C++ 示例： BasicIntegerParser

BasicIntegerParser 示例解析由非数字字符分隔的整数字符串。有关使用持续加载的此解析器的版本，请参阅 C++ 示例： ContinuousIntegerParser。

加载和使用示例

按如下所示加载并使用 BasicIntegerParser 示例。

=> CREATE LIBRARY BasicIntegerParserLib AS '/home/dbadmin/BIP.so';

=> CREATE PARSER BasicIntegerParser AS
LANGUAGE 'C++' NAME 'BasicIntegerParserFactory' LIBRARY BasicIntegerParserLib;

=> CREATE TABLE t (i integer);

=> COPY t FROM stdin WITH PARSER BasicIntegerParser();
0
1
2
3
4
5
\.

实施

BasicIntegerParser 类仅实施 API 中的 process() 方法。（它还实施了一个用于类型转换的 helper 方法。）此方法处理每一行输入，查找每一行的数字。当它前进到新行时，它会移动 input.offset 标记并检查输入状态。然后它写入输出。

    virtual StreamState process(ServerInterface &srvInterface, DataBuffer &input,
                InputState input_state) {
        // WARNING: This implementation is not trying for efficiency.
        // It is trying for simplicity, for demonstration purposes.

        size_t start = input.offset;
        const size_t end = input.size;

        do {
            bool found_newline = false;
            size_t numEnd = start;
            for (; numEnd < end; numEnd++) {
                if (input.buf[numEnd] < '0' || input.buf[numEnd] > '9') {
                    found_newline = true;
                    break;
                }
            }

            if (!found_newline) {
                input.offset = start;
                if (input_state == END_OF_FILE) {
                    // If we're at end-of-file,
                    // emit the last integer (if any) and return DONE.
                    if (start != end) {
                        writer->setInt(0, strToInt(input.buf + start, input.buf + numEnd));
                        writer->next();
                    }
                    return DONE;
                } else {
                    // Otherwise, we need more data.
                    return INPUT_NEEDED;
                }
            }

            writer->setInt(0, strToInt(input.buf + start, input.buf + numEnd));
            writer->next();

            start = numEnd + 1;
        } while (true);
    }
};

在工厂中，plan() 方法是无操作的；没有要检查的参数。prepare() 方法使用 SDK 提供的宏实例化解析器：

    virtual UDParser* prepare(ServerInterface &srvInterface,
            PerColumnParamReader &perColumnParamReader,
            PlanContext &planCtxt,
            const SizedColumnTypes &returnType) {

        return vt_createFuncObject<BasicIntegerParser>(srvInterface.allocator);
    }

getParserReturnType() 方法声明了单个输出：

    virtual void getParserReturnType(ServerInterface &srvInterface,
            PerColumnParamReader &perColumnParamReader,
            PlanContext &planCtxt,
            const SizedColumnTypes &argTypes,
            SizedColumnTypes &returnType) {
        // We only and always have a single integer column
        returnType.addInt(argTypes.getColumnName(0));
    }

对于所有用 C++ 编写的 UDx，该示例以注册其工厂结束：

RegisterFactory(BasicIntegerParserFactory);