tokenizer.js

"use strict";
const reportError = require("./lib/report_error").reportError;

// tokenizing lines longer than this makes editor very slow
var MAX_TOKEN_COUNT = 2000;
/**
 * This class takes a set of highlighting rules, and creates a tokenizer out of them. For more information, see [the wiki on extending highlighters](https://github.com/ajaxorg/ace/wiki/Creating-or-Extending-an-Edit-Mode#wiki-extendingTheHighlighter).
 **/
class Tokenizer {
    /**
     * Constructs a new tokenizer based on the given rules and flags.
     * @param {Object} rules The highlighting rules
     **/
    constructor(rules) {
        /**@type {RegExp}*/
        this.splitRegex;
        this.states = rules;

        this.regExps = {};
        this.matchMappings = {};
        for (var key in this.states) {
            var state = this.states[key];
            var ruleRegExps = [];
            var matchTotal = 0;
            var mapping = this.matchMappings[key] = {defaultToken: "text"};
            var flag = "g";

            var splitterRurles = [];
            for (var i = 0; i < state.length; i++) {
                var rule = state[i];
                if (rule.defaultToken)
                    mapping.defaultToken = rule.defaultToken;
                if (rule.caseInsensitive && flag.indexOf("i") === -1)
                    flag += "i";
                if (rule.unicode && flag.indexOf("u") === -1)
                    flag += "u";
                if (rule.regex == null)
                    continue;

                if (rule.regex instanceof RegExp)
                    rule.regex = rule.regex.toString().slice(1, -1);

                // Count number of matching groups. 2 extra groups from the full match
                // And the catch-all on the end (used to force a match);
                var adjustedregex = rule.regex;
                var matchcount = new RegExp("(?:(" + adjustedregex + ")|(.))").exec("a").length - 2;
                if (Array.isArray(rule.token)) {
                    if (rule.token.length == 1 || matchcount == 1) {
                        rule.token = rule.token[0];
                    } else if (matchcount - 1 != rule.token.length) {
                        this.reportError("number of classes and regexp groups doesn't match", {
                            rule: rule,
                            groupCount: matchcount - 1
                        });
                        rule.token = rule.token[0];
                    } else {
                        rule.tokenArray = rule.token;
                        rule.token = null;
                        rule.onMatch = this.$arrayTokens;
                    }
                } else if (typeof rule.token == "function" && !rule.onMatch) {
                    if (matchcount > 1)
                        rule.onMatch = this.$applyToken;
                    else
                        rule.onMatch = rule.token;
                }

                if (matchcount > 1) {
                    if (/\\\d/.test(rule.regex)) {
                        // Replace any backreferences and offset appropriately.
                        adjustedregex = rule.regex.replace(/\\([0-9]+)/g, function(match, digit) {
                            return "\\" + (parseInt(digit, 10) + matchTotal + 1);
                        });
                    } else {
                        matchcount = 1;
                        adjustedregex = this.removeCapturingGroups(rule.regex);
                    }
                    if (!rule.splitRegex && typeof rule.token != "string")
                        splitterRurles.push(rule); // flag will be known only at the very end
                }

                mapping[matchTotal] = i;
                matchTotal += matchcount;

                ruleRegExps.push(adjustedregex);

                // makes property access faster
                if (!rule.onMatch)
                    rule.onMatch = null;
            }

            if (!ruleRegExps.length) {
                mapping[0] = 0;
                ruleRegExps.push("$");
            }

            splitterRurles.forEach(function(rule) {
                rule.splitRegex = this.createSplitterRegexp(rule.regex, flag);
            }, this);

            this.regExps[key] = new RegExp("(" + ruleRegExps.join(")|(") + ")|($)", flag);
        }
    }

    /**
     * @param {number} m
     */
    $setMaxTokenCount(m) {
        MAX_TOKEN_COUNT = m | 0;
    }

    /**
     * @param {string} str
     * @return {import("../ace-internal").Ace.Token[]}
     */
    $applyToken(str) {
        var values = this.splitRegex.exec(str).slice(1);
        //@ts-ignore
        var types = this.token.apply(this, values);

        // required for compatibility with old modes
        if (typeof types === "string")
            return [{type: types, value: str}];

        var tokens = [];
        for (var i = 0, l = types.length; i < l; i++) {
            if (values[i])
                tokens[tokens.length] = {
                    type: types[i],
                    value: values[i]
                };
        }
        return tokens;
    }

    /**
     * @param {string} str
     * @return {import("../ace-internal").Ace.Token[] | string}
     */
    $arrayTokens(str) {
        if (!str)
            return [];
        var values = this.splitRegex.exec(str);
        if (!values)
            return "text";
        var tokens = [];
        //@ts-ignore
        var types = this.tokenArray;
        for (var i = 0, l = types.length; i < l; i++) {
            if (values[i + 1])
                tokens[tokens.length] = {
                    type: types[i],
                    value: values[i + 1]
                };
        }
        return tokens;
    }

    /**
     * @param {string} src
     * @returns {string}
     */
    removeCapturingGroups(src) {
        var r = src.replace(
            /\\.|\[(?:\\.|[^\\\]])*|\(\?[:=!<]|(\()/g,
            function(x, y) {return y ? "(?:" : x;}
        );
        return r;
    }

    /**
     * @param {string} src
     * @param {string} flag
     */
    createSplitterRegexp(src, flag) {
        if (src.indexOf("(?=") != -1) {
            var stack = 0;
            var inChClass = false;
            var lastCapture = {};
            src.replace(/(\\.)|(\((?:\?[=!])?)|(\))|([\[\]])/g, function(
                m, esc, parenOpen, parenClose, square, index
            ) {
                if (inChClass) {
                    inChClass = square != "]";
                } else if (square) {
                    inChClass = true;
                } else if (parenClose) {
                    if (stack == lastCapture.stack) {
                        lastCapture.end = index+1;
                        lastCapture.stack = -1;
                    }
                    stack--;
                } else if (parenOpen) {
                    stack++;
                    if (parenOpen.length != 1) {
                        lastCapture.stack = stack;
                        lastCapture.start = index;
                    }
                }
                return m;
            });

            if (lastCapture.end != null && /^\)*$/.test(src.substr(lastCapture.end)))
                src = src.substring(0, lastCapture.start) + src.substr(lastCapture.end);
        }
        
        // this is needed for regexps that can match in multiple ways
        if (src.charAt(0) != "^") src = "^" + src;
        if (src.charAt(src.length - 1) != "$") src += "$";
        
        return new RegExp(src, (flag||"").replace("g", ""));
    }

    /**
     * Returns an object containing two properties: `tokens`, which contains all the tokens; and `state`, the current state.
     * @param {string} line
     * @param {string | string[]} startState
     * @returns {{tokens:import("../ace-internal").Ace.Token[], state: string|string[]}}
     */
    getLineTokens(line, startState) {
        if (startState && typeof startState != "string") {
            /**@type {any[]}*/
            var stack = startState.slice(0);
            startState = stack[0];
            if (startState === "#tmp") {
                stack.shift();
                startState = stack.shift();
            }
        } else
            var stack = [];

        var currentState = /**@type{string}*/(startState) || "start";
        var state = this.states[currentState];
        if (!state) {
            currentState = "start";
            state = this.states[currentState];
        }
        var mapping = this.matchMappings[currentState];
        var re = this.regExps[currentState];
        re.lastIndex = 0;

        var match, tokens = [];
        var lastIndex = 0;
        var matchAttempts = 0;

        var token = {type: null, value: ""};

        while (match = re.exec(line)) {
            var type = mapping.defaultToken;
            var rule = null;
            var value = match[0];
            var index = re.lastIndex;

            if (index - value.length > lastIndex) {
                var skipped = line.substring(lastIndex, index - value.length);
                if (token.type == type) {
                    token.value += skipped;
                } else {
                    if (token.type)
                        tokens.push(token);
                    token = {type: type, value: skipped};
                }
            }

            for (var i = 0; i < match.length-2; i++) {
                if (match[i + 1] === undefined)
                    continue;

                rule = state[mapping[i]];

                if (rule.onMatch)
                    type = rule.onMatch(value, currentState, stack, line);
                else
                    type = rule.token;

                if (rule.next) {
                    if (typeof rule.next == "string") {
                        currentState = rule.next;
                    } else {
                        currentState = rule.next(currentState, stack);
                    }
                    
                    state = this.states[currentState];
                    if (!state) {
                        this.reportError("state doesn't exist", currentState);
                        currentState = "start";
                        state = this.states[currentState];
                    }
                    mapping = this.matchMappings[currentState];
                    lastIndex = index;
                    re = this.regExps[currentState];
                    re.lastIndex = index;
                }
                if (rule.consumeLineEnd)
                    lastIndex = index;
                break;
            }

            if (value) {
                if (typeof type === "string") {
                    if ((!rule || rule.merge !== false) && token.type === type) {
                        token.value += value;
                    } else {
                        if (token.type)
                            tokens.push(token);
                        token = {type: type, value: value};
                    }
                } else if (type) {
                    if (token.type)
                        tokens.push(token);
                    token = {type: null, value: ""};
                    for (var i = 0; i < type.length; i++)
                        tokens.push(type[i]);
                }
            }

            if (lastIndex == line.length)
                break;

            lastIndex = index;

            if (matchAttempts++ > MAX_TOKEN_COUNT) {
                if (matchAttempts > 2 * line.length) {
                    this.reportError("infinite loop with in ace tokenizer", {
                        startState: startState,
                        line: line
                    });
                }
                // chrome doens't show contents of text nodes with very long text
                while (lastIndex < line.length) {
                    if (token.type)
                        tokens.push(token);
                    token = {
                        value: line.substring(lastIndex, lastIndex += 500),
                        type: "overflow"
                    };
                }
                currentState = "start";
                stack = [];
                break;
            }
        }

        if (token.type)
            tokens.push(token);
        
        if (stack.length > 1) {
            if (stack[0] !== currentState)
                stack.unshift("#tmp", currentState);
        }
        return {
            tokens : tokens,
            state : stack.length ? stack : currentState
        };
    }
}

Tokenizer.prototype.reportError = reportError;
exports.Tokenizer = Tokenizer;