rust/clippy_lints/src/literal_digit_grouping.rs

381 lines
13 KiB
Rust
Raw Normal View History

2017-07-19 18:56:32 -05:00
//! Lints concerned with the grouping of digits with underscores in integral or
//! floating-point literal expressions.
use rustc::lint::*;
use syntax::ast::*;
use syntax_pos;
use utils::{span_help_and_lint, snippet_opt, in_external_macro};
2017-07-30 17:21:38 -05:00
/// **What it does:** Warns if a long integral or floating-point constant does
/// not contain underscores.
2017-07-19 18:56:32 -05:00
///
/// **Why is this bad?** Reading long numbers is difficult without separators.
///
/// **Known problems:** None.
///
/// **Example:**
///
/// ```rust
/// 61864918973511
/// ```
declare_lint! {
pub UNREADABLE_LITERAL,
Warn,
"long integer literal without underscores"
}
2017-07-30 17:21:38 -05:00
/// **What it does:** Warns if an integral or floating-point constant is
/// grouped inconsistently with underscores.
2017-07-19 18:56:32 -05:00
///
2017-07-30 17:21:38 -05:00
/// **Why is this bad?** Readers may incorrectly interpret inconsistently
/// grouped digits.
2017-07-19 18:56:32 -05:00
///
/// **Known problems:** None.
///
/// **Example:**
///
/// ```rust
/// 618_64_9189_73_511
/// ```
declare_lint! {
pub INCONSISTENT_DIGIT_GROUPING,
Warn,
"integer literals with digits grouped inconsistently"
}
2017-07-30 17:21:38 -05:00
/// **What it does:** Warns if the digits of an integral or floating-point
/// constant are grouped into groups that
2017-07-19 18:56:32 -05:00
/// are too large.
///
/// **Why is this bad?** Negatively impacts readability.
///
/// **Known problems:** None.
///
/// **Example:**
///
/// ```rust
/// 6186491_8973511
/// ```
declare_lint! {
pub LARGE_DIGIT_GROUPS,
Warn,
"grouping digits into groups that are too large"
}
#[derive(Debug)]
enum Radix {
Binary,
Octal,
Decimal,
Hexadecimal,
}
impl Radix {
/// Return a reasonable digit group size for this radix.
pub fn suggest_grouping(&self) -> usize {
match *self {
Radix::Binary | Radix::Hexadecimal => 4,
Radix::Octal | Radix::Decimal => 3,
}
}
}
#[derive(Debug)]
struct DigitInfo<'a> {
/// Characters of a literal between the radix prefix and type suffix.
pub digits: &'a str,
/// Which radix the literal was represented in.
pub radix: Radix,
/// The radix prefix, if present.
pub prefix: Option<&'a str>,
/// The type suffix, including preceding underscore if present.
pub suffix: Option<&'a str>,
/// True for floating-point literals.
pub float: bool,
}
impl<'a> DigitInfo<'a> {
pub fn new(lit: &str, float: bool) -> DigitInfo {
// Determine delimiter for radix prefix, if present, and radix.
let radix = if lit.starts_with("0x") {
Radix::Hexadecimal
} else if lit.starts_with("0b") {
Radix::Binary
} else if lit.starts_with("0o") {
Radix::Octal
} else {
Radix::Decimal
};
// Grab part of the literal after prefix, if present.
let (prefix, sans_prefix) = if let Radix::Decimal = radix {
(None, lit)
} else {
let (p, s) = lit.split_at(2);
(Some(p), s)
};
let mut last_d = '\0';
for (d_idx, d) in sans_prefix.char_indices() {
if !float && (d == 'i' || d == 'u') || float && d == 'f' {
let suffix_start = if last_d == '_' { d_idx - 1 } else { d_idx };
let (digits, suffix) = sans_prefix.split_at(suffix_start);
return DigitInfo {
digits: digits,
radix: radix,
prefix: prefix,
suffix: Some(suffix),
float: float,
};
}
last_d = d
}
// No suffix found
DigitInfo {
digits: sans_prefix,
radix: radix,
prefix: prefix,
suffix: None,
float: float,
}
}
/// Returns digits grouped in a sensible way.
fn grouping_hint(&self) -> String {
let group_size = self.radix.suggest_grouping();
if self.digits.contains('.') {
let mut parts = self.digits.split(".");
let int_part_hint = parts
.next()
.unwrap()
.chars()
.rev()
.filter(|&c| c != '_')
.collect::<Vec<_>>()
.chunks(group_size)
.map(|chunk| chunk.into_iter().rev().collect())
.rev()
.collect::<Vec<String>>()
.join("_");
let frac_part_hint = parts
.next()
.unwrap()
.chars()
.filter(|&c| c != '_')
.collect::<Vec<_>>()
.chunks(group_size)
.map(|chunk| chunk.into_iter().collect())
.collect::<Vec<String>>()
.join("_");
format!("{}.{}{}", int_part_hint, frac_part_hint, self.suffix.unwrap_or(""))
} else {
let hint = self.digits
.chars()
.rev()
.filter(|&c| c != '_')
.collect::<Vec<_>>()
.chunks(group_size)
.map(|chunk| chunk.into_iter().rev().collect())
.rev()
.collect::<Vec<String>>()
.join("_");
format!("{}{}{}", self.prefix.unwrap_or(""), hint, self.suffix.unwrap_or(""))
}
}
}
enum WarningType {
UnreadableLiteral,
InconsistentDigitGrouping,
LargeDigitGroups,
}
impl WarningType {
pub fn display(&self, grouping_hint: &str, cx: &EarlyContext, span: &syntax_pos::Span) {
match *self {
WarningType::UnreadableLiteral => {
span_help_and_lint(
cx,
UNREADABLE_LITERAL,
*span,
"long literal lacking separators",
&format!("consider: {}", grouping_hint),
)
},
WarningType::LargeDigitGroups => {
span_help_and_lint(
cx,
LARGE_DIGIT_GROUPS,
*span,
"digit groups should be smaller",
&format!("consider: {}", grouping_hint),
)
},
WarningType::InconsistentDigitGrouping => {
span_help_and_lint(
cx,
INCONSISTENT_DIGIT_GROUPING,
*span,
"digits grouped inconsistently by underscores",
&format!("consider: {}", grouping_hint),
)
},
};
}
}
2017-07-19 18:56:32 -05:00
#[derive(Copy, Clone)]
pub struct LiteralDigitGrouping;
impl LintPass for LiteralDigitGrouping {
fn get_lints(&self) -> LintArray {
lint_array!(UNREADABLE_LITERAL, INCONSISTENT_DIGIT_GROUPING, LARGE_DIGIT_GROUPS)
}
}
impl EarlyLintPass for LiteralDigitGrouping {
fn check_expr(&mut self, cx: &EarlyContext, expr: &Expr) {
if in_external_macro(cx, expr.span) {
return;
}
2017-07-22 18:54:44 -05:00
if let ExprKind::Lit(ref lit) = expr.node {
self.check_lit(cx, lit)
2017-07-19 18:56:32 -05:00
}
}
}
impl LiteralDigitGrouping {
fn check_lit(&self, cx: &EarlyContext, lit: &Lit) {
// Lint integral literals.
if_let_chain! {[
let LitKind::Int(..) = lit.node,
let Some(src) = snippet_opt(cx, lit.span),
let Some(firstch) = src.chars().next(),
char::to_digit(firstch, 10).is_some()
], {
let digits = LiteralDigitGrouping::get_digits(&src, false);
LiteralDigitGrouping::do_lint(digits, cx, &lit.span);
}}
// Lint floating-point literals.
if_let_chain! {[
let LitKind::Float(..) = lit.node,
let Some(src) = snippet_opt(cx, lit.span),
let Some(firstch) = src.chars().next(),
char::to_digit(firstch, 10).is_some()
], {
let digits: Vec<&str> = LiteralDigitGrouping::get_digits(&src, true)
.split_terminator('.')
.collect();
// Lint integral and fractional parts separately, and then check consistency of digit
// groups if both pass.
if let Some(integral_group_size) = LiteralDigitGrouping::do_lint(digits[0], cx, &lit.span) {
if digits.len() > 1 {
// Lint the fractional part of literal just like integral part, but reversed.
let fractional_part = &digits[1].chars().rev().collect::<String>();
if let Some(fractional_group_size) = LiteralDigitGrouping::do_lint(fractional_part, cx, &lit.span) {
let consistent = match (integral_group_size, fractional_group_size) {
// No groups on either side of decimal point - good to go.
(0, 0) => true,
// Integral part has grouped digits, fractional part does not.
(_, 0) => digits[1].len() <= integral_group_size,
// Fractional part has grouped digits, integral part does not.
(0, _) => digits[0].len() <= fractional_group_size,
// Both parts have grouped digits. Groups should be the same size.
(_, _) => integral_group_size == fractional_group_size,
};
if !consistent {
span_help_and_lint(cx, INCONSISTENT_DIGIT_GROUPING, lit.span,
"digits grouped inconsistently by underscores",
"consider making each group three or four digits");
}
}
}
}
}}
}
/// Returns the digits of an integral or floating-point literal.
fn get_digits(lit: &str, float: bool) -> &str {
// Determine delimiter for radix prefix, if present.
let mb_r = if lit.starts_with("0x") {
Some('x')
} else if lit.starts_with("0b") {
Some('b')
} else if lit.starts_with("0o") {
Some('o')
} else {
None
};
let has_suffix = !float && (lit.contains('i') || lit.contains('u')) || float && lit.contains('f');
// Grab part of literal between the radix prefix and type suffix.
let mut digits = if let Some(r) = mb_r {
lit.split(|c| c == 'i' || c == 'u' || c == r || (float && c == 'f')).nth(1).unwrap()
} else {
lit.split(|c| c == 'i' || c == 'u' || (float && c == 'f')).next().unwrap()
};
// If there was an underscore before type suffix, drop it.
if has_suffix && digits.chars().last().unwrap() == '_' {
digits = digits.split_at(digits.len() - 1).0;
}
digits
}
/// Performs lint on `digits` (no decimal point) and returns the group size. `None` is
/// returned when emitting a warning.
fn do_lint(digits: &str, cx: &EarlyContext, span: &syntax_pos::Span) -> Option<usize> {
// Grab underscore indices with respect to the units digit.
let underscore_positions: Vec<usize> = digits.chars().rev().enumerate()
.filter_map(|(idx, digit)|
if digit == '_' {
Some(idx)
} else {
None
})
.collect();
if underscore_positions.is_empty() {
// Check if literal needs underscores.
if digits.len() > 4 {
span_help_and_lint(cx, UNREADABLE_LITERAL, *span,
"long literal lacking separators",
"consider using underscores to make literal more readable");
return None;
} else {
return Some(0);
}
} else {
// Check consistency and the sizes of the groups.
let group_size = underscore_positions[0];
let consistent = underscore_positions
.windows(2)
.all(|ps| ps[1] - ps[0] == group_size + 1)
// number of digits to the left of the last group cannot be bigger than group size.
&& (digits.len() - underscore_positions.last().unwrap() <= group_size + 1);
if !consistent {
span_help_and_lint(cx, INCONSISTENT_DIGIT_GROUPING, *span,
"digits grouped inconsistently by underscores",
"consider making each group three or four digits");
return None;
} else if group_size > 4 {
span_help_and_lint(cx, LARGE_DIGIT_GROUPS, *span,
"digit groups should be smaller",
"consider using groups of three or four digits");
return None;
}
return Some(group_size);
}
}
}