clippy_lints/src/literal_representation.rs

   1 // Copyright 2014-2018 The Rust Project Developers. See the COPYRIGHT
   2 // file at the top-level directory of this distribution.
   3 //
   4 // Licensed under the Apache License, Version 2.0 <LICENSE-APACHE or
   5 // http://www.apache.org/licenses/LICENSE-2.0> or the MIT license
   6 // <LICENSE-MIT or http://opensource.org/licenses/MIT>, at your
   7 // option. This file may not be copied, modified, or distributed
   8 // except according to those terms.
   9
  10 //! Lints concerned with the grouping of digits with underscores in integral or
  11 //! floating-point literal expressions.
  12
  13 use crate::utils::{snippet_opt, span_lint_and_sugg};
  14 use if_chain::if_chain;
  15 use rustc::lint::{in_external_macro, EarlyContext, EarlyLintPass, LintArray, LintContext, LintPass};
  16 use rustc::{declare_tool_lint, lint_array};
  17 use rustc_errors::Applicability;
  18 use syntax::ast::*;
  19 use syntax_pos;
  20
  21 /// **What it does:** Warns if a long integral or floating-point constant does
  22 /// not contain underscores.
  23 ///
  24 /// **Why is this bad?** Reading long numbers is difficult without separators.
  25 ///
  26 /// **Known problems:** None.
  27 ///
  28 /// **Example:**
  29 ///
  30 /// ```rust
  31 /// 61864918973511
  32 /// ```
  33 declare_clippy_lint! {
  34     pub UNREADABLE_LITERAL,
  35     style,
  36     "long integer literal without underscores"
  37 }
  38
  39 /// **What it does:** Warns for mistyped suffix in literals
  40 ///
  41 /// **Why is this bad?** This is most probably a typo
  42 ///
  43 /// **Known problems:**
  44 ///             - Recommends a signed suffix, even though the number might be too big and an unsigned
  45 ///             suffix is required
  46 ///             - Does not match on `_128` since that is a valid grouping for decimal and octal numbers
  47 ///
  48 /// **Example:**
  49 ///
  50 /// ```rust
  51 /// 2_32
  52 /// ```
  53 declare_clippy_lint! {
  54     pub MISTYPED_LITERAL_SUFFIXES,
  55     correctness,
  56     "mistyped literal suffix"
  57 }
  58
  59 /// **What it does:** Warns if an integral or floating-point constant is
  60 /// grouped inconsistently with underscores.
  61 ///
  62 /// **Why is this bad?** Readers may incorrectly interpret inconsistently
  63 /// grouped digits.
  64 ///
  65 /// **Known problems:** None.
  66 ///
  67 /// **Example:**
  68 ///
  69 /// ```rust
  70 /// 618_64_9189_73_511
  71 /// ```
  72 declare_clippy_lint! {
  73     pub INCONSISTENT_DIGIT_GROUPING,
  74     style,
  75     "integer literals with digits grouped inconsistently"
  76 }
  77
  78 /// **What it does:** Warns if the digits of an integral or floating-point
  79 /// constant are grouped into groups that
  80 /// are too large.
  81 ///
  82 /// **Why is this bad?** Negatively impacts readability.
  83 ///
  84 /// **Known problems:** None.
  85 ///
  86 /// **Example:**
  87 ///
  88 /// ```rust
  89 /// 6186491_8973511
  90 /// ```
  91 declare_clippy_lint! {
  92     pub LARGE_DIGIT_GROUPS,
  93     pedantic,
  94     "grouping digits into groups that are too large"
  95 }
  96
  97 /// **What it does:** Warns if there is a better representation for a numeric literal.
  98 ///
  99 /// **Why is this bad?** Especially for big powers of 2 a hexadecimal representation is more
 100 /// readable than a decimal representation.
 101 ///
 102 /// **Known problems:** None.
 103 ///
 104 /// **Example:**
 105 ///
 106 /// `255` => `0xFF`
 107 /// `65_535` => `0xFFFF`
 108 /// `4_042_322_160` => `0xF0F0_F0F0`
 109 declare_clippy_lint! {
 110     pub DECIMAL_LITERAL_REPRESENTATION,
 111     restriction,
 112     "using decimal representation when hexadecimal would be better"
 113 }
 114
 115 #[derive(Debug, PartialEq)]
 116 pub(super) enum Radix {
 117     Binary,
 118     Octal,
 119     Decimal,
 120     Hexadecimal,
 121 }
 122
 123 impl Radix {
 124     /// Return a reasonable digit group size for this radix.
 125     crate fn suggest_grouping(&self) -> usize {
 126         match *self {
 127             Radix::Binary | Radix::Hexadecimal => 4,
 128             Radix::Octal | Radix::Decimal => 3,
 129         }
 130     }
 131 }
 132
 133 #[derive(Debug)]
 134 pub(super) struct DigitInfo<'a> {
 135     /// Characters of a literal between the radix prefix and type suffix.
 136     crate digits: &'a str,
 137     /// Which radix the literal was represented in.
 138     crate radix: Radix,
 139     /// The radix prefix, if present.
 140     crate prefix: Option<&'a str>,
 141     /// The type suffix, including preceding underscore if present.
 142     crate suffix: Option<&'a str>,
 143     /// True for floating-point literals.
 144     crate float: bool,
 145 }
 146
 147 impl<'a> DigitInfo<'a> {
 148     crate fn new(lit: &'a str, float: bool) -> Self {
 149         // Determine delimiter for radix prefix, if present, and radix.
 150         let radix = if lit.starts_with("0x") {
 151             Radix::Hexadecimal
 152         } else if lit.starts_with("0b") {
 153             Radix::Binary
 154         } else if lit.starts_with("0o") {
 155             Radix::Octal
 156         } else {
 157             Radix::Decimal
 158         };
 159
 160         // Grab part of the literal after prefix, if present.
 161         let (prefix, sans_prefix) = if let Radix::Decimal = radix {
 162             (None, lit)
 163         } else {
 164             let (p, s) = lit.split_at(2);
 165             (Some(p), s)
 166         };
 167
 168         let len = sans_prefix.len();
 169         let mut last_d = '\0';
 170         for (d_idx, d) in sans_prefix.char_indices() {
 171             let suffix_start = if last_d == '_' { d_idx - 1 } else { d_idx };
 172             if float
 173                 && (d == 'f'
 174                     || is_possible_float_suffix_index(&sans_prefix, suffix_start, len)
 175                     || ((d == 'E' || d == 'e') && !has_possible_float_suffix(&sans_prefix)))
 176                 || !float && (d == 'i' || d == 'u' || is_possible_suffix_index(&sans_prefix, suffix_start, len))
 177             {
 178                 let (digits, suffix) = sans_prefix.split_at(suffix_start);
 179                 return Self {
 180                     digits,
 181                     radix,
 182                     prefix,
 183                     suffix: Some(suffix),
 184                     float,
 185                 };
 186             }
 187             last_d = d
 188         }
 189
 190         // No suffix found
 191         Self {
 192             digits: sans_prefix,
 193             radix,
 194             prefix,
 195             suffix: None,
 196             float,
 197         }
 198     }
 199
 200     /// Returns literal formatted in a sensible way.
 201     crate fn grouping_hint(&self) -> String {
 202         let group_size = self.radix.suggest_grouping();
 203         if self.digits.contains('.') {
 204             let mut parts = self.digits.split('.');
 205             let int_part_hint = parts
 206                 .next()
 207                 .expect("split always returns at least one element")
 208                 .chars()
 209                 .rev()
 210                 .filter(|&c| c != '_')
 211                 .collect::<Vec<_>>()
 212                 .chunks(group_size)
 213                 .map(|chunk| chunk.iter().rev().collect())
 214                 .rev()
 215                 .collect::<Vec<String>>()
 216                 .join("_");
 217             let frac_part_hint = parts
 218                 .next()
 219                 .expect("already checked that there is a `.`")
 220                 .chars()
 221                 .filter(|&c| c != '_')
 222                 .collect::<Vec<_>>()
 223                 .chunks(group_size)
 224                 .map(|chunk| chunk.iter().collect())
 225                 .collect::<Vec<String>>()
 226                 .join("_");
 227             let suffix_hint = match self.suffix {
 228                 Some(suffix) if is_mistyped_float_suffix(suffix) => format!("_f{}", &suffix[1..]),
 229                 Some(suffix) => suffix.to_string(),
 230                 None => String::new(),
 231             };
 232             format!("{}.{}{}", int_part_hint, frac_part_hint, suffix_hint)
 233         } else if self.float && (self.digits.contains('E') || self.digits.contains('e')) {
 234             let which_e = if self.digits.contains('E') { 'E' } else { 'e' };
 235             let parts: Vec<&str> = self.digits.split(which_e).collect();
 236             let filtered_digits_vec_0 = parts[0].chars().filter(|&c| c != '_').rev().collect::<Vec<_>>();
 237             let filtered_digits_vec_1 = parts[1].chars().filter(|&c| c != '_').rev().collect::<Vec<_>>();
 238             let before_e_hint = filtered_digits_vec_0
 239                 .chunks(group_size)
 240                 .map(|chunk| chunk.iter().rev().collect())
 241                 .rev()
 242                 .collect::<Vec<String>>()
 243                 .join("_");
 244             let after_e_hint = filtered_digits_vec_1
 245                 .chunks(group_size)
 246                 .map(|chunk| chunk.iter().rev().collect())
 247                 .rev()
 248                 .collect::<Vec<String>>()
 249                 .join("_");
 250             let suffix_hint = match self.suffix {
 251                 Some(suffix) if is_mistyped_float_suffix(suffix) => format!("_f{}", &suffix[1..]),
 252                 Some(suffix) => suffix.to_string(),
 253                 None => String::new(),
 254             };
 255             format!(
 256                 "{}{}{}{}{}",
 257                 self.prefix.unwrap_or(""),
 258                 before_e_hint,
 259                 which_e,
 260                 after_e_hint,
 261                 suffix_hint
 262             )
 263         } else {
 264             let filtered_digits_vec = self.digits.chars().filter(|&c| c != '_').rev().collect::<Vec<_>>();
 265             let mut hint = filtered_digits_vec
 266                 .chunks(group_size)
 267                 .map(|chunk| chunk.iter().rev().collect())
 268                 .rev()
 269                 .collect::<Vec<String>>()
 270                 .join("_");
 271             // Forces hexadecimal values to be grouped by 4 being filled with zeroes (e.g 0x00ab_cdef)
 272             let nb_digits_to_fill = filtered_digits_vec.len() % 4;
 273             if self.radix == Radix::Hexadecimal && nb_digits_to_fill != 0 {
 274                 hint = format!("{:0>4}{}", &hint[..nb_digits_to_fill], &hint[nb_digits_to_fill..]);
 275             }
 276             let suffix_hint = match self.suffix {
 277                 Some(suffix) if is_mistyped_suffix(suffix) => format!("_i{}", &suffix[1..]),
 278                 Some(suffix) => suffix.to_string(),
 279                 None => String::new(),
 280             };
 281             format!("{}{}{}", self.prefix.unwrap_or(""), hint, suffix_hint)
 282         }
 283     }
 284 }
 285
 286 enum WarningType {
 287     UnreadableLiteral,
 288     InconsistentDigitGrouping,
 289     LargeDigitGroups,
 290     DecimalRepresentation,
 291     MistypedLiteralSuffix,
 292 }
 293
 294 impl WarningType {
 295     crate fn display(&self, grouping_hint: &str, cx: &EarlyContext<'_>, span: syntax_pos::Span) {
 296         match self {
 297             WarningType::MistypedLiteralSuffix => span_lint_and_sugg(
 298                 cx,
 299                 MISTYPED_LITERAL_SUFFIXES,
 300                 span,
 301                 "mistyped literal suffix",
 302                 "did you mean to write",
 303                 grouping_hint.to_string(),
 304                 Applicability::MaybeIncorrect,
 305             ),
 306             WarningType::UnreadableLiteral => span_lint_and_sugg(
 307                 cx,
 308                 UNREADABLE_LITERAL,
 309                 span,
 310                 "long literal lacking separators",
 311                 "consider",
 312                 grouping_hint.to_owned(),
 313                 Applicability::MachineApplicable,
 314             ),
 315             WarningType::LargeDigitGroups => span_lint_and_sugg(
 316                 cx,
 317                 LARGE_DIGIT_GROUPS,
 318                 span,
 319                 "digit groups should be smaller",
 320                 "consider",
 321                 grouping_hint.to_owned(),
 322                 Applicability::MachineApplicable,
 323             ),
 324             WarningType::InconsistentDigitGrouping => span_lint_and_sugg(
 325                 cx,
 326                 INCONSISTENT_DIGIT_GROUPING,
 327                 span,
 328                 "digits grouped inconsistently by underscores",
 329                 "consider",
 330                 grouping_hint.to_owned(),
 331                 Applicability::MachineApplicable,
 332             ),
 333             WarningType::DecimalRepresentation => span_lint_and_sugg(
 334                 cx,
 335                 DECIMAL_LITERAL_REPRESENTATION,
 336                 span,
 337                 "integer literal has a better hexadecimal representation",
 338                 "consider",
 339                 grouping_hint.to_owned(),
 340                 Applicability::MachineApplicable,
 341             ),
 342         };
 343     }
 344 }
 345
 346 #[derive(Copy, Clone)]
 347 pub struct LiteralDigitGrouping;
 348
 349 impl LintPass for LiteralDigitGrouping {
 350     fn get_lints(&self) -> LintArray {
 351         lint_array!(
 352             UNREADABLE_LITERAL,
 353             INCONSISTENT_DIGIT_GROUPING,
 354             LARGE_DIGIT_GROUPS,
 355             MISTYPED_LITERAL_SUFFIXES,
 356         )
 357     }
 358 }
 359
 360 impl EarlyLintPass for LiteralDigitGrouping {
 361     fn check_expr(&mut self, cx: &EarlyContext<'_>, expr: &Expr) {
 362         if in_external_macro(cx.sess(), expr.span) {
 363             return;
 364         }
 365
 366         if let ExprKind::Lit(ref lit) = expr.node {
 367             self.check_lit(cx, lit)
 368         }
 369     }
 370 }
 371
 372 impl LiteralDigitGrouping {
 373     fn check_lit(self, cx: &EarlyContext<'_>, lit: &Lit) {
 374         match lit.node {
 375             LitKind::Int(..) => {
 376                 // Lint integral literals.
 377                 if_chain! {
 378                     if let Some(src) = snippet_opt(cx, lit.span);
 379                     if let Some(firstch) = src.chars().next();
 380                     if char::to_digit(firstch, 10).is_some();
 381                     then {
 382                         let digit_info = DigitInfo::new(&src, false);
 383                         let _ = Self::do_lint(digit_info.digits, digit_info.suffix).map_err(|warning_type| {
 384                             warning_type.display(&digit_info.grouping_hint(), cx, lit.span)
 385                         });
 386                     }
 387                 }
 388             },
 389             LitKind::Float(..) | LitKind::FloatUnsuffixed(..) => {
 390                 // Lint floating-point literals.
 391                 if_chain! {
 392                     if let Some(src) = snippet_opt(cx, lit.span);
 393                     if let Some(firstch) = src.chars().next();
 394                     if char::to_digit(firstch, 10).is_some();
 395                     then {
 396                         let digit_info = DigitInfo::new(&src, true);
 397                         // Separate digits into integral and fractional parts.
 398                         let parts: Vec<&str> = digit_info
 399                             .digits
 400                             .split_terminator('.')
 401                             .collect();
 402
 403                         // Lint integral and fractional parts separately, and then check consistency of digit
 404                         // groups if both pass.
 405                         let _ = Self::do_lint(parts[0], digit_info.suffix)
 406                             .map(|integral_group_size| {
 407                                 if parts.len() > 1 {
 408                                     // Lint the fractional part of literal just like integral part, but reversed.
 409                                     let fractional_part = &parts[1].chars().rev().collect::<String>();
 410                                     let _ = Self::do_lint(fractional_part, None)
 411                                         .map(|fractional_group_size| {
 412                                             let consistent = Self::parts_consistent(integral_group_size,
 413                                                                                     fractional_group_size,
 414                                                                                     parts[0].len(),
 415                                                                                     parts[1].len());
 416                                                 if !consistent {
 417                                                     WarningType::InconsistentDigitGrouping.display(
 418                                                         &digit_info.grouping_hint(),
 419                                                         cx,
 420                                                         lit.span,
 421                                                     );
 422                                                 }
 423                                         })
 424                                     .map_err(|warning_type| warning_type.display(&digit_info.grouping_hint(),
 425                                     cx,
 426                                     lit.span));
 427                                 }
 428                             })
 429                         .map_err(|warning_type| warning_type.display(&digit_info.grouping_hint(), cx, lit.span));
 430                     }
 431                 }
 432             },
 433             _ => (),
 434         }
 435     }
 436
 437     /// Given the sizes of the digit groups of both integral and fractional
 438     /// parts, and the length
 439     /// of both parts, determine if the digits have been grouped consistently.
 440     fn parts_consistent(int_group_size: usize, frac_group_size: usize, int_size: usize, frac_size: usize) -> bool {
 441         match (int_group_size, frac_group_size) {
 442             // No groups on either side of decimal point - trivially consistent.
 443             (0, 0) => true,
 444             // Integral part has grouped digits, fractional part does not.
 445             (_, 0) => frac_size <= int_group_size,
 446             // Fractional part has grouped digits, integral part does not.
 447             (0, _) => int_size <= frac_group_size,
 448             // Both parts have grouped digits. Groups should be the same size.
 449             (_, _) => int_group_size == frac_group_size,
 450         }
 451     }
 452
 453     /// Performs lint on `digits` (no decimal point) and returns the group
 454     /// size on success or `WarningType` when emitting a warning.
 455     fn do_lint(digits: &str, suffix: Option<&str>) -> Result<usize, WarningType> {
 456         if let Some(suffix) = suffix {
 457             if is_mistyped_suffix(suffix) {
 458                 return Err(WarningType::MistypedLiteralSuffix);
 459             }
 460         }
 461         // Grab underscore indices with respect to the units digit.
 462         let underscore_positions: Vec<usize> = digits
 463             .chars()
 464             .rev()
 465             .enumerate()
 466             .filter_map(|(idx, digit)| if digit == '_' { Some(idx) } else { None })
 467             .collect();
 468
 469         if underscore_positions.is_empty() {
 470             // Check if literal needs underscores.
 471             if digits.len() > 5 {
 472                 Err(WarningType::UnreadableLiteral)
 473             } else {
 474                 Ok(0)
 475             }
 476         } else {
 477             // Check consistency and the sizes of the groups.
 478             let group_size = underscore_positions[0];
 479             let consistent = underscore_positions
 480                 .windows(2)
 481                 .all(|ps| ps[1] - ps[0] == group_size + 1)
 482                 // number of digits to the left of the last group cannot be bigger than group size.
 483                 && (digits.len() - underscore_positions.last()
 484                                                        .expect("there's at least one element") <= group_size + 1);
 485
 486             if !consistent {
 487                 return Err(WarningType::InconsistentDigitGrouping);
 488             } else if group_size > 4 {
 489                 return Err(WarningType::LargeDigitGroups);
 490             }
 491             Ok(group_size)
 492         }
 493     }
 494 }
 495
 496 #[derive(Copy, Clone)]
 497 pub struct LiteralRepresentation {
 498     threshold: u64,
 499 }
 500
 501 impl LintPass for LiteralRepresentation {
 502     fn get_lints(&self) -> LintArray {
 503         lint_array!(DECIMAL_LITERAL_REPRESENTATION)
 504     }
 505 }
 506
 507 impl EarlyLintPass for LiteralRepresentation {
 508     fn check_expr(&mut self, cx: &EarlyContext<'_>, expr: &Expr) {
 509         if in_external_macro(cx.sess(), expr.span) {
 510             return;
 511         }
 512
 513         if let ExprKind::Lit(ref lit) = expr.node {
 514             self.check_lit(cx, lit)
 515         }
 516     }
 517 }
 518
 519 impl LiteralRepresentation {
 520     pub fn new(threshold: u64) -> Self {
 521         Self { threshold }
 522     }
 523     fn check_lit(self, cx: &EarlyContext<'_>, lit: &Lit) {
 524         // Lint integral literals.
 525         if_chain! {
 526             if let LitKind::Int(..) = lit.node;
 527             if let Some(src) = snippet_opt(cx, lit.span);
 528             if let Some(firstch) = src.chars().next();
 529             if char::to_digit(firstch, 10).is_some();
 530             then {
 531                 let digit_info = DigitInfo::new(&src, false);
 532                 if digit_info.radix == Radix::Decimal {
 533                     let val = digit_info.digits
 534                         .chars()
 535                         .filter(|&c| c != '_')
 536                         .collect::<String>()
 537                         .parse::<u128>().unwrap();
 538                     if val < u128::from(self.threshold) {
 539                         return
 540                     }
 541                     let hex = format!("{:#X}", val);
 542                     let digit_info = DigitInfo::new(&hex[..], false);
 543                     let _ = Self::do_lint(digit_info.digits).map_err(|warning_type| {
 544                         warning_type.display(&digit_info.grouping_hint(), cx, lit.span)
 545                     });
 546                 }
 547             }
 548         }
 549     }
 550
 551     fn do_lint(digits: &str) -> Result<(), WarningType> {
 552         if digits.len() == 1 {
 553             // Lint for 1 digit literals, if someone really sets the threshold that low
 554             if digits == "1"
 555                 || digits == "2"
 556                 || digits == "4"
 557                 || digits == "8"
 558                 || digits == "3"
 559                 || digits == "7"
 560                 || digits == "F"
 561             {
 562                 return Err(WarningType::DecimalRepresentation);
 563             }
 564         } else if digits.len() < 4 {
 565             // Lint for Literals with a hex-representation of 2 or 3 digits
 566             let f = &digits[0..1]; // first digit
 567             let s = &digits[1..]; // suffix
 568
 569             // Powers of 2
 570             if ((f.eq("1") || f.eq("2") || f.eq("4") || f.eq("8")) && s.chars().all(|c| c == '0'))
 571                 // Powers of 2 minus 1
 572                 || ((f.eq("1") || f.eq("3") || f.eq("7") || f.eq("F")) && s.chars().all(|c| c == 'F'))
 573             {
 574                 return Err(WarningType::DecimalRepresentation);
 575             }
 576         } else {
 577             // Lint for Literals with a hex-representation of 4 digits or more
 578             let f = &digits[0..1]; // first digit
 579             let m = &digits[1..digits.len() - 1]; // middle digits, except last
 580             let s = &digits[1..]; // suffix
 581
 582             // Powers of 2 with a margin of +15/-16
 583             if ((f.eq("1") || f.eq("2") || f.eq("4") || f.eq("8")) && m.chars().all(|c| c == '0'))
 584                 || ((f.eq("1") || f.eq("3") || f.eq("7") || f.eq("F")) && m.chars().all(|c| c == 'F'))
 585                 // Lint for representations with only 0s and Fs, while allowing 7 as the first
 586                 // digit
 587                 || ((f.eq("7") || f.eq("F")) && s.chars().all(|c| c == '0' || c == 'F'))
 588             {
 589                 return Err(WarningType::DecimalRepresentation);
 590             }
 591         }
 592
 593         Ok(())
 594     }
 595 }
 596
 597 fn is_mistyped_suffix(suffix: &str) -> bool {
 598     ["_8", "_16", "_32", "_64"].contains(&suffix)
 599 }
 600
 601 fn is_possible_suffix_index(lit: &str, idx: usize, len: usize) -> bool {
 602     ((len > 3 && idx == len - 3) || (len > 2 && idx == len - 2)) && is_mistyped_suffix(lit.split_at(idx).1)
 603 }
 604
 605 fn is_mistyped_float_suffix(suffix: &str) -> bool {
 606     ["_32", "_64"].contains(&suffix)
 607 }
 608
 609 fn is_possible_float_suffix_index(lit: &str, idx: usize, len: usize) -> bool {
 610     (len > 3 && idx == len - 3) && is_mistyped_float_suffix(lit.split_at(idx).1)
 611 }
 612
 613 fn has_possible_float_suffix(lit: &str) -> bool {
 614     lit.ends_with("_32") || lit.ends_with("_64")
 615 }