1 use clippy_utils::consts::{constant, Constant};
2 use clippy_utils::diagnostics::{span_lint, span_lint_and_help};
3 use clippy_utils::{match_def_path, paths};
4 use if_chain::if_chain;
5 use rustc_ast::ast::{LitKind, StrStyle};
6 use rustc_hir::{BorrowKind, Expr, ExprKind};
7 use rustc_lint::{LateContext, LateLintPass};
8 use rustc_session::{declare_lint_pass, declare_tool_lint};
9 use rustc_span::source_map::{BytePos, Span};
10 use std::convert::TryFrom;
12 declare_clippy_lint! {
14 /// Checks [regex](https://crates.io/crates/regex) creation
15 /// (with `Regex::new`, `RegexBuilder::new`, or `RegexSet::new`) for correct
18 /// ### Why is this bad?
19 /// This will lead to a runtime panic.
25 #[clippy::version = "pre 1.29.0"]
28 "invalid regular expressions"
31 declare_clippy_lint! {
33 /// Checks for trivial [regex](https://crates.io/crates/regex)
34 /// creation (with `Regex::new`, `RegexBuilder::new`, or `RegexSet::new`).
36 /// ### Why is this bad?
37 /// Matching the regex can likely be replaced by `==` or
38 /// `str::starts_with`, `str::ends_with` or `std::contains` or other `str`
41 /// ### Known problems
42 /// If the same regex is going to be applied to multiple
43 /// inputs, the precomputations done by `Regex` construction can give
44 /// significantly better performance than any of the `str`-based methods.
48 /// Regex::new("^foobar")
50 #[clippy::version = "pre 1.29.0"]
53 "trivial regular expressions"
56 declare_lint_pass!(Regex => [INVALID_REGEX, TRIVIAL_REGEX]);
58 impl<'tcx> LateLintPass<'tcx> for Regex {
59 fn check_expr(&mut self, cx: &LateContext<'tcx>, expr: &'tcx Expr<'_>) {
61 if let ExprKind::Call(fun, args) = expr.kind;
62 if let ExprKind::Path(ref qpath) = fun.kind;
64 if let Some(def_id) = cx.qpath_res(qpath, fun.hir_id).opt_def_id();
66 if match_def_path(cx, def_id, &paths::REGEX_NEW) ||
67 match_def_path(cx, def_id, &paths::REGEX_BUILDER_NEW) {
68 check_regex(cx, &args[0], true);
69 } else if match_def_path(cx, def_id, &paths::REGEX_BYTES_NEW) ||
70 match_def_path(cx, def_id, &paths::REGEX_BYTES_BUILDER_NEW) {
71 check_regex(cx, &args[0], false);
72 } else if match_def_path(cx, def_id, &paths::REGEX_SET_NEW) {
73 check_set(cx, &args[0], true);
74 } else if match_def_path(cx, def_id, &paths::REGEX_BYTES_SET_NEW) {
75 check_set(cx, &args[0], false);
82 #[allow(clippy::cast_possible_truncation)] // truncation very unlikely here
84 fn str_span(base: Span, c: regex_syntax::ast::Span, offset: u16) -> Span {
85 let offset = u32::from(offset);
86 let end = base.lo() + BytePos(u32::try_from(c.end.offset).expect("offset too large") + offset);
87 let start = base.lo() + BytePos(u32::try_from(c.start.offset).expect("offset too large") + offset);
88 assert!(start <= end);
89 Span::new(start, end, base.ctxt(), base.parent())
92 fn const_str<'tcx>(cx: &LateContext<'tcx>, e: &'tcx Expr<'_>) -> Option<String> {
93 constant(cx, cx.typeck_results(), e).and_then(|(c, _)| match c {
94 Constant::Str(s) => Some(s),
99 fn is_trivial_regex(s: ®ex_syntax::hir::Hir) -> Option<&'static str> {
100 use regex_syntax::hir::Anchor::{EndText, StartText};
101 use regex_syntax::hir::HirKind::{Alternation, Anchor, Concat, Empty, Literal};
103 let is_literal = |e: &[regex_syntax::hir::Hir]| e.iter().all(|e| matches!(*e.kind(), Literal(_)));
106 Empty | Anchor(_) => Some("the regex is unlikely to be useful as it is"),
107 Literal(_) => Some("consider using `str::contains`"),
108 Alternation(ref exprs) => {
109 if exprs.iter().all(|e| e.kind().is_empty()) {
110 Some("the regex is unlikely to be useful as it is")
115 Concat(ref exprs) => match (exprs[0].kind(), exprs[exprs.len() - 1].kind()) {
116 (&Anchor(StartText), &Anchor(EndText)) if exprs[1..(exprs.len() - 1)].is_empty() => {
117 Some("consider using `str::is_empty`")
119 (&Anchor(StartText), &Anchor(EndText)) if is_literal(&exprs[1..(exprs.len() - 1)]) => {
120 Some("consider using `==` on `str`s")
122 (&Anchor(StartText), &Literal(_)) if is_literal(&exprs[1..]) => Some("consider using `str::starts_with`"),
123 (&Literal(_), &Anchor(EndText)) if is_literal(&exprs[1..(exprs.len() - 1)]) => {
124 Some("consider using `str::ends_with`")
126 _ if is_literal(exprs) => Some("consider using `str::contains`"),
133 fn check_set<'tcx>(cx: &LateContext<'tcx>, expr: &'tcx Expr<'_>, utf8: bool) {
135 if let ExprKind::AddrOf(BorrowKind::Ref, _, expr) = expr.kind;
136 if let ExprKind::Array(exprs) = expr.kind;
139 check_regex(cx, expr, utf8);
145 fn check_regex<'tcx>(cx: &LateContext<'tcx>, expr: &'tcx Expr<'_>, utf8: bool) {
146 let mut parser = regex_syntax::ParserBuilder::new()
148 .allow_invalid_utf8(!utf8)
151 if let ExprKind::Lit(ref lit) = expr.kind {
152 if let LitKind::Str(ref r, style) = lit.node {
154 let offset = if let StrStyle::Raw(n) = style { 2 + n } else { 1 };
155 match parser.parse(r) {
157 if let Some(repl) = is_trivial_regex(&r) {
158 span_lint_and_help(cx, TRIVIAL_REGEX, expr.span, "trivial regex", None, repl);
161 Err(regex_syntax::Error::Parse(e)) => {
165 str_span(expr.span, *e.span(), offset),
166 &format!("regex syntax error: {}", e.kind()),
169 Err(regex_syntax::Error::Translate(e)) => {
173 str_span(expr.span, *e.span(), offset),
174 &format!("regex syntax error: {}", e.kind()),
178 span_lint(cx, INVALID_REGEX, expr.span, &format!("regex syntax error: {}", e));
182 } else if let Some(r) = const_str(cx, expr) {
183 match parser.parse(&r) {
185 if let Some(repl) = is_trivial_regex(&r) {
186 span_lint_and_help(cx, TRIVIAL_REGEX, expr.span, "trivial regex", None, repl);
189 Err(regex_syntax::Error::Parse(e)) => {
194 &format!("regex syntax error on position {}: {}", e.span().start.offset, e.kind()),
197 Err(regex_syntax::Error::Translate(e)) => {
202 &format!("regex syntax error on position {}: {}", e.span().start.offset, e.kind()),
206 span_lint(cx, INVALID_REGEX, expr.span, &format!("regex syntax error: {}", e));