@xapp/arachne-utils/lib/ArachneURLPattern.js

"use strict";
/*! Copyright (c) 2020, XAPP AI */
Object.defineProperty(exports, "__esModule", { value: true });
exports.ArachneURLPattern = void 0;
/**
 * Parses PURL into Regex string.
 *
 * From apify-js SDK
 *
 * {@link https://github.com/apify/apify-js/blob/64a8c1036adc26d7b901187c9b24667e333096d1/src/pseudo_url.js#L10}
 * @ignore
 */
function parsePurl(purl) {
    const trimmedPurl = purl.trim();
    if (trimmedPurl.length === 0)
        throw new Error(`Cannot parse PURL '${trimmedPurl}': it must be a non-empty string`);
    let regex = "^";
    let openBrackets = 0;
    let insideNegation = false;
    let negationPattern = "";
    try {
        for (let i = 0; i < trimmedPurl.length; i++) {
            const ch = trimmedPurl.charAt(i);
            // console.log(`Inside loop char ${ch} insideNegation?${insideNegation}`);
            if (ch === "[" && ++openBrackets === 1) {
                regex += "("; // Start regex capture group
            }
            else if (ch === "]" && openBrackets > 0 && --openBrackets === 0) {
                if (insideNegation) {
                    regex += `(?!${negationPattern})`; // Correct placement of negative lookahead
                    insideNegation = false;
                    negationPattern = "";
                }
                regex += ")";
            }
            else if (openBrackets > 0) {
                if (ch === "!" && !insideNegation) {
                    insideNegation = true;
                }
                else if (insideNegation) {
                    negationPattern += ch;
                }
                else {
                    regex += ch;
                }
            }
            else {
                const code = ch.charCodeAt(0);
                if ((code >= 48 && code <= 57) || (code >= 65 && code <= 90) || (code >= 97 && code <= 122)) {
                    regex += ch;
                }
                else {
                    regex += `\\x${code.toString(16).padStart(2, "0")}`;
                }
            }
        }
        regex += "$";
    }
    catch (err) {
        throw new Error(`Cannot parse PURL '${purl}': ${err}`);
    }
    return regex;
}
/**
 * A URL pattern leveraging RegEx.
 *
 * {@link https://sdk.apify.com/docs/api/pseudo-url}
 */
class ArachneURLPattern {
    constructor(url) {
        if (url instanceof RegExp) {
            this.regExp = url;
        }
        else if (typeof url === "string") {
            this.regExp = new RegExp(parsePurl(url), "i");
            // console.log(this.regExp);
        }
        else {
            throw new Error("A URL is required to instantiate the URL pattern.");
        }
    }
    matches(url) {
        return typeof url === "string" && url.match(this.regExp) !== null;
    }
}
exports.ArachneURLPattern = ArachneURLPattern;
//# sourceMappingURL=ArachneURLPattern.js.map