clippy_lints/src/non_expressive_names.rs

   1 use crate::utils::{span_lint, span_lint_and_then};
   2 use rustc_lint::{EarlyContext, EarlyLintPass};
   3 use rustc_session::{declare_tool_lint, impl_lint_pass};
   4 use rustc_span::source_map::Span;
   5 use rustc_span::symbol::SymbolStr;
   6 use std::cmp::Ordering;
   7 use syntax::ast::*;
   8 use syntax::attr;
   9 use syntax::visit::{walk_block, walk_expr, walk_pat, Visitor};
  10
  11 declare_clippy_lint! {
  12     /// **What it does:** Checks for names that are very similar and thus confusing.
  13     ///
  14     /// **Why is this bad?** It's hard to distinguish between names that differ only
  15     /// by a single character.
  16     ///
  17     /// **Known problems:** None?
  18     ///
  19     /// **Example:**
  20     /// ```ignore
  21     /// let checked_exp = something;
  22     /// let checked_expr = something_else;
  23     /// ```
  24     pub SIMILAR_NAMES,
  25     pedantic,
  26     "similarly named items and bindings"
  27 }
  28
  29 declare_clippy_lint! {
  30     /// **What it does:** Checks for too many variables whose name consists of a
  31     /// single character.
  32     ///
  33     /// **Why is this bad?** It's hard to memorize what a variable means without a
  34     /// descriptive name.
  35     ///
  36     /// **Known problems:** None?
  37     ///
  38     /// **Example:**
  39     /// ```ignore
  40     /// let (a, b, c, d, e, f, g) = (...);
  41     /// ```
  42     pub MANY_SINGLE_CHAR_NAMES,
  43     style,
  44     "too many single character bindings"
  45 }
  46
  47 declare_clippy_lint! {
  48     /// **What it does:** Checks if you have variables whose name consists of just
  49     /// underscores and digits.
  50     ///
  51     /// **Why is this bad?** It's hard to memorize what a variable means without a
  52     /// descriptive name.
  53     ///
  54     /// **Known problems:** None?
  55     ///
  56     /// **Example:**
  57     /// ```rust
  58     /// let _1 = 1;
  59     /// let ___1 = 1;
  60     /// let __1___2 = 11;
  61     /// ```
  62     pub JUST_UNDERSCORES_AND_DIGITS,
  63     style,
  64     "unclear name"
  65 }
  66
  67 #[derive(Copy, Clone)]
  68 pub struct NonExpressiveNames {
  69     pub single_char_binding_names_threshold: u64,
  70 }
  71
  72 impl_lint_pass!(NonExpressiveNames => [SIMILAR_NAMES, MANY_SINGLE_CHAR_NAMES, JUST_UNDERSCORES_AND_DIGITS]);
  73
  74 struct ExistingName {
  75     interned: SymbolStr,
  76     span: Span,
  77     len: usize,
  78     whitelist: &'static [&'static str],
  79 }
  80
  81 struct SimilarNamesLocalVisitor<'a, 'tcx> {
  82     names: Vec<ExistingName>,
  83     cx: &'a EarlyContext<'tcx>,
  84     lint: &'a NonExpressiveNames,
  85
  86     /// A stack of scopes containing the single-character bindings in each scope.
  87     single_char_names: Vec<Vec<Ident>>,
  88 }
  89
  90 impl<'a, 'tcx> SimilarNamesLocalVisitor<'a, 'tcx> {
  91     fn check_single_char_names(&self) {
  92         let num_single_char_names = self.single_char_names.iter().flatten().count();
  93         let threshold = self.lint.single_char_binding_names_threshold;
  94         if num_single_char_names as u64 >= threshold {
  95             let span = self
  96                 .single_char_names
  97                 .iter()
  98                 .flatten()
  99                 .map(|ident| ident.span)
 100                 .collect::<Vec<_>>();
 101             span_lint(
 102                 self.cx,
 103                 MANY_SINGLE_CHAR_NAMES,
 104                 span,
 105                 &format!(
 106                     "{} bindings with single-character names in scope",
 107                     num_single_char_names
 108                 ),
 109             );
 110         }
 111     }
 112 }
 113
 114 // this list contains lists of names that are allowed to be similar
 115 // the assumption is that no name is ever contained in multiple lists.
 116 #[rustfmt::skip]
 117 const WHITELIST: &[&[&str]] = &[
 118     &["parsed", "parser"],
 119     &["lhs", "rhs"],
 120     &["tx", "rx"],
 121     &["set", "get"],
 122     &["args", "arms"],
 123     &["qpath", "path"],
 124     &["lit", "lint"],
 125 ];
 126
 127 struct SimilarNamesNameVisitor<'a, 'tcx, 'b>(&'b mut SimilarNamesLocalVisitor<'a, 'tcx>);
 128
 129 impl<'a, 'tcx, 'b> Visitor<'tcx> for SimilarNamesNameVisitor<'a, 'tcx, 'b> {
 130     fn visit_pat(&mut self, pat: &'tcx Pat) {
 131         match pat.kind {
 132             PatKind::Ident(_, ident, _) => self.check_ident(ident),
 133             PatKind::Struct(_, ref fields, _) => {
 134                 for field in fields {
 135                     if !field.is_shorthand {
 136                         self.visit_pat(&field.pat);
 137                     }
 138                 }
 139             },
 140             // just go through the first pattern, as either all patterns
 141             // bind the same bindings or rustc would have errored much earlier
 142             PatKind::Or(ref pats) => self.visit_pat(&pats[0]),
 143             _ => walk_pat(self, pat),
 144         }
 145     }
 146     fn visit_mac(&mut self, _mac: &Mac) {
 147         // do not check macs
 148     }
 149 }
 150
 151 #[must_use]
 152 fn get_whitelist(interned_name: &str) -> Option<&'static [&'static str]> {
 153     for &allow in WHITELIST {
 154         if whitelisted(interned_name, allow) {
 155             return Some(allow);
 156         }
 157     }
 158     None
 159 }
 160
 161 #[must_use]
 162 fn whitelisted(interned_name: &str, list: &[&str]) -> bool {
 163     list.iter()
 164         .any(|&name| interned_name.starts_with(name) || interned_name.ends_with(name))
 165 }
 166
 167 impl<'a, 'tcx, 'b> SimilarNamesNameVisitor<'a, 'tcx, 'b> {
 168     fn check_short_ident(&mut self, ident: Ident) {
 169         // Ignore shadowing
 170         if self
 171             .0
 172             .single_char_names
 173             .iter()
 174             .flatten()
 175             .any(|id| id.name == ident.name)
 176         {
 177             return;
 178         }
 179
 180         if let Some(scope) = &mut self.0.single_char_names.last_mut() {
 181             scope.push(ident);
 182         }
 183     }
 184
 185     #[allow(clippy::too_many_lines)]
 186     fn check_ident(&mut self, ident: Ident) {
 187         let interned_name = ident.name.as_str();
 188         if interned_name.chars().any(char::is_uppercase) {
 189             return;
 190         }
 191         if interned_name.chars().all(|c| c.is_digit(10) || c == '_') {
 192             span_lint(
 193                 self.0.cx,
 194                 JUST_UNDERSCORES_AND_DIGITS,
 195                 ident.span,
 196                 "consider choosing a more descriptive name",
 197             );
 198             return;
 199         }
 200         let count = interned_name.chars().count();
 201         if count < 3 {
 202             if count == 1 {
 203                 self.check_short_ident(ident);
 204             }
 205             return;
 206         }
 207         for existing_name in &self.0.names {
 208             if whitelisted(&interned_name, existing_name.whitelist) {
 209                 continue;
 210             }
 211             let mut split_at = None;
 212             match existing_name.len.cmp(&count) {
 213                 Ordering::Greater => {
 214                     if existing_name.len - count != 1 || levenstein_not_1(&interned_name, &existing_name.interned) {
 215                         continue;
 216                     }
 217                 },
 218                 Ordering::Less => {
 219                     if count - existing_name.len != 1 || levenstein_not_1(&existing_name.interned, &interned_name) {
 220                         continue;
 221                     }
 222                 },
 223                 Ordering::Equal => {
 224                     let mut interned_chars = interned_name.chars();
 225                     let mut existing_chars = existing_name.interned.chars();
 226                     let first_i = interned_chars.next().expect("we know we have at least one char");
 227                     let first_e = existing_chars.next().expect("we know we have at least one char");
 228                     let eq_or_numeric = |(a, b): (char, char)| a == b || a.is_numeric() && b.is_numeric();
 229
 230                     if eq_or_numeric((first_i, first_e)) {
 231                         let last_i = interned_chars.next_back().expect("we know we have at least two chars");
 232                         let last_e = existing_chars.next_back().expect("we know we have at least two chars");
 233                         if eq_or_numeric((last_i, last_e)) {
 234                             if interned_chars
 235                                 .zip(existing_chars)
 236                                 .filter(|&ie| !eq_or_numeric(ie))
 237                                 .count()
 238                                 != 1
 239                             {
 240                                 continue;
 241                             }
 242                         } else {
 243                             let second_last_i = interned_chars
 244                                 .next_back()
 245                                 .expect("we know we have at least three chars");
 246                             let second_last_e = existing_chars
 247                                 .next_back()
 248                                 .expect("we know we have at least three chars");
 249                             if !eq_or_numeric((second_last_i, second_last_e))
 250                                 || second_last_i == '_'
 251                                 || !interned_chars.zip(existing_chars).all(eq_or_numeric)
 252                             {
 253                                 // allowed similarity foo_x, foo_y
 254                                 // or too many chars differ (foo_x, boo_y) or (foox, booy)
 255                                 continue;
 256                             }
 257                             split_at = interned_name.char_indices().rev().next().map(|(i, _)| i);
 258                         }
 259                     } else {
 260                         let second_i = interned_chars.next().expect("we know we have at least two chars");
 261                         let second_e = existing_chars.next().expect("we know we have at least two chars");
 262                         if !eq_or_numeric((second_i, second_e))
 263                             || second_i == '_'
 264                             || !interned_chars.zip(existing_chars).all(eq_or_numeric)
 265                         {
 266                             // allowed similarity x_foo, y_foo
 267                             // or too many chars differ (x_foo, y_boo) or (xfoo, yboo)
 268                             continue;
 269                         }
 270                         split_at = interned_name.chars().next().map(char::len_utf8);
 271                     }
 272                 },
 273             }
 274             span_lint_and_then(
 275                 self.0.cx,
 276                 SIMILAR_NAMES,
 277                 ident.span,
 278                 "binding's name is too similar to existing binding",
 279                 |diag| {
 280                     diag.span_note(existing_name.span, "existing binding defined here");
 281                     if let Some(split) = split_at {
 282                         diag.span_help(
 283                             ident.span,
 284                             &format!(
 285                                 "separate the discriminating character by an \
 286                                  underscore like: `{}_{}`",
 287                                 &interned_name[..split],
 288                                 &interned_name[split..]
 289                             ),
 290                         );
 291                     }
 292                 },
 293             );
 294             return;
 295         }
 296         self.0.names.push(ExistingName {
 297             whitelist: get_whitelist(&interned_name).unwrap_or(&[]),
 298             interned: interned_name,
 299             span: ident.span,
 300             len: count,
 301         });
 302     }
 303 }
 304
 305 impl<'a, 'b> SimilarNamesLocalVisitor<'a, 'b> {
 306     /// ensure scoping rules work
 307     fn apply<F: for<'c> Fn(&'c mut Self)>(&mut self, f: F) {
 308         let n = self.names.len();
 309         let single_char_count = self.single_char_names.len();
 310         f(self);
 311         self.names.truncate(n);
 312         self.single_char_names.truncate(single_char_count);
 313     }
 314 }
 315
 316 impl<'a, 'tcx> Visitor<'tcx> for SimilarNamesLocalVisitor<'a, 'tcx> {
 317     fn visit_local(&mut self, local: &'tcx Local) {
 318         if let Some(ref init) = local.init {
 319             self.apply(|this| walk_expr(this, &**init));
 320         }
 321         // add the pattern after the expression because the bindings aren't available
 322         // yet in the init
 323         // expression
 324         SimilarNamesNameVisitor(self).visit_pat(&*local.pat);
 325     }
 326     fn visit_block(&mut self, blk: &'tcx Block) {
 327         self.single_char_names.push(vec![]);
 328
 329         self.apply(|this| walk_block(this, blk));
 330
 331         self.check_single_char_names();
 332         self.single_char_names.pop();
 333     }
 334     fn visit_arm(&mut self, arm: &'tcx Arm) {
 335         self.single_char_names.push(vec![]);
 336
 337         self.apply(|this| {
 338             SimilarNamesNameVisitor(this).visit_pat(&arm.pat);
 339             this.apply(|this| walk_expr(this, &arm.body));
 340         });
 341
 342         self.check_single_char_names();
 343         self.single_char_names.pop();
 344     }
 345     fn visit_item(&mut self, _: &Item) {
 346         // do not recurse into inner items
 347     }
 348     fn visit_mac(&mut self, _mac: &Mac) {
 349         // do not check macs
 350     }
 351 }
 352
 353 impl EarlyLintPass for NonExpressiveNames {
 354     fn check_item(&mut self, cx: &EarlyContext<'_>, item: &Item) {
 355         if let ItemKind::Fn(ref sig, _, Some(ref blk)) = item.kind {
 356             do_check(self, cx, &item.attrs, &sig.decl, blk);
 357         }
 358     }
 359
 360     fn check_impl_item(&mut self, cx: &EarlyContext<'_>, item: &AssocItem) {
 361         if let AssocItemKind::Fn(ref sig, Some(ref blk)) = item.kind {
 362             do_check(self, cx, &item.attrs, &sig.decl, blk);
 363         }
 364     }
 365 }
 366
 367 fn do_check(lint: &mut NonExpressiveNames, cx: &EarlyContext<'_>, attrs: &[Attribute], decl: &FnDecl, blk: &Block) {
 368     if !attr::contains_name(attrs, sym!(test)) {
 369         let mut visitor = SimilarNamesLocalVisitor {
 370             names: Vec::new(),
 371             cx,
 372             lint,
 373             single_char_names: vec![vec![]],
 374         };
 375
 376         // initialize with function arguments
 377         for arg in &decl.inputs {
 378             SimilarNamesNameVisitor(&mut visitor).visit_pat(&arg.pat);
 379         }
 380         // walk all other bindings
 381         walk_block(&mut visitor, blk);
 382
 383         visitor.check_single_char_names();
 384     }
 385 }
 386
 387 /// Precondition: `a_name.chars().count() < b_name.chars().count()`.
 388 #[must_use]
 389 fn levenstein_not_1(a_name: &str, b_name: &str) -> bool {
 390     debug_assert!(a_name.chars().count() < b_name.chars().count());
 391     let mut a_chars = a_name.chars();
 392     let mut b_chars = b_name.chars();
 393     while let (Some(a), Some(b)) = (a_chars.next(), b_chars.next()) {
 394         if a == b {
 395             continue;
 396         }
 397         if let Some(b2) = b_chars.next() {
 398             // check if there's just one character inserted
 399             return a != b2 || a_chars.ne(b_chars);
 400         } else {
 401             // tuple
 402             // ntuple
 403             return true;
 404         }
 405     }
 406     // for item in items
 407     true
 408 }