clippy_lints/src/non_expressive_names.rs

   1 use crate::utils::{span_lint, span_lint_and_then};
   2 use rustc::impl_lint_pass;
   3 use rustc::lint::{EarlyContext, EarlyLintPass, LintArray, LintPass};
   4 use rustc_session::declare_tool_lint;
   5 use std::cmp::Ordering;
   6 use syntax::ast::*;
   7 use syntax::attr;
   8 use syntax::source_map::Span;
   9 use syntax::symbol::SymbolStr;
  10 use syntax::visit::{walk_block, walk_expr, walk_pat, Visitor};
  11
  12 declare_clippy_lint! {
  13     /// **What it does:** Checks for names that are very similar and thus confusing.
  14     ///
  15     /// **Why is this bad?** It's hard to distinguish between names that differ only
  16     /// by a single character.
  17     ///
  18     /// **Known problems:** None?
  19     ///
  20     /// **Example:**
  21     /// ```ignore
  22     /// let checked_exp = something;
  23     /// let checked_expr = something_else;
  24     /// ```
  25     pub SIMILAR_NAMES,
  26     pedantic,
  27     "similarly named items and bindings"
  28 }
  29
  30 declare_clippy_lint! {
  31     /// **What it does:** Checks for too many variables whose name consists of a
  32     /// single character.
  33     ///
  34     /// **Why is this bad?** It's hard to memorize what a variable means without a
  35     /// descriptive name.
  36     ///
  37     /// **Known problems:** None?
  38     ///
  39     /// **Example:**
  40     /// ```ignore
  41     /// let (a, b, c, d, e, f, g) = (...);
  42     /// ```
  43     pub MANY_SINGLE_CHAR_NAMES,
  44     style,
  45     "too many single character bindings"
  46 }
  47
  48 declare_clippy_lint! {
  49     /// **What it does:** Checks if you have variables whose name consists of just
  50     /// underscores and digits.
  51     ///
  52     /// **Why is this bad?** It's hard to memorize what a variable means without a
  53     /// descriptive name.
  54     ///
  55     /// **Known problems:** None?
  56     ///
  57     /// **Example:**
  58     /// ```rust
  59     /// let _1 = 1;
  60     /// let ___1 = 1;
  61     /// let __1___2 = 11;
  62     /// ```
  63     pub JUST_UNDERSCORES_AND_DIGITS,
  64     style,
  65     "unclear name"
  66 }
  67
  68 #[derive(Copy, Clone)]
  69 pub struct NonExpressiveNames {
  70     pub single_char_binding_names_threshold: u64,
  71 }
  72
  73 impl_lint_pass!(NonExpressiveNames => [SIMILAR_NAMES, MANY_SINGLE_CHAR_NAMES, JUST_UNDERSCORES_AND_DIGITS]);
  74
  75 struct ExistingName {
  76     interned: SymbolStr,
  77     span: Span,
  78     len: usize,
  79     whitelist: &'static [&'static str],
  80 }
  81
  82 struct SimilarNamesLocalVisitor<'a, 'tcx> {
  83     names: Vec<ExistingName>,
  84     cx: &'a EarlyContext<'tcx>,
  85     lint: &'a NonExpressiveNames,
  86
  87     /// A stack of scopes containing the single-character bindings in each scope.
  88     single_char_names: Vec<Vec<Ident>>,
  89 }
  90
  91 impl<'a, 'tcx> SimilarNamesLocalVisitor<'a, 'tcx> {
  92     fn check_single_char_names(&self) {
  93         let num_single_char_names = self.single_char_names.iter().flatten().count();
  94         let threshold = self.lint.single_char_binding_names_threshold;
  95         if num_single_char_names as u64 >= threshold {
  96             let span = self
  97                 .single_char_names
  98                 .iter()
  99                 .flatten()
 100                 .map(|ident| ident.span)
 101                 .collect::<Vec<_>>();
 102             span_lint(
 103                 self.cx,
 104                 MANY_SINGLE_CHAR_NAMES,
 105                 span,
 106                 &format!(
 107                     "{} bindings with single-character names in scope",
 108                     num_single_char_names
 109                 ),
 110             );
 111         }
 112     }
 113 }
 114
 115 // this list contains lists of names that are allowed to be similar
 116 // the assumption is that no name is ever contained in multiple lists.
 117 #[rustfmt::skip]
 118 const WHITELIST: &[&[&str]] = &[
 119     &["parsed", "parser"],
 120     &["lhs", "rhs"],
 121     &["tx", "rx"],
 122     &["set", "get"],
 123     &["args", "arms"],
 124     &["qpath", "path"],
 125     &["lit", "lint"],
 126 ];
 127
 128 struct SimilarNamesNameVisitor<'a, 'tcx, 'b>(&'b mut SimilarNamesLocalVisitor<'a, 'tcx>);
 129
 130 impl<'a, 'tcx, 'b> Visitor<'tcx> for SimilarNamesNameVisitor<'a, 'tcx, 'b> {
 131     fn visit_pat(&mut self, pat: &'tcx Pat) {
 132         match pat.kind {
 133             PatKind::Ident(_, ident, _) => self.check_ident(ident),
 134             PatKind::Struct(_, ref fields, _) => {
 135                 for field in fields {
 136                     if !field.is_shorthand {
 137                         self.visit_pat(&field.pat);
 138                     }
 139                 }
 140             },
 141             // just go through the first pattern, as either all patterns
 142             // bind the same bindings or rustc would have errored much earlier
 143             PatKind::Or(ref pats) => self.visit_pat(&pats[0]),
 144             _ => walk_pat(self, pat),
 145         }
 146     }
 147     fn visit_mac(&mut self, _mac: &Mac) {
 148         // do not check macs
 149     }
 150 }
 151
 152 #[must_use]
 153 fn get_whitelist(interned_name: &str) -> Option<&'static [&'static str]> {
 154     for &allow in WHITELIST {
 155         if whitelisted(interned_name, allow) {
 156             return Some(allow);
 157         }
 158     }
 159     None
 160 }
 161
 162 #[must_use]
 163 fn whitelisted(interned_name: &str, list: &[&str]) -> bool {
 164     list.iter()
 165         .any(|&name| interned_name.starts_with(name) || interned_name.ends_with(name))
 166 }
 167
 168 impl<'a, 'tcx, 'b> SimilarNamesNameVisitor<'a, 'tcx, 'b> {
 169     fn check_short_ident(&mut self, ident: Ident) {
 170         // Ignore shadowing
 171         if self
 172             .0
 173             .single_char_names
 174             .iter()
 175             .flatten()
 176             .any(|id| id.name == ident.name)
 177         {
 178             return;
 179         }
 180
 181         if let Some(scope) = &mut self.0.single_char_names.last_mut() {
 182             scope.push(ident);
 183         }
 184     }
 185
 186     #[allow(clippy::too_many_lines)]
 187     fn check_ident(&mut self, ident: Ident) {
 188         let interned_name = ident.name.as_str();
 189         if interned_name.chars().any(char::is_uppercase) {
 190             return;
 191         }
 192         if interned_name.chars().all(|c| c.is_digit(10) || c == '_') {
 193             span_lint(
 194                 self.0.cx,
 195                 JUST_UNDERSCORES_AND_DIGITS,
 196                 ident.span,
 197                 "consider choosing a more descriptive name",
 198             );
 199             return;
 200         }
 201         let count = interned_name.chars().count();
 202         if count < 3 {
 203             if count == 1 {
 204                 self.check_short_ident(ident);
 205             }
 206             return;
 207         }
 208         for existing_name in &self.0.names {
 209             if whitelisted(&interned_name, existing_name.whitelist) {
 210                 continue;
 211             }
 212             let mut split_at = None;
 213             match existing_name.len.cmp(&count) {
 214                 Ordering::Greater => {
 215                     if existing_name.len - count != 1 || levenstein_not_1(&interned_name, &existing_name.interned) {
 216                         continue;
 217                     }
 218                 },
 219                 Ordering::Less => {
 220                     if count - existing_name.len != 1 || levenstein_not_1(&existing_name.interned, &interned_name) {
 221                         continue;
 222                     }
 223                 },
 224                 Ordering::Equal => {
 225                     let mut interned_chars = interned_name.chars();
 226                     let mut existing_chars = existing_name.interned.chars();
 227                     let first_i = interned_chars.next().expect("we know we have at least one char");
 228                     let first_e = existing_chars.next().expect("we know we have at least one char");
 229                     let eq_or_numeric = |(a, b): (char, char)| a == b || a.is_numeric() && b.is_numeric();
 230
 231                     if eq_or_numeric((first_i, first_e)) {
 232                         let last_i = interned_chars.next_back().expect("we know we have at least two chars");
 233                         let last_e = existing_chars.next_back().expect("we know we have at least two chars");
 234                         if eq_or_numeric((last_i, last_e)) {
 235                             if interned_chars
 236                                 .zip(existing_chars)
 237                                 .filter(|&ie| !eq_or_numeric(ie))
 238                                 .count()
 239                                 != 1
 240                             {
 241                                 continue;
 242                             }
 243                         } else {
 244                             let second_last_i = interned_chars
 245                                 .next_back()
 246                                 .expect("we know we have at least three chars");
 247                             let second_last_e = existing_chars
 248                                 .next_back()
 249                                 .expect("we know we have at least three chars");
 250                             if !eq_or_numeric((second_last_i, second_last_e))
 251                                 || second_last_i == '_'
 252                                 || !interned_chars.zip(existing_chars).all(eq_or_numeric)
 253                             {
 254                                 // allowed similarity foo_x, foo_y
 255                                 // or too many chars differ (foo_x, boo_y) or (foox, booy)
 256                                 continue;
 257                             }
 258                             split_at = interned_name.char_indices().rev().next().map(|(i, _)| i);
 259                         }
 260                     } else {
 261                         let second_i = interned_chars.next().expect("we know we have at least two chars");
 262                         let second_e = existing_chars.next().expect("we know we have at least two chars");
 263                         if !eq_or_numeric((second_i, second_e))
 264                             || second_i == '_'
 265                             || !interned_chars.zip(existing_chars).all(eq_or_numeric)
 266                         {
 267                             // allowed similarity x_foo, y_foo
 268                             // or too many chars differ (x_foo, y_boo) or (xfoo, yboo)
 269                             continue;
 270                         }
 271                         split_at = interned_name.chars().next().map(char::len_utf8);
 272                     }
 273                 },
 274             }
 275             span_lint_and_then(
 276                 self.0.cx,
 277                 SIMILAR_NAMES,
 278                 ident.span,
 279                 "binding's name is too similar to existing binding",
 280                 |diag| {
 281                     diag.span_note(existing_name.span, "existing binding defined here");
 282                     if let Some(split) = split_at {
 283                         diag.span_help(
 284                             ident.span,
 285                             &format!(
 286                                 "separate the discriminating character by an \
 287                                  underscore like: `{}_{}`",
 288                                 &interned_name[..split],
 289                                 &interned_name[split..]
 290                             ),
 291                         );
 292                     }
 293                 },
 294             );
 295             return;
 296         }
 297         self.0.names.push(ExistingName {
 298             whitelist: get_whitelist(&interned_name).unwrap_or(&[]),
 299             interned: interned_name,
 300             span: ident.span,
 301             len: count,
 302         });
 303     }
 304 }
 305
 306 impl<'a, 'b> SimilarNamesLocalVisitor<'a, 'b> {
 307     /// ensure scoping rules work
 308     fn apply<F: for<'c> Fn(&'c mut Self)>(&mut self, f: F) {
 309         let n = self.names.len();
 310         let single_char_count = self.single_char_names.len();
 311         f(self);
 312         self.names.truncate(n);
 313         self.single_char_names.truncate(single_char_count);
 314     }
 315 }
 316
 317 impl<'a, 'tcx> Visitor<'tcx> for SimilarNamesLocalVisitor<'a, 'tcx> {
 318     fn visit_local(&mut self, local: &'tcx Local) {
 319         if let Some(ref init) = local.init {
 320             self.apply(|this| walk_expr(this, &**init));
 321         }
 322         // add the pattern after the expression because the bindings aren't available
 323         // yet in the init
 324         // expression
 325         SimilarNamesNameVisitor(self).visit_pat(&*local.pat);
 326     }
 327     fn visit_block(&mut self, blk: &'tcx Block) {
 328         self.single_char_names.push(vec![]);
 329
 330         self.apply(|this| walk_block(this, blk));
 331
 332         self.check_single_char_names();
 333         self.single_char_names.pop();
 334     }
 335     fn visit_arm(&mut self, arm: &'tcx Arm) {
 336         self.single_char_names.push(vec![]);
 337
 338         self.apply(|this| {
 339             SimilarNamesNameVisitor(this).visit_pat(&arm.pat);
 340             this.apply(|this| walk_expr(this, &arm.body));
 341         });
 342
 343         self.check_single_char_names();
 344         self.single_char_names.pop();
 345     }
 346     fn visit_item(&mut self, _: &Item) {
 347         // do not recurse into inner items
 348     }
 349     fn visit_mac(&mut self, _mac: &Mac) {
 350         // do not check macs
 351     }
 352 }
 353
 354 impl EarlyLintPass for NonExpressiveNames {
 355     fn check_item(&mut self, cx: &EarlyContext<'_>, item: &Item) {
 356         if let ItemKind::Fn(ref sig, _, ref blk) = item.kind {
 357             do_check(self, cx, &item.attrs, &sig.decl, blk);
 358         }
 359     }
 360
 361     fn check_impl_item(&mut self, cx: &EarlyContext<'_>, item: &AssocItem) {
 362         if let AssocItemKind::Fn(ref sig, Some(ref blk)) = item.kind {
 363             do_check(self, cx, &item.attrs, &sig.decl, blk);
 364         }
 365     }
 366 }
 367
 368 fn do_check(lint: &mut NonExpressiveNames, cx: &EarlyContext<'_>, attrs: &[Attribute], decl: &FnDecl, blk: &Block) {
 369     if !attr::contains_name(attrs, sym!(test)) {
 370         let mut visitor = SimilarNamesLocalVisitor {
 371             names: Vec::new(),
 372             cx,
 373             lint,
 374             single_char_names: vec![vec![]],
 375         };
 376
 377         // initialize with function arguments
 378         for arg in &decl.inputs {
 379             SimilarNamesNameVisitor(&mut visitor).visit_pat(&arg.pat);
 380         }
 381         // walk all other bindings
 382         walk_block(&mut visitor, blk);
 383
 384         visitor.check_single_char_names();
 385     }
 386 }
 387
 388 /// Precondition: `a_name.chars().count() < b_name.chars().count()`.
 389 #[must_use]
 390 fn levenstein_not_1(a_name: &str, b_name: &str) -> bool {
 391     debug_assert!(a_name.chars().count() < b_name.chars().count());
 392     let mut a_chars = a_name.chars();
 393     let mut b_chars = b_name.chars();
 394     while let (Some(a), Some(b)) = (a_chars.next(), b_chars.next()) {
 395         if a == b {
 396             continue;
 397         }
 398         if let Some(b2) = b_chars.next() {
 399             // check if there's just one character inserted
 400             return a != b2 || a_chars.ne(b_chars);
 401         } else {
 402             // tuple
 403             // ntuple
 404             return true;
 405         }
 406     }
 407     // for item in items
 408     true
 409 }