stylelint-scss
Version:
A collection of SCSS specific rules for stylelint
185 lines (174 loc) • 6.79 kB
JavaScript
/**
* Finds comments, both CSS comments and double slash ones, in a CSS string
* This helper exists because PostCSS drops some inline comments (those
* between seelctors, property values, etc.)
* https://github.com/postcss/postcss/issues/845#issuecomment-232306259
*
* @param [string] rawString -- the source raw CSS string
* @return [array] array of objects with these props:
* � type -- "css" or "double-slash"
* � source: { start, end }
* IMPORTANT: the function itself considers \r as a character, and counts
* it for `start` and `end`. But if their values are passed to PostCSS's
* result.warn(), than "\r\n" is consideren ONE CHAR (newline)!
* � raws
* raws.startToken -- `/*`, `/**`, `/**!`, etc.
* raws.left -- whitespace after the comment opening marker
* raws.text -- the full comment, including markers (//, /*)
* raws.right -- whitespace before the comment closing marker
* raws.endToken -- `*\/`, `**\/` for CSS comments
* � text -- the comment text only, excluding //, /*, trailing whitespaces
* � inlineAfter -- true, if there is something before the comment on
* the same line
* � inlineBefore -- true, if there is something after the comment on
* the same line
*/
export default function findCommentsInRaws(rawString) {
const result = []
let comment = {}
// Keeps track of which structure the parser is inside (string, comment,
// url function, parens). E.g., /* comment */ inside a string doesn't
// constitute a comment, so as url(//path)
const modesEntered = [{
mode: "normal",
character: null,
}]
let commentStart = null
// postcss-scss transforms //-comments into CSS comments, like so:
// `// comment` -> `/* comment*/`. So to have a correct intex we need to
// keep track on the added `*/` sequences
let offset = 0
for (let i = 0; i < rawString.length; i++) {
const character = rawString[i]
const prevChar = i > 0 ? rawString[i - 1] : null
const nextChar = i + 1 < rawString.length ? rawString[i + 1] : null
const lastModeIndex = modesEntered.length - 1
const mode = modesEntered[lastModeIndex].mode
switch (character) {
// If entering/exiting a string
case "\"":
case "'": {
if (mode === "comment") { break }
if (mode === "string" &&
modesEntered[lastModeIndex].character === character &&
prevChar !== "\\"
) {
// Exiting a string
modesEntered.pop()
} else {
// Entering a string
modesEntered.push({
mode: "string",
character,
})
}
break
}
// Entering url, other function or parens (only url matters)
case "(": {
if (mode === "comment" || mode === "string") { break }
const functionName =
/(?:^|(?:\n)|(?:\r)|(?:\s-)|[:\s,.(){}*+/%])([a-zA-Z0-9_-]*)$/
.exec(rawString.substring(0, i))[1]
modesEntered.push({
mode: functionName === "url" ? "url" : "parens",
character: "(",
})
break
}
// Exiting url, other function or parens
case ")": {
if (mode === "comment" || mode === "string") { break }
modesEntered.pop()
break
}
// checking for comment
case "/": {
// Break if the / is inside a comment because we leap over the second
// slash in // and in */, so the / is not from a marker. Also break
// if inside a string
if (mode === "comment" || mode === "string") { break }
if (nextChar === "*") {
modesEntered.push({
mode: "comment",
character: "/*",
})
comment = {
type: "css",
source: { start: i + offset },
inlineAfter: rawString.substring(0, i).search(/\n\s*$/) === -1,
}
commentStart = i
// Skip the next loop as the * is already checked
i++
} else if (nextChar === "/") {
// `url(//path/to/file)` has no comment
if (mode === "url") { break }
modesEntered.push({
mode: "comment",
character: "//",
})
comment = {
type: "double-slash",
source: { start: i + offset },
inlineAfter: rawString.substring(0, i).search(/\n\s*$/) === -1,
}
commentStart = i
// Skip the next loop as the second slash in // is already checked
i++
}
break
}
// Might be a closing */
case "*": {
if (mode === "comment" && modesEntered[lastModeIndex].character === "/*" && nextChar === "/") {
comment.source.end = i + 1 + offset
const commentRaw = rawString.substring(commentStart, i + 2)
const matches = /^(\/\*+[!#]{0,1})(\s*)([\s\S]*?)(\s*?)(\*+\/)$/.exec(commentRaw)
modesEntered.pop()
comment.raws = {
startToken: matches[1],
left: matches[2],
text: commentRaw,
right: matches[4],
endToken: matches[5],
}
comment.text = matches[3]
comment.inlineBefore = rawString.substring(i + 2).search(/^\s*?\S+\s*?\n/) !== -1
result.push(Object.assign({}, comment))
comment = {}
// Skip the next loop as the / in */ is already checked
i++
}
break
}
default: {
const isNewline = character === "\r" && rawString[i + 1] === "\n" ||
character === "\n" && rawString[i - 1] !== "\r"
// //-comments end before newline and if the code string ends
if (isNewline || i === rawString.length - 1) {
if (mode === "comment" && modesEntered[lastModeIndex].character === "//") {
comment.source.end = (isNewline ? i - 1 : i) + offset
const commentRaw = rawString.substring(commentStart, isNewline ? i : i + 1)
const matches = /^(\/+)(\s*)(.*?)(\s*?)$/.exec(commentRaw)
modesEntered.pop()
comment.raws = {
startToken: matches[1],
left: matches[2],
text: commentRaw,
right: matches[4],
}
comment.text = matches[3]
comment.inlineBefore = false
result.push(Object.assign({}, comment))
comment = {}
// Compensate for the `*/` added by postcss-scss
offset += 2
}
}
break
}
}
}
return result
}