PineScriptCooker

PineScriptCooker is a high-performance, modular backtesting and scripting engine written in modern C++. It features a custom stack-based virtual machine and supports multiple popular trading script languages by compiling them into a common bytecode format.

Features

• Multi-Language Frontend: Compiles scripts from PineScript, EasyLanguage, and Hithink/TDX.
• Custom Virtual Machine: A lightweight, efficient stack-based VM (PineVM) designed for executing trading logic over time-series data.
• Modular Compiler Design: Utilizes the classic Lexer -> Parser -> AST -> Code Generator pipeline for each language, making it easy to extend or improve.
• Pluggable Data Sources: An abstracted data layer (DataSource) supports different data inputs, including in-memory mock data for testing and CSV or JSON files for real market data.
• High-Performance Data Handling: Leverages the DuckDB library for fast, in-process analytical queries on CSV or JSON files.
• Strong Portability: Support exporting to javascript and python environments.

Architecture

The engine is designed in three distinct stages, orchestrated by main.cpp:

1. Compilation (Frontend): The user-selected source code (e.g., PineScript) is processed by its corresponding compiler. This involves lexical analysis (Lexer), syntax analysis (Parser to build an AST), and code generation (Compiler using a Visitor pattern) to produce universal Bytecode.
2. Data Preparation: A DataSource is created based on user choice (e.g., CSVDataSource). It loads market data and prepares it for the VM by registering the necessary time series (open, high, low, close, etc.).
3. Execution (Backend): The PineVM is instantiated. It loads the Bytecode and the data series. The execute() method then iterates through all data bars, running the bytecode for each bar to simulate the script's logic.

graph TD
    subgraph "用户接口 (User Interface)"
        direction LR
        UserInput["用户输入<br>(选择语言, 数据源)"] --> Main["main.cpp<br>(应用入口)"]
    end

    subgraph "编译阶段 <br>(Compilation Frontend)"
        direction LR
        style Bytecode fill:#bbf,stroke:#333,stroke-width:2px
        
        Main -- "1\. 选择编译器" --> CompilerChoice
        
        subgraph CompilerChoice["语言编译器"]
            direction TB
            PineCompiler["PineScript<br>Compiler"]
            ELCompiler["EasyLanguage<br>Compiler"]
            HithinkCompiler["Hithink<br>Compiler"]
        end

        CompilerChoice -- "2\. 处理源码" --> Bytecode
        
        subgraph ProcessingFlow["编译流程 (以PineScript为例)"]
            SourceString["源码字符串<br>'ma = ta.sma(close, 14)'"]
            SourceString --> Lexer["PineLexer<br>(词法分析)"]
            Lexer --> Tokens["Token序列<br>[IDENTIFIER(ma), EQUAL, ... ]"]
            Tokens --> Parser["PineParser<br>(语法分析)"]
            Parser --> AST["抽象语法树 (AST)<br>(AssignmentStmt)"]
            AST --> CodeGen["PineCompiler<br>(代码生成 / AST Visitor)"]
            CodeGen --> Bytecode["通用字节码<br>(Bytecode)"]
        end
    end

    subgraph "数据准备<br> (Data Preparation)"
        direction LR
        Main -- "创建" --> PushSources["数据源 (PushSource)"]
        PushSources -.推送数据.-> PineVM
        Main -- "创建" --> DataSources["数据源 (DataSource)"]
        DataSources -- "加载数据" --> PineVM
    end

    subgraph "执行阶段<br> (Execution Backend)"
        direction LR
        style PineVM fill:#9f9,stroke:#333,stroke-width:2px
        Bytecode -- "加载指令" --> PineVM["PineVM<br>(虚拟机)"]
        PineVM -- "执行字节码" --> Result["执行结果<br>(指标序列)"]
    end

Loading