socraticode/dist/services/gdscript-syntax.js

SocratiCode — MCP server for local codebase indexing, semantic search, and code dependency graphs. All private, all local via Docker.

// SPDX-License-Identifier: AGPL-3.0-only
// Copyright (C) 2026 Giancarlo Erra - Altaire Limited
const IDENTIFIER_START = /^[_\p{ID_Start}]$/u;
const IDENTIFIER_CONTINUE = /^[_\p{ID_Continue}]$/u;
/** Return one Unicode code point without splitting surrogate pairs. */
function codePointAt(source, index) {
    const codePoint = source.codePointAt(index);
    return codePoint === undefined ? "" : String.fromCodePoint(codePoint);
}
/**
 * Tokenize only the GDScript constructs needed by import and class-name fallbacks.
 * Comments and whitespace are discarded, while quoted strings remain single tokens.
 */
export function tokenizeGdscript(source) {
    const tokens = [];
    let index = 0;
    while (index < source.length) {
        const current = codePointAt(source, index);
        if (/^\s$/u.test(current)) {
            index += current.length;
            continue;
        }
        if (current === "#") {
            while (index < source.length && source[index] !== "\n" && source[index] !== "\r") {
                index++;
            }
            continue;
        }
        const rawPrefix = (current === "r" || current === "R")
            && (source[index + 1] === "\"" || source[index + 1] === "'");
        const quoteIndex = rawPrefix ? index + 1 : index;
        const quote = source[quoteIndex];
        if (quote === "\"" || quote === "'") {
            const start = index;
            const triple = source.slice(quoteIndex, quoteIndex + 3) === quote.repeat(3);
            const delimiterLength = triple ? 3 : 1;
            index = quoteIndex + delimiterLength;
            while (index < source.length) {
                // Raw strings still recognize an escaped matching quote (and `\\\\`).
                // Skip the escaped code point before checking for the closing delimiter
                // so `r"quoted: \\" text"` remains one string token.
                if (source[index] === "\\") {
                    index += Math.min(2, source.length - index);
                    continue;
                }
                if (source.slice(index, index + delimiterLength) === quote.repeat(delimiterLength)) {
                    index += delimiterLength;
                    break;
                }
                if (!triple && (source[index] === "\n" || source[index] === "\r"))
                    break;
                index++;
            }
            tokens.push({ kind: "string", text: source.slice(start, index) });
            continue;
        }
        if (IDENTIFIER_START.test(current)) {
            const start = index;
            index += current.length;
            while (index < source.length) {
                const next = codePointAt(source, index);
                if (!IDENTIFIER_CONTINUE.test(next))
                    break;
                index += next.length;
            }
            tokens.push({ kind: "identifier", text: source.slice(start, index) });
            continue;
        }
        tokens.push({ kind: "punctuation", text: current });
        index += current.length;
    }
    return tokens;
}
/** Find a real class_name declaration while ignoring comments and string bodies. */
export function extractClassNameFromGdscript(source) {
    const tokens = tokenizeGdscript(source);
    for (let index = 0; index < tokens.length - 1; index++) {
        const token = tokens[index];
        if (token.kind !== "identifier" || token.text !== "class_name")
            continue;
        if (tokens[index - 1]?.text === ".")
            continue;
        const name = tokens[index + 1];
        if (name?.kind === "identifier")
            return name.text;
    }
    return null;
}
//# sourceMappingURL=gdscript-syntax.js.map