Apache Parquet C++ 项目教程

1、项目介绍

Apache Parquet-CPP 是 Apache Parquet 项目的 C++ 实现，它是一个高效的列式存储格式，特别适用于大数据处理。Parquet 格式支持复杂的嵌套数据结构，并且通过列式存储优化了数据压缩和查询性能。该项目已经被合并到 Apache Arrow 项目中，但仍然可以在 GitHub 上找到其历史版本。

2、项目快速启动

环境准备

确保你的系统已经安装了以下工具和库：

• CMake
• C++ 编译器（如 GCC 或 Clang）
• Git

克隆项目

git clone https://github.com/apache/parquet-cpp.git
cd parquet-cpp

构建项目

mkdir build
cd build
cmake ..
make

示例代码

以下是一个简单的示例代码，展示如何读取和写入 Parquet 文件：

#include <arrow/api.h>
#include <parquet/arrow/reader.h>
#include <parquet/arrow/writer.h>

int main() {
    // 初始化 Arrow 和 Parquet
    arrow::Status status = arrow::Status::OK();
    std::shared_ptr<arrow::MemoryPool> pool = arrow::default_memory_pool();

    // 创建一个简单的表格
    std::shared_ptr<arrow::Array> array;
    arrow::Int64Builder builder(pool);
    builder.Append(1);
    builder.Append(2);
    builder.Append(3);
    status = builder.Finish(&array);
    if (!status.ok()) {
        std::cerr << "Failed to build array: " << status.ToString() << std::endl;
        return -1;
    }

    std::shared_ptr<arrow::Schema> schema = arrow::schema({arrow::field("int", arrow::int64())});
    std::shared_ptr<arrow::Table> table = arrow::Table::Make(schema, {array});

    // 写入 Parquet 文件
    std::shared_ptr<arrow::io::FileOutputStream> out_file;
    status = arrow::io::FileOutputStream::Open("example.parquet", &out_file);
    if (!status.ok()) {
        std::cerr << "Failed to open file for writing: " << status.ToString() << std::endl;
        return -1;
    }

    std::shared_ptr<parquet::arrow::FileWriter> writer;
    status = parquet::arrow::FileWriter::Open(*schema, pool, out_file, &writer);
    if (!status.ok()) {
        std::cerr << "Failed to create Parquet writer: " << status.ToString() << std::endl;
        return -1;
    }

    status = writer->WriteTable(*table, 2);
    if (!status.ok()) {
        std::cerr << "Failed to write table: " << status.ToString() << std::endl;
        return -1;
    }

    writer->Close();
    out_file->Close();

    // 读取 Parquet 文件
    std::shared_ptr<arrow::io::ReadableFile> in_file;
    status = arrow::io::ReadableFile::Open("example.parquet", pool, &in_file);
    if (!status.ok()) {
        std::cerr << "Failed to open file for reading: " << status.ToString() << std::endl;
        return -1;
    }

    std::shared_ptr<parquet::arrow::FileReader> reader;
    status = parquet::arrow::OpenFile(in_file, pool, &reader);
    if (!status.ok()) {
        std::cerr << "Failed to create Parquet reader: " << status.ToString() << std::endl;
        return -1;
    }

    std::shared_ptr<arrow::Table> read_table;
    status = reader->ReadTable(&read_table);
    if (!status.ok()) {
        std::cerr << "Failed to read table: " << status.ToString() << std::endl;
        return -1;
    }

    std::cout << "Read table: " << read_table->ToString()