(mut self) -> Self {
 788            self.simple_text = true;
 789            self
 790        }
 791    }
 792
 793    impl<T: Rng> Iterator for RandomCharIter<T> {
 794        type Item = char;
 795
 796        fn next(&mut self) -> Option<Self::Item> {
 797            if self.simple_text {
 798                return if self.rng.gen_range(0..100) < 5 {
 799                    Some('\n')
 800                } else {
 801                    Some(self.rng.gen_range(b'a'..b'z' + 1).into())
 802                };
 803            }
 804
 805            match self.rng.gen_range(0..100) {
 806                // whitespace
 807                0..=19 => [' ', '\n', '\r', '\t'].choose(&mut self.rng).copied(),
 808                // two-byte greek letters
 809                20..=32 => char::from_u32(self.rng.gen_range(('Ξ±' as u32)..('Ο‰' as u32 + 1))),
 810                // // three-byte characters
 811                33..=45 => ['βœ‹', 'βœ…', '❌', '❎', '⭐']
 812                    .choose(&mut self.rng)
 813                    .copied(),
 814                // // four-byte characters
 815                46..=58 => ['🍐', 'πŸ€', 'πŸ—', 'πŸŽ‰'].choose(&mut self.rng).copied(),
 816                // ascii letters
 817                _ => Some(self.rng.gen_range(b'a'..b'z' + 1).into()),
 818            }
 819        }
 820    }
 821}
 822#[cfg(any(test, feature = "test-support"))]
 823pub use rng::RandomCharIter;
 824/// Get an embedded file as a string.
 825pub fn asset_str<A: rust_embed::RustEmbed>(path: &str) -> Cow<'static, str> {
 826    match A::get(path).expect(path).data {
 827        Cow::Borrowed(bytes) => Cow::Borrowed(std::str::from_utf8(bytes).unwrap()),
 828        Cow::Owned(bytes) => Cow::Owned(String::from_utf8(bytes).unwrap()),
 829    }
 830}
 831
 832/// Expands to an immediately-invoked function expression. Good for using the ? operator
 833/// in functions which do not return an Option or Result.
 834///
 835/// Accepts a normal block, an async block, or an async move block.
 836#[macro_export]
 837macro_rules! maybe {
 838    ($block:block) => {
 839        (|| $block)()
 840    };
 841    (async $block:block) => {
 842        (|| async $block)()
 843    };
 844    (async move $block:block) => {
 845        (|| async move $block)()
 846    };
 847}
 848
 849pub trait RangeExt<T> {
 850    fn sorted(&self) -> Self;
 851    fn to_inclusive(&self) -> RangeInclusive<T>;
 852    fn overlaps(&self, other: &Range<T>) -> bool;
 853    fn contains_inclusive(&self, other: &Range<T>) -> bool;
 854}
 855
 856impl<T: Ord + Clone> RangeExt<T> for Range<T> {
 857    fn sorted(&self) -> Self {
 858        cmp::min(&self.start, &self.end).clone()..cmp::max(&self.start, &self.end).clone()
 859    }
 860
 861    fn to_inclusive(&self) -> RangeInclusive<T> {
 862        self.start.clone()..=self.end.clone()
 863    }
 864
 865    fn overlaps(&self, other: &Range<T>) -> bool {
 866        self.start < other.end && other.start < self.end
 867    }
 868
 869    fn contains_inclusive(&self, other: &Range<T>) -> bool {
 870        self.start <= other.start && other.end <= self.end
 871    }
 872}
 873
 874impl<T: Ord + Clone> RangeExt<T> for RangeInclusive<T> {
 875    fn sorted(&self) -> Self {
 876        cmp::min(self.start(), self.end()).clone()..=cmp::max(self.start(), self.end()).clone()
 877    }
 878
 879    fn to_inclusive(&self) -> RangeInclusive<T> {
 880        self.clone()
 881    }
 882
 883    fn overlaps(&self, other: &Range<T>) -> bool {
 884        self.start() < &other.end && &other.start <= self.end()
 885    }
 886
 887    fn contains_inclusive(&self, other: &Range<T>) -> bool {
 888        self.start() <= &other.start && &other.end <= self.end()
 889    }
 890}
 891
 892/// A way to sort strings with starting numbers numerically first, falling back to alphanumeric one,
 893/// case-insensitive.
 894///
 895/// This is useful for turning regular alphanumerically sorted sequences as `1-abc, 10, 11-def, .., 2, 21-abc`
 896/// into `1-abc, 2, 10, 11-def, .., 21-abc`
 897#[derive(Debug, PartialEq, Eq)]
 898pub struct NumericPrefixWithSuffix<'a>(Option<u64>, &'a str);
 899
 900impl<'a> NumericPrefixWithSuffix<'a> {
 901    pub fn from_numeric_prefixed_str(str: &'a str) -> Self {
 902        let i = str.chars().take_while(|c| c.is_ascii_digit()).count();
 903        let (prefix, remainder) = str.split_at(i);
 904
 905        let prefix = prefix.parse().ok();
 906        Self(prefix, remainder)
 907    }
 908}
 909
 910/// When dealing with equality, we need to consider the case of the strings to achieve strict equality
 911/// to handle cases like "a" < "A" instead of "a" == "A".
 912impl Ord for NumericPrefixWithSuffix<'_> {
 913    fn cmp(&self, other: &Self) -> Ordering {
 914        match (self.0, other.0) {
 915            (None, None) => UniCase::new(self.1)
 916                .cmp(&UniCase::new(other.1))
 917                .then_with(|| self.1.cmp(other.1).reverse()),
 918            (None, Some(_)) => Ordering::Greater,
 919            (Some(_), None) => Ordering::Less,
 920            (Some(a), Some(b)) => a.cmp(&b).then_with(|| {
 921                UniCase::new(self.1)
 922                    .cmp(&UniCase::new(other.1))
 923                    .then_with(|| self.1.cmp(other.1).reverse())
 924            }),
 925        }
 926    }
 927}
 928
 929impl PartialOrd for NumericPrefixWithSuffix<'_> {
 930    fn partial_cmp(&self, other: &Self) -> Option<Ordering> {
 931        Some(self.cmp(other))
 932    }
 933}
 934
 935/// Capitalizes the first character of a string.
 936///
 937/// This function takes a string slice as input and returns a new `String` with the first character
 938/// capitalized.
 939///
 940/// # Examples
 941///
 942/// ```
 943/// use util::capitalize;
 944///
 945/// assert_eq!(capitalize("hello"), "Hello");
 946/// assert_eq!(capitalize("WORLD"), "WORLD");
 947/// assert_eq!(capitalize(""), "");
 948/// ```
 949pub fn capitalize(str: &str) -> String {
 950    let mut chars = str.chars();
 951    match chars.next() {
 952        None => String::new(),
 953        Some(first_char) => first_char.to_uppercase().collect::<String>() + chars.as_str(),
 954    }
 955}
 956
 957fn emoji_regex() -> &'static Regex {
 958    static EMOJI_REGEX: LazyLock<Regex> =
 959        LazyLock::new(|| Regex::new("(\\p{Emoji}|\u{200D})").unwrap());
 960    &EMOJI_REGEX
 961}
 962
 963/// Returns true if the given string consists of emojis only.
 964/// E.g. "πŸ‘¨β€πŸ‘©β€πŸ‘§β€πŸ‘§πŸ‘‹" will return true, but "πŸ‘‹!" will return false.
 965pub fn word_consists_of_emojis(s: &str) -> bool {
 966    let mut prev_end = 0;
 967    for capture in emoji_regex().find_iter(s) {
 968        if capture.start() != prev_end {
 969            return false;
 970        }
 971        prev_end = capture.end();
 972    }
 973    prev_end == s.len()
 974}
 975
 976/// Similar to `str::split`, but also provides byte-offset ranges of the results. Unlike
 977/// `str::split`, this is not generic on pattern types and does not return an `Iterator`.
 978pub fn split_str_with_ranges(s: &str, pat: impl Fn(char) -> bool) -> Vec<(Range<usize>, &str)> {
 979    let mut result = Vec::new();
 980    let mut start = 0;
 981
 982    for (i, ch) in s.char_indices() {
 983        if pat(ch) {
 984            if i > start {
 985                result.push((start..i, &s[start..i]));
 986            }
 987            start = i + ch.len_utf8();
 988        }
 989    }
 990
 991    if s.len() > start {
 992        result.push((start..s.len(), &s[start..s.len()]));
 993    }
 994
 995    result
 996}
 997
 998pub fn default<D: Default>() -> D {
 999    Default::default()
1000}
1001
1002pub fn get_system_shell() -> String {
1003    #[cfg(target_os = "windows")]
1004    {
1005        get_windows_system_shell()
1006    }
1007
1008    #[cfg(not(target_os = "windows"))]
1009    {
1010        std::env::var("SHELL").unwrap_or("/bin/sh".to_string())
1011    }
1012}
1013
1014#[derive(Debug)]
1015pub enum ConnectionResult<O> {
1016    Timeout,
1017    ConnectionReset,
1018    Result(anyhow::Result<O>),
1019}
1020
1021impl<O> ConnectionResult<O> {
1022    pub fn into_response(self) -> anyhow::Result<O> {
1023        match self {
1024            ConnectionResult::Timeout => anyhow::bail!("Request timed out"),
1025            ConnectionResult::ConnectionReset => anyhow::bail!("Server reset the connection"),
1026            ConnectionResult::Result(r) => r,
1027        }
1028    }
1029}
1030
1031impl<O> From<anyhow::Result<O>> for ConnectionResult<O> {
1032    fn from(result: anyhow::Result<O>) -> Self {
1033        ConnectionResult::Result(result)
1034    }
1035}
1036
1037#[cfg(test)]
1038mod tests {
1039    use super::*;
1040
1041    #[test]
1042    fn test_extend_sorted() {
1043        let mut vec = vec![];
1044
1045        extend_sorted(&mut vec, vec![21, 17, 13, 8, 1, 0], 5, |a, b| b.cmp(a));
1046        assert_eq!(vec, &[21, 17, 13, 8, 1]);
1047
1048        extend_sorted(&mut vec, vec![101, 19, 17, 8, 2], 8, |a, b| b.cmp(a));
1049        assert_eq!(vec, &[101, 21, 19, 17, 13, 8, 2, 1]);
1050
1051        extend_sorted(&mut vec, vec![1000, 19, 17, 9, 5], 8, |a, b| b.cmp(a));
1052        assert_eq!(vec, &[1000, 101, 21, 19, 17, 13, 9, 8]);
1053    }
1054
1055    #[test]
1056    fn test_truncate_to_bottom_n_sorted_by() {
1057        let mut vec: Vec<u32> = vec![5, 2, 3, 4, 1];
1058        truncate_to_bottom_n_sorted_by(&mut vec, 10, &u32::cmp);
1059        assert_eq!(vec, &[1, 2, 3, 4, 5]);
1060
1061        vec = vec![5, 2, 3, 4, 1];
1062        truncate_to_bottom_n_sorted_by(&mut vec, 5, &u32::cmp);
1063        assert_eq!(vec, &[1, 2, 3, 4, 5]);
1064
1065        vec = vec![5, 2, 3, 4, 1];
1066        truncate_to_bottom_n_sorted_by(&mut vec, 4, &u32::cmp);
1067        assert_eq!(vec, &[1, 2, 3, 4]);
1068
1069        vec = vec![5, 2, 3, 4, 1];
1070        truncate_to_bottom_n_sorted_by(&mut vec, 1, &u32::cmp);
1071        assert_eq!(vec, &[1]);
1072
1073        vec = vec![5, 2, 3, 4, 1];
1074        truncate_to_bottom_n_sorted_by(&mut vec, 0, &u32::cmp);
1075        assert!(vec.is_empty());
1076    }
1077
1078    #[test]
1079    fn test_iife() {
1080        fn option_returning_function() -> Option<()> {
1081            None
1082        }
1083
1084        let foo = maybe!({
1085            option_returning_function()?;
1086            Some(())
1087        });
1088
1089        assert_eq!(foo, None);
1090    }
1091
1092    #[test]
1093    fn test_truncate_and_trailoff() {
1094        assert_eq!(truncate_and_trailoff("", 5), "");
1095        assert_eq!(truncate_and_trailoff("aaaaaa", 7), "aaaaaa");
1096        assert_eq!(truncate_and_trailoff("aaaaaa", 6), "aaaaaa");
1097        assert_eq!(truncate_and_trailoff("aaaaaa", 5), "aaaaa…");
1098        assert_eq!(truncate_and_trailoff("èèèèèè", 7), "èèèèèè");
1099        assert_eq!(truncate_and_trailoff("èèèèèè", 6), "èèèèèè");
1100        assert_eq!(truncate_and_trailoff("èèèèèè", 5), "èèèèè…");
1101    }
1102
1103    #[test]
1104    fn test_truncate_and_remove_front() {
1105        assert_eq!(truncate_and_remove_front("", 5), "");
1106        assert_eq!(truncate_and_remove_front("aaaaaa", 7), "aaaaaa");
1107        assert_eq!(truncate_and_remove_front("aaaaaa", 6), "aaaaaa");
1108        assert_eq!(truncate_and_remove_front("aaaaaa", 5), "…aaaaa");
1109        assert_eq!(truncate_and_remove_front("èèèèèè", 7), "èèèèèè");
1110        assert_eq!(truncate_and_remove_front("èèèèèè", 6), "èèèèèè");
1111        assert_eq!(truncate_and_remove_front("èèèèèè", 5), "…èèèèè");
1112    }
1113
1114    #[test]
1115    fn test_numeric_prefix_str_method() {
1116        let target = "1a";
1117        assert_eq!(
1118            NumericPrefixWithSuffix::from_numeric_prefixed_str(target),
1119            NumericPrefixWithSuffix(Some(1), "a")
1120        );
1121
1122        let target = "12ab";
1123        assert_eq!(
1124            NumericPrefixWithSuffix::from_numeric_prefixed_str(target),
1125            NumericPrefixWithSuffix(Some(12), "ab")
1126        );
1127
1128        let target = "12_ab";
1129        assert_eq!(
1130            NumericPrefixWithSuffix::from_numeric_prefixed_str(target),
1131            NumericPrefixWithSuffix(Some(12), "_ab")
1132        );
1133
1134        let target = "1_2ab";
1135        assert_eq!(
1136            NumericPrefixWithSuffix::from_numeric_prefixed_str(target),
1137            NumericPrefixWithSuffix(Some(1), "_2ab")
1138        );
1139
1140        let target = "1.2";
1141        assert_eq!(
1142            NumericPrefixWithSuffix::from_numeric_prefixed_str(target),
1143            NumericPrefixWithSuffix(Some(1), ".2")
1144        );
1145
1146        let target = "1.2_a";
1147        assert_eq!(
1148            NumericPrefixWithSuffix::from_numeric_prefixed_str(target),
1149            NumericPrefixWithSuffix(Some(1), ".2_a")
1150        );
1151
1152        let target = "12.2_a";
1153        assert_eq!(
1154            NumericPrefixWithSuffix::from_numeric_prefixed_str(target),
1155            NumericPrefixWithSuffix(Some(12), ".2_a")
1156        );
1157
1158        let target = "12a.2_a";
1159        assert_eq!(
1160            NumericPrefixWithSuffix::from_numeric_prefixed_str(target),
1161            NumericPrefixWithSuffix(Some(12), "a.2_a")
1162        );
1163    }
1164
1165    #[test]
1166    fn test_numeric_prefix_with_suffix() {
1167        let mut sorted = vec!["1-abc", "10", "11def", "2", "21-abc"];
1168        sorted.sort_by_key(|s| NumericPrefixWithSuffix::from_numeric_prefixed_str(s));
1169        assert_eq!(sorted, ["1-abc", "2", "10", "11def", "21-abc"]);
1170
1171        for numeric_prefix_less in ["numeric_prefix_less", "aaa", "~β„’Β£"] {
1172            assert_eq!(
1173                NumericPrefixWithSuffix::from_numeric_prefixed_str(numeric_prefix_less),
1174                NumericPrefixWithSuffix(None, numeric_prefix_less),
1175                "String without numeric prefix `{numeric_prefix_less}` should not be converted into NumericPrefixWithSuffix"
1176            )
1177        }
1178    }
1179
1180    #[test]
1181    fn test_word_consists_of_emojis() {
1182        let words_to_test = vec![
1183            ("πŸ‘¨β€πŸ‘©β€πŸ‘§β€πŸ‘§πŸ‘‹πŸ₯’", true),
1184            ("πŸ‘‹", true),
1185            ("!πŸ‘‹", false),
1186            ("πŸ‘‹!", false),
1187            ("πŸ‘‹ ", false),
1188            (" πŸ‘‹", false),
1189            ("Test", false),
1190        ];
1191
1192        for (text, expected_result) in words_to_test {
1193            assert_eq!(word_consists_of_emojis(text), expected_result);
1194        }
1195    }
1196
1197    #[test]
1198    fn test_truncate_lines_and_trailoff() {
1199        let text = r#"Line 1
1200Line 2
1201Line 3"#;
1202
1203        assert_eq!(
1204            truncate_lines_and_trailoff(text, 2),
1205            r#"Line 1
1206…"#
1207        );
1208
1209        assert_eq!(
1210            truncate_lines_and_trailoff(text, 3),
1211            r#"Line 1
1212Line 2
1213…"#
1214        );
1215
1216        assert_eq!(
1217            truncate_lines_and_trailoff(text, 4),
1218            r#"Line 1
1219Line 2
1220Line 3"#
1221        );
1222    }
1223
1224    #[test]
1225    fn test_expanded_and_wrapped_usize_range() {
1226        // Neither wrap
1227        assert_eq!(
1228            expanded_and_wrapped_usize_range(2..4, 1, 1, 8).collect::<Vec<usize>>(),
1229            (1..5).collect::<Vec<usize>>()
1230        );
1231        // Start wraps
1232        assert_eq!(
1233            expanded_and_wrapped_usize_range(2..4, 3, 1, 8).collect::<Vec<usize>>(),
1234            ((0..5).chain(7..8)).collect::<Vec<usize>>()
1235        );
1236        // Start wraps all the way around
1237        assert_eq!(
1238            expanded_and_wrapped_usize_range(2..4, 5, 1, 8).collect::<Vec<usize>>(),
1239            (0..8).collect::<Vec<usize>>()
1240        );
1241        // Start wraps all the way around and past 0
1242        assert_eq!(
1243            expanded_and_wrapped_usize_range(2..4, 10, 1, 8).collect::<Vec<usize>>(),
1244            (0..8).collect::<Vec<usize>>()
1245        );
1246        // End wraps
1247        assert_eq!(
1248            expanded_and_wrapped_usize_range(3..5, 1, 4, 8).collect::<Vec<usize>>(),
1249            (0..1).chain(2..8).collect::<Vec<usize>>()
1250        );
1251        // End wraps all the way around
1252        assert_eq!(
1253            expanded_and_wrapped_usize_range(3..5, 1, 5, 8).collect::<Vec<usize>>(),
1254            (0..8).collect::<Vec<usize>>()
1255        );
1256        // End wraps all the way around and past the end
1257        assert_eq!(
1258            expanded_and_wrapped_usize_range(3..5, 1, 10, 8).collect::<Vec<usize>>(),
1259            (0..8).collect::<Vec<usize>>()
1260        );
1261        // Both start and end wrap
1262        assert_eq!(
1263            expanded_and_wrapped_usize_range(3..5, 4, 4, 8).collect::<Vec<usize>>(),
1264            (0..8).collect::<Vec<usize>>()
1265        );
1266    }
1267
1268    #[test]
1269    fn test_wrapped_usize_outward_from() {
1270        // No wrapping
1271        assert_eq!(
1272            wrapped_usize_outward_from(4, 2, 2, 10).collect::<Vec<usize>>(),
1273            vec![4, 5, 3, 6, 2]
1274        );
1275        // Wrapping at end
1276        assert_eq!(
1277            wrapped_usize_outward_from(8, 2, 3, 10).collect::<Vec<usize>>(),
1278            vec![8, 9, 7, 0, 6, 1]
1279        );
1280        // Wrapping at start
1281        assert_eq!(
1282            wrapped_usize_outward_from(1, 3, 2, 10).collect::<Vec<usize>>(),
1283            vec![1, 2, 0, 3, 9, 8]
1284        );
1285        // All values wrap around
1286        assert_eq!(
1287            wrapped_usize_outward_from(5, 10, 10, 8).collect::<Vec<usize>>(),
1288            vec![5, 6, 4, 7, 3, 0, 2, 1]
1289        );
1290        // None before / after
1291        assert_eq!(
1292            wrapped_usize_outward_from(3, 0, 0, 8).collect::<Vec<usize>>(),
1293            vec![3]
1294        );
1295        // Starting point already wrapped
1296        assert_eq!(
1297            wrapped_usize_outward_from(15, 2, 2, 10).collect::<Vec<usize>>(),
1298            vec![5, 6, 4, 7, 3]
1299        );
1300        // wrap_length of 0
1301        assert_eq!(
1302            wrapped_usize_outward_from(4, 2, 2, 0).collect::<Vec<usize>>(),
1303            Vec::<usize>::new()
1304        );
1305    }
1306
1307    #[test]
1308    fn test_split_with_ranges() {
1309        let input = "hi";
1310        let result = split_str_with_ranges(input, |c| c == ' ');
1311
1312        assert_eq!(result.len(), 1);
1313        assert_eq!(result[0], (0..2, "hi"));
1314
1315        let input = "hΓ©lloπŸ¦€world";
1316        let result = split_str_with_ranges(input, |c| c == 'πŸ¦€');
1317
1318        assert_eq!(result.len(), 2);
1319        assert_eq!(result[0], (0..6, "hΓ©llo")); // 'Γ©' is 2 bytes
1320        assert_eq!(result[1], (10..15, "world")); // 'πŸ¦€' is 4 bytes
1321    }
1322}