1 // Copyright 2014-2018 The Rust Project Developers. See the COPYRIGHT
2 // file at the top-level directory of this distribution.
4 // Licensed under the Apache License, Version 2.0 <LICENSE-APACHE or
5 // http://www.apache.org/licenses/LICENSE-2.0> or the MIT license
6 // <LICENSE-MIT or http://opensource.org/licenses/MIT>, at your
7 // option. This file may not be copied, modified, or distributed
8 // except according to those terms.
12 use crate::rustc::hir::*;
13 use crate::rustc::lint::{LateContext, LateLintPass, LintArray, LintPass};
14 use crate::rustc::{declare_tool_lint, lint_array};
15 use crate::rustc_data_structures::fx::FxHashSet;
16 use if_chain::if_chain;
17 use crate::syntax::ast::{LitKind, NodeId, StrStyle};
18 use crate::syntax::source_map::{BytePos, Span};
19 use crate::utils::{is_expn_of, match_def_path, match_type, opt_def_id, paths, span_help_and_lint, span_lint};
20 use crate::consts::{constant, Constant};
21 use std::convert::TryFrom;
23 /// **What it does:** Checks [regex](https://crates.io/crates/regex) creation
24 /// (with `Regex::new`,`RegexBuilder::new` or `RegexSet::new`) for correct
27 /// **Why is this bad?** This will lead to a runtime panic.
29 /// **Known problems:** None.
35 declare_clippy_lint! {
38 "invalid regular expressions"
41 /// **What it does:** Checks for trivial [regex](https://crates.io/crates/regex)
42 /// creation (with `Regex::new`, `RegexBuilder::new` or `RegexSet::new`).
44 /// **Why is this bad?** Matching the regex can likely be replaced by `==` or
45 /// `str::starts_with`, `str::ends_with` or `std::contains` or other `str`
48 /// **Known problems:** None.
52 /// Regex::new("^foobar")
54 declare_clippy_lint! {
57 "trivial regular expressions"
60 /// **What it does:** Checks for usage of `regex!(_)` which (as of now) is
61 /// usually slower than `Regex::new(_)` unless called in a loop (which is a bad
64 /// **Why is this bad?** Performance, at least for now. The macro version is
65 /// likely to catch up long-term, but for now the dynamic version is faster.
67 /// **Known problems:** None.
73 declare_clippy_lint! {
76 "use of `regex!(_)` instead of `Regex::new(_)`"
79 #[derive(Clone, Default)]
81 spans: FxHashSet<Span>,
85 impl LintPass for Pass {
86 fn get_lints(&self) -> LintArray {
87 lint_array!(INVALID_REGEX, REGEX_MACRO, TRIVIAL_REGEX)
91 impl<'a, 'tcx> LateLintPass<'a, 'tcx> for Pass {
92 fn check_crate(&mut self, _: &LateContext<'a, 'tcx>, _: &'tcx Crate) {
96 fn check_block(&mut self, cx: &LateContext<'a, 'tcx>, block: &'tcx Block) {
98 if self.last.is_none();
99 if let Some(ref expr) = block.expr;
100 if match_type(cx, cx.tables.expr_ty(expr), &paths::REGEX);
101 if let Some(span) = is_expn_of(expr.span, "regex");
103 if !self.spans.contains(&span) {
107 "`regex!(_)` found. \
108 Please use `Regex::new(_)`, which is faster for now.");
109 self.spans.insert(span);
111 self.last = Some(block.id);
116 fn check_block_post(&mut self, _: &LateContext<'a, 'tcx>, block: &'tcx Block) {
117 if self.last.map_or(false, |id| block.id == id) {
122 fn check_expr(&mut self, cx: &LateContext<'a, 'tcx>, expr: &'tcx Expr) {
124 if let ExprKind::Call(ref fun, ref args) = expr.node;
125 if let ExprKind::Path(ref qpath) = fun.node;
127 if let Some(def_id) = opt_def_id(cx.tables.qpath_def(qpath, fun.hir_id));
129 if match_def_path(cx.tcx, def_id, &paths::REGEX_NEW) ||
130 match_def_path(cx.tcx, def_id, &paths::REGEX_BUILDER_NEW) {
131 check_regex(cx, &args[0], true);
132 } else if match_def_path(cx.tcx, def_id, &paths::REGEX_BYTES_NEW) ||
133 match_def_path(cx.tcx, def_id, &paths::REGEX_BYTES_BUILDER_NEW) {
134 check_regex(cx, &args[0], false);
135 } else if match_def_path(cx.tcx, def_id, &paths::REGEX_SET_NEW) {
136 check_set(cx, &args[0], true);
137 } else if match_def_path(cx.tcx, def_id, &paths::REGEX_BYTES_SET_NEW) {
138 check_set(cx, &args[0], false);
145 #[allow(clippy::cast_possible_truncation)] // truncation very unlikely here
146 fn str_span(base: Span, c: regex_syntax::ast::Span, offset: u16) -> Span {
147 let offset = u32::from(offset);
148 let end = base.lo() + BytePos(u32::try_from(c.end.offset).expect("offset too large") + offset);
149 let start = base.lo() + BytePos(u32::try_from(c.start.offset).expect("offset too large") + offset);
150 assert!(start <= end);
151 Span::new(start, end, base.ctxt())
154 fn const_str<'a, 'tcx>(cx: &LateContext<'a, 'tcx>, e: &'tcx Expr) -> Option<String> {
155 constant(cx, cx.tables, e).and_then(|(c, _)| match c {
156 Constant::Str(s) => Some(s),
161 fn is_trivial_regex(s: ®ex_syntax::hir::Hir) -> Option<&'static str> {
162 use regex_syntax::hir::HirKind::*;
163 use regex_syntax::hir::Anchor::*;
165 let is_literal = |e: &[regex_syntax::hir::Hir]| e.iter().all(|e| match *e.kind() {
172 Anchor(_) => Some("the regex is unlikely to be useful as it is"),
173 Literal(_) => Some("consider using `str::contains`"),
174 Alternation(ref exprs) => if exprs.iter().all(|e| e.kind().is_empty()) {
175 Some("the regex is unlikely to be useful as it is")
179 Concat(ref exprs) => match (exprs[0].kind(), exprs[exprs.len() - 1].kind()) {
180 (&Anchor(StartText), &Anchor(EndText)) if exprs[1..(exprs.len() - 1)].is_empty() => Some("consider using `str::is_empty`"),
181 (&Anchor(StartText), &Anchor(EndText)) if is_literal(&exprs[1..(exprs.len() - 1)]) => Some("consider using `==` on `str`s"),
182 (&Anchor(StartText), &Literal(_)) if is_literal(&exprs[1..]) => Some("consider using `str::starts_with`"),
183 (&Literal(_), &Anchor(EndText)) if is_literal(&exprs[1..(exprs.len() - 1)]) => Some("consider using `str::ends_with`"),
184 _ if is_literal(exprs) => Some("consider using `str::contains`"),
191 fn check_set<'a, 'tcx>(cx: &LateContext<'a, 'tcx>, expr: &'tcx Expr, utf8: bool) {
193 if let ExprKind::AddrOf(_, ref expr) = expr.node;
194 if let ExprKind::Array(ref exprs) = expr.node;
197 check_regex(cx, expr, utf8);
203 fn check_regex<'a, 'tcx>(cx: &LateContext<'a, 'tcx>, expr: &'tcx Expr, utf8: bool) {
204 let mut parser = regex_syntax::ParserBuilder::new()
206 .allow_invalid_utf8(!utf8)
209 if let ExprKind::Lit(ref lit) = expr.node {
210 if let LitKind::Str(ref r, style) = lit.node {
212 let offset = if let StrStyle::Raw(n) = style { 2 + n } else { 1 };
213 match parser.parse(r) {
214 Ok(r) => if let Some(repl) = is_trivial_regex(&r) {
223 Err(regex_syntax::Error::Parse(e)) => {
227 str_span(expr.span, *e.span(), offset),
228 &format!("regex syntax error: {}", e.kind()),
231 Err(regex_syntax::Error::Translate(e)) => {
235 str_span(expr.span, *e.span(), offset),
236 &format!("regex syntax error: {}", e.kind()),
244 &format!("regex syntax error: {}", e),
249 } else if let Some(r) = const_str(cx, expr) {
250 match parser.parse(&r) {
251 Ok(r) => if let Some(repl) = is_trivial_regex(&r) {
260 Err(regex_syntax::Error::Parse(e)) => {
265 &format!("regex syntax error on position {}: {}", e.span().start.offset, e.kind()),
268 Err(regex_syntax::Error::Translate(e)) => {
273 &format!("regex syntax error on position {}: {}", e.span().start.offset, e.kind()),
281 &format!("regex syntax error: {}", e),