File Explorer

/var/lang/lib/node_modules/npm/node_modules/postcss-selector-parser/dist

This explorer reads the filesystem of the server it runs on, so /workspace/user isn't present here. Browsing and the terminal still work against this server's own disk from /.

tokenize.js9.5 KB · 301 lines
"use strict";var __createBinding = (this && this.__createBinding) || (Object.create ? (function(o, m, k, k2) {    if (k2 === undefined) k2 = k;    var desc = Object.getOwnPropertyDescriptor(m, k);    if (!desc || ("get" in desc ? !m.__esModule : desc.writable || desc.configurable)) {      desc = { enumerable: true, get: function() { return m[k]; } };    }    Object.defineProperty(o, k2, desc);}) : (function(o, m, k, k2) {    if (k2 === undefined) k2 = k;    o[k2] = m[k];}));var __setModuleDefault = (this && this.__setModuleDefault) || (Object.create ? (function(o, v) {    Object.defineProperty(o, "default", { enumerable: true, value: v });}) : function(o, v) {    o["default"] = v;});var __importStar = (this && this.__importStar) || (function () {    var ownKeys = function(o) {        ownKeys = Object.getOwnPropertyNames || function (o) {            var ar = [];            for (var k in o) if (Object.prototype.hasOwnProperty.call(o, k)) ar[ar.length] = k;            return ar;        };        return ownKeys(o);    };    return function (mod) {        if (mod && mod.__esModule) return mod;        var result = {};        if (mod != null) for (var k = ownKeys(mod), i = 0; i < k.length; i++) if (k[i] !== "default") __createBinding(result, mod, k[i]);        __setModuleDefault(result, mod);        return result;    };})();var _a, _b;Object.defineProperty(exports, "__esModule", { value: true });exports.FIELDS = void 0;exports.default = tokenize;var t = __importStar(require("./tokenTypes"));var unescapable = (_a = {},    _a[t.tab] = true,    _a[t.newline] = true,    _a[t.cr] = true,    _a[t.feed] = true,    _a);var wordDelimiters = (_b = {},    _b[t.space] = true,    _b[t.tab] = true,    _b[t.newline] = true,    _b[t.cr] = true,    _b[t.feed] = true,    _b[t.ampersand] = true,    _b[t.asterisk] = true,    _b[t.bang] = true,    _b[t.comma] = true,    _b[t.colon] = true,    _b[t.semicolon] = true,    _b[t.openParenthesis] = true,    _b[t.closeParenthesis] = true,    _b[t.openSquare] = true,    _b[t.closeSquare] = true,    _b[t.singleQuote] = true,    _b[t.doubleQuote] = true,    _b[t.plus] = true,    _b[t.pipe] = true,    _b[t.tilde] = true,    _b[t.greaterThan] = true,    _b[t.equals] = true,    _b[t.dollar] = true,    _b[t.caret] = true,    _b[t.slash] = true,    _b);var hex = {};var hexChars = "0123456789abcdefABCDEF";for (var i = 0; i < hexChars.length; i++) {    hex[hexChars.charCodeAt(i)] = true;}/** *  Returns the last index of the bar css word * @param {string} css The string in which the word begins * @param {number} start The index into the string where word's first letter occurs */function consumeWord(css, start) {    var next = start;    var code;    do {        code = css.charCodeAt(next);        if (wordDelimiters[code]) {            return next - 1;        }        else if (code === t.backslash) {            next = consumeEscape(css, next) + 1;        }        else {            // All other characters are part of the word            next++;        }    } while (next < css.length);    return next - 1;}/** *  Returns the last index of the escape sequence * @param {string} css The string in which the sequence begins * @param {number} start The index into the string where escape character (`\`) occurs. */function consumeEscape(css, start) {    var next = start;    var code = css.charCodeAt(next + 1);    if (unescapable[code]) {        // just consume the escape char    }    else if (hex[code]) {        var hexDigits = 0;        // consume up to 6 hex chars        do {            next++;            hexDigits++;            code = css.charCodeAt(next + 1);        } while (hex[code] && hexDigits < 6);        // if fewer than 6 hex chars, a trailing space ends the escape        if (hexDigits < 6 && code === t.space) {            next++;        }    }    else {        // the next char is part of the current word        next++;    }    return next;}exports.FIELDS = {    TYPE: 0,    START_LINE: 1,    START_COL: 2,    END_LINE: 3,    END_COL: 4,    START_POS: 5,    END_POS: 6,};function tokenize(input) {    var tokens = [];    var css = input.css.valueOf();    var length = css.length;    var offset = -1;    var line = 1;    var start = 0;    var end = 0;    var code, content, endColumn, endLine, escaped, escapePos, last, lines, next, nextLine, nextOffset, quote, tokenType;    function unclosed(what, fix) {        if (input.safe) {            // fyi: this is never set to true.            css += fix;            next = css.length - 1;        }        else {            throw input.error("Unclosed " + what, line, start - offset, start);        }    }    while (start < length) {        code = css.charCodeAt(start);        if (code === t.newline) {            offset = start;            line += 1;        }        switch (code) {            case t.space:            case t.tab:            case t.newline:            case t.cr:            case t.feed:                next = start;                do {                    next += 1;                    code = css.charCodeAt(next);                    if (code === t.newline) {                        offset = next;                        line += 1;                    }                } while (code === t.space ||                    code === t.newline ||                    code === t.tab ||                    code === t.cr ||                    code === t.feed);                tokenType = t.space;                endLine = line;                endColumn = next - offset - 1;                end = next;                break;            case t.plus:            case t.greaterThan:            case t.tilde:            case t.pipe:                next = start;                do {                    next += 1;                    code = css.charCodeAt(next);                } while (code === t.plus || code === t.greaterThan || code === t.tilde || code === t.pipe);                tokenType = t.combinator;                endLine = line;                endColumn = start - offset;                end = next;                break;            // Consume these characters as single tokens.            case t.asterisk:            case t.ampersand:            case t.bang:            case t.comma:            case t.equals:            case t.dollar:            case t.caret:            case t.openSquare:            case t.closeSquare:            case t.colon:            case t.semicolon:            case t.openParenthesis:            case t.closeParenthesis:                next = start;                tokenType = code;                endLine = line;                endColumn = start - offset;                end = next + 1;                break;            case t.singleQuote:            case t.doubleQuote:                quote = code === t.singleQuote ? "'" : '"';                next = start;                do {                    escaped = false;                    next = css.indexOf(quote, next + 1);                    if (next === -1) {                        unclosed("quote", quote);                    }                    escapePos = next;                    while (css.charCodeAt(escapePos - 1) === t.backslash) {                        escapePos -= 1;                        escaped = !escaped;                    }                } while (escaped);                tokenType = t.str;                endLine = line;                endColumn = start - offset;                end = next + 1;                break;            default:                if (code === t.slash && css.charCodeAt(start + 1) === t.asterisk) {                    next = css.indexOf("*/", start + 2) + 1;                    if (next === 0) {                        unclosed("comment", "*/");                    }                    content = css.slice(start, next + 1);                    lines = content.split("\n");                    last = lines.length - 1;                    if (last > 0) {                        nextLine = line + last;                        nextOffset = next - lines[last].length;                    }                    else {                        nextLine = line;                        nextOffset = offset;                    }                    tokenType = t.comment;                    line = nextLine;                    endLine = nextLine;                    endColumn = next - nextOffset;                }                else if (code === t.slash) {                    next = start;                    tokenType = code;                    endLine = line;                    endColumn = start - offset;                    end = next + 1;                }                else {                    next = consumeWord(css, start);                    tokenType = t.word;                    endLine = line;                    endColumn = next - offset;                }                end = next + 1;                break;        }        // Ensure that the token structure remains consistent        tokens.push([            tokenType, // [0] Token type            line, // [1] Starting line            start - offset, // [2] Starting column            endLine, // [3] Ending line            endColumn, // [4] Ending column            start, // [5] Start position / Source index            end, // [6] End position        ]);        // Reset offset for the next token        if (nextOffset) {            offset = nextOffset;            nextOffset = null;        }        start = end;    }    return tokens;}//# sourceMappingURL=tokenize.js.map