myworkspace/LinkSyncExtension/utils/query-engine.js

// LinkSync Query Engine
// Client-side query parser and executor for building queries

const QueryEngine = {
    tokenize(expression) {
        const tokens = [];
        let pos = 0;
        const len = expression.length;

        while (pos < len) {
            const ch = expression[pos];

            if (ch === " " || ch === "\t") {
                pos++;
                continue;
            }

            if (ch === "(") {
                tokens.push({ type: "LPAREN", value: "(" });
                pos++;
                continue;
            }

            if (ch === ")") {
                tokens.push({ type: "RPAREN", value: ")" });
                pos++;
                continue;
            }

            if (ch === ",") {
                tokens.push({ type: "COMMA", value: "," });
                pos++;
                continue;
            }

            if (expression.substring(pos, pos + 3) === "AND") {
                tokens.push({ type: "OPERATOR", value: "AND" });
                pos += 3;
                continue;
            }

            if (expression.substring(pos, pos + 2) === "OR") {
                tokens.push({ type: "OPERATOR", value: "OR" });
                pos += 2;
                continue;
            }

            if (expression.substring(pos, pos + 3) === "XOR") {
                tokens.push({ type: "OPERATOR", value: "XOR" });
                pos += 3;
                continue;
            }

            if (ch === "'" || ch === '"') {
                const quote = ch;
                pos++;
                let value = "";
                while (pos < len && expression[pos] !== quote) {
                    value += expression[pos];
                    pos++;
                }
                pos++;
                tokens.push({ type: "TERM", value });
                continue;
            }

            if (/[a-zA-Z0-9_.\-]/.test(ch)) {
                let value = "";
                while (pos < len && /[a-zA-Z0-9_.\-:\/?&=%]/.test(expression[pos])) {
                    value += expression[pos];
                    pos++;
                }

                if (value.includes(":")) {
                    const [field, ...rest] = value.split(":");
                    const fieldValue = rest.join(":");
                    const knownFields = ["url", "tag", "title", "description", "path", "id"];
                    if (knownFields.includes(field.toLowerCase())) {
                        tokens.push({ type: "FIELD", value: field.toUpperCase() });
                        tokens.push({ type: "TERM", value: fieldValue });
                        continue;
                    }
                }

                tokens.push({ type: "TERM", value });
                continue;
            }

            pos++;
        }

        return tokens;
    },

    parse(expression) {
        if (!expression || !expression.trim()) return null;

        const tokens = this.tokenize(expression);
        if (tokens.length === 0) return null;

        const parser = new QueryParserState(tokens);
        return parser.parseOr();
    },

    buildQueryString(ast) {
        if (!ast) return "";

        switch (ast.type) {
            case "TERM":
                return ast.value;
            case "TERM_SET":
                return `(${ast.values.join(", ")})`;
            case "FIELD":
                return `${ast.field.toLowerCase()}:${ast.value}`;
            case "AND":
            case "OR":
            case "XOR":
                return `(${this.buildQueryString(ast.left)} ${ast.type} ${this.buildQueryString(ast.right)})`;
            default:
                return "";
        }
    },

    validate(expression) {
        try {
            const ast = this.parse(expression);
            return { valid: true, ast };
        } catch (e) {
            return { valid: false, error: e.message };
        }
    },
};

class QueryParserState {
    constructor(tokens) {
        this.tokens = tokens;
        this.pos = 0;
    }

    current() {
        return this.pos < this.tokens.length ? this.tokens[this.pos] : null;
    }

    advance() {
        const token = this.current();
        this.pos++;
        return token;
    }

    expect(type) {
        const token = this.current();
        if (!token) throw new Error(`Expected ${type}, got end of input`);
        if (token.type !== type) throw new Error(`Expected ${type}, got ${token.type}`);
        return this.advance();
    }

    parseOr() {
        let left = this.parseAnd();
        while (this.current() && this.current().type === "OPERATOR" && this.current().value === "OR") {
            this.advance();
            const right = this.parseAnd();
            left = { type: "OR", left, right };
        }
        return left;
    }

    parseAnd() {
        let left = this.parseXor();
        while (this.current() && this.current().type === "OPERATOR" && this.current().value === "AND") {
            this.advance();
            const right = this.parseXor();
            left = { type: "AND", left, right };
        }
        return left;
    }

    parseXor() {
        let left = this.parsePrimary();
        while (this.current() && this.current().type === "OPERATOR" && this.current().value === "XOR") {
            this.advance();
            const right = this.parsePrimary();
            left = { type: "XOR", left, right };
        }
        return left;
    }

    parsePrimary() {
        const token = this.current();
        if (!token) throw new Error("Unexpected end of input");

        if (token.type === "LPAREN") {
            this.advance();
            const node = this.parseOr();
            this.expect("RPAREN");
            return node;
        }

        if (token.type === "FIELD") {
            const field = this.advance().value;
            const valueToken = this.current();
            if (valueToken && valueToken.type === "TERM") {
                this.advance();
                return { type: "FIELD", field, value: valueToken.value };
            }
            return { type: "FIELD", field, value: "" };
        }

        if (token.type === "TERM") {
            return this.parseTerm();
        }

        throw new Error(`Unexpected token: ${token.value}`);
    }

    parseTerm() {
        const token = this.advance();
        const next = this.current();

        if (next && next.type === "COMMA") {
            const values = [token.value];
            while (this.current() && this.current().type === "COMMA") {
                this.advance();
                const termToken = this.current();
                if (termToken && termToken.type === "TERM") {
                    values.push(this.advance().value);
                }
            }
            return { type: "TERM_SET", values };
        }

        return { type: "TERM", value: token.value };
    }
}