clippy_lints/src/literal_representation.rs

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