src/tools/clippy/clippy_lints/src/regex.rs

   1 use regex_syntax;
   2 use rustc::hir::*;
   3 use rustc::lint::*;
   4 use rustc::ty;
   5 use rustc::middle::const_val::ConstVal;
   6 use rustc_const_eval::ConstContext;
   7 use rustc::ty::subst::Substs;
   8 use std::collections::HashSet;
   9 use std::error::Error;
  10 use syntax::ast::{LitKind, NodeId};
  11 use syntax::codemap::{BytePos, Span};
  12 use syntax::symbol::InternedString;
  13 use utils::{is_expn_of, match_def_path, match_type, opt_def_id, paths, span_help_and_lint, span_lint};
  14
  15 /// **What it does:** Checks [regex](https://crates.io/crates/regex) creation
  16 /// (with `Regex::new`,`RegexBuilder::new` or `RegexSet::new`) for correct
  17 /// regex syntax.
  18 ///
  19 /// **Why is this bad?** This will lead to a runtime panic.
  20 ///
  21 /// **Known problems:** None.
  22 ///
  23 /// **Example:**
  24 /// ```rust
  25 /// Regex::new("|")
  26 /// ```
  27 declare_lint! {
  28     pub INVALID_REGEX,
  29     Deny,
  30     "invalid regular expressions"
  31 }
  32
  33 /// **What it does:** Checks for trivial [regex](https://crates.io/crates/regex)
  34 /// creation (with `Regex::new`, `RegexBuilder::new` or `RegexSet::new`).
  35 ///
  36 /// **Why is this bad?** Matching the regex can likely be replaced by `==` or
  37 /// `str::starts_with`, `str::ends_with` or `std::contains` or other `str`
  38 /// methods.
  39 ///
  40 /// **Known problems:** None.
  41 ///
  42 /// **Example:**
  43 /// ```rust
  44 /// Regex::new("^foobar")
  45 /// ```
  46 declare_lint! {
  47     pub TRIVIAL_REGEX,
  48     Warn,
  49     "trivial regular expressions"
  50 }
  51
  52 /// **What it does:** Checks for usage of `regex!(_)` which (as of now) is
  53 /// usually slower than `Regex::new(_)` unless called in a loop (which is a bad
  54 /// idea anyway).
  55 ///
  56 /// **Why is this bad?** Performance, at least for now. The macro version is
  57 /// likely to catch up long-term, but for now the dynamic version is faster.
  58 ///
  59 /// **Known problems:** None.
  60 ///
  61 /// **Example:**
  62 /// ```rust
  63 /// regex!("foo|bar")
  64 /// ```
  65 declare_lint! {
  66     pub REGEX_MACRO,
  67     Warn,
  68     "use of `regex!(_)` instead of `Regex::new(_)`"
  69 }
  70
  71 #[derive(Clone, Default)]
  72 pub struct Pass {
  73     spans: HashSet<Span>,
  74     last: Option<NodeId>,
  75 }
  76
  77 impl LintPass for Pass {
  78     fn get_lints(&self) -> LintArray {
  79         lint_array!(INVALID_REGEX, REGEX_MACRO, TRIVIAL_REGEX)
  80     }
  81 }
  82
  83 impl<'a, 'tcx> LateLintPass<'a, 'tcx> for Pass {
  84     fn check_crate(&mut self, _: &LateContext<'a, 'tcx>, _: &'tcx Crate) {
  85         self.spans.clear();
  86     }
  87
  88     fn check_block(&mut self, cx: &LateContext<'a, 'tcx>, block: &'tcx Block) {
  89         if_chain! {
  90             if self.last.is_none();
  91             if let Some(ref expr) = block.expr;
  92             if match_type(cx, cx.tables.expr_ty(expr), &paths::REGEX);
  93             if let Some(span) = is_expn_of(expr.span, "regex");
  94             then {
  95                 if !self.spans.contains(&span) {
  96                     span_lint(cx,
  97                               REGEX_MACRO,
  98                               span,
  99                               "`regex!(_)` found. \
 100                               Please use `Regex::new(_)`, which is faster for now.");
 101                     self.spans.insert(span);
 102                 }
 103                 self.last = Some(block.id);
 104             }
 105         }
 106     }
 107
 108     fn check_block_post(&mut self, _: &LateContext<'a, 'tcx>, block: &'tcx Block) {
 109         if self.last.map_or(false, |id| block.id == id) {
 110             self.last = None;
 111         }
 112     }
 113
 114     fn check_expr(&mut self, cx: &LateContext<'a, 'tcx>, expr: &'tcx Expr) {
 115         if_chain! {
 116             if let ExprCall(ref fun, ref args) = expr.node;
 117             if let ExprPath(ref qpath) = fun.node;
 118             if args.len() == 1;
 119             if let Some(def_id) = opt_def_id(cx.tables.qpath_def(qpath, fun.hir_id));
 120             then {
 121                 if match_def_path(cx.tcx, def_id, &paths::REGEX_NEW) ||
 122                    match_def_path(cx.tcx, def_id, &paths::REGEX_BUILDER_NEW) {
 123                     check_regex(cx, &args[0], true);
 124                 } else if match_def_path(cx.tcx, def_id, &paths::REGEX_BYTES_NEW) ||
 125                    match_def_path(cx.tcx, def_id, &paths::REGEX_BYTES_BUILDER_NEW) {
 126                     check_regex(cx, &args[0], false);
 127                 } else if match_def_path(cx.tcx, def_id, &paths::REGEX_SET_NEW) {
 128                     check_set(cx, &args[0], true);
 129                 } else if match_def_path(cx.tcx, def_id, &paths::REGEX_BYTES_SET_NEW) {
 130                     check_set(cx, &args[0], false);
 131                 }
 132             }
 133         }
 134     }
 135 }
 136
 137 #[allow(cast_possible_truncation)]
 138 fn str_span(base: Span, s: &str, c: usize) -> Span {
 139     let mut si = s.char_indices().skip(c);
 140
 141     match (si.next(), si.next()) {
 142         (Some((l, _)), Some((h, _))) => {
 143             Span::new(base.lo() + BytePos(l as u32), base.lo() + BytePos(h as u32), base.ctxt())
 144         },
 145         _ => base,
 146     }
 147 }
 148
 149 fn const_str<'a, 'tcx>(cx: &LateContext<'a, 'tcx>, e: &'tcx Expr) -> Option<InternedString> {
 150     let parent_item = cx.tcx.hir.get_parent(e.id);
 151     let parent_def_id = cx.tcx.hir.local_def_id(parent_item);
 152     let substs = Substs::identity_for_item(cx.tcx, parent_def_id);
 153     match ConstContext::new(cx.tcx, cx.param_env.and(substs), cx.tables).eval(e) {
 154         Ok(&ty::Const {
 155             val: ConstVal::Str(r),
 156             ..
 157         }) => Some(r),
 158         _ => None,
 159     }
 160 }
 161
 162 fn is_trivial_regex(s: &regex_syntax::Expr) -> Option<&'static str> {
 163     use regex_syntax::Expr;
 164
 165     match *s {
 166         Expr::Empty | Expr::StartText | Expr::EndText => Some("the regex is unlikely to be useful as it is"),
 167         Expr::Literal { .. } => Some("consider using `str::contains`"),
 168         Expr::Concat(ref exprs) => match exprs.len() {
 169             2 => match (&exprs[0], &exprs[1]) {
 170                 (&Expr::StartText, &Expr::EndText) => Some("consider using `str::is_empty`"),
 171                 (&Expr::StartText, &Expr::Literal { .. }) => Some("consider using `str::starts_with`"),
 172                 (&Expr::Literal { .. }, &Expr::EndText) => Some("consider using `str::ends_with`"),
 173                 _ => None,
 174             },
 175             3 => if let (&Expr::StartText, &Expr::Literal { .. }, &Expr::EndText) = (&exprs[0], &exprs[1], &exprs[2]) {
 176                 Some("consider using `==` on `str`s")
 177             } else {
 178                 None
 179             },
 180             _ => None,
 181         },
 182         _ => None,
 183     }
 184 }
 185
 186 fn check_set<'a, 'tcx>(cx: &LateContext<'a, 'tcx>, expr: &'tcx Expr, utf8: bool) {
 187     if_chain! {
 188         if let ExprAddrOf(_, ref expr) = expr.node;
 189         if let ExprArray(ref exprs) = expr.node;
 190         then {
 191             for expr in exprs {
 192                 check_regex(cx, expr, utf8);
 193             }
 194         }
 195     }
 196 }
 197
 198 fn check_regex<'a, 'tcx>(cx: &LateContext<'a, 'tcx>, expr: &'tcx Expr, utf8: bool) {
 199     let builder = regex_syntax::ExprBuilder::new().unicode(utf8);
 200
 201     if let ExprLit(ref lit) = expr.node {
 202         if let LitKind::Str(ref r, _) = lit.node {
 203             let r = &r.as_str();
 204             match builder.parse(r) {
 205                 Ok(r) => if let Some(repl) = is_trivial_regex(&r) {
 206                     span_help_and_lint(
 207                         cx,
 208                         TRIVIAL_REGEX,
 209                         expr.span,
 210                         "trivial regex",
 211                         &format!("consider using {}", repl),
 212                     );
 213                 },
 214                 Err(e) => {
 215                     span_lint(
 216                         cx,
 217                         INVALID_REGEX,
 218                         str_span(expr.span, r, e.position()),
 219                         &format!("regex syntax error: {}", e.description()),
 220                     );
 221                 },
 222             }
 223         }
 224     } else if let Some(r) = const_str(cx, expr) {
 225         match builder.parse(&r) {
 226             Ok(r) => if let Some(repl) = is_trivial_regex(&r) {
 227                 span_help_and_lint(
 228                     cx,
 229                     TRIVIAL_REGEX,
 230                     expr.span,
 231                     "trivial regex",
 232                     &format!("consider using {}", repl),
 233                 );
 234             },
 235             Err(e) => {
 236                 span_lint(
 237                     cx,
 238                     INVALID_REGEX,
 239                     expr.span,
 240                     &format!("regex syntax error on position {}: {}", e.position(), e.description()),
 241                 );
 242             },
 243         }
 244     }
 245 }