eslint/lib/rules/no-useless-backreference.js

   1 /**
   2  * @fileoverview Rule to disallow useless backreferences in regular expressions
   3  * @author Milos Djermanovic
   4  */
   5
   6 "use strict";
   7
   8 //------------------------------------------------------------------------------
   9 // Requirements
  10 //------------------------------------------------------------------------------
  11
  12 const { CALL, CONSTRUCT, ReferenceTracker, getStringIfConstant } = require("eslint-utils");
  13 const { RegExpParser, visitRegExpAST } = require("regexpp");
  14
  15 //------------------------------------------------------------------------------
  16 // Helpers
  17 //------------------------------------------------------------------------------
  18
  19 const parser = new RegExpParser();
  20
  21 /**
  22  * Finds the path from the given `regexpp` AST node to the root node.
  23  * @param {regexpp.Node} node Node.
  24  * @returns {regexpp.Node[]} Array that starts with the given node and ends with the root node.
  25  */
  26 function getPathToRoot(node) {
  27     const path = [];
  28     let current = node;
  29
  30     do {
  31         path.push(current);
  32         current = current.parent;
  33     } while (current);
  34
  35     return path;
  36 }
  37
  38 /**
  39  * Determines whether the given `regexpp` AST node is a lookaround node.
  40  * @param {regexpp.Node} node Node.
  41  * @returns {boolean} `true` if it is a lookaround node.
  42  */
  43 function isLookaround(node) {
  44     return node.type === "Assertion" &&
  45         (node.kind === "lookahead" || node.kind === "lookbehind");
  46 }
  47
  48 /**
  49  * Determines whether the given `regexpp` AST node is a negative lookaround node.
  50  * @param {regexpp.Node} node Node.
  51  * @returns {boolean} `true` if it is a negative lookaround node.
  52  */
  53 function isNegativeLookaround(node) {
  54     return isLookaround(node) && node.negate;
  55 }
  56
  57 //------------------------------------------------------------------------------
  58 // Rule Definition
  59 //------------------------------------------------------------------------------
  60
  61 /** @type {import('../shared/types').Rule} */
  62 module.exports = {
  63     meta: {
  64         type: "problem",
  65
  66         docs: {
  67             description: "Disallow useless backreferences in regular expressions",
  68             recommended: true,
  69             url: "https://eslint.org/docs/rules/no-useless-backreference"
  70         },
  71
  72         schema: [],
  73
  74         messages: {
  75             nested: "Backreference '{{ bref }}' will be ignored. It references group '{{ group }}' from within that group.",
  76             forward: "Backreference '{{ bref }}' will be ignored. It references group '{{ group }}' which appears later in the pattern.",
  77             backward: "Backreference '{{ bref }}' will be ignored. It references group '{{ group }}' which appears before in the same lookbehind.",
  78             disjunctive: "Backreference '{{ bref }}' will be ignored. It references group '{{ group }}' which is in another alternative.",
  79             intoNegativeLookaround: "Backreference '{{ bref }}' will be ignored. It references group '{{ group }}' which is in a negative lookaround."
  80         }
  81     },
  82
  83     create(context) {
  84
  85         /**
  86          * Checks and reports useless backreferences in the given regular expression.
  87          * @param {ASTNode} node Node that represents regular expression. A regex literal or RegExp constructor call.
  88          * @param {string} pattern Regular expression pattern.
  89          * @param {string} flags Regular expression flags.
  90          * @returns {void}
  91          */
  92         function checkRegex(node, pattern, flags) {
  93             let regExpAST;
  94
  95             try {
  96                 regExpAST = parser.parsePattern(pattern, 0, pattern.length, flags.includes("u"));
  97             } catch {
  98
  99                 // Ignore regular expressions with syntax errors
 100                 return;
 101             }
 102
 103             visitRegExpAST(regExpAST, {
 104                 onBackreferenceEnter(bref) {
 105                     const group = bref.resolved,
 106                         brefPath = getPathToRoot(bref),
 107                         groupPath = getPathToRoot(group);
 108                     let messageId = null;
 109
 110                     if (brefPath.includes(group)) {
 111
 112                         // group is bref's ancestor => bref is nested ('nested reference') => group hasn't matched yet when bref starts to match.
 113                         messageId = "nested";
 114                     } else {
 115
 116                         // Start from the root to find the lowest common ancestor.
 117                         let i = brefPath.length - 1,
 118                             j = groupPath.length - 1;
 119
 120                         do {
 121                             i--;
 122                             j--;
 123                         } while (brefPath[i] === groupPath[j]);
 124
 125                         const indexOfLowestCommonAncestor = j + 1,
 126                             groupCut = groupPath.slice(0, indexOfLowestCommonAncestor),
 127                             commonPath = groupPath.slice(indexOfLowestCommonAncestor),
 128                             lowestCommonLookaround = commonPath.find(isLookaround),
 129                             isMatchingBackward = lowestCommonLookaround && lowestCommonLookaround.kind === "lookbehind";
 130
 131                         if (!isMatchingBackward && bref.end <= group.start) {
 132
 133                             // bref is left, group is right ('forward reference') => group hasn't matched yet when bref starts to match.
 134                             messageId = "forward";
 135                         } else if (isMatchingBackward && group.end <= bref.start) {
 136
 137                             // the opposite of the previous when the regex is matching backward in a lookbehind context.
 138                             messageId = "backward";
 139                         } else if (groupCut[groupCut.length - 1].type === "Alternative") {
 140
 141                             // group's and bref's ancestor nodes below the lowest common ancestor are sibling alternatives => they're disjunctive.
 142                             messageId = "disjunctive";
 143                         } else if (groupCut.some(isNegativeLookaround)) {
 144
 145                             // group is in a negative lookaround which isn't bref's ancestor => group has already failed when bref starts to match.
 146                             messageId = "intoNegativeLookaround";
 147                         }
 148                     }
 149
 150                     if (messageId) {
 151                         context.report({
 152                             node,
 153                             messageId,
 154                             data: {
 155                                 bref: bref.raw,
 156                                 group: group.raw
 157                             }
 158                         });
 159                     }
 160                 }
 161             });
 162         }
 163
 164         return {
 165             "Literal[regex]"(node) {
 166                 const { pattern, flags } = node.regex;
 167
 168                 checkRegex(node, pattern, flags);
 169             },
 170             Program() {
 171                 const scope = context.getScope(),
 172                     tracker = new ReferenceTracker(scope),
 173                     traceMap = {
 174                         RegExp: {
 175                             [CALL]: true,
 176                             [CONSTRUCT]: true
 177                         }
 178                     };
 179
 180                 for (const { node } of tracker.iterateGlobalReferences(traceMap)) {
 181                     const [patternNode, flagsNode] = node.arguments,
 182                         pattern = getStringIfConstant(patternNode, scope),
 183                         flags = getStringIfConstant(flagsNode, scope);
 184
 185                     if (typeof pattern === "string") {
 186                         checkRegex(node, pattern, flags || "");
 187                     }
 188                 }
 189             }
 190         };
 191     }
 192 };