|
| 1 | +import type { Expression } from "estree" |
| 2 | +import type { RegExpVisitor } from "regexpp/visitor" |
| 3 | +import type { |
| 4 | + CharacterClassElement, |
| 5 | + UnicodePropertyCharacterSet, |
| 6 | +} from "regexpp/ast" |
| 7 | +import { |
| 8 | + CP_DIGIT_ZERO, |
| 9 | + CP_SPACE, |
| 10 | + createRule, |
| 11 | + defineRegexpVisitor, |
| 12 | + getRegexpLocation, |
| 13 | + getRegexpRange, |
| 14 | +} from "../utils" |
| 15 | + |
| 16 | +type CharacterClassElementKind = "\\w" | "\\d" | "\\s" | "\\p" | "*" |
| 17 | +const DEFAULT_ORDER: CharacterClassElementKind[] = [ |
| 18 | + "\\s", |
| 19 | + "\\w", |
| 20 | + "\\d", |
| 21 | + "\\p", |
| 22 | + "*", |
| 23 | +] |
| 24 | + |
| 25 | +/** |
| 26 | + * Get kind of CharacterClassElement for given CharacterClassElement |
| 27 | + */ |
| 28 | +function getCharacterClassElementKind( |
| 29 | + node: CharacterClassElement, |
| 30 | +): CharacterClassElementKind { |
| 31 | + if (node.type === "CharacterSet") { |
| 32 | + return node.kind === "word" |
| 33 | + ? "\\w" |
| 34 | + : node.kind === "digit" |
| 35 | + ? "\\d" |
| 36 | + : node.kind === "space" |
| 37 | + ? "\\s" |
| 38 | + : "\\p" |
| 39 | + } |
| 40 | + return "*" |
| 41 | +} |
| 42 | + |
| 43 | +export default createRule("order-in-character-class", { |
| 44 | + meta: { |
| 45 | + docs: { |
| 46 | + description: "enforces elements order in character class", |
| 47 | + recommended: false, |
| 48 | + }, |
| 49 | + fixable: "code", |
| 50 | + schema: [ |
| 51 | + { |
| 52 | + type: "object", |
| 53 | + properties: { |
| 54 | + order: { |
| 55 | + type: "array", |
| 56 | + items: { enum: ["\\w", "\\d", "\\s", "\\p", "*"] }, |
| 57 | + }, |
| 58 | + }, |
| 59 | + additionalProperties: false, |
| 60 | + }, |
| 61 | + ], |
| 62 | + messages: { |
| 63 | + sortElements: |
| 64 | + "Expected character class elements to be in ascending order. '{{next}}' should be before '{{prev}}'.", |
| 65 | + }, |
| 66 | + type: "layout", |
| 67 | + }, |
| 68 | + create(context) { |
| 69 | + const sourceCode = context.getSourceCode() |
| 70 | + const orderOption: { |
| 71 | + "*": number |
| 72 | + "\\w"?: number |
| 73 | + "\\d"?: number |
| 74 | + "\\s"?: number |
| 75 | + "\\p"?: number |
| 76 | + } = { "*": Infinity } |
| 77 | + |
| 78 | + ;((context.options[0]?.order ?? |
| 79 | + DEFAULT_ORDER) as CharacterClassElementKind[]).forEach((o, i) => { |
| 80 | + orderOption[o] = i + 1 |
| 81 | + }) |
| 82 | + |
| 83 | + /** |
| 84 | + * Create visitor |
| 85 | + * @param node |
| 86 | + */ |
| 87 | + function createVisitor(node: Expression): RegExpVisitor.Handlers { |
| 88 | + return { |
| 89 | + onCharacterClassEnter(ccNode) { |
| 90 | + const prevList: CharacterClassElement[] = [] |
| 91 | + for (const next of ccNode.elements) { |
| 92 | + if (prevList.length) { |
| 93 | + const prev = prevList[0] |
| 94 | + if (!isValidOrder(prev, next)) { |
| 95 | + let moveTarget = prev |
| 96 | + for (const p of prevList) { |
| 97 | + if (isValidOrder(p, next)) { |
| 98 | + break |
| 99 | + } else { |
| 100 | + moveTarget = p |
| 101 | + } |
| 102 | + } |
| 103 | + context.report({ |
| 104 | + node, |
| 105 | + loc: getRegexpLocation( |
| 106 | + sourceCode, |
| 107 | + node, |
| 108 | + next, |
| 109 | + ), |
| 110 | + messageId: "sortElements", |
| 111 | + data: { |
| 112 | + next: next.raw, |
| 113 | + prev: moveTarget.raw, |
| 114 | + }, |
| 115 | + *fix(fixer) { |
| 116 | + const nextRange = getRegexpRange( |
| 117 | + sourceCode, |
| 118 | + node, |
| 119 | + next, |
| 120 | + ) |
| 121 | + const targetRange = getRegexpRange( |
| 122 | + sourceCode, |
| 123 | + node, |
| 124 | + moveTarget, |
| 125 | + ) |
| 126 | + if (!targetRange || !nextRange) { |
| 127 | + return |
| 128 | + } |
| 129 | + yield fixer.insertTextBeforeRange( |
| 130 | + targetRange, |
| 131 | + next.raw, |
| 132 | + ) |
| 133 | + |
| 134 | + yield fixer.removeRange(nextRange) |
| 135 | + }, |
| 136 | + }) |
| 137 | + } |
| 138 | + } |
| 139 | + prevList.unshift(next) |
| 140 | + } |
| 141 | + }, |
| 142 | + } |
| 143 | + } |
| 144 | + |
| 145 | + /* eslint-disable complexity -- X( */ |
| 146 | + /** |
| 147 | + * Check that the two given CharacterClassElements are in a valid order. |
| 148 | + */ |
| 149 | + function isValidOrder( |
| 150 | + /* eslint-enable complexity -- X( */ |
| 151 | + prev: CharacterClassElement, |
| 152 | + next: CharacterClassElement, |
| 153 | + ) { |
| 154 | + const prevKind = getCharacterClassElementKind(prev) |
| 155 | + const nextKind = getCharacterClassElementKind(next) |
| 156 | + const prevOrder = orderOption[prevKind] ?? orderOption["*"] |
| 157 | + const nextOrder = orderOption[nextKind] ?? orderOption["*"] |
| 158 | + if (prevOrder < nextOrder) { |
| 159 | + return true |
| 160 | + } else if (prevOrder > nextOrder) { |
| 161 | + return false |
| 162 | + } |
| 163 | + if (prev.type === "CharacterSet" && prev.kind === "property") { |
| 164 | + if (next.type === "CharacterSet") { |
| 165 | + if (next.kind === "property") { |
| 166 | + return isValidOrderForUnicodePropertyCharacterSet( |
| 167 | + prev, |
| 168 | + next, |
| 169 | + ) |
| 170 | + } |
| 171 | + // e.g. /[\p{ASCII}\d]/ |
| 172 | + return false |
| 173 | + } |
| 174 | + // e.g. /[\p{ASCII}a]/ |
| 175 | + return true |
| 176 | + } else if ( |
| 177 | + next.type === "CharacterSet" && |
| 178 | + next.kind === "property" |
| 179 | + ) { |
| 180 | + if (prev.type === "CharacterSet") { |
| 181 | + // e.g. /[\d\p{ASCII}]/ |
| 182 | + return true |
| 183 | + } |
| 184 | + // e.g. /[a\p{ASCII}]/ |
| 185 | + return false |
| 186 | + } |
| 187 | + if (prev.type === "CharacterSet" && next.type === "CharacterSet") { |
| 188 | + if (prev.kind === "word" && next.kind === "digit") { |
| 189 | + return true |
| 190 | + } |
| 191 | + if (prev.kind === "digit" && next.kind === "word") { |
| 192 | + return false |
| 193 | + } |
| 194 | + } |
| 195 | + const prevCP = getTargetCodePoint(prev) |
| 196 | + const nextCP = getTargetCodePoint(next) |
| 197 | + if (prevCP <= nextCP) { |
| 198 | + return true |
| 199 | + } |
| 200 | + return false |
| 201 | + } |
| 202 | + |
| 203 | + /** |
| 204 | + * Check that the two given UnicodePropertyCharacterSet are in a valid order. |
| 205 | + */ |
| 206 | + function isValidOrderForUnicodePropertyCharacterSet( |
| 207 | + prev: UnicodePropertyCharacterSet, |
| 208 | + next: UnicodePropertyCharacterSet, |
| 209 | + ) { |
| 210 | + if (prev.key < next.key) { |
| 211 | + return true |
| 212 | + } else if (prev.key > next.key) { |
| 213 | + return false |
| 214 | + } |
| 215 | + if (prev.value) { |
| 216 | + if (next.value) { |
| 217 | + if (prev.value <= next.value) { |
| 218 | + return true |
| 219 | + } |
| 220 | + return false |
| 221 | + } |
| 222 | + return false |
| 223 | + } |
| 224 | + return true |
| 225 | + } |
| 226 | + |
| 227 | + /** |
| 228 | + * Gets the target code point for a given element. |
| 229 | + */ |
| 230 | + function getTargetCodePoint( |
| 231 | + node: Exclude<CharacterClassElement, UnicodePropertyCharacterSet>, |
| 232 | + ) { |
| 233 | + if (node.type === "CharacterSet") { |
| 234 | + if (node.kind === "digit" || node.kind === "word") { |
| 235 | + return CP_DIGIT_ZERO |
| 236 | + } |
| 237 | + if (node.kind === "space") { |
| 238 | + return CP_SPACE |
| 239 | + } |
| 240 | + return Infinity |
| 241 | + } |
| 242 | + if (node.type === "CharacterClassRange") { |
| 243 | + return node.min.value |
| 244 | + } |
| 245 | + return node.value |
| 246 | + } |
| 247 | + |
| 248 | + return defineRegexpVisitor(context, { |
| 249 | + createVisitor, |
| 250 | + }) |
| 251 | + }, |
| 252 | +}) |
0 commit comments